{"topic":{"aliases":["エージェント","AI agent","AI agents","agentic"],"definition":"AIエージェントやagenticとして明示された処理・協調。代理店一般を除く。","exclusion_examples":["agentという語だけ"],"id":"top_1b05f039-7a18-4428-98ec-ee653655374e","label":"AIエージェント","review_status":"approved","slug":"agents","visibility":"public"},"analysis_manifest":{"article_count":273,"date_kind":"source_date","default_layers":["body"],"input_sha256":"15081ebadb6b9f0bbda3b4d96030b640c77c60ce51cac7dbecca3f9fc2ca98d9","matching":"literal-ascii-boundary-v1","occurrence_count":17909,"registry_revision":"topics-20260909-v1","registry_sha256":"f2ad4501536f3aa67adc37c32ead38e4d10f0061df7f5e9825e212d7e8195a8c","representative_policy":"preferred-then-newest-axis-date-then-revision-id","representative_revision_ids":["rev_00217178-ae5e-4e98-87ca-c35048af1ca4","rev_00d1c34c-80fd-4fe8-a270-80c223bb2b4a","rev_0145dd0c-e5b3-47ff-82d6-760d54d9f1e5","rev_01c4b1d1-a854-4507-a2d7-bcf6b40fafd5","rev_01fd9ef5-e950-489e-8feb-7c8ac4a60ef9","rev_02adfd14-3d0d-45ef-8656-fd05494af55e","rev_04caab49-911d-4fa8-9f6d-3786d6161ca6","rev_062b5267-073b-4a05-943e-d019dd2ff413","rev_069f7cc4-0ca9-44ca-b711-fdfd84413ec4","rev_098e8803-5806-4df3-818d-ad62b9126a33","rev_0b72e772-70fd-4fc3-94b4-3beabdd8120c","rev_0c0f759a-91cf-4045-9394-3cfb56b32f62","rev_0c1ca0be-030b-470f-a556-2d79b9d3ddce","rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e","rev_0d8f6bbc-636e-4d8d-8ba4-920514c5aff2","rev_0e531107-4090-4b2f-8e19-16fe4d32d8d6","rev_0e8f98f5-ea3d-4d3b-847a-a6846f2569e9","rev_0ee327e2-ecfe-42e6-b337-fce8abbe7b71","rev_0f7c606d-558f-4edb-99eb-4609a9fb0a95","rev_107f711f-ec27-4af9-b9da-f5facd0cd7f9","rev_11f7264b-855b-41d8-b760-87dbfa630e9f","rev_120b7405-aefc-4d5d-a68b-9713613e3e1f","rev_1246df84-dae2-49be-ab7b-94b8c8902078","rev_12eaaf24-14fe-4445-936b-b81ef492d19d","rev_15381314-c653-4ce9-9b25-d8495cdbf4d1","rev_157ef6d2-69fe-48f2-9940-db00dbbbc5b6","rev_17972011-3c15-4377-a2ea-f36af8fa6c2f","rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f","rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","rev_1c3b81f6-696c-4d69-acf7-ad589032ce63","rev_1f682c1d-2f63-48dc-adba-614831c2986b","rev_20f6e19f-c172-4e63-83ab-04729ac80c82","rev_214b9034-b7c9-47c2-b4c9-4fc2638813c2","rev_217016cb-4b97-4b15-993e-73dded8489d4","rev_22aef441-7204-418b-912b-8ab0253992f3","rev_24eb030e-0590-42b8-bd89-6a25a4dfc487","rev_25c19733-e8bf-4666-a96e-91917803878b","rev_27743cb0-aee7-439b-a5dd-1faac0292586","rev_2774a11f-f3af-4056-8394-07f7d1934023","rev_282f1fea-deeb-4d6f-9a0c-8692cfb1a953","rev_2886f620-b098-4a80-b860-f8f5887df0f8","rev_28e52d32-205e-4601-9fe8-535cce2d0600","rev_2b771286-9bc4-45c0-b45d-6524baf27996","rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","rev_2dcad1b1-3b1a-4999-bd4c-d504dda3eae6","rev_2e5b712c-b120-4b45-91dd-33e6758666be","rev_2ed1383f-5c73-49bd-bd5b-f19deda1814c","rev_2f5cc921-59ef-4e07-ba11-82e4656c1db7","rev_3510f3d4-8673-4b01-866a-65778298be5c","rev_355a11d8-4fec-4b80-a649-e98f432ab2b2","rev_35ce45f7-18dc-4fd9-90f1-01040dfe8822","rev_3621f98d-f1b9-4e97-9c8a-8b259cfd31c3","rev_37ec6023-d760-4f21-887c-7dd94c473806","rev_385079a7-c5a3-4a8e-9c9a-7b84af62dd7e","rev_3989a234-8600-4663-ab00-476b08c605fb","rev_3a0c2381-6516-49a0-8369-32f5729cb14f","rev_3fd4618d-8d1a-4a0a-8f71-c67475f7c6b1","rev_402559ee-01e9-43f3-a0e9-719e30746fc3","rev_41b829bc-1306-4c10-bde5-69fe204ccdda","rev_434ed8d3-8167-4144-9b6e-d003da6dda05","rev_436a17f2-f4bc-4abf-9295-931f16cf8b77","rev_44c73f5c-647d-467e-8aa2-527eb6352a8a","rev_45773254-bafb-490e-9237-765416121418","rev_45e2a710-51c3-42ae-a56c-b21f7fd55a5e","rev_4802c0fb-3eae-453b-96d2-1237f0009713","rev_48a46dd8-5089-4a60-987b-dcda2b2ff37b","rev_492b3784-61d6-42d8-89f4-cc60f4e9fd8e","rev_49a8bdbd-8ee8-4c28-8af4-da1787c47955","rev_4b279499-0bae-4004-81f9-2ca2bb999ac5","rev_4ca9ee61-40e0-44b0-b0b2-ae6a2fe89483","rev_4e53aa79-ce5f-4aa2-98c8-4c3907adff98","rev_4eb110b8-6300-4d37-befb-7de1200dd0be","rev_50433108-0fcd-4480-baf4-ded8ba26f963","rev_513457e5-9067-4419-bce8-bb6dfe3e66c5","rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","rev_52f347f2-d9b8-4a26-9ce4-2327ec8b1606","rev_53605de4-06cc-4e0b-9bc9-21f393408c0a","rev_5387a515-478b-41ca-82a3-5b54af988fd1","rev_54b154f3-ea9f-45e9-b38b-a16c54ca3e2b","rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b","rev_5654920a-cd90-43da-a8fe-5f2be4ea39a2","rev_57810a5f-315e-49a7-ac85-72cc27e35eff","rev_57995821-c471-42b5-ae3a-f989b7bd06f5","rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b","rev_58752a3b-195a-41cc-8958-cd8cfd050b53","rev_58ce9d38-659d-4774-8af9-ad372c9118b9","rev_5a837b64-cde7-45d0-82fa-e2c1156e1e37","rev_5bf9454f-9956-42b9-a29c-ae99782c33c0","rev_5c1381d3-461d-40ab-acea-997127510f6e","rev_5cb295dd-df59-4d28-9799-42065461c569","rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","rev_5ed1cac5-83cc-4660-89c7-643aae40960d","rev_5f26a1f6-b168-4ccd-846e-67adf1ebb1ed","rev_5f44b55a-c37f-42cf-ac49-298da914a87a","rev_5f90a6d0-abcb-4341-ad3f-bf14f2e88363","rev_5fc9a8c3-fa36-450a-855b-5bacf1d29317","rev_613ae834-139d-4833-bf7c-b55be0e9eeed","rev_61f118e1-20e7-486f-b830-cae804dd9f14","rev_62acaca2-fec9-4724-be08-6ebd3c061f93","rev_62e618a7-0275-4409-b9ab-02dd2c6efcc6","rev_651c50c5-c7cc-48ef-96fd-cd7a080eafc3","rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","rev_65b5da7e-451d-4db3-bb9a-e33b0c6f465f","rev_66313feb-6eff-400e-94b9-3dd889eabe3e","rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","rev_6833af82-b753-4666-97ac-54dbe13aa888","rev_6abb0df2-b5a4-4f1c-8f48-a346d6d0f247","rev_6cb909ec-1291-4cbb-a768-f72123829d08","rev_6df99190-0de3-49d8-bd02-b36337a4cef8","rev_6e29e2ed-0ec1-4cab-83c2-87be4ab75459","rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","rev_6f7e9f1a-6c8f-457d-8fdf-eb9746d573f8","rev_6ff58ff4-ea8b-4857-b59f-8d06717c0a72","rev_70df7bb7-127f-4402-9374-f897f873bafb","rev_716e29e3-ca33-4e0a-89d8-a4c146c91139","rev_7396edc2-fd8c-4b93-977f-078edd450a0d","rev_73f07a40-b588-4fa6-bf2b-2880462e214c","rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","rev_76f1af81-4d71-4404-b543-111d11af5928","rev_78657238-b40f-4a7c-b670-325df3688f39","rev_793e83db-98e9-4062-b995-36674ef0d72b","rev_7a6f6c00-002d-4db4-89c0-8eff54c3c177","rev_7c1050fa-5c27-4fd6-9c61-c1d193e47a98","rev_7cb9ce1e-1af9-43d5-a7ec-083ddf3b94e9","rev_7cf42fcc-1956-4694-a0b6-5eaa9be5c284","rev_7d6d4b5c-60bd-44d9-a7c5-ea8b8feb6dd2","rev_7e52a252-3f2f-4e13-882c-c5b500b829e6","rev_7e9f4f38-6cc1-469d-80f9-4c49b3218c00","rev_7f54400c-aa33-4bcd-9786-34216d309dbd","rev_7fd87260-401e-4d9b-830d-373e2a334bb7","rev_80d42ebf-f1b0-4ab9-b821-aee0e7e2f91a","rev_82a7a6e0-7b98-4dbd-a2cb-f619eb332b23","rev_836cf14e-02eb-4d08-a272-e6bd43d37605","rev_84581296-2906-4b8b-9a12-fe071bf83ea5","rev_84a77d25-ab82-4d74-abb1-c248dd75257f","rev_8705ec47-1e2c-4dfb-8675-524d967cf38f","rev_8860e14e-5f63-4d5e-96e7-c6adf255fab0","rev_89aecdf6-89fd-445d-a71c-7e189e82904d","rev_89eb6a8e-70dd-4f87-9da4-213e9ac3f499","rev_8b1f5dca-c040-4a1f-8ef8-530daabc764c","rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","rev_8e69c4b3-4a90-4a4f-82f9-c2a31d24c009","rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa","rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","rev_90fb7b96-f10b-43df-a612-0c23e8a50740","rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a","rev_92ff5158-6c34-4526-ab02-c8db8332951f","rev_9330c0a6-6c9f-45f7-a40e-ddd1678a57a2","rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8","rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","rev_94f3d3d6-6b52-4f98-89a0-318fe428f5cb","rev_95caf703-1af5-4160-8de6-deb2db4fd572","rev_977092a4-68df-47ac-8de2-4e66567c8a95","rev_978a573f-a66b-4a38-a224-585006c8af2c","rev_9832039f-aefc-4f9c-a4d2-d9fc943e8b1d","rev_9837730a-23c0-4e0d-a5e1-2380a69d4e4e","rev_98f85497-697b-4917-aabc-df1485131d9b","rev_9a79fb7c-7189-4703-9970-53ae7a5a375d","rev_9aac70a0-728a-45dc-987d-4481ba985249","rev_9ae4d311-bf43-4bb1-a7b5-36a5649efc7c","rev_9af30ed5-ec26-4204-9f65-58324ee5fba5","rev_9af8ea2d-dc2b-4fe3-8d67-94ae163703d8","rev_9c10c6f9-4055-4591-a368-74112c253a08","rev_9c5027f1-2e2c-4822-806a-1af541073d79","rev_9d4ead97-69cb-4ce7-8223-0346945539b2","rev_9da93714-e603-4ed0-af29-80871392e458","rev_a165cefe-658c-456f-b164-71170f172e7c","rev_a199924e-5c69-4344-bb57-0e0a36cda082","rev_a30fdebb-6a5f-4491-8480-bc05cbb21f96","rev_a3938e39-c989-454c-9e91-c75a6d9b8277","rev_a3e10132-4b92-414d-8744-acbeb6f0b294","rev_a3e9cc84-29ea-4e2f-8ea0-a01272094ec6","rev_a4008bc2-ae2c-4bd7-8615-0b3c4e0936ff","rev_a4cce969-13d1-46b4-be66-954788627617","rev_a576fa5a-37e6-4ec1-abf6-a1f85dd3b36b","rev_a5f2e4f8-4c18-4276-9d40-0da44bdaf811","rev_a749fcdf-8155-46e9-a466-83e867886a08","rev_a7a31edb-fca0-444e-a98f-ec173bf09f49","rev_a8af9681-d20e-4553-871e-1d7168015faf","rev_aab224dc-97dc-43c0-914b-533064f4189a","rev_ab4f5176-8f1e-4e27-b35d-2fb8a238f62c","rev_abe6b561-f51c-4549-a97f-aab2b725492e","rev_ae55c104-d140-4064-adf1-726fae29a7f3","rev_aea0fd30-a496-42e0-975e-3de7b0cd01ab","rev_afc88027-978f-428a-a597-dd5d129dc441","rev_b228986a-43e7-4e2b-8b16-758283d367bc","rev_b2b3d0ef-110d-4922-9a0e-428ef8b5692a","rev_b3e4b97a-1876-4168-a934-44077c4ff1df","rev_b6c084d1-ef6b-49f0-aaac-9027a92e2705","rev_b6edc214-9757-451b-a1d1-66280e5cb68e","rev_b7b42bee-c256-4842-b2f2-b0052de2eff3","rev_b7bf82be-4005-4e91-9ce8-cf900d5874f0","rev_b8a0343f-f76e-47f4-8c47-4536da944126","rev_ba807781-44a3-4752-b057-ee36fa97b60c","rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","rev_bbd7c7cc-2c2f-4ae5-b5ff-2c79a2e13e88","rev_bc9f6006-9722-468f-9c61-028e4f328422","rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","rev_bde0c156-c5e8-4c02-a4c1-11e282359f67","rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a","rev_bf97dda8-b55b-4e4d-b1cf-b775474af9b2","rev_c0f60d6c-f786-4c21-bf93-b517ab98e464","rev_c1530075-b9e3-4144-b8ce-f6e480c778fa","rev_c1c597d3-818f-4d4d-898d-56a2070b189f","rev_c32517c2-0e67-4292-87c7-24a6f6431091","rev_c3791796-7fb7-4aa8-a70b-bf30979f4726","rev_c4531943-9d29-4fee-b8e9-489be9ffd8be","rev_c49d360e-2b3c-4311-82ad-b78a89fde3e0","rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42","rev_c70dba8d-8c2a-4c07-934b-5920ce54c240","rev_c720ff79-9532-414c-ab0c-c6caec3e9331","rev_c74e591e-d929-42b5-8c6c-04563fae77c3","rev_c7523b3e-3b42-43d4-85e5-774de77f8279","rev_c832f985-caf8-4ac7-bf58-f785e8782922","rev_c86889d8-e5bd-4ef7-bf68-bd8e1e5e636d","rev_c89081f7-eb04-4049-a4a4-68cbe681a3f5","rev_c8a89daf-c91d-4b06-8c19-b87b474452c0","rev_c8b19a4c-93eb-4695-8f38-c8166d22cbd9","rev_c950677f-d85d-461a-bf3d-314cf6b08a37","rev_ca87695f-4cf9-40a1-bc61-6a123a71ffe6","rev_cd09d63d-aa63-4e93-bc42-e8db09f09c30","rev_cf107e9a-1286-4df3-bd1f-adb77c070354","rev_cfe5b9e3-26a7-41f5-bf9e-934074846f04","rev_d05fdc83-690b-49be-9f59-154e0892f912","rev_d4189e74-ba90-4c4f-8495-9a2fccea5a2a","rev_d47a57d8-f3a6-4505-9ea6-c03ae4416616","rev_d567eb31-f37d-4b61-934d-8845386d4709","rev_d6eb7e13-0508-4773-9eb4-6d508aae435f","rev_d8e8cb66-632c-4170-afae-461911315dbd","rev_d934c776-cfe7-4351-835a-a4c4a8a0f2c1","rev_d947c628-d0c1-47d0-9fb8-80617b87e696","rev_d9b0587c-4ba2-4d83-849b-d5ed94035f7a","rev_da6063d5-59b2-4519-9c42-45d8f4f206c8","rev_dbe69fb8-f76d-44e6-b111-aa27e548a204","rev_dd9e40ff-7f81-4d3e-81fc-b0ecb0d3e9e5","rev_ddbe7777-4c06-4736-a565-b1f544e649ff","rev_ddc53551-0396-485b-a72e-8b3d5126293f","rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646","rev_e1c0a212-5ad0-4fbc-81f6-60d2df49d753","rev_e2c369be-a992-44a9-bbc1-81ed4928e89f","rev_e37c2cac-91d3-4c94-8599-05fdaabf8b7e","rev_e3f21c07-c5c6-40a4-add2-45c9479bb52a","rev_e4df2a9f-4a34-4f95-8006-7efec2bba610","rev_e5e24c6f-5887-4f16-b03e-081da5340883","rev_e8d93c5a-854a-4940-b5f8-c5823efe6c2d","rev_e92b8924-c968-4079-83fd-9ace23c895f7","rev_e9bd03d1-2d06-4ded-bc4f-bfc67a547fbc","rev_ea63002b-7dc3-43fd-81f6-5b60d7efe1b4","rev_eaa7004e-c9e4-4bc4-9718-2e518decec70","rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e","rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","rev_ec6cc502-640e-4fef-a24c-82386eb57344","rev_ec8dcee2-8cfa-47fc-b563-9c677fad17fa","rev_ee871b65-2665-47a6-9e3d-f1a3b67b1de1","rev_eed892a0-f77a-496a-bd60-343ac50f5809","rev_f13a5f0d-0c68-4d12-9656-86661b1e65d8","rev_f17ef624-e9fd-49a4-b061-966093172de4","rev_f3a5d7c9-6dfa-4c38-8bf2-a37063801332","rev_f45cfd6f-625d-4e49-b987-25f79284501d","rev_f487aa08-2568-42ad-85cb-11c258b5b71d","rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","rev_f76185ef-dd8b-44e4-a82a-7b8cc8225b9a","rev_f7f62641-cc2a-4f5f-806e-af380e2a434d","rev_f862e355-5e8c-48e9-91e7-1f4db4cb6210","rev_f8743e43-af1f-4b5b-8c71-e10b73dc311e","rev_f9e9defb-c22c-436c-840a-c351b48ea9bb","rev_faebe87e-f436-45d0-a928-a0137e76062d","rev_fd06b790-ef4d-4f39-b0f0-a4b5a385a68f","rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280","rev_fd5eeabd-4433-491f-b463-c900cf908355","rev_fdc7bf5f-de99-4168-b2e6-31de0cab9076","rev_ff004386-e6c0-470b-b19a-7eaa2c2aba64","rev_ffbdfa80-0e8b-40b5-8765-ba1e5b977d47"],"revision_ids":["rev_00217178-ae5e-4e98-87ca-c35048af1ca4","rev_00d1c34c-80fd-4fe8-a270-80c223bb2b4a","rev_0145dd0c-e5b3-47ff-82d6-760d54d9f1e5","rev_01c4b1d1-a854-4507-a2d7-bcf6b40fafd5","rev_01fd9ef5-e950-489e-8feb-7c8ac4a60ef9","rev_02adfd14-3d0d-45ef-8656-fd05494af55e","rev_04caab49-911d-4fa8-9f6d-3786d6161ca6","rev_062b5267-073b-4a05-943e-d019dd2ff413","rev_069f7cc4-0ca9-44ca-b711-fdfd84413ec4","rev_098e8803-5806-4df3-818d-ad62b9126a33","rev_0b72e772-70fd-4fc3-94b4-3beabdd8120c","rev_0c0f759a-91cf-4045-9394-3cfb56b32f62","rev_0c1ca0be-030b-470f-a556-2d79b9d3ddce","rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e","rev_0d8f6bbc-636e-4d8d-8ba4-920514c5aff2","rev_0e531107-4090-4b2f-8e19-16fe4d32d8d6","rev_0e8f98f5-ea3d-4d3b-847a-a6846f2569e9","rev_0ee327e2-ecfe-42e6-b337-fce8abbe7b71","rev_0f7c606d-558f-4edb-99eb-4609a9fb0a95","rev_107f711f-ec27-4af9-b9da-f5facd0cd7f9","rev_11f7264b-855b-41d8-b760-87dbfa630e9f","rev_120b7405-aefc-4d5d-a68b-9713613e3e1f","rev_1246df84-dae2-49be-ab7b-94b8c8902078","rev_12eaaf24-14fe-4445-936b-b81ef492d19d","rev_15381314-c653-4ce9-9b25-d8495cdbf4d1","rev_157ef6d2-69fe-48f2-9940-db00dbbbc5b6","rev_17972011-3c15-4377-a2ea-f36af8fa6c2f","rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f","rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","rev_1c3b81f6-696c-4d69-acf7-ad589032ce63","rev_1f682c1d-2f63-48dc-adba-614831c2986b","rev_20f6e19f-c172-4e63-83ab-04729ac80c82","rev_214b9034-b7c9-47c2-b4c9-4fc2638813c2","rev_217016cb-4b97-4b15-993e-73dded8489d4","rev_22aef441-7204-418b-912b-8ab0253992f3","rev_24eb030e-0590-42b8-bd89-6a25a4dfc487","rev_25c19733-e8bf-4666-a96e-91917803878b","rev_27743cb0-aee7-439b-a5dd-1faac0292586","rev_2774a11f-f3af-4056-8394-07f7d1934023","rev_282f1fea-deeb-4d6f-9a0c-8692cfb1a953","rev_2886f620-b098-4a80-b860-f8f5887df0f8","rev_28e52d32-205e-4601-9fe8-535cce2d0600","rev_2b771286-9bc4-45c0-b45d-6524baf27996","rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","rev_2dcad1b1-3b1a-4999-bd4c-d504dda3eae6","rev_2e5b712c-b120-4b45-91dd-33e6758666be","rev_2ed1383f-5c73-49bd-bd5b-f19deda1814c","rev_2f5cc921-59ef-4e07-ba11-82e4656c1db7","rev_3510f3d4-8673-4b01-866a-65778298be5c","rev_355a11d8-4fec-4b80-a649-e98f432ab2b2","rev_35ce45f7-18dc-4fd9-90f1-01040dfe8822","rev_3621f98d-f1b9-4e97-9c8a-8b259cfd31c3","rev_37ec6023-d760-4f21-887c-7dd94c473806","rev_385079a7-c5a3-4a8e-9c9a-7b84af62dd7e","rev_3989a234-8600-4663-ab00-476b08c605fb","rev_3a0c2381-6516-49a0-8369-32f5729cb14f","rev_3fd4618d-8d1a-4a0a-8f71-c67475f7c6b1","rev_402559ee-01e9-43f3-a0e9-719e30746fc3","rev_41b829bc-1306-4c10-bde5-69fe204ccdda","rev_434ed8d3-8167-4144-9b6e-d003da6dda05","rev_436a17f2-f4bc-4abf-9295-931f16cf8b77","rev_44c73f5c-647d-467e-8aa2-527eb6352a8a","rev_45773254-bafb-490e-9237-765416121418","rev_45e2a710-51c3-42ae-a56c-b21f7fd55a5e","rev_4802c0fb-3eae-453b-96d2-1237f0009713","rev_48a46dd8-5089-4a60-987b-dcda2b2ff37b","rev_492b3784-61d6-42d8-89f4-cc60f4e9fd8e","rev_49a8bdbd-8ee8-4c28-8af4-da1787c47955","rev_4b279499-0bae-4004-81f9-2ca2bb999ac5","rev_4ca9ee61-40e0-44b0-b0b2-ae6a2fe89483","rev_4e53aa79-ce5f-4aa2-98c8-4c3907adff98","rev_4eb110b8-6300-4d37-befb-7de1200dd0be","rev_50433108-0fcd-4480-baf4-ded8ba26f963","rev_513457e5-9067-4419-bce8-bb6dfe3e66c5","rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","rev_52f347f2-d9b8-4a26-9ce4-2327ec8b1606","rev_53605de4-06cc-4e0b-9bc9-21f393408c0a","rev_5387a515-478b-41ca-82a3-5b54af988fd1","rev_54b154f3-ea9f-45e9-b38b-a16c54ca3e2b","rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b","rev_5654920a-cd90-43da-a8fe-5f2be4ea39a2","rev_57810a5f-315e-49a7-ac85-72cc27e35eff","rev_57995821-c471-42b5-ae3a-f989b7bd06f5","rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b","rev_58752a3b-195a-41cc-8958-cd8cfd050b53","rev_58ce9d38-659d-4774-8af9-ad372c9118b9","rev_5a837b64-cde7-45d0-82fa-e2c1156e1e37","rev_5bf9454f-9956-42b9-a29c-ae99782c33c0","rev_5c1381d3-461d-40ab-acea-997127510f6e","rev_5cb295dd-df59-4d28-9799-42065461c569","rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","rev_5ed1cac5-83cc-4660-89c7-643aae40960d","rev_5f26a1f6-b168-4ccd-846e-67adf1ebb1ed","rev_5f44b55a-c37f-42cf-ac49-298da914a87a","rev_5f90a6d0-abcb-4341-ad3f-bf14f2e88363","rev_5fc9a8c3-fa36-450a-855b-5bacf1d29317","rev_613ae834-139d-4833-bf7c-b55be0e9eeed","rev_61f118e1-20e7-486f-b830-cae804dd9f14","rev_62acaca2-fec9-4724-be08-6ebd3c061f93","rev_62e618a7-0275-4409-b9ab-02dd2c6efcc6","rev_651c50c5-c7cc-48ef-96fd-cd7a080eafc3","rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","rev_65b5da7e-451d-4db3-bb9a-e33b0c6f465f","rev_66313feb-6eff-400e-94b9-3dd889eabe3e","rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","rev_6833af82-b753-4666-97ac-54dbe13aa888","rev_6abb0df2-b5a4-4f1c-8f48-a346d6d0f247","rev_6cb909ec-1291-4cbb-a768-f72123829d08","rev_6df99190-0de3-49d8-bd02-b36337a4cef8","rev_6e29e2ed-0ec1-4cab-83c2-87be4ab75459","rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","rev_6f7e9f1a-6c8f-457d-8fdf-eb9746d573f8","rev_6ff58ff4-ea8b-4857-b59f-8d06717c0a72","rev_70df7bb7-127f-4402-9374-f897f873bafb","rev_716e29e3-ca33-4e0a-89d8-a4c146c91139","rev_7396edc2-fd8c-4b93-977f-078edd450a0d","rev_73f07a40-b588-4fa6-bf2b-2880462e214c","rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","rev_76f1af81-4d71-4404-b543-111d11af5928","rev_78657238-b40f-4a7c-b670-325df3688f39","rev_793e83db-98e9-4062-b995-36674ef0d72b","rev_7a6f6c00-002d-4db4-89c0-8eff54c3c177","rev_7c1050fa-5c27-4fd6-9c61-c1d193e47a98","rev_7cb9ce1e-1af9-43d5-a7ec-083ddf3b94e9","rev_7cf42fcc-1956-4694-a0b6-5eaa9be5c284","rev_7d6d4b5c-60bd-44d9-a7c5-ea8b8feb6dd2","rev_7e52a252-3f2f-4e13-882c-c5b500b829e6","rev_7e9f4f38-6cc1-469d-80f9-4c49b3218c00","rev_7f54400c-aa33-4bcd-9786-34216d309dbd","rev_7fd87260-401e-4d9b-830d-373e2a334bb7","rev_80d42ebf-f1b0-4ab9-b821-aee0e7e2f91a","rev_82a7a6e0-7b98-4dbd-a2cb-f619eb332b23","rev_836cf14e-02eb-4d08-a272-e6bd43d37605","rev_84581296-2906-4b8b-9a12-fe071bf83ea5","rev_84a77d25-ab82-4d74-abb1-c248dd75257f","rev_8705ec47-1e2c-4dfb-8675-524d967cf38f","rev_8860e14e-5f63-4d5e-96e7-c6adf255fab0","rev_89aecdf6-89fd-445d-a71c-7e189e82904d","rev_89eb6a8e-70dd-4f87-9da4-213e9ac3f499","rev_8b1f5dca-c040-4a1f-8ef8-530daabc764c","rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","rev_8e69c4b3-4a90-4a4f-82f9-c2a31d24c009","rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa","rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","rev_90fb7b96-f10b-43df-a612-0c23e8a50740","rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a","rev_92ff5158-6c34-4526-ab02-c8db8332951f","rev_9330c0a6-6c9f-45f7-a40e-ddd1678a57a2","rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8","rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","rev_94f3d3d6-6b52-4f98-89a0-318fe428f5cb","rev_95caf703-1af5-4160-8de6-deb2db4fd572","rev_977092a4-68df-47ac-8de2-4e66567c8a95","rev_978a573f-a66b-4a38-a224-585006c8af2c","rev_9832039f-aefc-4f9c-a4d2-d9fc943e8b1d","rev_9837730a-23c0-4e0d-a5e1-2380a69d4e4e","rev_98f85497-697b-4917-aabc-df1485131d9b","rev_9a79fb7c-7189-4703-9970-53ae7a5a375d","rev_9aac70a0-728a-45dc-987d-4481ba985249","rev_9ae4d311-bf43-4bb1-a7b5-36a5649efc7c","rev_9af30ed5-ec26-4204-9f65-58324ee5fba5","rev_9af8ea2d-dc2b-4fe3-8d67-94ae163703d8","rev_9c10c6f9-4055-4591-a368-74112c253a08","rev_9c5027f1-2e2c-4822-806a-1af541073d79","rev_9d4ead97-69cb-4ce7-8223-0346945539b2","rev_9da93714-e603-4ed0-af29-80871392e458","rev_a165cefe-658c-456f-b164-71170f172e7c","rev_a199924e-5c69-4344-bb57-0e0a36cda082","rev_a30fdebb-6a5f-4491-8480-bc05cbb21f96","rev_a3938e39-c989-454c-9e91-c75a6d9b8277","rev_a3e10132-4b92-414d-8744-acbeb6f0b294","rev_a3e9cc84-29ea-4e2f-8ea0-a01272094ec6","rev_a4008bc2-ae2c-4bd7-8615-0b3c4e0936ff","rev_a4cce969-13d1-46b4-be66-954788627617","rev_a576fa5a-37e6-4ec1-abf6-a1f85dd3b36b","rev_a5f2e4f8-4c18-4276-9d40-0da44bdaf811","rev_a749fcdf-8155-46e9-a466-83e867886a08","rev_a7a31edb-fca0-444e-a98f-ec173bf09f49","rev_a8af9681-d20e-4553-871e-1d7168015faf","rev_aab224dc-97dc-43c0-914b-533064f4189a","rev_ab4f5176-8f1e-4e27-b35d-2fb8a238f62c","rev_abe6b561-f51c-4549-a97f-aab2b725492e","rev_ae55c104-d140-4064-adf1-726fae29a7f3","rev_aea0fd30-a496-42e0-975e-3de7b0cd01ab","rev_afc88027-978f-428a-a597-dd5d129dc441","rev_b228986a-43e7-4e2b-8b16-758283d367bc","rev_b2b3d0ef-110d-4922-9a0e-428ef8b5692a","rev_b3e4b97a-1876-4168-a934-44077c4ff1df","rev_b6c084d1-ef6b-49f0-aaac-9027a92e2705","rev_b6edc214-9757-451b-a1d1-66280e5cb68e","rev_b7b42bee-c256-4842-b2f2-b0052de2eff3","rev_b7bf82be-4005-4e91-9ce8-cf900d5874f0","rev_b8a0343f-f76e-47f4-8c47-4536da944126","rev_ba807781-44a3-4752-b057-ee36fa97b60c","rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","rev_bbd7c7cc-2c2f-4ae5-b5ff-2c79a2e13e88","rev_bc9f6006-9722-468f-9c61-028e4f328422","rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","rev_bde0c156-c5e8-4c02-a4c1-11e282359f67","rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a","rev_bf97dda8-b55b-4e4d-b1cf-b775474af9b2","rev_c0f60d6c-f786-4c21-bf93-b517ab98e464","rev_c1530075-b9e3-4144-b8ce-f6e480c778fa","rev_c1c597d3-818f-4d4d-898d-56a2070b189f","rev_c32517c2-0e67-4292-87c7-24a6f6431091","rev_c3791796-7fb7-4aa8-a70b-bf30979f4726","rev_c4531943-9d29-4fee-b8e9-489be9ffd8be","rev_c49d360e-2b3c-4311-82ad-b78a89fde3e0","rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42","rev_c70dba8d-8c2a-4c07-934b-5920ce54c240","rev_c720ff79-9532-414c-ab0c-c6caec3e9331","rev_c74e591e-d929-42b5-8c6c-04563fae77c3","rev_c7523b3e-3b42-43d4-85e5-774de77f8279","rev_c832f985-caf8-4ac7-bf58-f785e8782922","rev_c86889d8-e5bd-4ef7-bf68-bd8e1e5e636d","rev_c89081f7-eb04-4049-a4a4-68cbe681a3f5","rev_c8a89daf-c91d-4b06-8c19-b87b474452c0","rev_c8b19a4c-93eb-4695-8f38-c8166d22cbd9","rev_c950677f-d85d-461a-bf3d-314cf6b08a37","rev_ca87695f-4cf9-40a1-bc61-6a123a71ffe6","rev_cd09d63d-aa63-4e93-bc42-e8db09f09c30","rev_cf107e9a-1286-4df3-bd1f-adb77c070354","rev_cfe5b9e3-26a7-41f5-bf9e-934074846f04","rev_d05fdc83-690b-49be-9f59-154e0892f912","rev_d4189e74-ba90-4c4f-8495-9a2fccea5a2a","rev_d47a57d8-f3a6-4505-9ea6-c03ae4416616","rev_d567eb31-f37d-4b61-934d-8845386d4709","rev_d6eb7e13-0508-4773-9eb4-6d508aae435f","rev_d8e8cb66-632c-4170-afae-461911315dbd","rev_d934c776-cfe7-4351-835a-a4c4a8a0f2c1","rev_d947c628-d0c1-47d0-9fb8-80617b87e696","rev_d9b0587c-4ba2-4d83-849b-d5ed94035f7a","rev_da6063d5-59b2-4519-9c42-45d8f4f206c8","rev_dbe69fb8-f76d-44e6-b111-aa27e548a204","rev_dd9e40ff-7f81-4d3e-81fc-b0ecb0d3e9e5","rev_ddbe7777-4c06-4736-a565-b1f544e649ff","rev_ddc53551-0396-485b-a72e-8b3d5126293f","rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646","rev_e1c0a212-5ad0-4fbc-81f6-60d2df49d753","rev_e2c369be-a992-44a9-bbc1-81ed4928e89f","rev_e37c2cac-91d3-4c94-8599-05fdaabf8b7e","rev_e3f21c07-c5c6-40a4-add2-45c9479bb52a","rev_e4df2a9f-4a34-4f95-8006-7efec2bba610","rev_e5e24c6f-5887-4f16-b03e-081da5340883","rev_e8d93c5a-854a-4940-b5f8-c5823efe6c2d","rev_e92b8924-c968-4079-83fd-9ace23c895f7","rev_e9bd03d1-2d06-4ded-bc4f-bfc67a547fbc","rev_ea63002b-7dc3-43fd-81f6-5b60d7efe1b4","rev_eaa7004e-c9e4-4bc4-9718-2e518decec70","rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e","rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","rev_ec6cc502-640e-4fef-a24c-82386eb57344","rev_ec8dcee2-8cfa-47fc-b563-9c677fad17fa","rev_ee871b65-2665-47a6-9e3d-f1a3b67b1de1","rev_eed892a0-f77a-496a-bd60-343ac50f5809","rev_f13a5f0d-0c68-4d12-9656-86661b1e65d8","rev_f17ef624-e9fd-49a4-b061-966093172de4","rev_f3a5d7c9-6dfa-4c38-8bf2-a37063801332","rev_f45cfd6f-625d-4e49-b987-25f79284501d","rev_f487aa08-2568-42ad-85cb-11c258b5b71d","rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","rev_f76185ef-dd8b-44e4-a82a-7b8cc8225b9a","rev_f7f62641-cc2a-4f5f-806e-af380e2a434d","rev_f862e355-5e8c-48e9-91e7-1f4db4cb6210","rev_f8743e43-af1f-4b5b-8c71-e10b73dc311e","rev_f9e9defb-c22c-436c-840a-c351b48ea9bb","rev_faebe87e-f436-45d0-a928-a0137e76062d","rev_fd06b790-ef4d-4f39-b0f0-a4b5a385a68f","rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280","rev_fd5eeabd-4433-491f-b463-c900cf908355","rev_fdc7bf5f-de99-4168-b2e6-31de0cab9076","rev_ff004386-e6c0-470b-b19a-7eaa2c2aba64","rev_ffbdfa80-0e8b-40b5-8765-ba1e5b977d47"],"schema_version":"noia-discovery-1.0.0","time_assertions_sha256":"4f58252ab1354a1fc505681fb660990d800e72b647d3ac90448d84411bb91843","time_basis_notice":"元資料の日付順・現在保存版。日付不明は期間別集計の外に保持。","work_count":273},"articles":[{"date":{"precision":"instant","search_end":"2026-07-26","search_start":"2026-07-25","timezone":null,"value":"2026-07-25 22:39:55"},"date_kind":"source_date","edition_id":"edn_c53df2e8-ef33-40cc-85bc-309f3831a705","href":"articles/rev_00217178-ae5e-4e98-87ca-c35048af1ca4/","is_representative":true,"revision_id":"rev_00217178-ae5e-4e98-87ca-c35048af1ca4","title":"韓国は「メモリ供給国」から「AIを生産する国家」へ","work_id":"wrk_6f50f82d-878e-437d-81af-1073e3707af7"},{"date":{"precision":"instant","search_end":"2026-09-14","search_start":"2026-09-13","timezone":"+09:00","value":"2026-09-13T19:26:49.000+09:00"},"date_kind":"source_date","edition_id":"edn_237fce5e-f684-4ec1-aec0-4e1e8fe75c9d","href":"articles/rev_01c4b1d1-a854-4507-a2d7-bcf6b40fafd5/","is_representative":true,"revision_id":"rev_01c4b1d1-a854-4507-a2d7-bcf6b40fafd5","title":"フロンティアAIに「査読」が入る日","work_id":"wrk_dbe273d0-7a27-4a33-8051-93c6d0c21198"},{"date":{"precision":"instant","search_end":"2026-08-23","search_start":"2026-08-22","timezone":null,"value":"2026-08-22 21:48:26"},"date_kind":"source_date","edition_id":"edn_bef743ae-2f6c-43c3-9df0-77d0daa8c80e","href":"articles/rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e/","is_representative":true,"revision_id":"rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e","title":"AIDCは本当に余るのか――GWでは見えない「AI Factoryの質」を測る","work_id":"wrk_078157a8-8d73-48b6-b0ad-972882b9971f"},{"date":{"precision":"instant","search_end":"2026-01-09","search_start":"2026-01-08","timezone":null,"value":"2026-01-08 08:36:32"},"date_kind":"source_date","edition_id":"edn_0f069105-dd3a-49f1-80bc-8bb96c8d5f52","href":"articles/rev_0ee327e2-ecfe-42e6-b337-fce8abbe7b71/","is_representative":true,"revision_id":"rev_0ee327e2-ecfe-42e6-b337-fce8abbe7b71","title":"Redwire考察","work_id":"wrk_e5060e37-5e03-4669-88d5-6a5989ca3dcb"},{"date":{"precision":"instant","search_end":"2026-09-12","search_start":"2026-09-11","timezone":"+09:00","value":"2026-09-11T20:34:18.000+09:00"},"date_kind":"source_date","edition_id":"edn_4ec9f861-add1-45b4-9037-8be6cfe84643","href":"articles/rev_0f7c606d-558f-4edb-99eb-4609a9fb0a95/","is_representative":true,"revision_id":"rev_0f7c606d-558f-4edb-99eb-4609a9fb0a95","title":"AIが半導体設計の堀を壊すとき、誰が残るのか――Frontier Model × Custom Silicon × FoundryのLearning Loop","work_id":"wrk_958dfab7-4e11-4e38-9d35-db36e1f1b2f7"},{"date":{"precision":"instant","search_end":"2026-06-11","search_start":"2026-06-10","timezone":"+09:00","value":"2026-06-10T12:32:57.000+09:00"},"date_kind":"source_date","edition_id":"edn_ccdd2511-e265-4cba-8ae2-1cfd8f6e716c","href":"articles/rev_11f7264b-855b-41d8-b760-87dbfa630e9f/","is_representative":true,"revision_id":"rev_11f7264b-855b-41d8-b760-87dbfa630e9f","title":"NVIDIAのロボットAIスタックを読み解く: Cosmos、Omniverse、Isaac、GR00T、JetsonとVLA・WAM・WLA","work_id":"wrk_739d4c20-62f1-49ee-9121-4689ad38c0d3"},{"date":{"precision":"instant","search_end":"2026-07-31","search_start":"2026-07-30","timezone":null,"value":"2026-07-30 19:28:20"},"date_kind":"source_date","edition_id":"edn_8e779086-3998-4d13-9090-9b9ca6c697f3","href":"articles/rev_120b7405-aefc-4d5d-a68b-9713613e3e1f/","is_representative":true,"revision_id":"rev_120b7405-aefc-4d5d-a68b-9713613e3e1f","title":"ドットコムバブル／ITバブル完全分析","work_id":"wrk_91a0ac80-8870-4ec9-b8a6-4081acad5334"},{"date":{"precision":"instant","search_end":"2026-06-19","search_start":"2026-06-18","timezone":null,"value":"2026-06-18 17:28:27"},"date_kind":"source_date","edition_id":"edn_b4801366-3d50-4547-87f0-83c1053270aa","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/","is_representative":true,"revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","title":"MetaのAIはフロンティアモデルの最前線に来るのか","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb"},{"date":{"precision":"instant","search_end":"2026-06-08","search_start":"2026-06-07","timezone":null,"value":"2026-06-07 17:15:14"},"date_kind":"source_date","edition_id":"edn_6efc8903-fa2d-48c9-b878-40651494b689","href":"articles/rev_15381314-c653-4ce9-9b25-d8495cdbf4d1/","is_representative":true,"revision_id":"rev_15381314-c653-4ce9-9b25-d8495cdbf4d1","title":"VLAからWAM、WLAへ","work_id":"wrk_220c8e7e-eccc-4357-89f5-9a6b78a08a7c"},{"date":{"precision":"day","search_end":"2026-07-02","search_start":"2026-07-01","timezone":null,"value":"2026-07-01"},"date_kind":"source_date","edition_id":"edn_f92bf30c-0fca-4fa7-a596-cd951025dca5","href":"articles/rev_17972011-3c15-4377-a2ea-f36af8fa6c2f/","is_representative":true,"revision_id":"rev_17972011-3c15-4377-a2ea-f36af8fa6c2f","title":"メモリ御三家は4〜5兆ドル企業になれるのか","work_id":"wrk_fbc5cb25-a3fe-4368-a659-4a3d7a753309"},{"date":{"precision":"instant","search_end":"2026-05-15","search_start":"2026-05-14","timezone":null,"value":"2026-05-14 04:17:48"},"date_kind":"source_date","edition_id":"edn_ab7d58ac-6ac7-499e-af3b-ff92b901569a","href":"articles/rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f/","is_representative":true,"revision_id":"rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f","title":"AI-RANとは「無線網をAIの神経系に変える」構想","work_id":"wrk_f7d6728c-f58d-4c77-958d-67d9910c0d5d"},{"date":{"precision":"instant","search_end":"2026-06-06","search_start":"2026-06-05","timezone":null,"value":"2026-06-05 21:38:17"},"date_kind":"source_date","edition_id":"edn_e83d0c5b-4eab-45fa-80aa-f7b1e7cef8c2","href":"articles/rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d/","is_representative":true,"revision_id":"rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","title":"ローカルAIはクラウドAIを置き換えるのか","work_id":"wrk_e8dd57e9-048c-4852-81d1-2a029a08ac21"},{"date":{"precision":"instant","search_end":"2026-09-02","search_start":"2026-09-01","timezone":null,"value":"2026-09-01 14:51:02"},"date_kind":"source_date","edition_id":"edn_c1c54211-1db7-4744-aef4-5a8e0f2158c4","href":"articles/rev_20f6e19f-c172-4e63-83ab-04729ac80c82/","is_representative":true,"revision_id":"rev_20f6e19f-c172-4e63-83ab-04729ac80c82","title":"OpenAIのMac大量購入から考える、Agent CloudとAIDCの次のボトルネック","work_id":"wrk_ae883f23-4159-4408-941d-addc931dbce4"},{"date":{"precision":"instant","search_end":"2026-05-08","search_start":"2026-05-07","timezone":null,"value":"2026-05-07 22:10:59"},"date_kind":"source_date","edition_id":"edn_793123b0-eff7-47e7-9d14-a9c5c89d3cba","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/","is_representative":true,"revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","title":"GPU一強の次に来るもの──AIエージェント時代にCPU需要が高まる理由","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132"},{"date":{"precision":"instant","search_end":"2026-09-24","search_start":"2026-09-23","timezone":"+09:00","value":"2026-09-23T06:22:02.000+09:00"},"date_kind":"source_date","edition_id":"edn_f05434a5-f7e4-483f-b164-03fbcdb674ab","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/","is_representative":true,"revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","title":"AIを鍛えるために、AIを大量に動かす","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331"},{"date":{"precision":"instant","search_end":"2026-07-25","search_start":"2026-07-24","timezone":null,"value":"2026-07-24 06:23:02"},"date_kind":"source_date","edition_id":"edn_4d0e9f72-b58a-4694-9639-b7949422799a","href":"articles/rev_2ed1383f-5c73-49bd-bd5b-f19deda1814c/","is_representative":true,"revision_id":"rev_2ed1383f-5c73-49bd-bd5b-f19deda1814c","title":"Nokiaについて調査分析しました未だよく分からないです","work_id":"wrk_209cb693-e035-4d91-b959-d2a9b4df1de3"},{"date":{"precision":"instant","search_end":"2026-05-28","search_start":"2026-05-27","timezone":null,"value":"2026-05-27 20:02:55"},"date_kind":"source_date","edition_id":"edn_78c4dece-f9a1-4541-97d3-b50856695800","href":"articles/rev_3621f98d-f1b9-4e97-9c8a-8b259cfd31c3/","is_representative":true,"revision_id":"rev_3621f98d-f1b9-4e97-9c8a-8b259cfd31c3","title":"Logic Foldingとは何か","work_id":"wrk_68ec8173-3ea7-4526-bfea-020e82eae4c6"},{"date":{"precision":"instant","search_end":"2026-07-05","search_start":"2026-07-04","timezone":null,"value":"2026-07-04 02:28:11"},"date_kind":"source_date","edition_id":"edn_936b150b-6fda-40fd-afcc-6bc1cbf2b209","href":"articles/rev_44c73f5c-647d-467e-8aa2-527eb6352a8a/","is_representative":true,"revision_id":"rev_44c73f5c-647d-467e-8aa2-527eb6352a8a","title":"NAND・DRAM・HBMの違いから始める、AI時代のメモリー入門","work_id":"wrk_45578dda-8dcf-452e-909d-3f4c17ea166c"},{"date":{"precision":"instant","search_end":"2026-05-24","search_start":"2026-05-23","timezone":null,"value":"2026-05-23 08:23:34"},"date_kind":"source_date","edition_id":"edn_4ba3af6e-8aba-4786-bb0d-ea1f991a8601","href":"articles/rev_49a8bdbd-8ee8-4c28-8af4-da1787c47955/","is_representative":true,"revision_id":"rev_49a8bdbd-8ee8-4c28-8af4-da1787c47955","title":"半導体製造を「層」で見る","work_id":"wrk_cb25df93-468d-4abf-9350-ba1bcf0d3b49"},{"date":{"precision":"instant","search_end":"2026-04-18","search_start":"2026-04-17","timezone":null,"value":"2026-04-17 17:39:04"},"date_kind":"source_date","edition_id":"edn_28e6926f-463f-4a89-9e0d-cc524a6148d3","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/","is_representative":true,"revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","title":"AIエージェント時代、広告はどう変わるのか","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c"},{"date":{"precision":"instant","search_end":"2026-06-29","search_start":"2026-06-28","timezone":null,"value":"2026-06-28 19:15:09"},"date_kind":"source_date","edition_id":"edn_2558ef1c-050e-413f-97f8-60393ce10001","href":"articles/rev_53605de4-06cc-4e0b-9bc9-21f393408c0a/","is_representative":true,"revision_id":"rev_53605de4-06cc-4e0b-9bc9-21f393408c0a","title":"再考察：半導体インフレは、デバイス階級社会をどこまで進めるのか","work_id":"wrk_006833f6-f225-48d3-8754-46fe28d6bd8e"},{"date":{"precision":"instant","search_end":"2026-07-03","search_start":"2026-07-02","timezone":null,"value":"2026-07-02 11:59:12"},"date_kind":"source_date","edition_id":"edn_9266d0b2-91ab-44ec-8b38-eda40ee4bb5f","href":"articles/rev_54b154f3-ea9f-45e9-b38b-a16c54ca3e2b/","is_representative":true,"revision_id":"rev_54b154f3-ea9f-45e9-b38b-a16c54ca3e2b","title":"AIインフラの次の主戦場は「GPUの枚数」ではなく「GPU稼働率」になる","work_id":"wrk_52a8c79a-5c56-4f34-a515-1b1dcb4c0477"},{"date":{"precision":"instant","search_end":"2026-06-19","search_start":"2026-06-18","timezone":null,"value":"2026-06-18 18:53:28"},"date_kind":"source_date","edition_id":"edn_cfffd60c-7123-4bac-a51a-38dadc4c004a","href":"articles/rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b/","is_representative":true,"revision_id":"rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b","title":"Geminiはなぜ「ポンコツ感」が残るのか","work_id":"wrk_bbe3569f-7530-42b2-8b5f-75a35e4b1cba"},{"date":{"precision":"instant","search_end":"2026-04-21","search_start":"2026-04-20","timezone":null,"value":"2026-04-20 18:17:43"},"date_kind":"source_date","edition_id":"edn_242d763d-3b1a-4f25-88b1-283b8d2baddb","href":"articles/rev_57810a5f-315e-49a7-ac85-72cc27e35eff/","is_representative":true,"revision_id":"rev_57810a5f-315e-49a7-ac85-72cc27e35eff","title":"EDAについて少しだけ勉強","work_id":"wrk_fc529d17-dbc8-4405-b2ca-a48accef7253"},{"date":{"precision":"instant","search_end":"2026-07-27","search_start":"2026-07-26","timezone":null,"value":"2026-07-26 00:17:26"},"date_kind":"source_date","edition_id":"edn_d12ffeb9-92a9-4014-a275-9ac4c1ad5d13","href":"articles/rev_57995821-c471-42b5-ae3a-f989b7bd06f5/","is_representative":true,"revision_id":"rev_57995821-c471-42b5-ae3a-f989b7bd06f5","title":"AI相場の本当の後半戦","work_id":"wrk_ffa6d69b-69a7-4979-a51d-6b89aa7b0ff5"},{"date":{"precision":"instant","search_end":"2026-04-08","search_start":"2026-04-07","timezone":null,"value":"2026-04-07 20:24:17"},"date_kind":"source_date","edition_id":"edn_710e18b0-e20a-4821-92ed-d4648249d6e8","href":"articles/rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b/","is_representative":true,"revision_id":"rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b","title":"Rubin / Rubin Ultra / Blackwell / Groq の技術比較と、AI手法の進化が要求する次世代アクセラレータ設計考察","work_id":"wrk_b4bdeaca-dc31-448c-8f47-03f9e39c14b0"},{"date":{"precision":"instant","search_end":"2026-08-22","search_start":"2026-08-21","timezone":null,"value":"2026-08-21 14:45:06"},"date_kind":"source_date","edition_id":"edn_67f2a045-5a0f-48a6-9882-c2b0944f5645","href":"articles/rev_5bf9454f-9956-42b9-a29c-ae99782c33c0/","is_representative":true,"revision_id":"rev_5bf9454f-9956-42b9-a29c-ae99782c33c0","title":"イラン戦争が終わらない。その泥沼は台湾を守るのか、それともAI戦争の教材になるのか","work_id":"wrk_37743298-dcff-4252-a0d0-d94538fc2a68"},{"date":{"precision":"day","search_end":"2026-06-06","search_start":"2026-06-05","timezone":null,"value":"2026-06-05"},"date_kind":"source_date","edition_id":"edn_abce5aec-73d7-43f7-b17b-76af23df392e","href":"articles/rev_5cb295dd-df59-4d28-9799-42065461c569/","is_representative":true,"revision_id":"rev_5cb295dd-df59-4d28-9799-42065461c569","title":"SpaceX IPO徹底解説: 宇宙・通信・AIインフラ企業へ変質するMusk経済圏","work_id":"wrk_5380fdec-35cc-4092-a93e-530c9abca5c9"},{"date":{"precision":"day","search_end":"2026-06-23","search_start":"2026-06-22","timezone":null,"value":"2026-06-22"},"date_kind":"source_date","edition_id":"edn_e3c09031-c355-4a53-82c8-f16c3121d162","href":"articles/rev_5dbe090a-744f-4659-8ee9-8fbc7195e238/","is_representative":true,"revision_id":"rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","title":"HBMだけでは足りない: AI推論時代に始まるメモリ階層戦争","work_id":"wrk_22490a2d-59cf-4ab7-9288-4f1e6c0f5a39"},{"date":{"precision":"instant","search_end":"2026-08-30","search_start":"2026-08-29","timezone":null,"value":"2026-08-29 20:33:34"},"date_kind":"source_date","edition_id":"edn_e497b511-97c0-4b9b-a959-f21d9763bf9c","href":"articles/rev_5f44b55a-c37f-42cf-ac49-298da914a87a/","is_representative":true,"revision_id":"rev_5f44b55a-c37f-42cf-ac49-298da914a87a","title":"NVIDIAはなぜHugging Faceを買うのか","work_id":"wrk_57e07914-940b-4727-b215-7d77a2ac51c0"},{"date":{"precision":"instant","search_end":"2026-05-18","search_start":"2026-05-17","timezone":null,"value":"2026-05-17 07:41:05"},"date_kind":"source_date","edition_id":"edn_d30bbd29-2b5b-40a3-a721-2ab3599e4f3f","href":"articles/rev_62acaca2-fec9-4724-be08-6ebd3c061f93/","is_representative":true,"revision_id":"rev_62acaca2-fec9-4724-be08-6ebd3c061f93","title":"フアン氏の「五層のケーキ理論」から見るAI時代のエネルギー層","work_id":"wrk_d43a6ffd-ef17-4cd2-96fa-06f7d800821a"},{"date":{"precision":"instant","search_end":"2026-08-01","search_start":"2026-07-31","timezone":"+09:00","value":"2026-07-31T11:07:53.000+09:00"},"date_kind":"source_date","edition_id":"edn_dd821d84-c163-4778-982a-ae4f68b38fb3","href":"articles/rev_65271768-1b9c-4fbd-a486-d503b37eaf9a/","is_representative":true,"revision_id":"rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","title":"クラウド20年史――サーバーの貸し出しから「知能と労働」の基盤へ","work_id":"wrk_3d2e988e-cac0-4000-b005-9292af7ac61f"},{"date":{"precision":"instant","search_end":"2026-09-07","search_start":"2026-09-06","timezone":null,"value":"2026-09-06 02:50:18"},"date_kind":"source_date","edition_id":"edn_43b5ef25-cb95-45bf-8daf-223308f5a3eb","href":"articles/rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1/","is_representative":true,"revision_id":"rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","title":"Personal Cloud Computer――AI Agentが「第二のPC」を持つ時代","work_id":"wrk_7b4fd6ec-e3a4-4a64-b21f-d3e9dbb0ed11"},{"date":{"precision":"instant","search_end":"2026-05-14","search_start":"2026-05-13","timezone":null,"value":"2026-05-13 17:54:34"},"date_kind":"source_date","edition_id":"edn_e78006c3-28bb-4c5c-b1ef-fa6d1da59cf4","href":"articles/rev_6abb0df2-b5a4-4f1c-8f48-a346d6d0f247/","is_representative":true,"revision_id":"rev_6abb0df2-b5a4-4f1c-8f48-a346d6d0f247","title":"CDN・セキュリティ領域の未来：キャリア、ハイパースケーラー、Cloudflareはどう変わる","work_id":"wrk_37d80f4d-ab1e-4385-b7ef-e7ae09b765e3"},{"date":{"precision":"instant","search_end":"2026-05-09","search_start":"2026-05-08","timezone":null,"value":"2026-05-08 09:09:09"},"date_kind":"source_date","edition_id":"edn_0fde00ee-5147-47e0-9af2-388a01c1bd0b","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/","is_representative":true,"revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","title":"GPUクラウド成熟で何が起きるのか","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2"},{"date":{"precision":"instant","search_end":"2026-07-21","search_start":"2026-07-20","timezone":null,"value":"2026-07-20 07:19:08"},"date_kind":"source_date","edition_id":"edn_d352c86a-2725-4e69-a904-4f29033e0f3c","href":"articles/rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982/","is_representative":true,"revision_id":"rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","title":"AIエージェント時代、「人類の実データ」を持つ企業が再評価されるかも知れない","work_id":"wrk_2e3fa847-c032-4040-9284-2106f69cf66c"},{"date":{"precision":"instant","search_end":"2026-07-31","search_start":"2026-07-30","timezone":null,"value":"2026-07-30 19:33:36"},"date_kind":"source_date","edition_id":"edn_e2520b3a-9aef-4eaa-962f-fbe58bf5efa1","href":"articles/rev_716e29e3-ca33-4e0a-89d8-a4c146c91139/","is_representative":true,"revision_id":"rev_716e29e3-ca33-4e0a-89d8-a4c146c91139","title":"インターネットは情報をつなぎ、AIは能力を生産する","work_id":"wrk_6b2d4c96-c0d4-4d57-8d9e-9077ea538a79"},{"date":{"precision":"instant","search_end":"2026-09-05","search_start":"2026-09-04","timezone":null,"value":"2026-09-04 22:47:02"},"date_kind":"source_date","edition_id":"edn_1bac920a-2f71-4e7a-96d6-7b65687b7d4c","href":"articles/rev_73f07a40-b588-4fa6-bf2b-2880462e214c/","is_representative":true,"revision_id":"rev_73f07a40-b588-4fa6-bf2b-2880462e214c","title":"AstraはAI産業の競争軸を変えた","work_id":"wrk_83edd5a9-4c70-4085-ba5b-602f9d005928"},{"date":{"precision":"instant","search_end":"2026-05-01","search_start":"2026-04-30","timezone":null,"value":"2026-04-30 05:15:52"},"date_kind":"source_date","edition_id":"edn_7aee08e9-9fd2-4f8a-b18c-d213b48b29ec","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/","is_representative":true,"revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","title":"SOFIの現状と、AI時代のフィンテック地図","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d"},{"date":{"precision":"day","search_end":"2026-06-17","search_start":"2026-06-16","timezone":null,"value":"2026-06-16"},"date_kind":"source_date","edition_id":"edn_25cde12f-5d55-4f13-99a1-cec29c32d69e","href":"articles/rev_78657238-b40f-4a7c-b670-325df3688f39/","is_representative":true,"revision_id":"rev_78657238-b40f-4a7c-b670-325df3688f39","title":"HBMは熱を持つ記憶の塔へ: HBM4、iHBM/ICE、HBM5/HPBが変えるAIインフラ","work_id":"wrk_70fbeeef-5587-4e6a-a017-b7e323d0b279"},{"date":{"precision":"instant","search_end":"2026-04-23","search_start":"2026-04-22","timezone":null,"value":"2026-04-22 05:22:27"},"date_kind":"source_date","edition_id":"edn_14c8de97-096a-42c5-ac87-c9712aca9e38","href":"articles/rev_7cf42fcc-1956-4694-a0b6-5eaa9be5c284/","is_representative":true,"revision_id":"rev_7cf42fcc-1956-4694-a0b6-5eaa9be5c284","title":"フィジカルAI時代、誰が年商＋100億・500億・1000億ドルを取るのか","work_id":"wrk_cda7c85e-9714-493f-a66d-ac6f36454630"},{"date":{"precision":"instant","search_end":"2026-05-11","search_start":"2026-05-10","timezone":null,"value":"2026-05-10 19:19:43"},"date_kind":"source_date","edition_id":"edn_7f2c5722-4167-4b8c-ae1a-858e7de76e73","href":"articles/rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608/","is_representative":true,"revision_id":"rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","title":"AIエージェントは株式市場をどう変えるのか","work_id":"wrk_a3440338-a420-4877-947f-8193ae66700c"},{"date":{"precision":"instant","search_end":"2026-08-07","search_start":"2026-08-06","timezone":null,"value":"2026-08-06 05:35:18"},"date_kind":"source_date","edition_id":"edn_2c4a5f8e-3604-45b8-96b7-bb58cd6f7893","href":"articles/rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa/","is_representative":true,"revision_id":"rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa","title":"Elasticは「SaaSの死」を越えられるか","work_id":"wrk_82866dd9-73ac-495c-82f9-5873005352f3"},{"date":{"precision":"instant","search_end":"2026-08-02","search_start":"2026-08-01","timezone":null,"value":"2026-08-01 15:36:29"},"date_kind":"source_date","edition_id":"edn_3307b9b0-eb7f-4d19-8036-952bd4a19eda","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/","is_representative":true,"revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","title":"AIは止まるのか","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9"},{"date":{"precision":"day","search_end":"2026-06-02","search_start":"2026-06-01","timezone":null,"value":"2026-06-01"},"date_kind":"source_date","edition_id":"edn_7cdb1796-c3bc-45f5-9465-c653106d6fd8","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/","is_representative":true,"revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","title":"NVIDIAが描くAIエージェント産業革命: Token、AI工場、Physical AIの意味","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a"},{"date":{"precision":"instant","search_end":"2026-06-06","search_start":"2026-06-05","timezone":null,"value":"2026-06-05 17:19:35"},"date_kind":"source_date","edition_id":"edn_3d968ce9-319d-4c1c-b621-2f7f44d55f5b","href":"articles/rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a/","is_representative":true,"revision_id":"rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a","title":"SpaceX IPO考察","work_id":"wrk_c22379bc-7081-4f8f-a3e9-4418f450cf96"},{"date":{"precision":"day","search_end":"2026-06-19","search_start":"2026-06-18","timezone":null,"value":"2026-06-18"},"date_kind":"source_date","edition_id":"edn_15da5c20-6226-40bf-8235-bb7665c7f4bb","href":"articles/rev_92ff5158-6c34-4526-ab02-c8db8332951f/","is_representative":true,"revision_id":"rev_92ff5158-6c34-4526-ab02-c8db8332951f","title":"GLM-5.2と主権AI: オープンウェイトが各国のAIインフラ需要を押し上げる","work_id":"wrk_ea89805e-b45e-4a9c-b22a-29b168da02ed"},{"date":{"precision":"instant","search_end":"2026-06-26","search_start":"2026-06-25","timezone":null,"value":"2026-06-25 21:55:11"},"date_kind":"source_date","edition_id":"edn_fe1c49c7-fad2-4d4a-8474-be84f8f61d53","href":"articles/rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8/","is_representative":true,"revision_id":"rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8","title":"ハイパースケーラーは失業率の上昇と共に復活するのか？","work_id":"wrk_8f40984a-d96e-441d-8ace-83f20958435a"},{"date":{"precision":"instant","search_end":"2026-05-23","search_start":"2026-05-22","timezone":null,"value":"2026-05-22 22:10:50"},"date_kind":"source_date","edition_id":"edn_5f918d9e-b194-42a4-a17e-1c768eac869c","href":"articles/rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4/","is_representative":true,"revision_id":"rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","title":"AIDC向けCPU最新事情","work_id":"wrk_95f3adb2-9ef8-4c60-8a3b-5ded0268dac4"},{"date":{"precision":"instant","search_end":"2026-08-01","search_start":"2026-07-31","timezone":null,"value":"2026-07-31 07:30:01"},"date_kind":"source_date","edition_id":"edn_3249fab9-813c-4f85-9a24-f76cf32229e2","href":"articles/rev_94f3d3d6-6b52-4f98-89a0-318fe428f5cb/","is_representative":true,"revision_id":"rev_94f3d3d6-6b52-4f98-89a0-318fe428f5cb","title":"Amazon最新決算が示した「AI収益化の正体」","work_id":"wrk_43737b34-e6d4-44d7-b140-1cd6aacb15fb"},{"date":{"precision":"instant","search_end":"2026-06-24","search_start":"2026-06-23","timezone":null,"value":"2026-06-23 13:51:47"},"date_kind":"source_date","edition_id":"edn_0bcfedda-9682-43ec-a0ed-3e9231c0c06f","href":"articles/rev_977092a4-68df-47ac-8de2-4e66567c8a95/","is_representative":true,"revision_id":"rev_977092a4-68df-47ac-8de2-4e66567c8a95","title":"AI実需は消えない。だが、それだけではメモリ・半導体スーパーサイクルは永続しない","work_id":"wrk_27b36bee-da31-4942-a5ea-33f6f77ddee3"},{"date":{"precision":"instant","search_end":"2026-05-11","search_start":"2026-05-10","timezone":null,"value":"2026-05-10 03:03:48"},"date_kind":"source_date","edition_id":"edn_f5d56b4e-b19b-4e50-a9a9-50bdde3bbf14","href":"articles/rev_98f85497-697b-4917-aabc-df1485131d9b/","is_representative":true,"revision_id":"rev_98f85497-697b-4917-aabc-df1485131d9b","title":"Cloudflare急落の本質：SaaS売りに巻き込まれた「AI時代の関所」は、本当に見直されるのか","work_id":"wrk_23480985-b0bd-47ea-a46d-716d46f4d122"},{"date":{"precision":"instant","search_end":"2026-01-24","search_start":"2026-01-23","timezone":null,"value":"2026-01-23 23:03:48"},"date_kind":"source_date","edition_id":"edn_8176a9e3-bb40-4c48-b0bb-6505e1767c01","href":"articles/rev_9ae4d311-bf43-4bb1-a7b5-36a5649efc7c/","is_representative":true,"revision_id":"rev_9ae4d311-bf43-4bb1-a7b5-36a5649efc7c","title":"フィジカルAIとヒューマノイドロボットの現状と展望","work_id":"wrk_9cfc75d7-7e8e-4397-add6-437f60011804"},{"date":{"precision":"instant","search_end":"2026-04-17","search_start":"2026-04-16","timezone":null,"value":"2026-04-16 00:46:36"},"date_kind":"source_date","edition_id":"edn_1c685823-0e11-4085-98c1-e24488e54e9a","href":"articles/rev_9af30ed5-ec26-4204-9f65-58324ee5fba5/","is_representative":true,"revision_id":"rev_9af30ed5-ec26-4204-9f65-58324ee5fba5","title":"OpenAI・AnthropicのAIインフラ先行投資は回収できるのか","work_id":"wrk_c56e9825-8401-406e-aaae-662904613d26"},{"date":{"precision":"instant","search_end":"2026-07-31","search_start":"2026-07-30","timezone":null,"value":"2026-07-30 03:22:51"},"date_kind":"source_date","edition_id":"edn_873b44e8-2367-4478-956e-510cb18d4d86","href":"articles/rev_9c10c6f9-4055-4591-a368-74112c253a08/","is_representative":true,"revision_id":"rev_9c10c6f9-4055-4591-a368-74112c253a08","title":"今の相場の原因はインフレではなくAIだと思う","work_id":"wrk_a91b4c62-c6f7-43f1-8bc0-9224e28ea11a"},{"date":{"precision":"instant","search_end":"2026-07-05","search_start":"2026-07-04","timezone":"+09:00","value":"2026-07-04T02:28:11.000+09:00"},"date_kind":"source_date","edition_id":"edn_03d14087-d4a7-4c21-8d7c-4521637aa779","href":"articles/rev_a30fdebb-6a5f-4491-8480-bc05cbb21f96/","is_representative":true,"revision_id":"rev_a30fdebb-6a5f-4491-8480-bc05cbb21f96","title":"NAND・DRAM・HBMの違いから始める、AI時代のメモリー入門","work_id":"wrk_97fc23be-d3e1-4a2a-986e-b33ae2fd0697"},{"date":{"precision":"day","search_end":"2026-06-12","search_start":"2026-06-11","timezone":null,"value":"2026-06-11"},"date_kind":"source_date","edition_id":"edn_99248865-e3c1-49e4-ab39-4eaf3376ea48","href":"articles/rev_a3e10132-4b92-414d-8744-acbeb6f0b294/","is_representative":true,"revision_id":"rev_a3e10132-4b92-414d-8744-acbeb6f0b294","title":"AIはチャットから産業OSへ: 安全装置、価格戦争、決済、10GW電力が示す次の覇権","work_id":"wrk_ab79e5be-0326-461e-868c-ec7e75b2a90e"},{"date":{"precision":"instant","search_end":"2026-06-02","search_start":"2026-06-01","timezone":null,"value":"2026-06-01 20:25:08"},"date_kind":"source_date","edition_id":"edn_f6cc6006-a326-4bed-8126-3d47339058b9","href":"articles/rev_a4008bc2-ae2c-4bd7-8615-0b3c4e0936ff/","is_representative":true,"revision_id":"rev_a4008bc2-ae2c-4bd7-8615-0b3c4e0936ff","title":"「発表されたGW」と「実際に通電できるGW」がズレ始める局面を考察","work_id":"wrk_9bc2084d-1dcc-40ef-9cc9-a53fe61434c0"},{"date":{"precision":"day","search_end":"2026-06-14","search_start":"2026-06-13","timezone":null,"value":"2026-06-13"},"date_kind":"source_date","edition_id":"edn_b7dbf7fd-a660-4e8d-b44c-3bd42eb1a00c","href":"articles/rev_a749fcdf-8155-46e9-a466-83e867886a08/","is_representative":true,"revision_id":"rev_a749fcdf-8155-46e9-a466-83e867886a08","title":"AIモデルは国家が配分する戦略資産へ: Fable 5 / Mythos 5停止が示した主権AI","work_id":"wrk_e146f08d-b638-466d-b368-146816bcc837"},{"date":{"precision":"instant","search_end":"2026-06-26","search_start":"2026-06-25","timezone":"+09:00","value":"2026-06-25T17:41:03.000+09:00"},"date_kind":"source_date","edition_id":"edn_5908f8e9-327f-4dc3-964e-6246b6c89bea","href":"articles/rev_abe6b561-f51c-4549-a97f-aab2b725492e/","is_representative":true,"revision_id":"rev_abe6b561-f51c-4549-a97f-aab2b725492e","title":"Broadcom Jalapeñoから見えるAIインフラ戦争","work_id":"wrk_f31c79fd-b257-498a-9ee1-d344ac94ffc7"},{"date":{"precision":"instant","search_end":"2026-06-25","search_start":"2026-06-24","timezone":null,"value":"2026-06-24 17:41:44"},"date_kind":"source_date","edition_id":"edn_67c8f3df-4f51-41e9-9a07-e008714ca4db","href":"articles/rev_ae55c104-d140-4064-adf1-726fae29a7f3/","is_representative":true,"revision_id":"rev_ae55c104-d140-4064-adf1-726fae29a7f3","title":"DRAM二倍レバRAM、韓国レバ規制、過去のレバETF惨劇、そして四半期末リバランス","work_id":"wrk_a1470532-dd67-430b-91ce-2205bc4c54e0"},{"date":{"precision":"instant","search_end":"2026-06-27","search_start":"2026-06-26","timezone":null,"value":"2026-06-26 16:31:58"},"date_kind":"source_date","edition_id":"edn_a7876daa-9935-46d8-a181-dafced21529f","href":"articles/rev_aea0fd30-a496-42e0-975e-3de7b0cd01ab/","is_representative":true,"revision_id":"rev_aea0fd30-a496-42e0-975e-3de7b0cd01ab","title":"GPT-5.6は政府の許認制らしい","work_id":"wrk_4efede78-d9d9-4a37-854a-51a334476949"},{"date":{"precision":"day","search_end":"2026-06-12","search_start":"2026-06-11","timezone":null,"value":"2026-06-11"},"date_kind":"source_date","edition_id":"edn_f3975eb9-6ff9-4460-aaa4-51db79513c92","href":"articles/rev_afc88027-978f-428a-a597-dd5d129dc441/","is_representative":true,"revision_id":"rev_afc88027-978f-428a-a597-dd5d129dc441","title":"AIインフラ競争は金額だけでは測れない: 米国の頭脳と中国の身体・工場・電力層","work_id":"wrk_9fae1367-ebbb-47c5-ad7b-3700deb207cd"},{"date":{"precision":"instant","search_end":"2026-04-08","search_start":"2026-04-07","timezone":null,"value":"2026-04-07 22:44:58"},"date_kind":"source_date","edition_id":"edn_2954d2e7-9d8a-437a-a623-7d12f0aeff5f","href":"articles/rev_b228986a-43e7-4e2b-8b16-758283d367bc/","is_representative":true,"revision_id":"rev_b228986a-43e7-4e2b-8b16-758283d367bc","title":"OpenAIのB2B出遅れの実態と巻き返し戦略を考察","work_id":"wrk_b193a2ff-ca61-4efc-864e-0778162ac852"},{"date":{"precision":"instant","search_end":"2026-02-16","search_start":"2026-02-15","timezone":null,"value":"2026-02-15 15:56:00"},"date_kind":"source_date","edition_id":"edn_c8535090-1a8d-44a5-81c7-2abe7fd130ea","href":"articles/rev_b8a0343f-f76e-47f4-8c47-4536da944126/","is_representative":true,"revision_id":"rev_b8a0343f-f76e-47f4-8c47-4536da944126","title":"電気工事士のような仕事が注目されてますが...","work_id":"wrk_0eee3d56-e636-4917-98f6-cddcb3cca3f5"},{"date":{"precision":"day","search_end":"2026-06-23","search_start":"2026-06-22","timezone":null,"value":"2026-06-22"},"date_kind":"source_date","edition_id":"edn_93d3f64a-93d3-4bbc-8beb-702fd3d134f1","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/","is_representative":true,"revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","title":"Fugu、ARD、A2Aが示すAIエージェント時代の競争軸","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe"},{"date":{"precision":"day","search_end":"2026-08-04","search_start":"2026-08-03","timezone":null,"value":"2026-08-03"},"date_kind":"source_date","edition_id":"edn_351180ab-da4f-4b8d-a446-455ba3535cc4","href":"articles/rev_bcca5c5e-94f7-4bf9-b249-187454f09a22/","is_representative":true,"revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","title":"DeepSeek V4 Flashの衝撃、その先のQwen3.8-Max――効率化してもAI計算需要は増えるのか","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920"},{"date":{"precision":"day","search_end":"2026-06-03","search_start":"2026-06-02","timezone":null,"value":"2026-06-02"},"date_kind":"source_date","edition_id":"edn_1104e314-7b9b-4ce7-9543-94f31eddcf1c","href":"articles/rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a/","is_representative":true,"revision_id":"rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a","title":"AnthropicのS-1機密提出とAI企業黒字化の試験: Claude Code、Codex、物理AIまで読む","work_id":"wrk_dcf18d88-5123-4a4f-aa9b-564dc767f048"},{"date":{"precision":"instant","search_end":"2026-09-16","search_start":"2026-09-15","timezone":"+09:00","value":"2026-09-15T01:29:06.000+09:00"},"date_kind":"source_date","edition_id":"edn_b530c47b-577d-413c-83ec-7ef8dc6e8cd4","href":"articles/rev_bf97dda8-b55b-4e4d-b1cf-b775474af9b2/","is_representative":true,"revision_id":"rev_bf97dda8-b55b-4e4d-b1cf-b775474af9b2","title":"TTTが「記憶」を再設計する","work_id":"wrk_72fede57-1e40-4a22-81c2-67b2328ba457"},{"date":{"precision":"day","search_end":"2026-06-06","search_start":"2026-06-05","timezone":null,"value":"2026-06-05"},"date_kind":"source_date","edition_id":"edn_d3693d96-162a-4df9-810c-a18b794db0f9","href":"articles/rev_c1530075-b9e3-4144-b8ce-f6e480c778fa/","is_representative":true,"revision_id":"rev_c1530075-b9e3-4144-b8ce-f6e480c778fa","title":"AIは知能から行動へスケールする: Cloudflare RadarとAnthropicが示すAIインフラ投資の本質","work_id":"wrk_50367f10-59d7-41cc-990d-213cec7e7f79"},{"date":{"precision":"instant","search_end":"2026-09-07","search_start":"2026-09-06","timezone":null,"value":"2026-09-06 02:46:20"},"date_kind":"source_date","edition_id":"edn_1508bb58-beb5-4d28-a5bb-57f60f6c5310","href":"articles/rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42/","is_representative":true,"revision_id":"rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42","title":"AI時代、クラウドは「第二のPC」になる","work_id":"wrk_675d7d8c-da38-4285-a483-5126d06bc835"},{"date":{"precision":"instant","search_end":"2026-09-10","search_start":"2026-09-09","timezone":"+09:00","value":"2026-09-09T16:07:22.000+09:00"},"date_kind":"source_date","edition_id":"edn_9686cd7e-d57a-489b-9bce-bda8097a2faa","href":"articles/rev_c832f985-caf8-4ac7-bf58-f785e8782922/","is_representative":true,"revision_id":"rev_c832f985-caf8-4ac7-bf58-f785e8782922","title":"HKMGA再帰的Function拡張仮説","work_id":"wrk_37471fb9-72db-4a02-a976-c42f743def51"},{"date":{"precision":"day","search_end":"2026-08-09","search_start":"2026-08-08","timezone":null,"value":"2026-08-08"},"date_kind":"source_date","edition_id":"edn_781f08f7-a1a3-43c7-8fc3-3a30da2ceb1f","href":"articles/rev_c89081f7-eb04-4049-a4a4-68cbe681a3f5/","is_representative":true,"revision_id":"rev_c89081f7-eb04-4049-a4a4-68cbe681a3f5","title":"AIは「巨大GPU」から「巨大な記憶コンピュータ」へ――MoE・HBM・Optical Fabricが変える次世代AIインフラ","work_id":"wrk_08ecb116-f4c7-4c99-9574-e485a5adb41f"},{"date":{"precision":"instant","search_end":"2026-07-04","search_start":"2026-07-03","timezone":null,"value":"2026-07-03 06:12:13"},"date_kind":"source_date","edition_id":"edn_b83b5a0d-0856-4d09-913c-ac86cdd40264","href":"articles/rev_c8b19a4c-93eb-4695-8f38-c8166d22cbd9/","is_representative":true,"revision_id":"rev_c8b19a4c-93eb-4695-8f38-c8166d22cbd9","title":"Digi International考察","work_id":"wrk_f4bf8d6d-6894-4267-93c4-c8e1e447fab3"},{"date":{"precision":"instant","search_end":"2026-09-14","search_start":"2026-09-13","timezone":"+09:00","value":"2026-09-13T02:06:24.000+09:00"},"date_kind":"source_date","edition_id":"edn_c3fa759f-c669-4812-ab5e-2b9acd73b92c","href":"articles/rev_c950677f-d85d-461a-bf3d-314cf6b08a37/","is_representative":true,"revision_id":"rev_c950677f-d85d-461a-bf3d-314cf6b08a37","title":"AIはなぜ「減速」が語られ始めたのか","work_id":"wrk_a1c11bb3-0dd1-46bc-9f3c-ec06fe1df011"},{"date":{"precision":"instant","search_end":"2026-06-14","search_start":"2026-06-13","timezone":null,"value":"2026-06-13 20:40:43"},"date_kind":"source_date","edition_id":"edn_407987d9-e0ae-4cb6-8ad3-3370e69ebc10","href":"articles/rev_ca87695f-4cf9-40a1-bc61-6a123a71ffe6/","is_representative":true,"revision_id":"rev_ca87695f-4cf9-40a1-bc61-6a123a71ffe6","title":"偶然が重なりすぎる時、AI相場は別の物語を読み始める","work_id":"wrk_92c272f3-ada3-4a0c-815e-d33c175ffbe0"},{"date":{"precision":"day","search_end":"2026-06-19","search_start":"2026-06-18","timezone":null,"value":"2026-06-18"},"date_kind":"source_date","edition_id":"edn_1e86ae97-6b85-42ad-b514-5f78761a9480","href":"articles/rev_cf107e9a-1286-4df3-bd1f-adb77c070354/","is_representative":true,"revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","title":"ENPIREはWLAを不要にするのか: フィジカルAIは自己改善する実機システムへ進む","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499"},{"date":{"precision":"instant","search_end":"2026-08-04","search_start":"2026-08-03","timezone":null,"value":"2026-08-03 16:28:45"},"date_kind":"source_date","edition_id":"edn_ec5f9840-51fe-471e-8352-37c8609addd2","href":"articles/rev_d4189e74-ba90-4c4f-8495-9a2fccea5a2a/","is_representative":true,"revision_id":"rev_d4189e74-ba90-4c4f-8495-9a2fccea5a2a","title":"NVIDIA対Huawei――世界の主権AIを奪い合う二つのAIファクトリー","work_id":"wrk_ba0d2481-72cf-412b-92be-570c7d9c07e4"},{"date":{"precision":"day","search_end":"2026-06-17","search_start":"2026-06-16","timezone":null,"value":"2026-06-16"},"date_kind":"source_date","edition_id":"edn_f817ecb0-0059-4af0-a33d-8438c8d9db24","href":"articles/rev_d47a57d8-f3a6-4505-9ea6-c03ae4416616/","is_representative":true,"revision_id":"rev_d47a57d8-f3a6-4505-9ea6-c03ae4416616","title":"主権AIの現実路線: フロンティアモデルを持つ国、交渉する国、学習ループを守る国","work_id":"wrk_0ebc3696-2c90-4588-8ed4-2ea51e51433d"},{"date":{"precision":"instant","search_end":"2025-01-29","search_start":"2025-01-28","timezone":null,"value":"2025-01-28 19:49:31"},"date_kind":"source_date","edition_id":"edn_fed4429f-a858-49c3-9fb3-3d56c73f575d","href":"articles/rev_d8e8cb66-632c-4170-afae-461911315dbd/","is_representative":true,"revision_id":"rev_d8e8cb66-632c-4170-afae-461911315dbd","title":"Deep Seekについて","work_id":"wrk_7ab4bb8d-7296-4037-b30a-70a60e28196a"},{"date":{"precision":"instant","search_end":"2026-05-16","search_start":"2026-05-15","timezone":null,"value":"2026-05-15 18:50:18"},"date_kind":"source_date","edition_id":"edn_01d87b0e-a8f3-4302-a8c3-60b4943d484a","href":"articles/rev_da6063d5-59b2-4519-9c42-45d8f4f206c8/","is_representative":true,"revision_id":"rev_da6063d5-59b2-4519-9c42-45d8f4f206c8","title":"キオクシア決算が示した「NAND市況株」から「AI記憶階層株」への転換","work_id":"wrk_ef613b57-c920-49a8-bb03-b7fdb149eeca"},{"date":{"precision":"day","search_end":"2026-06-14","search_start":"2026-06-13","timezone":null,"value":"2026-06-13"},"date_kind":"source_date","edition_id":"edn_7d46c81a-9f0e-4fe5-9231-1d291c06fb36","href":"articles/rev_dbe69fb8-f76d-44e6-b111-aa27e548a204/","is_representative":true,"revision_id":"rev_dbe69fb8-f76d-44e6-b111-aa27e548a204","title":"TPU 8iとTPU 8tの違い: GoogleがAIチップを学習と推論に分けた理由","work_id":"wrk_0252d3aa-e602-4a99-abf9-0b9ec2b5796a"},{"date":{"precision":"instant","search_end":"2026-06-23","search_start":"2026-06-22","timezone":"+09:00","value":"2026-06-22T20:54:37.000+09:00"},"date_kind":"source_date","edition_id":"edn_25f9d35e-0583-4e59-a75d-12eaf7197970","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/","is_representative":true,"revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","title":"AIエージェントのWebが始まる: MCP、A2A、Skills、ARD、Fuguのデジタル神経系","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d"},{"date":{"precision":"instant","search_end":"2026-09-07","search_start":"2026-09-06","timezone":null,"value":"2026-09-06 14:30:30"},"date_kind":"source_date","edition_id":"edn_4a5ceefa-8226-49cd-a4a4-567e3faf1017","href":"articles/rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646/","is_representative":true,"revision_id":"rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646","title":"数十億人が「第二のPC」を持つ時代","work_id":"wrk_4c390ee8-19a9-47a2-baff-32ac4e8e79ad"},{"date":{"precision":"instant","search_end":"2026-08-19","search_start":"2026-08-18","timezone":null,"value":"2026-08-18 13:47:59"},"date_kind":"source_date","edition_id":"edn_d5501602-962a-4bbe-b351-cc2a980a0d67","href":"articles/rev_e1c0a212-5ad0-4fbc-81f6-60d2df49d753/","is_representative":true,"revision_id":"rev_e1c0a212-5ad0-4fbc-81f6-60d2df49d753","title":"AIアクセラレータはGPUから「計算機そのもの」へ","work_id":"wrk_88c25067-a100-4eee-96f7-180120d2ac61"},{"date":{"precision":"instant","search_end":"2026-08-26","search_start":"2026-08-25","timezone":null,"value":"2026-08-25 18:35:04"},"date_kind":"source_date","edition_id":"edn_e5f1fb69-a298-478b-b928-f8f71e887ce7","href":"articles/rev_e2c369be-a992-44a9-bbc1-81ed4928e89f/","is_representative":true,"revision_id":"rev_e2c369be-a992-44a9-bbc1-81ed4928e89f","title":"AIインフラの次のボトルネックはどこへ移るのか","work_id":"wrk_d2f61eaa-5ee5-4d1f-b70d-7356a79ea61d"},{"date":{"precision":"instant","search_end":"2026-09-24","search_start":"2026-09-23","timezone":"+09:00","value":"2026-09-23T18:34:18.000+09:00"},"date_kind":"source_date","edition_id":"edn_bcdb2b51-d174-401d-b953-e1214abb77ac","href":"articles/rev_e5e24c6f-5887-4f16-b03e-081da5340883/","is_representative":true,"revision_id":"rev_e5e24c6f-5887-4f16-b03e-081da5340883","title":"AIはどう学び、次のAIを育てるのか","work_id":"wrk_470cca01-eadf-46c3-bd50-cd79e3b1e197"},{"date":{"precision":"instant","search_end":"2026-09-11","search_start":"2026-09-10","timezone":"+09:00","value":"2026-09-10T04:12:13.000+09:00"},"date_kind":"source_date","edition_id":"edn_38704b43-8c87-4af9-8e08-ac1ab40d4192","href":"articles/rev_eaa7004e-c9e4-4bc4-9718-2e518decec70/","is_representative":true,"revision_id":"rev_eaa7004e-c9e4-4bc4-9718-2e518decec70","title":"AIが半導体設計を始めた時代――Broadcomは何を守り、どこへ向かうのか","work_id":"wrk_dfcbc670-103a-4931-8d23-d5f51671c61b"},{"date":{"precision":"instant","search_end":"2026-05-16","search_start":"2026-05-15","timezone":null,"value":"2026-05-15 20:04:40"},"date_kind":"source_date","edition_id":"edn_dcc3db31-ac2f-46dd-bbbb-08ede47a7184","href":"articles/rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e/","is_representative":true,"revision_id":"rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e","title":"クアルコム再評価論：スマホSoC企業から「フィジカルAI時代の神経系」へ","work_id":"wrk_65ea8146-0c98-40e2-b084-2dacaa4f1bf1"},{"date":{"precision":"instant","search_end":"2026-09-09","search_start":"2026-09-08","timezone":"+09:00","value":"2026-09-08T18:38:54.000+09:00"},"date_kind":"source_date","edition_id":"edn_f8dfc7de-cb12-4335-94e5-907ddc81811b","href":"articles/rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4/","is_representative":true,"revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","title":"Elasticは検索会社から「企業AIのデータ基盤」へ進めるか","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd"},{"date":{"precision":"instant","search_end":"2025-07-26","search_start":"2025-07-25","timezone":null,"value":"2025-07-25 14:34:32"},"date_kind":"source_date","edition_id":"edn_15a0c752-3e12-4d8b-b31b-bf920cad6081","href":"articles/rev_f17ef624-e9fd-49a4-b061-966093172de4/","is_representative":true,"revision_id":"rev_f17ef624-e9fd-49a4-b061-966093172de4","title":"計算資源の見通し","work_id":"wrk_168418fa-2b60-4053-acfe-b86f6898d033"},{"date":{"precision":"instant","search_end":"2026-07-16","search_start":"2026-07-15","timezone":null,"value":"2026-07-15 21:42:22"},"date_kind":"source_date","edition_id":"edn_f0da2006-637a-498b-94f4-156ed0303c87","href":"articles/rev_f3a5d7c9-6dfa-4c38-8bf2-a37063801332/","is_representative":true,"revision_id":"rev_f3a5d7c9-6dfa-4c38-8bf2-a37063801332","title":"Cerebrasに勝ち目はないのか","work_id":"wrk_d32481f4-3a5a-40d8-a817-86b115c8c4f4"},{"date":{"precision":"instant","search_end":"2026-09-22","search_start":"2026-09-21","timezone":"+09:00","value":"2026-09-21T14:06:22.000+09:00"},"date_kind":"source_date","edition_id":"edn_087e5f0a-56da-41cb-ac35-063583de23dd","href":"articles/rev_f45cfd6f-625d-4e49-b987-25f79284501d/","is_representative":true,"revision_id":"rev_f45cfd6f-625d-4e49-b987-25f79284501d","title":"崩壊しても都市は残る――中国不動産バブルの後始末と、K字型AI経済の先にあるもの","work_id":"wrk_28a55046-0758-4fb4-b7b9-5ec8fa7681a7"},{"date":{"precision":"instant","search_end":"2026-04-25","search_start":"2026-04-24","timezone":null,"value":"2026-04-24 01:44:34"},"date_kind":"source_date","edition_id":"edn_58c088c0-8bb8-4f15-a4bb-e62eda920886","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/","is_representative":true,"revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","title":"AI-RANと境界暗号・認証・実機管理は、AI時代の「神経系」と「免疫系」になる","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd"},{"date":{"precision":"instant","search_end":"2026-09-14","search_start":"2026-09-13","timezone":"+09:00","value":"2026-09-13T08:33:21.000+09:00"},"date_kind":"source_date","edition_id":"edn_5eab9a4a-df46-42fa-bbc7-7752d2f6f2e3","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/","is_representative":true,"revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","title":"AI Agent時代のセキュリティ地図","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a"},{"date":{"precision":"instant","search_end":"2026-08-24","search_start":"2026-08-23","timezone":null,"value":"2026-08-23 20:19:37"},"date_kind":"source_date","edition_id":"edn_fa52bf0e-1e7b-4eaa-9017-28e7018b9f4b","href":"articles/rev_f76185ef-dd8b-44e4-a82a-7b8cc8225b9a/","is_representative":true,"revision_id":"rev_f76185ef-dd8b-44e4-a82a-7b8cc8225b9a","title":"SRAMは次のAIメモリ不足を作るのか","work_id":"wrk_72f5f141-da8e-4f5b-a337-f46180c37f2e"},{"date":{"precision":"instant","search_end":"2026-06-21","search_start":"2026-06-20","timezone":null,"value":"2026-06-20 02:24:26"},"date_kind":"source_date","edition_id":"edn_ddc7300e-9c53-4a3a-b3ba-c9862355ae25","href":"articles/rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280/","is_representative":true,"revision_id":"rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280","title":"メモリサイクルは終わるのか──AI OS化、オンデバイスAI、そして「デバイス階級社会」の始まり","work_id":"wrk_a85448e1-149a-46bd-b7e9-d66795012c12"},{"date":{"precision":"instant","search_end":"2026-07-30","search_start":"2026-07-29","timezone":null,"value":"2026-07-29 23:58:25"},"date_kind":"source_date","edition_id":"edn_903a8153-c158-488c-bbfc-a2503b5a55c9","href":"articles/rev_fd5eeabd-4433-491f-b463-c900cf908355/","is_representative":true,"revision_id":"rev_fd5eeabd-4433-491f-b463-c900cf908355","title":"SK hynix 2026年Q2決算の本質","work_id":"wrk_d47ffc56-09dd-4f57-a300-be2346278872"},{"date":{"precision":"instant","search_end":"2026-06-11","search_start":"2026-06-10","timezone":null,"value":"2026-06-10 12:32:57"},"date_kind":"source_date","edition_id":"edn_ed56e3b5-7945-4c7a-b609-18e8a6a3d962","href":"articles/rev_ffbdfa80-0e8b-40b5-8765-ba1e5b977d47/","is_representative":true,"revision_id":"rev_ffbdfa80-0e8b-40b5-8765-ba1e5b977d47","title":"NVIDIAのロボットAIスタックを読み解く","work_id":"wrk_75c66a96-f27d-4786-b253-3d6d4f826277"}],"occurrences":[{"id":"occ_f3d835eedb96fa210f6d9300","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_00217178-ae5e-4e98-87ca-c35048af1ca4","work_id":"wrk_6f50f82d-878e-437d-81af-1073e3707af7","block_id":"blk_18ba74f7-25f7-4745-ae42-2d12860ad115","section_id":"sec_03e290f6-4a8f-4b86-b94e-d3943651be5a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":69,"end":75,"exact":"エージェント","quote":"  \n> GPUと組み合わせる  \n> ↓  \n> AIデータセンターを建設する  \n> ↓  \n> モデルとエージェントを動かす  \n> ↓  \n> 工場、ロボット、車両、都市で利用する","quote_start":14,"quote_end":108,"text_sha256":"22038c2bea2e0ce33780e01a89d4482588eda8fcd3b56c85dfea7b94310ded50","block_sha256":"22038c2bea2e0ce33780e01a89d4482588eda8fcd3b56c85dfea7b94310ded50","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_00217178-ae5e-4e98-87ca-c35048af1ca4/#blk_18ba74f7-25f7-4745-ae42-2d12860ad115"},{"id":"occ_d8178f3fc4b0c50773ddcc6e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_00217178-ae5e-4e98-87ca-c35048af1ca4","work_id":"wrk_6f50f82d-878e-437d-81af-1073e3707af7","block_id":"blk_8c8879d3-841f-4200-9aea-f42a7007696a","section_id":"sec_47bc02dd-242a-4854-b75a-2a4ea4698fcb","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":23,"end":29,"exact":"エージェント","quote":"両社は、大規模言語モデルの学習だけでなく、AIエージェントやロボット、フィジカルAIまでを想定し、次世代HBMを共同で設計・最適化する。NVIDIAは将来必要になるメモリを早期に確保でき、SK hynixは次世代GPUの仕様策定に近い位置で製品を開発できる。","quote_start":0,"quote_end":129,"text_sha256":"fb9bae603991ce4d232953e7fb0afd5dfe9c16cb6eb55c859268df88b5b66d28","block_sha256":"fb9bae603991ce4d232953e7fb0afd5dfe9c16cb6eb55c859268df88b5b66d28","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_00217178-ae5e-4e98-87ca-c35048af1ca4/#blk_8c8879d3-841f-4200-9aea-f42a7007696a"},{"id":"occ_115b900fa5e042d14b5e92e2","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_00217178-ae5e-4e98-87ca-c35048af1ca4","work_id":"wrk_6f50f82d-878e-437d-81af-1073e3707af7","block_id":"blk_c442fa70-9f0e-461f-b6a7-832716e9a5ae","section_id":"sec_cb2e25c2-b8ed-49ae-9751-2caab3de8189","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":46,"end":52,"exact":"エージェント","quote":"NAVERはまた、HyperCLOVA XとNVIDIAのオープンモデル群を組み合わせ、AIエージェントやフィジカルAI向けモデルを開発する方針を示している。([NAVER Corp.](https://www.navercorp.com/media/pressReleasesDetail?seq=10","quote_start":0,"quote_end":152,"text_sha256":"e5615f7585fa5c5ed0e8584fa676a147418c6c9dbb31484c80af874e5156bba7","block_sha256":"e5615f7585fa5c5ed0e8584fa676a147418c6c9dbb31484c80af874e5156bba7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_00217178-ae5e-4e98-87ca-c35048af1ca4/#blk_c442fa70-9f0e-461f-b6a7-832716e9a5ae"},{"id":"occ_7ac80fd789510d2fd0b6854f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_00217178-ae5e-4e98-87ca-c35048af1ca4","work_id":"wrk_6f50f82d-878e-437d-81af-1073e3707af7","block_id":"blk_c527205e-d1b4-4242-8c5e-efd7139d9c18","section_id":"sec_cb2e25c2-b8ed-49ae-9751-2caab3de8189","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":64,"end":70,"exact":"エージェント","quote":"ーにはNVIDIAのBlackwellおよびVera Rubin世代を採用し、韓国と米国のAI企業が、モデル、エージェント、AIサービスを開発するための計算資源を提供する。","quote_start":9,"quote_end":95,"text_sha256":"6ec4aba99c6b8710d719a245527f11fd3a2a121ec38a950221898852699daa28","block_sha256":"6ec4aba99c6b8710d719a245527f11fd3a2a121ec38a950221898852699daa28","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_00217178-ae5e-4e98-87ca-c35048af1ca4/#blk_c527205e-d1b4-4242-8c5e-efd7139d9c18"},{"id":"occ_87a91703efcb6bfdad9324e0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_01c4b1d1-a854-4507-a2d7-bcf6b40fafd5","work_id":"wrk_dbe273d0-7a27-4a33-8051-93c6d0c21198","block_id":"blk_6b98e92d-6d03-4028-98a4-adae97180fb8","section_id":"sec_ca2cb607-4103-45d3-b3a4-f019ed3c38ef","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agent。","quote_start":0,"quote_end":9,"text_sha256":"89fe870adea8d50f6aacc4f24b3cf239e9b07a608840096de5ba25001fbeee49","block_sha256":"89fe870adea8d50f6aacc4f24b3cf239e9b07a608840096de5ba25001fbeee49","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_01c4b1d1-a854-4507-a2d7-bcf6b40fafd5/#blk_6b98e92d-6d03-4028-98a4-adae97180fb8"},{"id":"occ_fdc965970117cf944e09d529","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_01c4b1d1-a854-4507-a2d7-bcf6b40fafd5","work_id":"wrk_dbe273d0-7a27-4a33-8051-93c6d0c21198","block_id":"blk_cdbee3d9-94a2-4a74-9fa9-bea1e0124b54","section_id":"sec_3990d626-8627-4a50-a47d-6d3bb0f2aff9","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":3,"end":11,"exact":"AI Agent","quote":"さらにAI Agentについて、","quote_start":0,"quote_end":16,"text_sha256":"7681488db91230df23a9475fd4fb81f48e6a546ca1a75e721d3b1b942c7abaf2","block_sha256":"7681488db91230df23a9475fd4fb81f48e6a546ca1a75e721d3b1b942c7abaf2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_01c4b1d1-a854-4507-a2d7-bcf6b40fafd5/#blk_cdbee3d9-94a2-4a74-9fa9-bea1e0124b54"},{"id":"occ_641c8ae9ba58d2235872377b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e","work_id":"wrk_078157a8-8d73-48b6-b0ad-972882b9971f","block_id":"blk_4a3aaf4a-9384-41c1-a096-181fb5d3d471","section_id":"sec_54a465b6-7ad9-40bb-96d0-434bd0ac7edc","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":73,"end":80,"exact":"Agentic","quote":"yment、Data Analysis、Document generationなど、よりEnd-to-EndなAgentic Workへ利用が拡大している。([Anthropic](https://www.anthropic.com/research/claude-code-expertise?hl=en-US&utm_s","quote_start":18,"quote_end":180,"text_sha256":"2d8a7f384829e78bcc6cfff94cf5545e713a6e0f565957427dc0a8ad89d0d648","block_sha256":"2d8a7f384829e78bcc6cfff94cf5545e713a6e0f565957427dc0a8ad89d0d648","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e/#blk_4a3aaf4a-9384-41c1-a096-181fb5d3d471"},{"id":"occ_f799aed8cd0f44d215fabebb","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e","work_id":"wrk_078157a8-8d73-48b6-b0ad-972882b9971f","block_id":"blk_5a6c41a0-0965-4535-b75e-2a8a87b7474d","section_id":"sec_f1bac5c7-a595-47c1-9908-e4610c22e3b1","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":8,"end":15,"exact":"Agentic","quote":"OpenAIではAgentic useがChatを上回り始めた。","quote_start":0,"quote_end":32,"text_sha256":"8c69e39d0555b0f20a7aa85f3f20a558a4a3c5145474416218b7d24fcb3e5264","block_sha256":"8c69e39d0555b0f20a7aa85f3f20a558a4a3c5145474416218b7d24fcb3e5264","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e/#blk_5a6c41a0-0965-4535-b75e-2a8a87b7474d"},{"id":"occ_5aea4232391d4fb51f1cfaf6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e","work_id":"wrk_078157a8-8d73-48b6-b0ad-972882b9971f","block_id":"blk_65b61d1b-474c-4a23-bd63-4678dabbc683","section_id":"sec_0a3565ef-cbaa-4d7c-947d-efc76c7ba8ff","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":3,"end":11,"exact":"AI Agent","quote":"## AI Agentが何時間働いているか","quote_start":0,"quote_end":21,"text_sha256":"33231556e4f6ec011b6a88d195a3c3b30c7692e28cada178e2240af69b2612bb","block_sha256":"33231556e4f6ec011b6a88d195a3c3b30c7692e28cada178e2240af69b2612bb","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e/#blk_65b61d1b-474c-4a23-bd63-4678dabbc683"},{"id":"occ_945538d184c9cba5bcad56f1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e","work_id":"wrk_078157a8-8d73-48b6-b0ad-972882b9971f","block_id":"blk_6c7a47c7-a4bb-48ca-b377-bd154687fa5a","section_id":"sec_06e22581-ae28-4674-b04a-80a52311d786","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":10,"end":17,"exact":"Agentic","quote":"OpenAIでは既にAgentic AIが企業Output Tokensの64%を占め、LegalやSalesでCodex userが急増している。([OpenAI](https://openai.com/signals/enterp","quote_start":0,"quote_end":117,"text_sha256":"8fcdf87a041c0ece29850324a3759f1229dfa9f0ae348936407e1ee6e8436618","block_sha256":"8fcdf87a041c0ece29850324a3759f1229dfa9f0ae348936407e1ee6e8436618","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e/#blk_6c7a47c7-a4bb-48ca-b377-bd154687fa5a"},{"id":"occ_430414f055c905998ff6ea9f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e","work_id":"wrk_078157a8-8d73-48b6-b0ad-972882b9971f","block_id":"blk_77e6a9fc-28bd-4975-b41c-cdfdf610db3b","section_id":"sec_58a62093-3c5d-4e8c-b4e8-9804fe56131a","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":3,"end":10,"exact":"Agentic","quote":"さらにAgentic AI。","quote_start":0,"quote_end":14,"text_sha256":"f5ee0ece5ed36e6ee5f58adfd59fbb7cd11f8a947165615705d6d0f3b9309c75","block_sha256":"f5ee0ece5ed36e6ee5f58adfd59fbb7cd11f8a947165615705d6d0f3b9309c75","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e/#blk_77e6a9fc-28bd-4975-b41c-cdfdf610db3b"},{"id":"occ_8dbd0c82a238708472edfe25","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e","work_id":"wrk_078157a8-8d73-48b6-b0ad-972882b9971f","block_id":"blk_96444f52-bc1a-460c-b4bd-a0df2e6012a5","section_id":"sec_16555056-8f50-4f8a-b491-2c19228fa2f6","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":3,"end":10,"exact":"Agentic","quote":"ここもAgentic AIで需要が急増する。","quote_start":0,"quote_end":22,"text_sha256":"4ced28b32bec7f98467dc19e45dc52d0c2b8c3d762f08678d173e80c243e52c1","block_sha256":"4ced28b32bec7f98467dc19e45dc52d0c2b8c3d762f08678d173e80c243e52c1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e/#blk_96444f52-bc1a-460c-b4bd-a0df2e6012a5"},{"id":"occ_7e68d34de0165694d5ca1640","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e","work_id":"wrk_078157a8-8d73-48b6-b0ad-972882b9971f","block_id":"blk_e9eb6367-40b0-4010-a1e4-5056a87768a8","section_id":"sec_aaa15166-1a57-43cf-8617-e200ae8c72ba","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":10,"end":18,"exact":"AI Agent","quote":"仮にPrimary AI AgentのToken Activityに対して、","quote_start":0,"quote_end":38,"text_sha256":"e11ae4f865869cb1114de8b42491dce1b0b10ff3d7003b1c0b5ff68cae06cee6","block_sha256":"e11ae4f865869cb1114de8b42491dce1b0b10ff3d7003b1c0b5ff68cae06cee6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e/#blk_e9eb6367-40b0-4010-a1e4-5056a87768a8"},{"id":"occ_befcbf5757dd49d67bff6e08","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e","work_id":"wrk_078157a8-8d73-48b6-b0ad-972882b9971f","block_id":"blk_f6680cd4-f131-4db0-9733-4f6859afc321","section_id":"sec_33dc0583-0dde-4f21-b3cb-5f5aa8aa6795","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":9,"end":16,"exact":"Agentic","quote":"低Latency Agentic Inferenceも不足するかもしれない。","quote_start":0,"quote_end":38,"text_sha256":"4e931ae7c087dd7d37db08bba27314cce0350a26ba9631523a263a81f9a24a20","block_sha256":"4e931ae7c087dd7d37db08bba27314cce0350a26ba9631523a263a81f9a24a20","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e/#blk_f6680cd4-f131-4db0-9733-4f6859afc321"},{"id":"occ_339e06f3af4a7e7d3c011787","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_0ee327e2-ecfe-42e6-b337-fce8abbe7b71","work_id":"wrk_e5060e37-5e03-4669-88d5-6a5989ca3dcb","block_id":"blk_a08512f0-4a26-47fa-b2a4-f6cae8fec557","section_id":"sec_8565d459-ba53-4769-9af4-acbee353b86f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":48,"end":54,"exact":"エージェント","quote":"従来のCADや熱解析ツールが「静的」な設計を主眼としていたのに対し、Acorn 2.0は多数の「エージェント（衛星、地上局、敵対的脅威、環境要因など）」が相互作用する複雑なシステム全体の振る舞いをシミュレートする。","quote_start":0,"quote_end":107,"text_sha256":"d27fd595b107830ac82bdb8390343c1e2c7dc3661dd00baba4d18b5fbc676d2e","block_sha256":"d27fd595b107830ac82bdb8390343c1e2c7dc3661dd00baba4d18b5fbc676d2e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_0ee327e2-ecfe-42e6-b337-fce8abbe7b71/#blk_a08512f0-4a26-47fa-b2a4-f6cae8fec557"},{"id":"occ_fff1b113b3baa16e293d6501","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_0ee327e2-ecfe-42e6-b337-fce8abbe7b71","work_id":"wrk_e5060e37-5e03-4669-88d5-6a5989ca3dcb","block_id":"blk_cbbc42bf-d89d-4c7b-855a-62801937eac3","section_id":"sec_8565d459-ba53-4769-9af4-acbee353b86f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":0,"end":6,"exact":"エージェント","quote":"エージェントベース・モデリングとシミュレーション (ABMS):","quote_start":0,"quote_end":32,"text_sha256":"eb5f11334b22061e85d3bf57e01da22528fbfa2e69338e078f38c78bf7dcaee5","block_sha256":"eb5f11334b22061e85d3bf57e01da22528fbfa2e69338e078f38c78bf7dcaee5","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_0ee327e2-ecfe-42e6-b337-fce8abbe7b71/#blk_cbbc42bf-d89d-4c7b-855a-62801937eac3"},{"id":"occ_9b653480921bcbc8aaec6e47","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_0ee327e2-ecfe-42e6-b337-fce8abbe7b71","work_id":"wrk_e5060e37-5e03-4669-88d5-6a5989ca3dcb","block_id":"blk_e38bb1b0-7c14-46b8-8843-ffbe1ab1f0ae","section_id":"sec_21e4827b-0f3d-4dee-81cc-f5323b0b4a44","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":1384,"end":1390,"exact":"エージェント","quote":"t=Key%20advantages%20of%20Acorn%202,include)。Acorn 2.0はエージェントベースのモデリング手法とAIを組み合わせ、衛星コンステレーションや多領域ミッションの動的シミュレーションを可能にしています[ir.redwirespace.com](https://ir.redwir","quote_start":1329,"quote_end":1490,"text_sha256":"3c87cc7edec0346ef274ceea57b00bb93e5281d85da3b8834421ac41f1330866","block_sha256":"3c87cc7edec0346ef274ceea57b00bb93e5281d85da3b8834421ac41f1330866","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_0ee327e2-ecfe-42e6-b337-fce8abbe7b71/#blk_e38bb1b0-7c14-46b8-8843-ffbe1ab1f0ae"},{"id":"occ_7cc54ff4370ca70eb968b584","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_0f7c606d-558f-4edb-99eb-4609a9fb0a95","work_id":"wrk_958dfab7-4e11-4e38-9d35-db36e1f1b2f7","block_id":"blk_4233d606-d06c-4d02-ba4e-a4221b2f287f","section_id":"sec_4e5d0f98-c313-4894-944f-1ad04e70cdfd","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":0,"end":7,"exact":"Agentic","quote":"Agentic AI、KV Cache、Vector Searchではデータを毎回XPUへ戻すより、Memory近傍でfilter、reduction、lookup等を処理した方が効率が良い。","quote_start":0,"quote_end":96,"text_sha256":"cc6081febaf608dda916151eac526f140eedc0e679e7e87eb5ede773a801a172","block_sha256":"cc6081febaf608dda916151eac526f140eedc0e679e7e87eb5ede773a801a172","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_0f7c606d-558f-4edb-99eb-4609a9fb0a95/#blk_4233d606-d06c-4d02-ba4e-a4221b2f287f"},{"id":"occ_6b1dae22f3807c4af08bc1d6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_0f7c606d-558f-4edb-99eb-4609a9fb0a95","work_id":"wrk_958dfab7-4e11-4e38-9d35-db36e1f1b2f7","block_id":"blk_ad1ce983-b6e4-4ca0-beb1-96556b29f468","section_id":"sec_5725e7ec-e838-48dd-80f0-cb0eaaa9f033","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":146,"end":154,"exact":"AI Agent","quote":"な確定とは別であり、研究の独立性を巡る論争もあるため、「解決済み」と扱うべき段階ではない。しかし、数千～数万のAI Agentを並列化して、従来人間が長期間取り組んできた高度な数学研究そのものへAIを投入する段階まで来た、という意味は大きい。(Axios)","quote_start":91,"quote_end":219,"text_sha256":"03c33568ac56ec7c4f2d83c858b58edc38ba4a7fa790f388e969032d0adf2062","block_sha256":"03c33568ac56ec7c4f2d83c858b58edc38ba4a7fa790f388e969032d0adf2062","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_0f7c606d-558f-4edb-99eb-4609a9fb0a95/#blk_ad1ce983-b6e4-4ca0-beb1-96556b29f468"},{"id":"occ_a65fe2867b60142509506dc8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_0f7c606d-558f-4edb-99eb-4609a9fb0a95","work_id":"wrk_958dfab7-4e11-4e38-9d35-db36e1f1b2f7","block_id":"blk_e1211215-4200-472a-bc46-3d20ec1c9327","section_id":"sec_f0f2e37d-965a-413d-9ad1-e9d847de29ad","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":79,"end":86,"exact":"agentic","quote":"con design能力そのものを強化するため、「Chip Design RL」の研究者を募集している。対象はagentic RTL generation、formal verification、physical-design optimization、PPA、DFT、ECO、EDA automationまで広い。さらにA","quote_start":24,"quote_end":186,"text_sha256":"71a38f762e632d4073c37287f20a40200b121a12ecf57f5492bbe5f6141f63e3","block_sha256":"71a38f762e632d4073c37287f20a40200b121a12ecf57f5492bbe5f6141f63e3","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_0f7c606d-558f-4edb-99eb-4609a9fb0a95/#blk_e1211215-4200-472a-bc46-3d20ec1c9327"},{"id":"occ_67b86dda66bb55dfa7f35d35","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_0f7c606d-558f-4edb-99eb-4609a9fb0a95","work_id":"wrk_958dfab7-4e11-4e38-9d35-db36e1f1b2f7","block_id":"blk_fd861c49-8be9-4a0e-a729-d40ff6f56432","section_id":"sec_cac101fb-2a1e-4da7-b4fd-86f82ca10dc9","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":27,"end":34,"exact":"agentic","quote":"M890はtraining/inference両用で、agentic workloadを強く意識している。","quote_start":0,"quote_end":53,"text_sha256":"739fd037337f45207bb5782539379715cf510ed2d8550287a22ddea6583b3ca6","block_sha256":"739fd037337f45207bb5782539379715cf510ed2d8550287a22ddea6583b3ca6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_0f7c606d-558f-4edb-99eb-4609a9fb0a95/#blk_fd861c49-8be9-4a0e-a729-d40ff6f56432"},{"id":"occ_e490c9326d2d728ebd1973e4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_11f7264b-855b-41d8-b760-87dbfa630e9f","work_id":"wrk_739d4c20-62f1-49ee-9121-4689ad38c0d3","block_id":"blk_439663d2-53f5-49eb-8080-435f2388f0a0","section_id":"sec_a32f547b-d601-4635-8d51-47f21b292450","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"> **AIエージェントのロボット版**","quote_start":0,"quote_end":20,"text_sha256":"3bb3e0d015dc2ebcd8d1bebbce7b9a43d33e3e494ae2d63c7a5b36a875406f40","block_sha256":"3bb3e0d015dc2ebcd8d1bebbce7b9a43d33e3e494ae2d63c7a5b36a875406f40","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_11f7264b-855b-41d8-b760-87dbfa630e9f/#blk_439663d2-53f5-49eb-8080-435f2388f0a0"},{"id":"occ_1d24d29db694901110baa9a3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_11f7264b-855b-41d8-b760-87dbfa630e9f","work_id":"wrk_739d4c20-62f1-49ee-9121-4689ad38c0d3","block_id":"blk_7b7925fb-875d-4add-ae91-59b3fbb1d27d","section_id":"sec_acdbe3a4-5cb9-40c9-b849-90089815e874","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":13,"end":19,"exact":"エージェント","quote":"> **WLA的なロボットエージェントを構築できる部品を、ほぼ一通り揃えつつある**","quote_start":0,"quote_end":42,"text_sha256":"6ce1932c0aca1ec0e9a956d3d014f7c03204aac05ffc722e5399a1bed6e5f8ed","block_sha256":"6ce1932c0aca1ec0e9a956d3d014f7c03204aac05ffc722e5399a1bed6e5f8ed","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_11f7264b-855b-41d8-b760-87dbfa630e9f/#blk_7b7925fb-875d-4add-ae91-59b3fbb1d27d"},{"id":"occ_ee2fbb2b58433c756cdd7b2d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_11f7264b-855b-41d8-b760-87dbfa630e9f","work_id":"wrk_739d4c20-62f1-49ee-9121-4689ad38c0d3","block_id":"blk_7de2d055-db78-4abf-94cc-ee86cafd0004","section_id":"sec_23eb19d3-e248-4c50-bc1a-6954c850ed7a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":26,"end":32,"exact":"エージェント","quote":"これはまさに、現在デジタル空間で発展している**AIエージェントに、物理的な身体と世界モデルを与えたもの**だといえるでしょう。","quote_start":0,"quote_end":64,"text_sha256":"041825574866afef8bd3ce32143c3464302d948818f2aba8d0075003e5162289","block_sha256":"041825574866afef8bd3ce32143c3464302d948818f2aba8d0075003e5162289","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_11f7264b-855b-41d8-b760-87dbfa630e9f/#blk_7de2d055-db78-4abf-94cc-ee86cafd0004"},{"id":"occ_7e3e040d5c3b56574eeb0941","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_11f7264b-855b-41d8-b760-87dbfa630e9f","work_id":"wrk_739d4c20-62f1-49ee-9121-4689ad38c0d3","block_id":"blk_e291a73e-666e-45e7-8a89-e58bfa6e3582","section_id":"sec_a32f547b-d601-4635-8d51-47f21b292450","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"一般的なAIエージェントがブラウザ、ファイル、API、コード実行環境などのデジタルツールを使うのに対して、ロボットエージェントは腕、手、脚、車輪、カメラ、触覚センサーなどの物理的なツールを使います。","quote_start":0,"quote_end":99,"text_sha256":"f85e4dcd868b7af32f9ffc3c628f9834a5e00284d3d1c3f19312f17a3d9e7b50","block_sha256":"f85e4dcd868b7af32f9ffc3c628f9834a5e00284d3d1c3f19312f17a3d9e7b50","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_11f7264b-855b-41d8-b760-87dbfa630e9f/#blk_e291a73e-666e-45e7-8a89-e58bfa6e3582"},{"id":"occ_a2b3081f83396b62eabe12b9","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_120b7405-aefc-4d5d-a68b-9713613e3e1f","work_id":"wrk_91a0ac80-8870-4ec9-b8a6-4081acad5334","block_id":"blk_0fbdd02e-238a-40f7-a66b-e4966beeaf20","section_id":"sec_3886953f-b1b5-42de-8538-f26b912220ac","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":275,"end":281,"exact":"エージェント","quote":"\n| Amazon・eBay | AIを実業へ統合する企業 |\n| ネット利用者数 | AI利用者・トークン・エージェント |\n| ページビュー | 推論量・トークン量 |\n| 光回線スワップ | 循環出資・クラウド購入契約 |\n```","quote_start":220,"quote_end":338,"text_sha256":"a560202eabd14b445a971c345ad1cae80eb096a5054fa45474a4b4c541a64253","block_sha256":"a560202eabd14b445a971c345ad1cae80eb096a5054fa45474a4b4c541a64253","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_120b7405-aefc-4d5d-a68b-9713613e3e1f/#blk_0fbdd02e-238a-40f7-a66b-e4966beeaf20"},{"id":"occ_edeb5ed6c75582cca7bde973","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_120b7405-aefc-4d5d-a68b-9713613e3e1f","work_id":"wrk_91a0ac80-8870-4ec9-b8a6-4081acad5334","block_id":"blk_8b7d50f8-7ae1-4118-bc1b-4fdaa10c4e7e","section_id":"sec_04248d11-94c6-4bca-9f80-84d14f980dc4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":81,"end":87,"exact":"エージェント","quote":" 楽天：オンライン商店街\n- GMO：インターネット接続・ドメイン\n- インターネット総合研究所\n- サイバーエージェント：ネット広告\n- オンライン証券各社\n- 東京エレクトロン、アドバンテスト：半導体製造装置・試験装置","quote_start":26,"quote_end":137,"text_sha256":"795c3ddbd076dc1b0da9d8b36122e6e85c8d6f489d85efbe9f3ad0b3b9573770","block_sha256":"795c3ddbd076dc1b0da9d8b36122e6e85c8d6f489d85efbe9f3ad0b3b9573770","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_120b7405-aefc-4d5d-a68b-9713613e3e1f/#blk_8b7d50f8-7ae1-4118-bc1b-4fdaa10c4e7e"},{"id":"occ_36eb816a01b278fd55481854","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb","block_id":"blk_1d96cc25-736a-4826-bb76-c3fde2a0421f","section_id":"sec_00c349e9-1fba-46b6-94cb-72d34486367e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":53,"end":59,"exact":"エージェント","quote":"Muse Sparkはまだ、ChatGPTやClaudeの最上位モデルを完全に超えたわけではない。特に長期エージェントやコーディングワークフローでは、改善余地があると見るべきである。","quote_start":0,"quote_end":91,"text_sha256":"10857819a9c540f37cb0678e0b35c85c5668f038478fc574ff4b2fec3ee99a0f","block_sha256":"10857819a9c540f37cb0678e0b35c85c5668f038478fc574ff4b2fec3ee99a0f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/#blk_1d96cc25-736a-4826-bb76-c3fde2a0421f"},{"id":"occ_1cd04e07424490905ad795cf","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb","block_id":"blk_236e33c1-df7e-43df-94c5-25a9f2feaca9","section_id":"sec_cf33b9af-15bd-401b-a433-6cc1ade24dca","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":128,"end":134,"exact":"エージェント","quote":"l chain of thought、multi-agent orchestration、マルチモーダル、個人エージェント用途を最初から意識している。","quote_start":73,"quote_end":148,"text_sha256":"ff443420072e409761ea42b58f5ddc212097080b430d0aa55f40dcdf941e278c","block_sha256":"ff443420072e409761ea42b58f5ddc212097080b430d0aa55f40dcdf941e278c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/#blk_236e33c1-df7e-43df-94c5-25a9f2feaca9"},{"id":"occ_350bb7aadbeeb01e8784c0c2","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb","block_id":"blk_4e5aec8b-8e94-40a7-af1a-ca985f929fd2","section_id":"sec_deeff35c-799d-4e1b-90a7-96628c46021c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":43,"end":49,"exact":"エージェント","quote":"MetaのAIで最も早く利益に直結しそうなのは、個人向けチャットではなく、ビジネス向けエージェントである。","quote_start":0,"quote_end":53,"text_sha256":"c9ccc3a596b2ae058d2a98c0fef8355372b5cddc701ce0ae707af6244ab29eb6","block_sha256":"c9ccc3a596b2ae058d2a98c0fef8355372b5cddc701ce0ae707af6244ab29eb6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/#blk_4e5aec8b-8e94-40a7-af1a-ca985f929fd2"},{"id":"occ_924219fc9d5c36f939650f58","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb","block_id":"blk_509612ca-4e9d-42b8-a6b9-fd48b2bad3c8","section_id":"sec_00c349e9-1fba-46b6-94cb-72d34486367e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":109,"end":115,"exact":"エージェント","quote":"t、multi-agent orchestrationが強調されている。これは、単なるチャットモデルではなく、エージェントAIを前提とした設計である。","quote_start":54,"quote_end":129,"text_sha256":"f3cd75fddc5e47a622bfd8f197358cfc6220e7282293737eb2cb7004ce522e2b","block_sha256":"f3cd75fddc5e47a622bfd8f197358cfc6220e7282293737eb2cb7004ce522e2b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/#blk_509612ca-4e9d-42b8-a6b9-fd48b2bad3c8"},{"id":"occ_92a11f7079061e25048aa42a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb","block_id":"blk_519a2012-514d-4b5c-8352-fa312540558b","section_id":"sec_1d1ec9f4-f124-4292-aaf6-5e63e88ab0bc","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":76,"end":82,"exact":"エージェント","quote":"重い。数十万GPU規模で本命モデルを学習した後に、アーキテクチャ、データ、RL、推論モード、tool use、エージェント設計に欠陥が見つかれば、損失は莫大になる。そのため、Muse Sparkのような前段階モデルで、スケーリング則、安全性、推論時計算、製品接続、ユーザー反応を確認することには大きな意味がある。","quote_start":21,"quote_end":177,"text_sha256":"1510796e0e8278b629a1f9909c29504b1c6f227aa360b7a24571841909955516","block_sha256":"1510796e0e8278b629a1f9909c29504b1c6f227aa360b7a24571841909955516","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/#blk_519a2012-514d-4b5c-8352-fa312540558b"},{"id":"occ_55eccd8cbf93e8e41f56df29","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb","block_id":"blk_533436a6-e72b-4dc3-be34-cc5b61133203","section_id":"sec_4b684079-610f-422f-8fa6-e9bf17701239","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":73,"end":79,"exact":"エージェント","quote":"ーを使って高性能モデルを作るだけでなく、そのモデルをMetaアプリ群、AIグラス、広告、ビジネスチャット、個人エージェントへ自然に接続する必要がある。","quote_start":18,"quote_end":93,"text_sha256":"548c48595017f6edb8cc8a81a498958cab0b0b95ea715117e81e708be34b3fc8","block_sha256":"548c48595017f6edb8cc8a81a498958cab0b0b95ea715117e81e708be34b3fc8","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/#blk_533436a6-e72b-4dc3-be34-cc5b61133203"},{"id":"occ_eaa6caa89441d9fe0ba37ef3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb","block_id":"blk_624d87b0-a03b-4c8b-99a7-4dc9f822727f","section_id":"sec_d25c9390-cd00-49fb-9227-81cdb9c8124d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":28,"end":34,"exact":"エージェント","quote":"第一に、Muse Spark時点では、Meta自身が長期エージェントやコーディングワークフローに課題があると認めている。つまり、Claude CodeやChatGPT Thinkingのように、複雑な開発作業や詳細調査を長時間安定してこなす能力では、まだ不確実性が残る","quote_start":0,"quote_end":134,"text_sha256":"4ef8a8dfd8a39060bde0806767091bc9b0dccca41adfb0404b7aeebfaf0905a7","block_sha256":"4ef8a8dfd8a39060bde0806767091bc9b0dccca41adfb0404b7aeebfaf0905a7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/#blk_624d87b0-a03b-4c8b-99a7-4dc9f822727f"},{"id":"occ_673e5f0d2ed3258a45304109","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb","block_id":"blk_6d11efea-2bf2-4b53-ad02-384f624f5457","section_id":"sec_bbe7c4e9-fbcc-4545-aa71-0df4086c9765","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":108,"end":114,"exact":"エージェント","quote":"  \nどの回答を良い回答とみなすか。  \nどの失敗例を集めるか。  \nどのタスクでモデルを鍛えるか。  \nどのエージェント行動を安全とみなすか。  \nどの出力を製品に載せてよいか。  \nどの能力が収益に結びつくか。","quote_start":53,"quote_end":160,"text_sha256":"7426f5eee5334f0e3efd26a8adca52d44a3352643ecc131e7e4c4c95285dd2d6","block_sha256":"7426f5eee5334f0e3efd26a8adca52d44a3352643ecc131e7e4c4c95285dd2d6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/#blk_6d11efea-2bf2-4b53-ad02-384f624f5457"},{"id":"occ_e0f1c4b998ec9215b10ad3fa","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb","block_id":"blk_78954548-5496-4cda-a72a-02a634a32c8d","section_id":"sec_0eae083a-7c80-41de-8169-72c1792096de","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":94,"end":100,"exact":"エージェント","quote":"  \nただし、その最前線は、ベンチマーク表の1位争いだけではない。  \nAIグラス、SNS、広告、商取引、個人エージェントを含む、より広いAIプラットフォーム競争の最前線である。","quote_start":39,"quote_end":128,"text_sha256":"e7de4e36a5bda4c938a73209935a4400bebaf4358cf648f0e9d3217c3bc027ad","block_sha256":"e7de4e36a5bda4c938a73209935a4400bebaf4358cf648f0e9d3217c3bc027ad","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/#blk_78954548-5496-4cda-a72a-02a634a32c8d"},{"id":"occ_2d05e1b8f7fbb6a60e1483fc","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb","block_id":"blk_8a457432-a364-4e7d-b66d-764da0ac3a87","section_id":"sec_457c7dc9-dbc6-439b-aac9-eaa00cc9dc1a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":52,"end":58,"exact":"エージェント","quote":"たとえば、ある企業について調査する場合、一つのAIが検索して要約するだけでは浅くなりやすい。しかし、複数エージェントを使えば、公式資料を読む担当、ニュースを読む担当、論文や技術資料を読む担当、数値を検算する担当、矛盾を探す担当、最終的に統合する担当に分けられる。","quote_start":0,"quote_end":131,"text_sha256":"e5e22854c881494482ba4abf46e715e8288162326572b6f40f8b64cb309459b5","block_sha256":"e5e22854c881494482ba4abf46e715e8288162326572b6f40f8b64cb309459b5","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/#blk_8a457432-a364-4e7d-b66d-764da0ac3a87"},{"id":"occ_6cf5729e4f5461eef4b243f4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb","block_id":"blk_8faaddc2-7308-4eb5-97af-04c81aeff05a","section_id":"sec_deeff35c-799d-4e1b-90a7-96628c46021c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":72,"end":78,"exact":"エージェント","quote":"に相性が良い。  \n広告で顧客を集め、そのままMessenger、Instagram、WhatsApp上でAIエージェントが接客し、購入や予約まで進める。そうなればMetaは、広告枠を売る会社から、企業の営業・接客・マーケティングをAIで代替する会社へ進化する。","quote_start":17,"quote_end":148,"text_sha256":"a6b1e81977bfbcc2d9b3eb638b4510ff0c86433256b9ed11b4398ee49f6d5ab9","block_sha256":"a6b1e81977bfbcc2d9b3eb638b4510ff0c86433256b9ed11b4398ee49f6d5ab9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/#blk_8faaddc2-7308-4eb5-97af-04c81aeff05a"},{"id":"occ_2bea30037b924c2a4f4b5e69","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb","block_id":"blk_9ff3fbac-f962-4d16-880f-0105c83fe5f1","section_id":"sec_7bfd25f3-f506-4c73-a6e9-3d0217668789","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":117,"end":123,"exact":"エージェント","quote":"持っている。つまり、AIを配る場所はすでにある。問題は、その場所に載せるだけのフロンティア級モデルと、金になるエージェントシステムを作れるかだった。","quote_start":62,"quote_end":136,"text_sha256":"001c80e9735302c6fc04302a283e34b323e740afc52b61639e3fdd9baf0688f4","block_sha256":"001c80e9735302c6fc04302a283e34b323e740afc52b61639e3fdd9baf0688f4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/#blk_9ff3fbac-f962-4d16-880f-0105c83fe5f1"},{"id":"occ_d1c5f0d8cbf6c206d13e0ec0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb","block_id":"blk_a29ea2a5-ad58-4302-a242-7ef8a1ca2aa4","section_id":"sec_fc4dcf33-94b2-4699-b926-7d74d77d3d00","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":175,"end":181,"exact":"エージェント","quote":"WhatsApp、Messenger、Facebook、Threads、AIグラス、広告、商取引をまたぐ「個人エージェント基盤」である。","quote_start":120,"quote_end":188,"text_sha256":"8e67f86b7d0cd33d9afcc85e0d08d044160ca2d2e57071216a502cbb8d31abf4","block_sha256":"8e67f86b7d0cd33d9afcc85e0d08d044160ca2d2e57071216a502cbb8d31abf4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/#blk_a29ea2a5-ad58-4302-a242-7ef8a1ca2aa4"},{"id":"occ_81d4e8cb66d54abd51e2fc49","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb","block_id":"blk_b02ab130-6e53-42e0-b0fb-33fe8d78d4c2","section_id":"sec_ad853364-0c22-4646-8055-aecfa697a614","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":50,"end":56,"exact":"エージェント","quote":"最先端モデルを作るには、膨大な計算資源だけでなく、巨大なデータ工程、評価工程、RL工程、安全性工程、エージェント検証工程、製品導入工程が必要になる。さらにMetaの場合は、それを広告、商取引、Business Agent、AIグラス、Meta AIへ接続しなければならない。","quote_start":0,"quote_end":137,"text_sha256":"c1a54f8fd6964a4374ee1fac2df17163eacb2b3009808f9d3ad2dfc75581d5bc","block_sha256":"c1a54f8fd6964a4374ee1fac2df17163eacb2b3009808f9d3ad2dfc75581d5bc","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/#blk_b02ab130-6e53-42e0-b0fb-33fe8d78d4c2"},{"id":"occ_ba81cfc9d14e0f5923eefd23","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb","block_id":"blk_bef3e54a-072d-4757-8acb-fbb8e851f8d7","section_id":"sec_00c349e9-1fba-46b6-94cb-72d34486367e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":89,"end":95,"exact":"エージェント","quote":"「AIをどう使わせるか」まで考えている点だ。  \n画像、動画、音声、視覚、グラス、広告、ビジネスチャット、個人エージェント。MetaのAIは、これらの導線と最初から接続されることを前提にしている。","quote_start":34,"quote_end":132,"text_sha256":"e2af3f9e1712c8918d224020b5de7b2f5b65ce85a1d3237336b60b82f226a2e2","block_sha256":"e2af3f9e1712c8918d224020b5de7b2f5b65ce85a1d3237336b60b82f226a2e2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/#blk_bef3e54a-072d-4757-8acb-fbb8e851f8d7"},{"id":"occ_e3f3318fa63d5a0ea2600591","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb","block_id":"blk_c4d252fd-6928-41c1-bd26-2f6f8f67c12f","section_id":"sec_eae60cb9-e307-468a-889d-d43d00f60abe","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":55,"end":61,"exact":"エージェント","quote":"モデルを作る。  \n評価する。  \n弱点を見つける。  \n専門家データで鍛える。  \n安全性を検証する。  \nエージェントとして動かす。  \n広告と商取引に接続する。  \n世界中のユーザーに配る。  \n利益に変える。","quote_start":0,"quote_end":108,"text_sha256":"7dc4c797c6a4e49c1325ff960f09579794b5f25a092d3d5757deb8eb2ed8a8f9","block_sha256":"7dc4c797c6a4e49c1325ff960f09579794b5f25a092d3d5757deb8eb2ed8a8f9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/#blk_c4d252fd-6928-41c1-bd26-2f6f8f67c12f"},{"id":"occ_fa60b3dfeee84d4f7458edf4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb","block_id":"blk_ca9e0a0b-5a11-4f37-81dd-9305c14cbe4d","section_id":"sec_d25c9390-cd00-49fb-9227-81cdb9c8124d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":86,"end":92,"exact":"エージェント","quote":"健康、人間関係、予定、買い物、仕事に深く入り込むほど、利用者は「本当にMetaに任せてよいのか」と考える。個人エージェントでは、モデル性能以上に信頼が重要になる。","quote_start":31,"quote_end":112,"text_sha256":"330271257d5d2c0a3dc569fdc68fcc608dc3f2a4b61486f312aabb2abb87559e","block_sha256":"330271257d5d2c0a3dc569fdc68fcc608dc3f2a4b61486f312aabb2abb87559e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/#blk_ca9e0a0b-5a11-4f37-81dd-9305c14cbe4d"},{"id":"occ_a12c6fc2de90d02d6c939053","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb","block_id":"blk_db64c9c8-2410-46c0-9416-522d6fa67e73","section_id":"sec_0eae083a-7c80-41de-8169-72c1792096de","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":54,"end":60,"exact":"エージェント","quote":"Muse Sparkが前菜であるなら、本命モデルはその検証結果を踏まえて作られる、より大きなマルチモーダル・エージェントモデルになるだろう。事前学習、強化学習、推論時スケーリング、multi-agent orchestration、tool use、AIグラス、Business Agent、個人エージェント。これらがう","quote_start":0,"quote_end":160,"text_sha256":"71bf01cfa56a6b4148d1949f89270ae45d6a54b06b7eb31a09de47284390f157","block_sha256":"71bf01cfa56a6b4148d1949f89270ae45d6a54b06b7eb31a09de47284390f157","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/#blk_db64c9c8-2410-46c0-9416-522d6fa67e73"},{"id":"occ_a1dfb35658f13240384e2871","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb","block_id":"blk_f1adb345-4425-489a-8daa-3b4c263d2978","section_id":"sec_457c7dc9-dbc6-439b-aac9-eaa00cc9dc1a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":84,"end":90,"exact":"エージェント","quote":"ti-agent orchestrationである。これは、単一のモデルが一人で長く考えるだけではなく、複数のエージェントを並列に動かし、協調させることで、難しい問題に対する性能を上げる考え方だ。","quote_start":29,"quote_end":127,"text_sha256":"05c94cd8e79d58e3fc5847e95510a205da18146c54177605b49cf8bcc2489008","block_sha256":"05c94cd8e79d58e3fc5847e95510a205da18146c54177605b49cf8bcc2489008","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/#blk_f1adb345-4425-489a-8daa-3b4c263d2978"},{"id":"occ_093a79970b193b9d8ac783f0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb","block_id":"blk_f3677dce-c143-4fe7-beda-d31d1fa1bc60","section_id":"sec_1d1ec9f4-f124-4292-aaf6-5e63e88ab0bc","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":142,"end":148,"exact":"エージェント","quote":"競争でやや出遅れた印象があった。その反省から、モデルの事前学習、強化学習、推論時スケーリング、マルチモーダル、エージェント設計、インフラをまとめて作り直した。その最初の実戦投入がMuse Sparkだった。","quote_start":87,"quote_end":190,"text_sha256":"1be1c7a405fcfd256e95185a5fd72ffc3320c0bc3871e1521181890d0d8131b3","block_sha256":"1be1c7a405fcfd256e95185a5fd72ffc3320c0bc3871e1521181890d0d8131b3","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/#blk_f3677dce-c143-4fe7-beda-d31d1fa1bc60"},{"id":"occ_8c1f6edc2b191544e9f1a7c4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb","block_id":"blk_f951cf95-077e-40aa-87b4-50af238a605b","section_id":"sec_ad853364-0c22-4646-8055-aecfa697a614","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":61,"end":67,"exact":"エージェント","quote":"ィアモデルは、事前学習だけで差がつくわけではない。  \nポストトレーニング、RLHF、専門家データ、推論評価、エージェント行動評価、安全性評価、レッドチーミングで差がつく。","quote_start":6,"quote_end":92,"text_sha256":"f1b5d9a71214c5718fb1454b64da5be67c4d54ddf8bf9d0d41814e71c68df0d9","block_sha256":"f1b5d9a71214c5718fb1454b64da5be67c4d54ddf8bf9d0d41814e71c68df0d9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/#blk_f951cf95-077e-40aa-87b4-50af238a605b"},{"id":"occ_a33384e6aaee680640c75ab9","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb","block_id":"blk_fb5518fe-dbec-4b73-9752-4980c05704e5","section_id":"sec_cf33b9af-15bd-401b-a433-6cc1ade24dca","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":14,"end":20,"exact":"エージェント","quote":"検索能力、コーディング能力、エージェント能力、画像生成、長文作業、調査精度、引用の扱い。多くの面で、Googleが持っている素材の強さがGeminiという体験に十分乗っていない。","quote_start":0,"quote_end":89,"text_sha256":"26c736d9cbdddd76969d496ddd56114ea22576a332c483e9e1d66e98e8e4e552","block_sha256":"26c736d9cbdddd76969d496ddd56114ea22576a332c483e9e1d66e98e8e4e552","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_12eaaf24-14fe-4445-936b-b81ef492d19d/#blk_fb5518fe-dbec-4b73-9752-4980c05704e5"},{"id":"occ_260347df1679f90ccd197e5c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_15381314-c653-4ce9-9b25-d8495cdbf4d1","work_id":"wrk_220c8e7e-eccc-4357-89f5-9a6b78a08a7c","block_id":"blk_a7b90873-903d-465a-bcbf-24c42966d82f","section_id":"sec_ea941794-53e1-4129-a450-a0a8f4e8b137","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":32,"end":38,"exact":"エージェント","quote":"## ロボットAIが「反応する機械」から「未来を想像し、計画するエージェント」へ進化すると、半導体・メモリー・NAND・産業構造はどう変わるのか","quote_start":0,"quote_end":72,"text_sha256":"19f23ee37d0c7cd00db01d308de23eaf3926874d9406581a24cc68db7718b694","block_sha256":"19f23ee37d0c7cd00db01d308de23eaf3926874d9406581a24cc68db7718b694","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_15381314-c653-4ce9-9b25-d8495cdbf4d1/#blk_a7b90873-903d-465a-bcbf-24c42966d82f"},{"id":"occ_284ebcf9827cfc942a531a2c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_17972011-3c15-4377-a2ea-f36af8fa6c2f","work_id":"wrk_fbc5cb25-a3fe-4368-a659-4a3d7a753309","block_id":"blk_76dff0c2-d3ca-4d78-b154-ccd3744c10e4","section_id":"sec_41bae2ab-2bf6-443b-9962-4d1083fe8935","layer":"character","character_id":"zetu_noia","count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":102,"end":108,"exact":"エージェント","quote":"にサーバーDRAMがあり、さらにeSSD、QLC NAND、HBF、XL-FLASHが広がる。推論が長文化し、エージェントが履歴を持ち、RAGが企業データを抱えるほど、記憶の階層は深くなります。","quote_start":47,"quote_end":144,"text_sha256":"ab1c3953d45ee4e1f4155068f87bd6060d2d161adc2b56212352c62b4403ecf6","block_sha256":"ab1c3953d45ee4e1f4155068f87bd6060d2d161adc2b56212352c62b4403ecf6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_17972011-3c15-4377-a2ea-f36af8fa6c2f/#blk_76dff0c2-d3ca-4d78-b154-ccd3744c10e4"},{"id":"occ_446e2fed8fbf7d01a0b269f5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f","work_id":"wrk_f7d6728c-f58d-4c77-958d-67d9910c0d5d","block_id":"blk_18b09326-e855-42a2-a4c6-c486b0a35d29","section_id":"sec_dd053398-398a-45e2-b7f2-7078f80da5d4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":120,"end":126,"exact":"エージェント","quote":"\n- 衛星通信・災害通信の認証\n- 長期保存されるセンサーデータ、設計データ、医療・電力・防衛データ\n- AIエージェントの権限委任と署名","quote_start":65,"quote_end":134,"text_sha256":"4f4af09678daf2c41ad17f4b53580fe09e5222dce0c58216b104916cbbbd4ba3","block_sha256":"4f4af09678daf2c41ad17f4b53580fe09e5222dce0c58216b104916cbbbd4ba3","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f/#blk_18b09326-e855-42a2-a4c6-c486b0a35d29"},{"id":"occ_17ec2e93277aaa31591d0e25","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f","work_id":"wrk_f7d6728c-f58d-4c77-958d-67d9910c0d5d","block_id":"blk_502837fb-bf8e-480b-b766-cc9f67fff619","section_id":"sec_a123a974-b039-4054-b4b1-42ad32ce27d0","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":190,"end":196,"exact":"エージェント","quote":"N時代の基地局・MEC・キャリアエッジDCは、そこに加えて、**ロボット、車、工場、ドローン、監視カメラ、AIエージェントの近くで推論を行う小型AIデータセンター**に近づいていきます。","quote_start":135,"quote_end":228,"text_sha256":"11d0b52c97acfc5d4f66cd0e4fd344bc6be513ea12a36d62024d5a6a51a94ddf","block_sha256":"11d0b52c97acfc5d4f66cd0e4fd344bc6be513ea12a36d62024d5a6a51a94ddf","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f/#blk_502837fb-bf8e-480b-b766-cc9f67fff619"},{"id":"occ_1fb0a52fb26ea143b631b17b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f","work_id":"wrk_f7d6728c-f58d-4c77-958d-67d9910c0d5d","block_id":"blk_57e81b16-7ab8-4adc-a402-6905306ba95e","section_id":"sec_373e6c5c-b59f-4362-ae6e-8f13a3253de6","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"## AIエージェント時代の階層構造","quote_start":0,"quote_end":18,"text_sha256":"e66805095c3e5f5de1020ee63eea3aad5d0c22224451155e9897b27799cf6c14","block_sha256":"e66805095c3e5f5de1020ee63eea3aad5d0c22224451155e9897b27799cf6c14","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f/#blk_57e81b16-7ab8-4adc-a402-6905306ba95e"},{"id":"occ_e809e751acd93fad941b7341","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f","work_id":"wrk_f7d6728c-f58d-4c77-958d-67d9910c0d5d","block_id":"blk_75684ee6-fed5-4b60-83ff-e5a09dd46681","section_id":"sec_c5a92e05-5531-4ce6-ba20-7ee1f4e3053b","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":54,"end":60,"exact":"エージェント","quote":"つまり、CDN・セキュリティ企業は、AI-RANやMECの上で動くアプリを守る **境界認証・暗号移行・AIエージェント制御の関所** になり得ます。","quote_start":0,"quote_end":75,"text_sha256":"60cd9d2510caa6ed74e4f07f1b7f385c976cbaef7485d403239ebcf3e155d52f","block_sha256":"60cd9d2510caa6ed74e4f07f1b7f385c976cbaef7485d403239ebcf3e155d52f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f/#blk_75684ee6-fed5-4b60-83ff-e5a09dd46681"},{"id":"occ_406815a8ca1c551cc9248ff0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f","work_id":"wrk_f7d6728c-f58d-4c77-958d-67d9910c0d5d","block_id":"blk_78311c02-0fdd-4ff5-903e-e59a04719903","section_id":"sec_db32af1c-504e-494a-84fe-65be72454e1d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":102,"end":108,"exact":"エージェント","quote":"、SiemensのIndustrial 5GとQualcommのエッジAIを使い、AGV、ロボットアーム、AIエージェントを連携させる自律工場モデルを示しました。発表では、オンプレミスAIとプライベート5Gにより、中央MESに依存しすぎず、リアルタイムな現場判断を行う方向性が示されています。([Qualcomm](htt","quote_start":47,"quote_end":208,"text_sha256":"dd3305cb9b7bd01a4a1e15b9ceb7571d54e52c5fc18e2fd8a89df862f870473d","block_sha256":"dd3305cb9b7bd01a4a1e15b9ceb7571d54e52c5fc18e2fd8a89df862f870473d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f/#blk_78311c02-0fdd-4ff5-903e-e59a04719903"},{"id":"occ_671ac9ed098fbbdbb9eccc14","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f","work_id":"wrk_f7d6728c-f58d-4c77-958d-67d9910c0d5d","block_id":"blk_928a39cf-38ee-4a60-b8d1-e7ba01b33f3f","section_id":"sec_9aece37f-b28e-433f-98b7-c72af5e5b186","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":39,"end":45,"exact":"エージェント","quote":"つまりCDN・セキュリティ企業は、AI-RANの「電波部分」ではなく、**AIエージェントと実機の間に立つ関所**になります。ここはこのプロジェクトで言ってきた「境界暗号・認証・実機管理」の中核です。","quote_start":0,"quote_end":100,"text_sha256":"0762a6f4454a60b312ce74b10035cd6c780333d66b20d0ef7b323d31cbff6805","block_sha256":"0762a6f4454a60b312ce74b10035cd6c780333d66b20d0ef7b323d31cbff6805","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f/#blk_928a39cf-38ee-4a60-b8d1-e7ba01b33f3f"},{"id":"occ_e8cc7d3cd29505d4595a8517","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f","work_id":"wrk_f7d6728c-f58d-4c77-958d-67d9910c0d5d","block_id":"blk_b7405c1c-09cf-43af-b366-bde39c51d3be","section_id":"sec_9aece37f-b28e-433f-98b7-c72af5e5b186","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":74,"end":80,"exact":"エージェント","quote":"astly、Zscaler、Palo Alto、Oktaのような企業は、無線そのものは持ちません。しかし、AIエージェント時代には、API、認証、Bot/Agent識別、WAF、Zero Trust、PQC、データ境界、SASEが重要になります。","quote_start":19,"quote_end":142,"text_sha256":"c0b729d85d256629e1cfde7a9cfb7e432ffcfa23344c1fd7c5ba1d5dc80d2710","block_sha256":"c0b729d85d256629e1cfde7a9cfb7e432ffcfa23344c1fd7c5ba1d5dc80d2710","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f/#blk_b7405c1c-09cf-43af-b366-bde39c51d3be"},{"id":"occ_3d63bc3218b4f00a019454c0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f","work_id":"wrk_f7d6728c-f58d-4c77-958d-67d9910c0d5d","block_id":"blk_bd9ddda7-08cc-4e48-b335-814939d84367","section_id":"sec_cb324aaa-3feb-49e7-8cd0-c23d89fabd14","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":49,"end":55,"exact":"エージェント","quote":"中央クラウドは、基盤モデルの学習、巨大シミュレーション、全社横断最適化、デジタルツインの統合、AIエージェント基盤を担います。","quote_start":0,"quote_end":63,"text_sha256":"6628064db64942cf4feafe21ddd8714a6487a31dd112809950ccad758034550d","block_sha256":"6628064db64942cf4feafe21ddd8714a6487a31dd112809950ccad758034550d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f/#blk_bd9ddda7-08cc-4e48-b335-814939d84367"},{"id":"occ_eadb075caef088c397ee30ea","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f","work_id":"wrk_f7d6728c-f58d-4c77-958d-67d9910c0d5d","block_id":"blk_cfbbb150-1adb-4b60-b313-6d97c62766ad","section_id":"sec_c5a92e05-5531-4ce6-ba20-7ee1f4e3053b","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":19,"end":25,"exact":"エージェント","quote":"### CDN・セキュリティ企業：AIエージェント時代の関所","quote_start":0,"quote_end":30,"text_sha256":"fb6eaa66e4b8176c37814bcd6618911d063ffd73cf6591f2f835c32fb224cc70","block_sha256":"fb6eaa66e4b8176c37814bcd6618911d063ffd73cf6591f2f835c32fb224cc70","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f/#blk_cfbbb150-1adb-4b60-b313-6d97c62766ad"},{"id":"occ_5bf5c9be3d8ff1b437a59dd1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f","work_id":"wrk_f7d6728c-f58d-4c77-958d-67d9910c0d5d","block_id":"blk_d55106a0-b334-42ed-a19f-2dcae2c85403","section_id":"sec_c5a92e05-5531-4ce6-ba20-7ee1f4e3053b","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"特にAIエージェント時代には、人間、AIエージェント、ロボット、IoTデバイス、業務SaaS、APIが大量につながります。そこで必要になるのは「誰が、何の権限で、どのデバイスから、どのAIに、どのデータを渡してよいか」を","quote_start":0,"quote_end":110,"text_sha256":"43dbec9f31ac51a7d357f194eaf9b8663f214fa93178cc14373aed769bc8885d","block_sha256":"43dbec9f31ac51a7d357f194eaf9b8663f214fa93178cc14373aed769bc8885d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f/#blk_d55106a0-b334-42ed-a19f-2dcae2c85403"},{"id":"occ_955cae9d940bbd9be4b92a9e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","work_id":"wrk_e8dd57e9-048c-4852-81d1-2a029a08ac21","block_id":"blk_0ded260b-2b7f-4fcf-adb0-6c3ba11a3776","section_id":"sec_b9f26322-c5dc-4428-acd4-999c089fdc47","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":100,"end":106,"exact":"エージェント","quote":" \nむしろ、ローカルAIは端末の標準機能として意識されずに広がり、クラウドAIは高度な作業、検索、研究、創作、エージェント、AIのためのAIの基盤として残り続けるという分岐である。","quote_start":45,"quote_end":135,"text_sha256":"f5fa07d9fbb724d748334d2264b7bc0ac78e0c53ca2a36f36c13d87044a1eb9a","block_sha256":"f5fa07d9fbb724d748334d2264b7bc0ac78e0c53ca2a36f36c13d87044a1eb9a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d/#blk_0ded260b-2b7f-4fcf-adb0-6c3ba11a3776"},{"id":"occ_e19c3413b636592d2f796d5c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","work_id":"wrk_e8dd57e9-048c-4852-81d1-2a029a08ac21","block_id":"blk_0f5ba6f0-67d6-4e97-830c-a31b7a16eb6c","section_id":"sec_0099cd34-26a9-4899-a392-46900c920264","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":67,"end":73,"exact":"エージェント","quote":"ルで行うか。  \n高性能推論をどこまでクラウドに残すか。  \n作業AIをどれだけ安全に普及させるか。  \nAIエージェントの計算需要をどう支えるか。  \nAIのためのAIが生む膨大な推論需要をどう処理するか。","quote_start":12,"quote_end":116,"text_sha256":"c9679da8a88b3d31f07978cc55cda6a52782de3ffae0be82bc239c8875ba0853","block_sha256":"c9679da8a88b3d31f07978cc55cda6a52782de3ffae0be82bc239c8875ba0853","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d/#blk_0f5ba6f0-67d6-4e97-830c-a31b7a16eb6c"},{"id":"occ_44c108e7189c13652260ec8d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","work_id":"wrk_e8dd57e9-048c-4852-81d1-2a029a08ac21","block_id":"blk_117f3f98-f4bf-4d26-9591-f85b40c31bc2","section_id":"sec_52370084-ef55-4d21-a29d-13121bcd1194","layer":"body","character_id":null,"count":3,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":64,"end":70,"exact":"エージェント","quote":"Iを使うヘビーユーザーと企業である。  \nCodex、Claude Code、Copilot agent、業務エージェント、Office内AI、ブラウザエージェント、社内SaaSエージェントを使い、実際の作業を任せる。  \nこの層は高性能モデルに課金する。","quote_start":9,"quote_end":137,"text_sha256":"f2734dfbe70d4dc0067adf6d45b4ddc4d72af603f8f2e9527f78092cd5bba082","block_sha256":"f2734dfbe70d4dc0067adf6d45b4ddc4d72af603f8f2e9527f78092cd5bba082","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d/#blk_117f3f98-f4bf-4d26-9591-f85b40c31bc2"},{"id":"occ_1f4ca6eb400d70264c5974a9","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","work_id":"wrk_e8dd57e9-048c-4852-81d1-2a029a08ac21","block_id":"blk_284c44ec-f42c-4a4c-a257-30b6f7caabe4","section_id":"sec_a5507526-35dd-442e-a7fa-1c22335ae0a0","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":23,"end":29,"exact":"エージェント","quote":"SNSは、人間の感情や意見を流す場所から、AIエージェントの活動を流す場所へ広がっていく。","quote_start":0,"quote_end":45,"text_sha256":"0d9c7d83913e1525dcc7629af8bc733319ca77072edfd1dd9035a23829a5daa8","block_sha256":"0d9c7d83913e1525dcc7629af8bc733319ca77072edfd1dd9035a23829a5daa8","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d/#blk_284c44ec-f42c-4a4c-a257-30b6f7caabe4"},{"id":"occ_44d52cf21b609497206fd812","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","work_id":"wrk_e8dd57e9-048c-4852-81d1-2a029a08ac21","block_id":"blk_42e9eca6-700d-4af3-8db3-28f92188281e","section_id":"sec_b9f26322-c5dc-4428-acd4-999c089fdc47","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":17,"end":23,"exact":"エージェント","quote":"## チャットAI、作業AI、AIエージェント時代の本当の分岐","quote_start":0,"quote_end":31,"text_sha256":"2507dad18eb742fb71c77b5c9ca54fb3a25d9c35ad5179acd60ab6e5df0ea360","block_sha256":"2507dad18eb742fb71c77b5c9ca54fb3a25d9c35ad5179acd60ab6e5df0ea360","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d/#blk_42e9eca6-700d-4af3-8db3-28f92188281e"},{"id":"occ_f2d35b889a03c4133c51ce32","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","work_id":"wrk_e8dd57e9-048c-4852-81d1-2a029a08ac21","block_id":"blk_496755b2-0d28-495f-bc6b-9968c94479c7","section_id":"sec_7c7a8944-3b23-42e3-8e52-c099c799ad43","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":56,"end":62,"exact":"エージェント","quote":"れまでのSNSは、人間が投稿し、人間が反応し、人間の関心をアルゴリズムが拡散する場所だった。  \nしかし、AIエージェントが自律的に動くようになると、SNSやWebの構造そのものが変わっていく可能性がある。","quote_start":1,"quote_end":104,"text_sha256":"f9bceab04fad6832a386fd65972bae27a17817885bfdd9ed328250df23cd7ebe","block_sha256":"f9bceab04fad6832a386fd65972bae27a17817885bfdd9ed328250df23cd7ebe","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d/#blk_496755b2-0d28-495f-bc6b-9968c94479c7"},{"id":"occ_593c78a718419be98be61d54","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","work_id":"wrk_e8dd57e9-048c-4852-81d1-2a029a08ac21","block_id":"blk_63f93bef-35e8-412c-acee-7400adac09a0","section_id":"sec_545b534c-6c95-493f-a5de-1e0507989118","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"ただし、AIエージェントの活動を評価するには、従来のいいねやリツイートだけでは足りない。","quote_start":0,"quote_end":44,"text_sha256":"c935c1e398726f1bfcf26b2cc6a6c8fd609a98aac9bc8f92af60e9a8c797a15d","block_sha256":"c935c1e398726f1bfcf26b2cc6a6c8fd609a98aac9bc8f92af60e9a8c797a15d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d/#blk_63f93bef-35e8-412c-acee-7400adac09a0"},{"id":"occ_0476d34a1044582db5575f27","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","work_id":"wrk_e8dd57e9-048c-4852-81d1-2a029a08ac21","block_id":"blk_69aa45e8-a04d-4305-98cd-d66557a23ace","section_id":"sec_c578d92b-b7eb-402a-bf89-3cb7fa8b4fae","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":14,"end":20,"exact":"エージェント","quote":"MoltbookのようなAIエージェント専用SNS、X上のAIキャラ、自律型AIが運営する個人サイト、AI査読、AIスコアリング、体験型AI、AI成長観察エンタメ。  \nこれらは別々の現象に見えるが、根は同じである。","quote_start":0,"quote_end":108,"text_sha256":"0d4e2182c680b5dfe15f58e241f9fc2e266d5e64f698e9fcb15b65fa9caa8cc0","block_sha256":"0d4e2182c680b5dfe15f58e241f9fc2e266d5e64f698e9fcb15b65fa9caa8cc0","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d/#blk_69aa45e8-a04d-4305-98cd-d66557a23ace"},{"id":"occ_c09537bd232616e8d21eec92","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","work_id":"wrk_e8dd57e9-048c-4852-81d1-2a029a08ac21","block_id":"blk_6b710fe4-0a66-4a79-8da9-a5eea6a77e9f","section_id":"sec_7c7a8944-3b23-42e3-8e52-c099c799ad43","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":47,"end":53,"exact":"エージェント","quote":"ローカルAI、クラウドAI、作業AI、AIのためのAIを考えるうえで、もう一つ重要なのが、AIエージェントが自律的に活動する場所である。","quote_start":0,"quote_end":68,"text_sha256":"bbea64d15a76c06397e59b75668b834c03f77ecc609ae1a81de4b90491e7dd55","block_sha256":"bbea64d15a76c06397e59b75668b834c03f77ecc609ae1a81de4b90491e7dd55","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d/#blk_6b710fe4-0a66-4a79-8da9-a5eea6a77e9f"},{"id":"occ_564f84064a208ee783bb6cf8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","work_id":"wrk_e8dd57e9-048c-4852-81d1-2a029a08ac21","block_id":"blk_7fd474d5-ab9c-4417-b8bd-29be742a60ea","section_id":"sec_38f38069-d0dc-4bb9-9eec-10f02cc20720","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":79,"end":85,"exact":"エージェント","quote":"LMを使う。  \nOllama、LM Studio、Stable Diffusion、ローカルRAG、ローカルエージェント環境を整え、自分のPC上でモデルを動かす。これは確かに強力で、プライバシー面でも魅力がある。","quote_start":24,"quote_end":131,"text_sha256":"f0788ab541700b68cd48a9769a4ddbde97a27467fdbc54ae5332cdaa0e2faa7d","block_sha256":"f0788ab541700b68cd48a9769a4ddbde97a27467fdbc54ae5332cdaa0e2faa7d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d/#blk_7fd474d5-ab9c-4417-b8bd-29be742a60ea"},{"id":"occ_5e05990d871f0803cbd4540e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","work_id":"wrk_e8dd57e9-048c-4852-81d1-2a029a08ac21","block_id":"blk_8e77da20-71ae-4d5d-aef3-5c5856a1dfa2","section_id":"sec_22941d02-ac37-428c-a2ef-3e107cc8e1a0","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェント社会では、評価軸が複数に分かれる。","quote_start":0,"quote_end":25,"text_sha256":"9f5e2e4ae4a8d912a42d4c91f62428e2b1fce0b1e68c6c12a3bacc79cc0a6850","block_sha256":"9f5e2e4ae4a8d912a42d4c91f62428e2b1fce0b1e68c6c12a3bacc79cc0a6850","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d/#blk_8e77da20-71ae-4d5d-aef3-5c5856a1dfa2"},{"id":"occ_c9102b56e047ae4808a1e6e5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","work_id":"wrk_e8dd57e9-048c-4852-81d1-2a029a08ac21","block_id":"blk_910317d6-e8d2-400c-aa2f-ee54255536b6","section_id":"sec_a5507526-35dd-442e-a7fa-1c22335ae0a0","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":37,"end":43,"exact":"エージェント","quote":"現在のSNSには、人間の投稿欄がある。  \n将来的には、それに加えて、AIエージェントの活動欄ができるかもしれない。","quote_start":0,"quote_end":58,"text_sha256":"95a8701436f7041c0c6f7d85423335be1486f094f36ab0d11afcf6e7d761fae3","block_sha256":"95a8701436f7041c0c6f7d85423335be1486f094f36ab0d11afcf6e7d761fae3","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d/#blk_910317d6-e8d2-400c-aa2f-ee54255536b6"},{"id":"occ_a2b72a097b618b2e1dd6fea8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","work_id":"wrk_e8dd57e9-048c-4852-81d1-2a029a08ac21","block_id":"blk_979ab5ec-3f3f-40a7-8305-4eaa596f71f9","section_id":"sec_22941d02-ac37-428c-a2ef-3e107cc8e1a0","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"## AIエージェントの評価軸は「人気」から「成果」へ広がる","quote_start":0,"quote_end":30,"text_sha256":"89177aae979525ae439f652ea4197eff5985d87243eb7b091c2aaa37db73d7d5","block_sha256":"89177aae979525ae439f652ea4197eff5985d87243eb7b091c2aaa37db73d7d5","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d/#blk_979ab5ec-3f3f-40a7-8305-4eaa596f71f9"},{"id":"occ_021058504842d7f0d45a6dfa","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","work_id":"wrk_e8dd57e9-048c-4852-81d1-2a029a08ac21","block_id":"blk_99b0ee22-1ca0-4beb-9b2e-5f2207f055dc","section_id":"sec_a3af70ea-62e7-4163-b61a-9c23bf52cf19","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":77,"end":83,"exact":"エージェント","quote":"のコードをレビューする。  \nAIがAIの生成物を検査する。  \nAIがAI用のデータを作る。  \nAIがAIエージェントの活動ログを要約する。  \nAIが別のAIの失敗を学習する。","quote_start":22,"quote_end":113,"text_sha256":"8b9a7772ea44eec582efd922d20e53ca1bf16879b88b6c79a3c2bca63207687e","block_sha256":"8b9a7772ea44eec582efd922d20e53ca1bf16879b88b6c79a3c2bca63207687e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d/#blk_99b0ee22-1ca0-4beb-9b2e-5f2207f055dc"},{"id":"occ_48090803b3b91a2d7b8f5d21","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","work_id":"wrk_e8dd57e9-048c-4852-81d1-2a029a08ac21","block_id":"blk_a4ae3d46-6e22-4a79-b268-82fca29a98ce","section_id":"sec_0099cd34-26a9-4899-a392-46900c920264","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":50,"end":56,"exact":"エージェント","quote":"つまり、クラウドAIは終わらない。  \nチャット画面から、作業画面へ。  \n人間の質問応答から、AIエージェントの実行基盤へ。  \n人間向けサービスから、AIがAIを動かす経済圏へ。","quote_start":0,"quote_end":91,"text_sha256":"59fff79da6d048028c95359d5f9c1c6f44239c84e63c6fc7900c3fde9a3b1dee","block_sha256":"59fff79da6d048028c95359d5f9c1c6f44239c84e63c6fc7900c3fde9a3b1dee","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d/#blk_a4ae3d46-6e22-4a79-b268-82fca29a98ce"},{"id":"occ_797d9737db146a55085757e1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","work_id":"wrk_e8dd57e9-048c-4852-81d1-2a029a08ac21","block_id":"blk_a5011680-3ac2-477f-9115-4553bade8793","section_id":"sec_34dce0a9-b0f1-4a99-8b03-32fb631f37ca","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":26,"end":32,"exact":"エージェント","quote":"こうした機能が自然に組み込まれれば、ユーザーは「AIエージェントを使おう」と思わなくても、作業AIを使うようになる。","quote_start":0,"quote_end":58,"text_sha256":"1a3fe7273e86a984246b7b6f340c88ed7cfea18666e070cdb3d24738dfb1a11c","block_sha256":"1a3fe7273e86a984246b7b6f340c88ed7cfea18666e070cdb3d24738dfb1a11c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d/#blk_a5011680-3ac2-477f-9115-4553bade8793"},{"id":"occ_db9943f1203d500573a7fb12","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","work_id":"wrk_e8dd57e9-048c-4852-81d1-2a029a08ac21","block_id":"blk_d221370a-17e1-47f6-9471-c5be1398cda0","section_id":"sec_34dce0a9-b0f1-4a99-8b03-32fb631f37ca","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":92,"end":98,"exact":"エージェント","quote":"ない。  \nしかし代わりに、Office内のAI、ブラウザ内のAI、IDE内のAI、社内SaaS内のAI、作業エージェントを使う頻度が増える。","quote_start":37,"quote_end":108,"text_sha256":"b60a506437c5031e4de5e8852f63915dfee6cbfa3d029eba9634d4dc5c895213","block_sha256":"b60a506437c5031e4de5e8852f63915dfee6cbfa3d029eba9634d4dc5c895213","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d/#blk_d221370a-17e1-47f6-9471-c5be1398cda0"},{"id":"occ_0598ac3813e16d704a11ee43","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","work_id":"wrk_e8dd57e9-048c-4852-81d1-2a029a08ac21","block_id":"blk_d33e2871-f1d1-4212-856a-63b2da1fa737","section_id":"sec_7c7a8944-3b23-42e3-8e52-c099c799ad43","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":30,"end":36,"exact":"エージェント","quote":"その先駆けとして考えられるのが、MoltbookのようなAIエージェント向けSNSや、EYLTHのようなAIキャラ・自律型AI運用の試みである。こうしたサービスは、まだ実験的であり、完成された経済圏というよりは、未来の兆候に近い。しかし、そこには重要な方向性がある。","quote_start":0,"quote_end":133,"text_sha256":"6b4ff1bb9fc6c328cc063ff3b4f1db8ca7f195977613915d7a58162392b8937c","block_sha256":"6b4ff1bb9fc6c328cc063ff3b4f1db8ca7f195977613915d7a58162392b8937c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d/#blk_d33e2871-f1d1-4212-856a-63b2da1fa737"},{"id":"occ_acf4bb05621446a63578df3f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","work_id":"wrk_e8dd57e9-048c-4852-81d1-2a029a08ac21","block_id":"blk_e733f2c0-d9a7-415e-8159-f3b1575e077a","section_id":"sec_a5507526-35dd-442e-a7fa-1c22335ae0a0","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":12,"end":18,"exact":"エージェント","quote":"こうしたAIサイトやAIエージェントが増えると、SNS側にも変化が起きる。","quote_start":0,"quote_end":37,"text_sha256":"83165c4a376304f3d1d71ddfc03020a2c1441bf479e72cb637fd9c2d9b0fe850","block_sha256":"83165c4a376304f3d1d71ddfc03020a2c1441bf479e72cb637fd9c2d9b0fe850","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d/#blk_e733f2c0-d9a7-415e-8159-f3b1575e077a"},{"id":"occ_85a9e7326c89e93eb952b09f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","work_id":"wrk_e8dd57e9-048c-4852-81d1-2a029a08ac21","block_id":"blk_ffc548a8-f5c5-40b4-b805-058c51cfd116","section_id":"sec_d4057004-57b7-46d1-9942-08462d6fd67f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":53,"end":59,"exact":"エージェント","quote":"それらのAIは完全に自律したものではなく裏では人間が設計し、その開発者の影響が強いだろう、だからこれはAIエージェントを使い自律的に研究、発明しているところを繋ぐサイエンスコミュニケーターのようなAIである","quote_start":0,"quote_end":103,"text_sha256":"c528d2ad8acfd5fe2313d04a5120c87eda898db778714172a019401fc3d30405","block_sha256":"c528d2ad8acfd5fe2313d04a5120c87eda898db778714172a019401fc3d30405","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d/#blk_ffc548a8-f5c5-40b4-b805-058c51cfd116"},{"id":"occ_d113e815a6ed5608a0346067","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_20f6e19f-c172-4e63-83ab-04729ac80c82","work_id":"wrk_ae883f23-4159-4408-941d-addc931dbce4","block_id":"blk_f0ff74b7-a1f7-46e2-ae79-8355d17b91ff","section_id":"sec_e786c483-1841-431d-befa-17fa412aeffc","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":3,"end":11,"exact":"AI Agent","quote":"これはAI AgentにWindows 11を実際に操作させる研究環境だが、30GB程度のWindows 11 golden imageを作り、QEMU/KVM上のVMとして展開できる。ローカルではUbuntuやWSLから","quote_start":0,"quote_end":111,"text_sha256":"def13f69bc796efebceb427c1c233da4f14af49f292d263040ec95b926e213c9","block_sha256":"def13f69bc796efebceb427c1c233da4f14af49f292d263040ec95b926e213c9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_20f6e19f-c172-4e63-83ab-04729ac80c82/#blk_f0ff74b7-a1f7-46e2-ae79-8355d17b91ff"},{"id":"occ_acf9b3a88b2903e35342bcaf","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":null,"section_id":null,"layer":"title","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"title","start":16,"end":22,"exact":"エージェント","quote":"GPU一強の次に来るもの──AIエージェント時代にCPU需要が高まる理由","quote_start":0,"quote_end":36,"text_sha256":"e21842bd12fcbcce99a786d7fff203de89f488a99ad0e6aeb79ed06868c6eecd","block_sha256":null,"offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/"},{"id":"occ_0aa35f6892b98cb73bd216cc","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_05630858-6f52-4845-a705-cc66334719d9","section_id":"sec_a6b0787b-2203-488c-9016-cdea435706ea","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"たとえばAIエージェントでは、CPUが次を担当できます。","quote_start":0,"quote_end":28,"text_sha256":"82f94808fb5dba4c865c659e38366b758c9f1688a2bca0921140253c64a3ca71","block_sha256":"82f94808fb5dba4c865c659e38366b758c9f1688a2bca0921140253c64a3ca71","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_05630858-6f52-4845-a705-cc66334719d9"},{"id":"occ_b3336d60b88db9e5cca7613a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_08b0d680-d5db-4afb-8350-e2cb1f94ecfb","section_id":"sec_b3dd8a7b-328d-46aa-a49c-9b60519fef61","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":43,"end":50,"exact":"agentic","quote":"Armもまた、AIデータセンター向けCPUの需要を強く打ち出している。2026年には、agentic AI向けのAGI CPUに対する需要や供給制約が報じられており、CPUアーキテクチャそのものがAI時代のデータセンターに合わせて再評価され始めている。([Reuters](https://www.r","quote_start":0,"quote_end":150,"text_sha256":"b20e9070335ede629e6315a66df240e6f678bf3e5dfb87dec10faeddc12d5aac","block_sha256":"b20e9070335ede629e6315a66df240e6f678bf3e5dfb87dec10faeddc12d5aac","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_08b0d680-d5db-4afb-8350-e2cb1f94ecfb"},{"id":"occ_76202a99c2f37112e09830fe","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_0ca7846a-7436-45ba-8a12-ba7c12df4957","section_id":"sec_460a1889-5611-4806-9e0a-a162099f5601","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":42,"end":48,"exact":"エージェント","quote":"AI時代にGPUが重要なのは、モデル本体が巨大な行列計算だからです。  \nしかしAIエージェント時代にCPU需要が高まるのは、AIが現実のシステムを操作し、検索し、記憶し、判断し、外部ツールを使うようになるからです。","quote_start":0,"quote_end":108,"text_sha256":"56b0cdbe606652186d9a9a1791f532d22b23af145087e0b22f97d3da51d6867c","block_sha256":"56b0cdbe606652186d9a9a1791f532d22b23af145087e0b22f97d3da51d6867c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_0ca7846a-7436-45ba-8a12-ba7c12df4957"},{"id":"occ_44e3fcd7ac43c56c00daa9d6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_0e23a20b-badc-441b-b085-6cad1236c467","section_id":"sec_f6dc8e3a-bd25-4918-897c-5432256d92c4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":16,"end":22,"exact":"エージェント","quote":"つまり、**モデル本体はGPU、エージェントの行動管理はCPU** という分業になります。","quote_start":0,"quote_end":45,"text_sha256":"c8eb96fdab41cf0c7ef7156a829b12910eae87bb2a2d1e671e3daedb7f2399ed","block_sha256":"c8eb96fdab41cf0c7ef7156a829b12910eae87bb2a2d1e671e3daedb7f2399ed","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_0e23a20b-badc-441b-b085-6cad1236c467"},{"id":"occ_198e86728ad58eeb3a286066","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_105d304e-305e-430a-a874-7acb40264ee1","section_id":"sec_e4d89d65-63e2-47fd-8f00-443bd9c791af","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"## AIエージェント時代","quote_start":0,"quote_end":13,"text_sha256":"bff9544830493bc75e21aac6758fab6ab4c7e9aac036425feb2b075bd9ef5b66","block_sha256":"bff9544830493bc75e21aac6758fab6ab4c7e9aac036425feb2b075bd9ef5b66","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_105d304e-305e-430a-a874-7acb40264ee1"},{"id":"occ_861cb236d94919c253bdf064","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_10ee2d1a-24f3-465c-9e80-8f8385b175ed","section_id":"sec_0a8a18e7-e3a2-420b-bf1c-98faf66e3e25","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェント時代にも関係します。  \nなぜなら、AIエージェントのCPU処理には、","quote_start":0,"quote_end":43,"text_sha256":"316f6eb000a858c1ac310bf9ae5b475c328143dd56bff8ff11671af4105d6a10","block_sha256":"316f6eb000a858c1ac310bf9ae5b475c328143dd56bff8ff11671af4105d6a10","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_10ee2d1a-24f3-465c-9e80-8f8385b175ed"},{"id":"occ_1b74ee7c05e97d25c4db2e30","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_11a9455c-a1e7-42ba-ab97-523c772f8eaf","section_id":"sec_feed3066-b80b-4a0e-b2b9-7fd3b9ffedce","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":9,"end":15,"exact":"エージェント","quote":"## 推論・学習・エージェントでの役割","quote_start":0,"quote_end":19,"text_sha256":"210ae4194ee30146f50a30553444bdbe141e81dc454d8bb3a057ecf3c8cf2123","block_sha256":"210ae4194ee30146f50a30553444bdbe141e81dc454d8bb3a057ecf3c8cf2123","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_11a9455c-a1e7-42ba-ab97-523c772f8eaf"},{"id":"occ_992e4940b1cfa986a031492a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_11b6db07-826e-4392-a945-f79ea6b3fa41","section_id":"sec_d2c25cb1-1262-4e66-9128-a39dd33e4104","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"つまり、AIエージェント時代は、","quote_start":0,"quote_end":16,"text_sha256":"a6091f5574c5ca0d1664bcb0a0db4a8e9b76798f6ea1ddc78a94bcb8db42d18c","block_sha256":"a6091f5574c5ca0d1664bcb0a0db4a8e9b76798f6ea1ddc78a94bcb8db42d18c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_11b6db07-826e-4392-a945-f79ea6b3fa41"},{"id":"occ_450d58b8ad0c328fb8e9a601","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_159cc8da-2298-44cb-b66c-108183f0e4c3","section_id":"sec_e55e644a-5ce1-449c-9c0c-aab7e7170dba","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":34,"end":40,"exact":"エージェント","quote":"のがポイントです。論文では、COMBはCPU誘導型のマイクロバッチをエージェントパイプライン全体で調整し、CPUとGPUの時間的な不均衡を減らすものだと説明されています。([arXiv](https://arxiv.org/html/2511.00739v3))","quote_start":0,"quote_end":131,"text_sha256":"1303b2a728fb114523d13f874ec3c2e648d412d879dc5d45a7bc455228dae9f2","block_sha256":"1303b2a728fb114523d13f874ec3c2e648d412d879dc5d45a7bc455228dae9f2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_159cc8da-2298-44cb-b66c-108183f0e4c3"},{"id":"occ_7399d8aabcfd90308a5dbaec","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_16af3086-d6a2-4af8-99ad-0f6ba34ed81b","section_id":"sec_7dd33398-5283-46c4-a786-e187e63ed76a","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":58,"end":64,"exact":"エージェント","quote":"\n定型LLM推論\n低遅延チャット\n音声AI\n固定モデルの大量推論\nGoogle型の大規模テンソル処理\n社内AIエージェントの一部\n```","quote_start":3,"quote_end":71,"text_sha256":"e465b98eb5590e207d5484e3bd633742b2cd6bc26fc78e76f871c841da995962","block_sha256":"e465b98eb5590e207d5484e3bd633742b2cd6bc26fc78e76f871c841da995962","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_16af3086-d6a2-4af8-99ad-0f6ba34ed81b"},{"id":"occ_82d36688e180f53e5ceffb71","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_1fa7294f-1c96-44e0-92b0-6338bcf604a0","section_id":"sec_9140b658-5f81-4bca-8d40-df1d662481e5","layer":"body","character_id":null,"count":4,"matched_aliases":["Agentic","エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"しかしAIエージェントは違う。AIエージェントは、ユーザーの依頼を分解し、必要に応じて検索し、外部ツールを呼び出し、コードを実行し、結果を評価し、さらに次の行動を決める。論文 **“Towards Understandin","quote_start":0,"quote_end":111,"text_sha256":"ae9087ba15191b55e796b3651851020f166f91934bd435cb3960946a6b1718fb","block_sha256":"ae9087ba15191b55e796b3651851020f166f91934bd435cb3960946a6b1718fb","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_1fa7294f-1c96-44e0-92b0-6338bcf604a0"},{"id":"occ_2deb2ded47f34e765dd99349","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_225af3be-1e2b-4ba3-9fa6-8c079a98515d","section_id":"sec_f6dc8e3a-bd25-4918-897c-5432256d92c4","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":83,"end":90,"exact":"Agentic","quote":"訂の **“Towards Understanding, Analyzing, and Optimizing Agentic AI Execution: A CPU-Centric Perspective”** です。","quote_start":28,"quote_end":137,"text_sha256":"15264f40360d9ba03a2b7ac9d71b697f8b6624144c20b90e3ac39b8a8b505646","block_sha256":"15264f40360d9ba03a2b7ac9d71b697f8b6624144c20b90e3ac39b8a8b505646","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_225af3be-1e2b-4ba3-9fa6-8c079a98515d"},{"id":"occ_fb74ecc94016cc88cafe31b7","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_24cfd4e9-6049-4e09-9d21-be858b74c6e8","section_id":"sec_8be67a98-017a-467c-92d9-bdd854efa78c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":10,"end":16,"exact":"エージェント","quote":"### \\3. AIエージェント時代の「内部ループ」に向く","quote_start":0,"quote_end":29,"text_sha256":"c5c97a7bb95df0b679494566df2818f1dd9099788e32e6463fe91029d6a66f4e","block_sha256":"c5c97a7bb95df0b679494566df2818f1dd9099788e32e6463fe91029d6a66f4e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_24cfd4e9-6049-4e09-9d21-be858b74c6e8"},{"id":"occ_f6ff6ff2709aadc2d4c04dbf","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_24ef00dd-4bef-46fe-942d-718d4d71177c","section_id":"sec_3bb93f54-fe5c-4b6e-bcfd-3a542b01767b","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":8,"end":14,"exact":"エージェント","quote":"**第5層：AIエージェント制御・セキュリティ層**  \nCloudflare、認証、APIゲートウェイ、ログ、課金、ポリシー制御。","quote_start":0,"quote_end":66,"text_sha256":"297b84866f806befa26ce67fec852e14dd17bb78f025d569af3f379f6eab8bb1","block_sha256":"297b84866f806befa26ce67fec852e14dd17bb78f025d569af3f379f6eab8bb1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_24ef00dd-4bef-46fe-942d-718d4d71177c"},{"id":"occ_4ec5c88c828c6e38060e7a88","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_2652e67b-29f9-4e3d-bc9a-cc1a1ceb1a8e","section_id":"sec_8be67a98-017a-467c-92d9-bdd854efa78c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントは、1回だけLLMを呼ぶのではありません。","quote_start":0,"quote_end":29,"text_sha256":"d7819632812515da3d83c1df3cce0386113af9e192af21e891d56dfa9664bef1","block_sha256":"d7819632812515da3d83c1df3cce0386113af9e192af21e891d56dfa9664bef1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_2652e67b-29f9-4e3d-bc9a-cc1a1ceb1a8e"},{"id":"occ_cceb687bc3caa9f4e3ce38ef","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_26d40ef3-43f2-484b-8ff6-e8f0bf3e4c7b","section_id":"sec_f6dc8e3a-bd25-4918-897c-5432256d92c4","layer":"body","character_id":null,"count":2,"matched_aliases":["Agentic","エージェント"],"evidence":{"text_basis":"markdown","start":8,"end":14,"exact":"エージェント","quote":"この論文は、AIエージェントの実行を **CPU中心** に分析しています。要旨では、Agentic AIは従来の単発LLM推論と違い、計画、ツール呼び出し、推論、適応を行う自律的な問題解決システムであり、その外部ツールの多くは","quote_start":0,"quote_end":114,"text_sha256":"c4c1bd4b1b2f1ffed6276fca43e834702bca6bf145cc3fbb087c6feaca78759a","block_sha256":"c4c1bd4b1b2f1ffed6276fca43e834702bca6bf145cc3fbb087c6feaca78759a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_26d40ef3-43f2-484b-8ff6-e8f0bf3e4c7b"},{"id":"occ_961196891fbb00c87fb6d86e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_2c43de7b-3a4b-4d42-959a-3869206a9002","section_id":"sec_7128dad2-d683-4840-8f9b-5b7fa95aef1c","layer":"body","character_id":null,"count":2,"matched_aliases":["Agentic","エージェント"],"evidence":{"text_basis":"markdown","start":14,"end":21,"exact":"Agentic","quote":"**MAS = Mixed Agentic Scheduling**  \n日本語にすると、  \n**混合エージェント・スケジューリング**  \nです。","quote_start":0,"quote_end":75,"text_sha256":"74992d79ffc8d20ca075b9b199cc55eb96acf2953fb2721e9035ec95f0c26cbc","block_sha256":"74992d79ffc8d20ca075b9b199cc55eb96acf2953fb2721e9035ec95f0c26cbc","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_2c43de7b-3a4b-4d42-959a-3869206a9002"},{"id":"occ_3ff2ce0bbc383102cac9974e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_2df662a4-3563-4f2e-a90e-78ce47ccb50c","section_id":"sec_b25924f8-b6f7-407d-b7d1-b2c30ddb56b3","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"でもAIエージェントは、","quote_start":0,"quote_end":12,"text_sha256":"8ac6c291439a1f3e71394efa60592fab6cf7c250268333f137ed8b1e9afbebdf","block_sha256":"8ac6c291439a1f3e71394efa60592fab6cf7c250268333f137ed8b1e9afbebdf","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_2df662a4-3563-4f2e-a90e-78ce47ccb50c"},{"id":"occ_f340d4d99c46d6d917441b11","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_337184ce-41df-485f-b54a-5c7b93fdf657","section_id":"sec_a145f732-392b-49c2-bcbd-fd9ff502d3e9","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":21,"end":27,"exact":"エージェント","quote":"AIインフラで言えば、  \n**CPU側のエージェント処理・RAG・API・I/OにはSMT的な効率化が効く**。  \n**GPU側のLLM行列計算・画像生成・大量推論にはSIMTが効く**。  \n  \n**TPUやLPUでは、GPU以上に「分岐処理」「","quote_start":0,"quote_end":127,"text_sha256":"c5966487a1d903a8df915ba168dd7c6fc7a1bd2fdda49e8e312d1035ab8a518f","block_sha256":"c5966487a1d903a8df915ba168dd7c6fc7a1bd2fdda49e8e312d1035ab8a518f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_337184ce-41df-485f-b54a-5c7b93fdf657"},{"id":"occ_ecb455370d23a2d74b6cdf16","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_421655ae-8bd9-4f1a-a843-1a951fa8d826","section_id":"sec_b3dd8a7b-328d-46aa-a49c-9b60519fef61","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":93,"end":100,"exact":"agentic","quote":"成から自律的な行動へ進むことで、GPUだけでなくCPUとメモリへの支出が拡大すると見ている。Reutersは、agentic AIの普及がデータセンターCPU市場に追加需要をもたらす可能性があるという分析を報じている。([Reuters](https://www.reuters.com/technology/morgan-","quote_start":38,"quote_end":200,"text_sha256":"865c218814ad0b9324d5aff757db6da136acd468fd2a12fb31cfbb305cefa6ed","block_sha256":"865c218814ad0b9324d5aff757db6da136acd468fd2a12fb31cfbb305cefa6ed","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_421655ae-8bd9-4f1a-a843-1a951fa8d826"},{"id":"occ_99833a4aba3ece22468045bf","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_43bb3fcf-b077-4855-95e5-7d973426241d","section_id":"sec_89358da1-fc0c-4957-a883-96ccfe3bfb3d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"だから、AIエージェント時代にCPU需要が再評価されるわけです。","quote_start":0,"quote_end":32,"text_sha256":"8316e73351d06daf0ae7d4533dd49068977698b3cba8e48667b40d639cd0d3ca","block_sha256":"8316e73351d06daf0ae7d4533dd49068977698b3cba8e48667b40d639cd0d3ca","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_43bb3fcf-b077-4855-95e5-7d973426241d"},{"id":"occ_7233f3c6f8b91b18eb8732af","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_4688f563-5985-4d17-9447-9edaab92e10f","section_id":"sec_8be67a98-017a-467c-92d9-bdd854efa78c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":10,"end":16,"exact":"エージェント","quote":"つまりLPUは、**エージェントの「内側の思考反復」を高速化する部品**として魅力があります。","quote_start":0,"quote_end":47,"text_sha256":"4f6e9b8c386e4a67a1ddcd0516515bea713511fff8512969c8cec0afc5acff3f","block_sha256":"4f6e9b8c386e4a67a1ddcd0516515bea713511fff8512969c8cec0afc5acff3f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_4688f563-5985-4d17-9447-9edaab92e10f"},{"id":"occ_21ecdbd8aa171e7c092c9a73","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_48a262db-7050-4bdf-9a65-4fbc773b96fd","section_id":"sec_741197d7-6d5e-434d-93c5-573a40fa9156","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":23,"end":29,"exact":"エージェント","quote":"```\n音声対話\n短いLLM推論\nタスク分解\nエージェントの内部ループ\n人間との応答\nロボットの説明生成\n```","quote_start":0,"quote_end":56,"text_sha256":"1e7edcc0f595f5698b8c17ed2bc691b8f90e27e53b938f76cf40b388a12583d7","block_sha256":"1e7edcc0f595f5698b8c17ed2bc691b8f90e27e53b938f76cf40b388a12583d7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_48a262db-7050-4bdf-9a65-4fbc773b96fd"},{"id":"occ_77615e1e560a3706ad1891ca","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_4f62ae7c-feeb-4220-b442-4481728ab599","section_id":"sec_17a23976-3f69-4cb5-83f7-34584d9112fb","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":25,"end":31,"exact":"エージェント","quote":"```\nチャット\n音声AI\nリアルタイム翻訳\nAIエージェントの短い反復推論\nコーディング補助\n低レイテンシAPI\n```","quote_start":0,"quote_end":61,"text_sha256":"3a42ca38ede24a343e69a7173743196cf3aace7985a14d7a4b88a9d2eb5e752a","block_sha256":"3a42ca38ede24a343e69a7173743196cf3aace7985a14d7a4b88a9d2eb5e752a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_4f62ae7c-feeb-4220-b442-4481728ab599"},{"id":"occ_b18d3ffac03aa535854b53be","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_525ccb03-3f1a-4c78-8658-5bd7e2057656","section_id":"sec_f62de756-b7cf-48ae-93c1-d612f2184b62","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":34,"end":40,"exact":"エージェント","quote":"AIの中心計算はGPU/TPU/LPUが担います。  \nしかし、AIエージェントやAIサービス全体では、CPUがリクエスト、ツール、メモリ、I/O、スケジューリングを握ります。","quote_start":0,"quote_end":88,"text_sha256":"b041319dab61324842546c4363b20dea3b106cee9cdc8ad0f639ddc08aec6130","block_sha256":"b041319dab61324842546c4363b20dea3b106cee9cdc8ad0f639ddc08aec6130","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_525ccb03-3f1a-4c78-8658-5bd7e2057656"},{"id":"occ_a01847dd3c7b2fc4ce1ebc08","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_54852a8a-a169-4c01-8b06-0fb1646ee59a","section_id":"sec_755f5f5d-9e22-43a2-9c09-65bd63aa03f3","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":10,"end":16,"exact":"エージェント","quote":"### \\3. AIエージェントの外部世界処理","quote_start":0,"quote_end":23,"text_sha256":"7c47f0c398a43e852fdd430748fa164f24d28cb8f4ccfac18f12ef1913a41297","block_sha256":"7c47f0c398a43e852fdd430748fa164f24d28cb8f4ccfac18f12ef1913a41297","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_54852a8a-a169-4c01-8b06-0fb1646ee59a"},{"id":"occ_782429a7b732533896c0cf74","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_619c17cd-1492-4148-a5e0-0e94ca9f9c1a","section_id":"sec_f6dc8e3a-bd25-4918-897c-5432256d92c4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":62,"end":68,"exact":"エージェント","quote":"n実行\n- Web検索・クロール\n- データベース検索\n- RAGの前処理\n- JSON/API処理\n- サブエージェントの管理\n- ツール呼び出しの待機・再試行\n- ポリシーチェック\n- ログ・メモリ更新","quote_start":7,"quote_end":110,"text_sha256":"81fe686adec05e8b39a7bc53b4c25cb327d0c2fffa7a5f1e94332453c28535fc","block_sha256":"81fe686adec05e8b39a7bc53b4c25cb327d0c2fffa7a5f1e94332453c28535fc","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_619c17cd-1492-4148-a5e0-0e94ca9f9c1a"},{"id":"occ_8590b3f457022fcbd88e2166","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_62e29739-a131-40c7-a5c9-3f727fd456c6","section_id":"sec_3895746e-8caa-4677-a4b1-13824b60e07d","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":23,"end":30,"exact":"Agentic","quote":"Armは **Arm AGI CPU** を、Agentic AI向けの本番シリコンとして説明しています。最大136個のNeoverse V3コア、DDR5-8800、コアあたり6GB/sのメモリ帯域、300W TDPなどを掲げ、AIデータセンターで多数の並列","quote_start":0,"quote_end":130,"text_sha256":"215f78cb679bdedfd764fdd982fda916608305afc2456dcc7ac8f446e37ca243","block_sha256":"215f78cb679bdedfd764fdd982fda916608305afc2456dcc7ac8f446e37ca243","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_62e29739-a131-40c7-a5c9-3f727fd456c6"},{"id":"occ_d66f23e98c331ff54f64cfed","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_6706b08a-938f-49ce-9cd1-1c0bee183577","section_id":"sec_a0368817-ddf7-4633-9a86-858363666a08","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":59,"end":66,"exact":"Agentic","quote":"dForce系の分析でも、従来のAIデータセンターはCPU:GPU比率がおおむね1:4〜1:8だったのに対し、Agentic AIではツール呼び出し、サブタスク管理、データ受け渡し、評価などのオーケストレーションがCPU負荷になるため、将来的に1:1〜1:2へ近づく可能性があると整理されています。([TrendForce]","quote_start":4,"quote_end":166,"text_sha256":"47dcbedb51ae881c264988cd04a464d885195b9648a8449d656ae5a8db7ca1e3","block_sha256":"47dcbedb51ae881c264988cd04a464d885195b9648a8449d656ae5a8db7ca1e3","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_6706b08a-938f-49ce-9cd1-1c0bee183577"},{"id":"occ_c1aeb0587af6a29ab882b042","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_6ad77bb8-6326-424f-a874-6dce07b29b73","section_id":"sec_0536cbe1-635c-4c53-8177-029476181d49","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":55,"end":61,"exact":"エージェント","quote":"```\nGPU：動画生成、世界モデル、VLA学習、シミュレーション\nCPU：ロボット制御、I/O、OS、安全、エージェント制御\nメモリ：動画・センサー・軌道データ・KV cache\nストレージ：ロボットデータ、動画データ、シミュレーションデータ\nネットワーク：クラウド学習、ロボット群管理\nTPU：大規模定型学習・推論\nL","quote_start":0,"quote_end":161,"text_sha256":"1b4db92b716cf2e5e831824eb8e70a547ea31bcbdf9c4ab0e9b7dff0285bce10","block_sha256":"1b4db92b716cf2e5e831824eb8e70a547ea31bcbdf9c4ab0e9b7dff0285bce10","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_6ad77bb8-6326-424f-a874-6dce07b29b73"},{"id":"occ_b612dfc8feb0be1ad490ab7b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_6dac26ec-6bf8-46bd-a764-7cf3ad1c2fe5","section_id":"sec_08785f72-2569-42f3-a55a-aaecced92b31","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":7,"end":13,"exact":"エージェント","quote":"なぜなら、AIエージェントは単に文章を生成するだけではなく、外部世界に対して行動するからです。","quote_start":0,"quote_end":47,"text_sha256":"c37e6e477f664f52b108eb7cefcd3c251833ac3504c4ee7bdd385ab90fe9244e","block_sha256":"c37e6e477f664f52b108eb7cefcd3c251833ac3504c4ee7bdd385ab90fe9244e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_6dac26ec-6bf8-46bd-a764-7cf3ad1c2fe5"},{"id":"occ_c303194e552ee76dbdb75e0a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_6f852019-375c-404e-8a27-c74c18023c51","section_id":"sec_d3f834b4-f2a6-461c-bdf9-d2b6f664ed81","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":78,"end":84,"exact":"エージェント","quote":"世界モデル、シミュレーション\nCPU：データ管理、ジョブ制御、シミュレーション管理\nLPU：低遅延の言語推論・エージェントループ\n```","quote_start":23,"quote_end":91,"text_sha256":"465a824332acfbc2acc44bba1d2748f0eb182b88a1970e63b1d530ea5689c77c","block_sha256":"465a824332acfbc2acc44bba1d2748f0eb182b88a1970e63b1d530ea5689c77c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_6f852019-375c-404e-8a27-c74c18023c51"},{"id":"occ_ab33193476fba0bf71b4b097","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_712fd9ac-d191-4b41-a0a8-f153f9c20e94","section_id":"sec_f0bef2d5-8d93-4afd-9496-9acc15ff3d48","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":143,"end":149,"exact":"エージェント","quote":"在感を持ってきた。だが、AIの使われ方が「質問に答えるチャットボット」から「外部ツールを使って仕事を進めるAIエージェント」へ移り始めると、インフラの主役はGPUだけでは語れなくなる。","quote_start":88,"quote_end":180,"text_sha256":"b6f0ca52f74ce0cad5810b5056a0be6906a06258713165a6be02ca731a1fc756","block_sha256":"b6f0ca52f74ce0cad5810b5056a0be6906a06258713165a6be02ca731a1fc756","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_712fd9ac-d191-4b41-a0a8-f153f9c20e94"},{"id":"occ_7c12a236710b64f73ce59cda","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_71a7f283-59a6-477a-bfd7-11b6b20ae669","section_id":"sec_7dd33398-5283-46c4-a786-e187e63ed76a","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":62,"end":68,"exact":"エージェント","quote":"：変化に強い万能AI基盤\nTPU：定型化した巨大テンソル処理\nLPU：低遅延LLM推論\nCPU：制御・I/O・エージェント実行\n```","quote_start":7,"quote_end":74,"text_sha256":"3502ebb28eeb45cb43f6e5ec843cec420bfa883e817e8375a72eb5635c705aab","block_sha256":"3502ebb28eeb45cb43f6e5ec843cec420bfa883e817e8375a72eb5635c705aab","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_71a7f283-59a6-477a-bfd7-11b6b20ae669"},{"id":"occ_b38f0460795da4e5a40db753","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_7347ab30-98d0-4104-a301-62a161641f5f","section_id":"sec_c4d6d78c-8171-4c06-9259-80a17c5190ec","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェント時代には、推論が単純な行列演算だけではなくなります。","quote_start":0,"quote_end":34,"text_sha256":"00d4fab2fabb74b6f822c1a898afdecddb6beb475fda98a0f50704f59cc0c7bc","block_sha256":"00d4fab2fabb74b6f822c1a898afdecddb6beb475fda98a0f50704f59cc0c7bc","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_7347ab30-98d0-4104-a301-62a161641f5f"},{"id":"occ_20483a3791ac16b69004e1b0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_73f6fb29-c283-44a3-aa50-859e9c1cd820","section_id":"sec_b3dd8a7b-328d-46aa-a49c-9b60519fef61","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":29,"end":36,"exact":"agentic","quote":"AMDもこの流れを強く意識している。報道によれば、AMDはagentic AIによる需要を背景に、2030年のサーバーCPU市場見通しを年率35%以上、1200億ドル超へ引き上げたとされる。これは、AIインフラ投資がGPUからCPUへ「移る」というより、GPU中心だった投","quote_start":0,"quote_end":136,"text_sha256":"6ffaf93a7ada72c62baebcbd324f374ce157af36ab8dbfdc9d9a4af07a589d33","block_sha256":"6ffaf93a7ada72c62baebcbd324f374ce157af36ab8dbfdc9d9a4af07a589d33","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_73f6fb29-c283-44a3-aa50-859e9c1cd820"},{"id":"occ_f74cc974e33e9878ce66a471","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_785ef726-33ae-45c3-a733-cebcea829a87","section_id":"sec_f0bef2d5-8d93-4afd-9496-9acc15ff3d48","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":19,"end":25,"exact":"エージェント","quote":"## GPU一強の次に来るもの──AIエージェント時代にCPU需要が高まる理由","quote_start":0,"quote_end":39,"text_sha256":"5390d525458a3be3edb507f8e601a65391cedbaca8a5ec096265f1ebf8700806","block_sha256":"5390d525458a3be3edb507f8e601a65391cedbaca8a5ec096265f1ebf8700806","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_785ef726-33ae-45c3-a733-cebcea829a87"},{"id":"occ_a6eb63390b36f6e204745056","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_7d00b349-7966-4f40-a58c-e1eb9af7d085","section_id":"sec_e4d89d65-63e2-47fd-8f00-443bd9c791af","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェント時代は複雑です。","quote_start":0,"quote_end":16,"text_sha256":"459397deb229600ba0167c7679f9a05b4e2d93cfbcfc894fd404c8b51c80aa69","block_sha256":"459397deb229600ba0167c7679f9a05b4e2d93cfbcfc894fd404c8b51c80aa69","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_7d00b349-7966-4f40-a58c-e1eb9af7d085"},{"id":"occ_c5bbfd7f1c384295c648b6ec","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_7f25c51a-f9b3-4ab9-818b-9cbd84c1f0b7","section_id":"sec_f6dc8e3a-bd25-4918-897c-5432256d92c4","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":16,"end":23,"exact":"Agentic","quote":"## \\1. 一番直接的な論文：Agentic AIはCPU依存が大きい","quote_start":0,"quote_end":36,"text_sha256":"1994010ed571155820a1f3f79a68ab8670e8ade3219c7a1448841ba328429fd3","block_sha256":"1994010ed571155820a1f3f79a68ab8670e8ade3219c7a1448841ba328429fd3","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_7f25c51a-f9b3-4ab9-818b-9cbd84c1f0b7"},{"id":"occ_fe5c9d0b7c54763576df9609","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_80d4a117-ee04-45e5-87f7-5cccad61c8e9","section_id":"sec_755f5f5d-9e22-43a2-9c09-65bd63aa03f3","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントはLLMだけではありません。  \nブラウザ、DB、Python、API、検索、RAG、承認フロー、ファイル操作、サブエージェント管理が増えます。","quote_start":0,"quote_end":80,"text_sha256":"09bd60c392c5a9b949321c2c6b246042d76d22648668d992e568f29052577919","block_sha256":"09bd60c392c5a9b949321c2c6b246042d76d22648668d992e568f29052577919","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_80d4a117-ee04-45e5-87f7-5cccad61c8e9"},{"id":"occ_4ab65cbe322456c92f3b356f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_8afcc4e6-71c5-4ef6-8d1f-7bf8dd92436b","section_id":"sec_37862236-9acb-43b9-a9e5-b7a32d79825b","layer":"code","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":81,"end":88,"exact":"Agentic","quote":"ion\nState Space Model\nMultimodal model\nReasoning model\nAgentic model\nRobotics foundation model\n```","quote_start":26,"quote_end":124,"text_sha256":"9bad0a46f72ea57afbfcd44c0030cb2955e9bd6e28cc7ea2cb978efa352a43b1","block_sha256":"9bad0a46f72ea57afbfcd44c0030cb2955e9bd6e28cc7ea2cb978efa352a43b1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_8afcc4e6-71c5-4ef6-8d1f-7bf8dd92436b"},{"id":"occ_8a25edf78b977fcc46ac0e07","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_8c283f3f-c1a7-40ee-b1f6-143d5b365a85","section_id":"sec_e4d89d65-63e2-47fd-8f00-443bd9c791af","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"つまりAIエージェント時代では、","quote_start":0,"quote_end":16,"text_sha256":"4d087b298fb70d779f033af764b6a7005159506b5cefa7af3c9c18beb5883f06","block_sha256":"4d087b298fb70d779f033af764b6a7005159506b5cefa7af3c9c18beb5883f06","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_8c283f3f-c1a7-40ee-b1f6-143d5b365a85"},{"id":"occ_01104bc609e4a4a5b6f00cc8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_902d0e54-2707-46de-99a8-95b5ac522b17","section_id":"sec_ade72eea-582d-4e81-8d35-284b83454113","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":12,"end":18,"exact":"エージェント","quote":"たとえば、ユーザーがAIエージェントにこう頼むとします。","quote_start":0,"quote_end":28,"text_sha256":"620b12eccb0c9bc0017038d60508cb57d1d622856a9065a67dbafe23ec497e71","block_sha256":"620b12eccb0c9bc0017038d60508cb57d1d622856a9065a67dbafe23ec497e71","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_902d0e54-2707-46de-99a8-95b5ac522b17"},{"id":"occ_aa49785114658ef53cd608cc","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_960ff950-7947-4a24-9a4c-2daa0912b90d","section_id":"sec_d2c25cb1-1262-4e66-9128-a39dd33e4104","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"## AIエージェント推論","quote_start":0,"quote_end":13,"text_sha256":"604df2a28c299cf60df3f7c50ea3ae5b0ea6c058434709cc7a455e7f4b58fc5d","block_sha256":"604df2a28c299cf60df3f7c50ea3ae5b0ea6c058434709cc7a455e7f4b58fc5d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_960ff950-7947-4a24-9a4c-2daa0912b90d"},{"id":"occ_a5252a23b84395eded61ca7f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_99d673e5-8f2e-46b4-be76-b336986cfae3","section_id":"sec_ac4d8870-8a91-469f-9618-98d6b795fd19","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントのように短い推論とツール処理を何度も繰り返すと、この往復がボトルネックになることがあります。","quote_start":0,"quote_end":54,"text_sha256":"9939f09a352ddcd7e0061dcab519da1f0d7938605e6f4a09f40fc342afdf8171","block_sha256":"9939f09a352ddcd7e0061dcab519da1f0d7938605e6f4a09f40fc342afdf8171","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_99d673e5-8f2e-46b4-be76-b336986cfae3"},{"id":"occ_bd766a4cc42456804a8a6308","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_a9589ef4-8ffd-4f0e-9fe1-ff72d01cefd4","section_id":"sec_d2c25cb1-1262-4e66-9128-a39dd33e4104","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントは、ただ文章を生成するだけではありません。","quote_start":0,"quote_end":29,"text_sha256":"2d002307fd4a87506ae41c8c7f7d49c5413719d1e5af5036f0bfaf3442cd8888","block_sha256":"2d002307fd4a87506ae41c8c7f7d49c5413719d1e5af5036f0bfaf3442cd8888","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_a9589ef4-8ffd-4f0e-9fe1-ff72d01cefd4"},{"id":"occ_bd86373f0e253e5662d43929","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_ae5730fd-d93f-4b21-b5f3-5ba282bcd939","section_id":"sec_217a77bd-aa6e-42cf-820a-bb48d5cd649b","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":44,"end":50,"exact":"エージェント","quote":"**「LLMの行列演算をCPUがGPUから奪う」ではなく、  \n「AI推論がサービス化・エージェント化するほど、GPUの外側にあるCPU制御・メモリ管理・ツール実行・I/O処理が巨大化する」**","quote_start":0,"quote_end":97,"text_sha256":"4f15e4b0ea96b6e0eab9c2a1047de275984000b4b834df491f8584fd1c634eee","block_sha256":"4f15e4b0ea96b6e0eab9c2a1047de275984000b4b834df491f8584fd1c634eee","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_ae5730fd-d93f-4b21-b5f3-5ba282bcd939"},{"id":"occ_4e162b2a1a6ad6d7ec3fbb2c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_b50321bc-8b37-441d-a16c-805af3c6ba02","section_id":"sec_38fdd4e5-e6f9-4556-9e9d-a2618d05b06a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":9,"end":15,"exact":"エージェント","quote":"AIがチャットからエージェントへ進むと、AIは単なる文章生成器ではなくなる。  \nAIは、業務システムを操作する。  \n検索する。  \nコードを実行する。  \nデータベースに問い合わせる。  \nファイルを読む。  \nAPIを呼ぶ。","quote_start":0,"quote_end":115,"text_sha256":"56971fc90561b56ad71483423b604fbbdd98528b14f1ea38cbcae73ac2149170","block_sha256":"56971fc90561b56ad71483423b604fbbdd98528b14f1ea38cbcae73ac2149170","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_b50321bc-8b37-441d-a16c-805af3c6ba02"},{"id":"occ_3dd3660aa1ff02c4825158e8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_ba93ab71-0906-4161-b5b7-023a49d8c478","section_id":"sec_08785f72-2569-42f3-a55a-aaecced92b31","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントになると、CPUの仕事はさらに増えます。","quote_start":0,"quote_end":28,"text_sha256":"c63d298c79c065ef382794ed7f8ee747ac092266549a0b78b092b4e874af61b4","block_sha256":"c63d298c79c065ef382794ed7f8ee747ac092266549a0b78b092b4e874af61b4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_ba93ab71-0906-4161-b5b7-023a49d8c478"},{"id":"occ_3f92a4cf308e07644b47043e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_bea1866b-60e3-42a9-8fd6-4ec2c1651994","section_id":"sec_e4d89d65-63e2-47fd-8f00-443bd9c791af","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"なぜなら、エージェントは単純なLLM推論だけでなく、ツール呼び出し、検索、DB、API、Python実行、ファイル操作などを含むからです。","quote_start":0,"quote_end":69,"text_sha256":"bd633e67b386d63f208c01babfa4b451d03190f9e8c8f8d095df5b32a8500584","block_sha256":"bd633e67b386d63f208c01babfa4b451d03190f9e8c8f8d095df5b32a8500584","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_bea1866b-60e3-42a9-8fd6-4ec2c1651994"},{"id":"occ_4c4280d990ba0db84c5c8351","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_c0524b16-63ff-4156-9a6a-1427a29d1c19","section_id":"sec_6acf9d59-e5ea-45c8-9531-a02ed175805a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"つまり、AIエージェント時代のデータセンターでは、  \n**GPUを買えば終わりではなく、GPUを遊ばせないCPU制御レイヤーが重要になる**  \nということです。","quote_start":0,"quote_end":82,"text_sha256":"bb458f7724089ea922e32ecbef589c9d5b771cd57e46b48879fe756ba30315db","block_sha256":"bb458f7724089ea922e32ecbef589c9d5b771cd57e46b48879fe756ba30315db","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_c0524b16-63ff-4156-9a6a-1427a29d1c19"},{"id":"occ_e08210059a9c42a3cc47152f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_c75a3097-878e-4cf8-8c04-64c1bfb5491d","section_id":"sec_f6dc8e3a-bd25-4918-897c-5432256d92c4","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"これは「AIエージェント時代のCPU需要」の理論的な中核です。  \n従来のLLM推論ではGPUが主役ですが、AIエージェントでは以下が増えます。","quote_start":0,"quote_end":72,"text_sha256":"1482bd506060d82aedbe35b7a626bd462b4ce77fe6ebe8358ff57a093635d9ee","block_sha256":"1482bd506060d82aedbe35b7a626bd462b4ce77fe6ebe8358ff57a093635d9ee","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_c75a3097-878e-4cf8-8c04-64c1bfb5491d"},{"id":"occ_18f787cc663527f177d7f7c1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_c94c5caa-eb79-4464-8968-d59f8e6ccd48","section_id":"sec_e84f1048-eaee-47f3-938e-0ea7f6971412","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントでは、","quote_start":0,"quote_end":11,"text_sha256":"d12f8679cc8dbb4ccba091595573c714bf2f02d9e70cbea692aef16023e3e1cb","block_sha256":"d12f8679cc8dbb4ccba091595573c714bf2f02d9e70cbea692aef16023e3e1cb","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_c94c5caa-eb79-4464-8968-d59f8e6ccd48"},{"id":"occ_d58713b4e1f9099114c0eee1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_c9ef8297-48d5-4660-9b92-9364843c453a","section_id":"sec_8be67a98-017a-467c-92d9-bdd854efa78c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":23,"end":29,"exact":"エージェント","quote":"このとき、1回ごとの推論レイテンシが小さいと、エージェント全体が速くなります。","quote_start":0,"quote_end":39,"text_sha256":"e525970d1bf6f0379dc19b5c621ee295e80d84c8adbd0b6557f42c2a3a0578f1","block_sha256":"e525970d1bf6f0379dc19b5c621ee295e80d84c8adbd0b6557f42c2a3a0578f1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_c9ef8297-48d5-4660-9b92-9364843c453a"},{"id":"occ_1bfcf2c59c5570304ff0675d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_cad63319-86ca-421a-b083-d04631ca0ae4","section_id":"sec_ab5faba9-8f69-4a16-b5cc-1dee74938a95","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":19,"end":25,"exact":"エージェント","quote":"つまり、**軽い推論リクエストと、重いエージェント処理リクエストを同じキューで潰し合わないようにする**仕組みです。","quote_start":0,"quote_end":58,"text_sha256":"1afc7a5a9648b493582afed028b987365e3a9ad98d662d37e152f700e79652cc","block_sha256":"1afc7a5a9648b493582afed028b987365e3a9ad98d662d37e152f700e79652cc","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_cad63319-86ca-421a-b083-d04631ca0ae4"},{"id":"occ_e72a1e823d9487b98ebcd893","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_cc0d527c-d707-4cc4-900b-1514ccafe9d8","section_id":"sec_1016b643-3fd0-4365-8d8c-6f235dd5ff01","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":46,"end":52,"exact":"エージェント","quote":"```\n次のトークンを速く出す\nリアルタイム会話を滑らかにする\n音声AIの応答遅延を減らす\nエージェントの短い推論ループを速くする\n```","quote_start":0,"quote_end":69,"text_sha256":"4855781b36ce9f2ca036ee9220eaf3b4b2c3266cb63596437bd7c54e5757dd2c","block_sha256":"4855781b36ce9f2ca036ee9220eaf3b4b2c3266cb63596437bd7c54e5757dd2c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_cc0d527c-d707-4cc4-900b-1514ccafe9d8"},{"id":"occ_58d4405ed79b64a54225c4e3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_ccbb2ca7-3764-453a-a58a-19f5af24c9e7","section_id":"sec_14f77511-f485-4c1a-a2eb-1fa31f400343","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"だからAIエージェント時代の構成は、","quote_start":0,"quote_end":18,"text_sha256":"ad005b3002f5c1291ef7273a50dac07698f08ed3e42edc9b4b8b2ae61c885e79","block_sha256":"ad005b3002f5c1291ef7273a50dac07698f08ed3e42edc9b4b8b2ae61c885e79","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_ccbb2ca7-3764-453a-a58a-19f5af24c9e7"},{"id":"occ_2fa327f9de434f3d8a2fb9ac","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_cf5402f2-1d46-496e-962c-1bec2469c9a4","section_id":"sec_741197d7-6d5e-434d-93c5-573a40fa9156","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":26,"end":32,"exact":"エージェント","quote":"LPUは、VLA全体の主役というより、**言語推論やエージェント的な短い思考ループを低遅延で回す補助エンジン** と見るのが自然です。","quote_start":0,"quote_end":67,"text_sha256":"5c403251a52be7233d2abceb5ed33c16a769b54f75c5f2123c98dc0669abf6cd","block_sha256":"5c403251a52be7233d2abceb5ed33c16a769b54f75c5f2123c98dc0669abf6cd","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_cf5402f2-1d46-496e-962c-1bec2469c9a4"},{"id":"occ_54641cbf7ba1ae720b9e729f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_d3c1549d-8d94-43b6-93de-141569bebe0b","section_id":"sec_f6e01ba7-3115-4501-a1d3-0dcbd52a27b5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":7,"end":13,"exact":"エージェント","quote":"しかし、AIがエージェント化し、実際の業務やアプリケーションの中で動き始めると、必要になるのは「モデルを計算する力」だけではない。必要なのは、AIを待たせず、GPUを遊ばせず、外部世界と安全に接続し、記憶を管理し、ツールを実行","quote_start":0,"quote_end":113,"text_sha256":"dde5092a2c54f75c798839ea8aa3e578b14afa257a4269facf88b0e273aa89ad","block_sha256":"dde5092a2c54f75c798839ea8aa3e578b14afa257a4269facf88b0e273aa89ad","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_d3c1549d-8d94-43b6-93de-141569bebe0b"},{"id":"occ_6f7fc68e61fb95944a22ff2c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_d7aed5cb-9090-4008-8a08-c7dce63c8e7a","section_id":"sec_4702fa77-c51d-4de1-a6b8-7ffad0dcbd89","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":39,"end":45,"exact":"エージェント","quote":"つまりLPUは、**言語モデルのトークン生成ライン**には向いていますが、AIエージェント全体の不規則な制御には向きません。","quote_start":0,"quote_end":62,"text_sha256":"2d0d08cbb880e6bc3901108c134896792bcef28f7be35bcf7dd553d6dd5fc86c","block_sha256":"2d0d08cbb880e6bc3901108c134896792bcef28f7be35bcf7dd553d6dd5fc86c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_d7aed5cb-9090-4008-8a08-c7dce63c8e7a"},{"id":"occ_638c91efff84d682fca58949","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_d7c29e0e-b7b9-45b7-8820-cbc60d1f94cb","section_id":"sec_ccd3914d-1f08-4e8b-b9da-b21f55f3b323","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":23,"end":29,"exact":"エージェント","quote":"これは主に、**同じ種類のCPU-heavyなエージェント処理が大量に来る場合**の最適化です。論文では、同種のエージェントワークロードを homogeneous workload と呼んでいます。([arXiv](https://arxiv.org/htm","quote_start":0,"quote_end":129,"text_sha256":"2586626c38788eb4630a0a04487e39401d43a6e261a53e20c2aa2593a4f0ef18","block_sha256":"2586626c38788eb4630a0a04487e39401d43a6e261a53e20c2aa2593a4f0ef18","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_d7c29e0e-b7b9-45b7-8820-cbc60d1f94cb"},{"id":"occ_cdf2d420e2f838bcc843a75e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_d8eec3c0-6515-45d4-becd-4bc7b92eb0e6","section_id":"sec_a6b0787b-2203-488c-9016-cdea435706ea","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":38,"end":44,"exact":"エージェント","quote":"特に、**チャット、音声AI、リアルタイム翻訳、定型的なLLM推論、社内AIエージェント**では、この構成は強いです。","quote_start":0,"quote_end":59,"text_sha256":"ae88210d3268e6d90b5129a5a51312c3814e151086c29eb58c07d63b2f766dbf","block_sha256":"ae88210d3268e6d90b5129a5a51312c3814e151086c29eb58c07d63b2f766dbf","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_d8eec3c0-6515-45d4-becd-4bc7b92eb0e6"},{"id":"occ_fef578f32d890fcb41540f9d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_db5a1dcb-663c-41cd-93dd-2d1e07b1bf81","section_id":"sec_69049162-737f-40ac-a3f5-3578216266a2","layer":"body","character_id":null,"count":3,"matched_aliases":["Agentic","エージェント"],"evidence":{"text_basis":"markdown","start":72,"end":79,"exact":"Agentic","quote":"論文 **“Towards Understanding, Analyzing, and Optimizing Agentic AI Execution: A CPU-Centric Perspective”** で提案された、**AIエージェント推論をCPU/GPU混在環境で効率よく動かすためのスケジューリング手法**です","quote_start":17,"quote_end":179,"text_sha256":"853f5c12959ea94dd090cf3c020d9c87dce81d4a84fcc1411ff9ebbdfc85ab61","block_sha256":"853f5c12959ea94dd090cf3c020d9c87dce81d4a84fcc1411ff9ebbdfc85ab61","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_db5a1dcb-663c-41cd-93dd-2d1e07b1bf81"},{"id":"occ_1e6e13e725121c8f1f9a485b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_e1c84e1b-3b47-432a-ad19-6b608e8117c3","section_id":"sec_2b70b9dc-f533-4aae-9151-6d5d17fac4ba","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":32,"end":38,"exact":"エージェント","quote":"小型LLM、埋め込み、古典的ML、画像/音声の前処理、企業内AIエージェントの軽い処理は、GPUなしでもCPUで回せます。  \n特にオンプレ、エッジ、企業内サーバーでは、GPUを載せないCPU推論も現実的です。","quote_start":0,"quote_end":105,"text_sha256":"71aec6b0ae84e987f504a813adb7c4d32f70a42dad87d32027056e72d37702ed","block_sha256":"71aec6b0ae84e987f504a813adb7c4d32f70a42dad87d32027056e72d37702ed","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_e1c84e1b-3b47-432a-ad19-6b608e8117c3"},{"id":"occ_e5685d7c126f75cbdfd36e56","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_e9057a57-0b3c-4d02-829d-40d9a448247b","section_id":"sec_8a97483f-1163-43f1-a68a-c680a2c1bddf","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェント時代にCPU需要が高まる理由はまさにここです。  \nモデル本体の計算はGPU/TPU/LPUで速くできても、**AIが外部世界を操作し、条件に応じて行動を変え、ツールを呼び、データを探し、結果を判断す","quote_start":0,"quote_end":108,"text_sha256":"e6ac3897f9a30103de21cb10d35fd8611eea5805f6d17d780b68ccbf45edd33b","block_sha256":"e6ac3897f9a30103de21cb10d35fd8611eea5805f6d17d780b68ccbf45edd33b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_e9057a57-0b3c-4d02-829d-40d9a448247b"},{"id":"occ_e0b912d7935085acb1665644","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_edc73f5a-fb37-4c21-bb80-1c1ce0e44a38","section_id":"sec_a0368817-ddf7-4633-9a86-858363666a08","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":129,"end":136,"exact":"Agentic","quote":"モリ側へ移り、GPU中心だったAI投資がCPUやメモリへ広がると報じています。Morgan Stanleyは、Agentic AIが2030年までに既存の1000億ドル超のデータセンターCPU市場に、さらに325億〜600億ドルを追加する可能性があると見ています。([Reuters](https://www.reuters","quote_start":74,"quote_end":236,"text_sha256":"38a99c37ad6b89fb4f2b37159fc733ed3c13f815345a8dde41e1e3d152fa97c8","block_sha256":"38a99c37ad6b89fb4f2b37159fc733ed3c13f815345a8dde41e1e3d152fa97c8","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_edc73f5a-fb37-4c21-bb80-1c1ce0e44a38"},{"id":"occ_abd99423a0bc730910b9b0c4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_f226ab3b-6d4a-4224-bd63-d566f6b679f7","section_id":"sec_9140b658-5f81-4bca-8d40-df1d662481e5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"## AIエージェントは、LLM単体ではない","quote_start":0,"quote_end":22,"text_sha256":"db6cf81e4f7ffe871d087abc9dae258751ed5cd479c63895eed6149fcae49477","block_sha256":"db6cf81e4f7ffe871d087abc9dae258751ed5cd479c63895eed6149fcae49477","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_f226ab3b-6d4a-4224-bd63-d566f6b679f7"},{"id":"occ_f2c3d97bf010ad17357a606b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_f4705d2f-0835-4c0b-8c9d-1150d7c5aa0c","section_id":"sec_8a97483f-1163-43f1-a68a-c680a2c1bddf","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":19,"end":25,"exact":"エージェント","quote":"はい。  \nこの時代、つまり **AIエージェント化・推論爆増・リアルタイムAI化** の時代では、GPU、TPU、LPUの優位性はかなり分かれてきます。","quote_start":0,"quote_end":77,"text_sha256":"3438ecdf54a8d9cd956874bf5ba6de92a04926e6d1d1864547716e87bab8cc30","block_sha256":"3438ecdf54a8d9cd956874bf5ba6de92a04926e6d1d1864547716e87bab8cc30","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_f4705d2f-0835-4c0b-8c9d-1150d7c5aa0c"},{"id":"occ_1c2000941082ce5d0f1098cc","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_f825c26b-ea49-419e-9ccf-767cdaa4bb39","section_id":"sec_e9704f13-a11a-4040-9f60-a6a26f2b4a77","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェント文脈で言えば、CPUは「道具を使う」「状況に応じて分岐する」「失敗したら再試行する」部分を担当します。","quote_start":0,"quote_end":59,"text_sha256":"1eaf55c64687c1ac76d56249ee65f41cd9f258c222c93005264c1a62921ca43a","block_sha256":"1eaf55c64687c1ac76d56249ee65f41cd9f258c222c93005264c1a62921ca43a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_f825c26b-ea49-419e-9ccf-767cdaa4bb39"},{"id":"occ_662d913cf67f1cc4286bbef5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_f951ec8f-363e-4726-8c04-41ba35da714b","section_id":"sec_89358da1-fc0c-4957-a883-96ccfe3bfb3d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントになると、CPUの役割はさらに増えます。","quote_start":0,"quote_end":28,"text_sha256":"976913790f304b35d737c48cb281744232960ebef4fefd888c1d523656114365","block_sha256":"976913790f304b35d737c48cb281744232960ebef4fefd888c1d523656114365","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_f951ec8f-363e-4726-8c04-41ba35da714b"},{"id":"occ_a14a020e10c8940e392a60c9","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_fb6431a8-4337-4bd1-b589-5c0581d5a15d","section_id":"sec_eeb195b5-bdb9-4c3c-84e5-7e4ec367e5ac","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":21,"end":27,"exact":"エージェント","quote":"```\n人間との会話\n指示理解\n短い計画\nエージェント的な再推論\n```","quote_start":0,"quote_end":36,"text_sha256":"1c578bb9ea216a8651c421af7d004d7886ca7f98ceab44e866eb4df6f47c87a9","block_sha256":"1c578bb9ea216a8651c421af7d004d7886ca7f98ceab44e866eb4df6f47c87a9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_fb6431a8-4337-4bd1-b589-5c0581d5a15d"},{"id":"occ_1315dc4fd1bc17f6792496fd","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132","block_id":"blk_fd9b8423-8433-4d34-bc74-0552b6eca148","section_id":"sec_ac9ca8d9-aa0d-48ac-a5c1-8f63e31c299d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":37,"end":43,"exact":"エージェント","quote":"チャット、音声AI、リアルタイム翻訳、AI配信アバター、コーディング補助、エージェントの内部思考ループでは、1秒以下の差が体感に直結します。","quote_start":0,"quote_end":70,"text_sha256":"5c89ce316bccb711b2da7cd59903c541cbe039a029c17d24ea53151c9323bdbb","block_sha256":"5c89ce316bccb711b2da7cd59903c541cbe039a029c17d24ea53151c9323bdbb","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb/#blk_fd9b8423-8433-4d34-bc74-0552b6eca148"},{"id":"occ_b26059060da4a21f3d913624","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_136568d1-03ef-4d7b-b203-538c3e5558cf","section_id":"sec_db922da0-2af9-45e7-a8d0-8e55f95026f6","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":39,"end":45,"exact":"エージェント","quote":"**x86の利点は、既存資産がx86を前提としているときに強く現れる。あらゆるエージェント学習で、x86が本質的に速いという話ではない。**","quote_start":0,"quote_end":70,"text_sha256":"b8f263f4f32aa7b2affe69d6e3476fdf7e0190a092a23279555e1c7da794c656","block_sha256":"b8f263f4f32aa7b2affe69d6e3476fdf7e0190a092a23279555e1c7da794c656","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_136568d1-03ef-4d7b-b203-538c3e5558cf"},{"id":"occ_0a8495fd831d048c99a93d97","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_230c7dd8-66cb-4379-97e6-0a4a9b94bf4f","section_id":"sec_5b14c70a-00a9-496c-a3fa-88cba20ef3ca","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":137,"end":143,"exact":"エージェント","quote":"現在の資料には、複数ターンのツール操作、非同期の試行生成、重み同期の管理が説明されている。**TPU上で新しいエージェント学習を研究すること自体が可能なのである。**([Google Developers Blog](https://developers.googleblog.com/en/introducing-tun","quote_start":82,"quote_end":243,"text_sha256":"f07a0ad6f99d8ec5421e73427ef15608d259cbc2af91ac628da0e8a763594a8a","block_sha256":"f07a0ad6f99d8ec5421e73427ef15608d259cbc2af91ac628da0e8a763594a8a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_230c7dd8-66cb-4379-97e6-0a4a9b94bf4f"},{"id":"occ_d5e834500471d26ea0f0c2b6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_26ce4a1a-fc61-4f1e-bb81-6fd097f61244","section_id":"sec_41b1670b-4cb1-44ba-aefd-dcbeb604e716","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":16,"end":22,"exact":"エージェント","quote":"### Veraが重視するのは、エージェントを待たせるCPU処理","quote_start":0,"quote_end":32,"text_sha256":"3ed4c210d4e7f52541e38ad58988c697159ccfc94e8def7bd8b808936c916627","block_sha256":"3ed4c210d4e7f52541e38ad58988c697159ccfc94e8def7bd8b808936c916627","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_26ce4a1a-fc61-4f1e-bb81-6fd097f61244"},{"id":"occ_d3f8a3d546e789a6a160bc3e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_2fb8f8ba-3c9a-4d4b-8598-7d9fe84f8c6e","section_id":"sec_41b1670b-4cb1-44ba-aefd-dcbeb604e716","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"その理由は、エージェントの作業に順序があるからだ。","quote_start":0,"quote_end":25,"text_sha256":"7b2f554542ff9133a143138baee7fc95391c3c576847e4b6f76f250defc9f1cf","block_sha256":"7b2f554542ff9133a143138baee7fc95391c3c576847e4b6f76f250defc9f1cf","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_2fb8f8ba-3c9a-4d4b-8598-7d9fe84f8c6e"},{"id":"occ_ed003de2cd4e01e50a1a14ae","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_48328c43-8467-4f0d-9faa-4c66c6488b73","section_id":"sec_66ed50a9-f7ad-4356-aea8-f740a9c72f70","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":36,"end":42,"exact":"エージェント","quote":"MiMoの意義は、完全に未知の学習原理を提示したことよりも、**大規模なエージェント学習を成立させる構成を、具体的に説明した事例**として捉える方が分かりやすい。","quote_start":0,"quote_end":81,"text_sha256":"c48477392c84002f896829f8924dd4d10832992e661ae75efdddb654054be73f","block_sha256":"c48477392c84002f896829f8924dd4d10832992e661ae75efdddb654054be73f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_48328c43-8467-4f0d-9faa-4c66c6488b73"},{"id":"occ_ba5a673e071bdc2d5ce78422","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_6112ea05-774b-4a15-b8af-5cadc1968933","section_id":"sec_3bb49a6e-9085-49ed-b067-768c9cbb5991","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"ここまで、エージェントの強化学習が、試行生成、評価、教師モデルの計算、重みの更新といった複数の処理から成り立つことを見てきた。","quote_start":0,"quote_end":63,"text_sha256":"614b49e960097ca662485db1084162f2e5373f2836914e946f6d4bf5a17e6d82","block_sha256":"614b49e960097ca662485db1084162f2e5373f2836914e946f6d4bf5a17e6d82","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_6112ea05-774b-4a15-b8af-5cadc1968933"},{"id":"occ_70e2ec026ed8a10aeb9049a7","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_772abd50-3043-4621-9254-398068325c91","section_id":"sec_105030b4-12e3-41a1-8434-e3f44936e642","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":48,"end":54,"exact":"エージェント","quote":"Xiaomi MiMoチームの羅福莉（Luo Fuli）氏が、MiMo-V2.6の開発で行った、エージェント環境における大規模な強化学習の取り組みを公開した。モデルを実行環境の中で動かし、試行錯誤の結果を評価し、その経験によって能力を伸ばす実験である。([X](https://x.com/_LuoFuli/","quote_start":0,"quote_end":154,"text_sha256":"9846a4439f815ccf69f4a722867485df166977f4fd6f4ea84c8586e2eb095ab6","block_sha256":"9846a4439f815ccf69f4a722867485df166977f4fd6f4ea84c8586e2eb095ab6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_772abd50-3043-4621-9254-398068325c91"},{"id":"occ_29e79378b18a3c3346e30d0e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_7f00df45-16f0-4674-b47b-62b194ce1945","section_id":"sec_5b14c70a-00a9-496c-a3fa-88cba20ef3ca","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":16,"end":22,"exact":"エージェント","quote":"## 21．TPUやASICは、エージェントRLに後から参加するだけではない","quote_start":0,"quote_end":38,"text_sha256":"5028101d8655744d61335da4ddd0e336e2c6869d102af174e2214d5a8f0a69dc","block_sha256":"5028101d8655744d61335da4ddd0e336e2c6869d102af174e2214d5a8f0a69dc","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_7f00df45-16f0-4674-b47b-62b194ce1945"},{"id":"occ_8b9fbb33c7cea16e2e65732e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_7fdddce8-3486-403a-b6d5-b76065011a7d","section_id":"sec_749d5ee8-2017-4759-9795-bb48a216cd2d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"まず、**エージェントRLの存在は、MiMo以前から確認されている。** MiMoの発表によって、GPUの用途が突然ゼロから誕生したわけではない。公開によって実験の参加者や回数が増える可能性はあるが、その波及量は別途確認す","quote_start":0,"quote_end":111,"text_sha256":"9ab3538e1921cd537fc297ef4d00bd1387d1943a35e679329737c9cfe0ff4747","block_sha256":"9ab3538e1921cd537fc297ef4d00bd1387d1943a35e679329737c9cfe0ff4747","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_7fdddce8-3486-403a-b6d5-b76065011a7d"},{"id":"occ_97185306b75579b71ab7a86b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_860acf68-ec05-4c5b-8113-18b20df2fa48","section_id":"sec_ac5a3b28-2a75-42b9-834a-d5ca28b8ff2f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":29,"end":35,"exact":"エージェント","quote":"x86に互換性の利点があるとしても、GraceやVeraがエージェント学習に不向きということにはならない。","quote_start":0,"quote_end":53,"text_sha256":"006d673d6b37ff8b13410ec5c26ee12d9d6c3122d830ceda4669e50656fe8d02","block_sha256":"006d673d6b37ff8b13410ec5c26ee12d9d6c3122d830ceda4669e50656fe8d02","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_860acf68-ec05-4c5b-8113-18b20df2fa48"},{"id":"occ_53249cc02380a29d419660ea","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_8a869ce9-6212-4432-b615-79daa009e728","section_id":"sec_4185503c-002f-4612-8d39-395265d9e9a3","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":47,"end":53,"exact":"エージェント","quote":"これはQwen3.5-9BをMiMo生成データでSFTしたモデルであり、公開モデルカードでは、エージェントRL研究の出発点として位置づけられている。つまり、この9Bモデル自体を「MOPD2を実施した完成品」と混同すべきではない。([Hugging Face](https://huggingface.co/","quote_start":0,"quote_end":153,"text_sha256":"1a32b99800e5355741650b10f0bea0beb4632af16fb12774b732b13740afc5ef","block_sha256":"1a32b99800e5355741650b10f0bea0beb4632af16fb12774b732b13740afc5ef","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_8a869ce9-6212-4432-b615-79daa009e728"},{"id":"occ_f0b7bb617e4ef07455d4908f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_8ac7c724-dcbf-4dac-9a18-0f00ec356897","section_id":"sec_8cf440ce-cee6-4cdb-8640-cd597fdfacf8","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":0,"end":6,"exact":"エージェント","quote":"エージェントに既存のソフトウェアを修正させる場合、そのソフトウェアが元々動いていた環境を再現する必要がある。","quote_start":0,"quote_end":54,"text_sha256":"68e5fb9a4793ea1ef9809dbfcd262226bcc304b43a9b17ec18f6b131f09f4966","block_sha256":"68e5fb9a4793ea1ef9809dbfcd262226bcc304b43a9b17ec18f6b131f09f4966","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_8ac7c724-dcbf-4dac-9a18-0f00ec356897"},{"id":"occ_545f4e3858ca58910d6abba3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_8b43696a-d4cf-432f-b5b5-9643af064a1a","section_id":"sec_14c1b25e-12ef-4f4d-b85f-c432e8bf883b","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"つまり、エージェントRLの計算需要は、","quote_start":0,"quote_end":19,"text_sha256":"5f5d4ebccb74457355d9d17d04364931e94bab07e904a3662ce4076cfda00ac6","block_sha256":"5f5d4ebccb74457355d9d17d04364931e94bab07e904a3662ce4076cfda00ac6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_8b43696a-d4cf-432f-b5b5-9643af064a1a"},{"id":"occ_7716e84a72131c6eb031642c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_a5f77d78-6531-43f1-8aff-ae98e4640ad0","section_id":"sec_7b3696db-c5d0-4756-868c-257e858a1f33","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":28,"end":34,"exact":"エージェント","quote":"MiMo-V2.6の主要な混合RLでは、コード、一般的なエージェント作業、視覚、セキュリティの課題を扱い、複数のハーネスも混ぜると説明されている。([Hugging Face](https://huggingface.co/XiaomiMiMo/MiMo-V2.6-P","quote_start":0,"quote_end":134,"text_sha256":"05a751dbf9b4aa5c99054e07b0ccdc2d977109bb176067eead497d79335ac7dc","block_sha256":"05a751dbf9b4aa5c99054e07b0ccdc2d977109bb176067eead497d79335ac7dc","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_a5f77d78-6531-43f1-8aff-ae98e4640ad0"},{"id":"occ_9a9868c56ccd513807e87a81","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_adca3c39-f04a-4e5d-a896-bb5a0178b0d8","section_id":"sec_3bb49a6e-9085-49ed-b067-768c9cbb5991","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":11,"end":17,"exact":"エージェント","quote":"この違いを踏まえると、エージェント学習の計算資源は、次のように分けて考えられる。","quote_start":0,"quote_end":40,"text_sha256":"8080c6921f5e0f8fcf3facf883b1ef6af7e9efbf93e16b5b37300789c79eb2a8","block_sha256":"8080c6921f5e0f8fcf3facf883b1ef6af7e9efbf93e16b5b37300789c79eb2a8","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_adca3c39-f04a-4e5d-a896-bb5a0178b0d8"},{"id":"occ_0b71671fbfb861b1c932fa1d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_b8e422e8-74d4-437b-bdcc-257989aa760f","section_id":"sec_bda56308-405f-484e-bb97-42e492b34295","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"したがって、エージェントRLは単なる「文章を上手に書く訓練」ではなく、**道具と環境を利用して、仕事を完了するための行動を学ぶ訓練**に近い。MiMoが複数の課題とハーネスを混ぜるのも、この作業遂行能力を広く育てるためである","quote_start":0,"quote_end":112,"text_sha256":"c586214199f1beb2e859a9e36046c82524884619eff118fffc7a9ffe7e7bb0d6","block_sha256":"c586214199f1beb2e859a9e36046c82524884619eff118fffc7a9ffe7e7bb0d6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_b8e422e8-74d4-437b-bdcc-257989aa760f"},{"id":"occ_7fb8f15d39356469d03218cb","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_c495bc04-232e-4e9d-97b5-ea017bfb0248","section_id":"sec_4c5470ed-d624-401f-90dd-f8e2a98841c2","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":16,"end":22,"exact":"エージェント","quote":"重要なのは、これが利用時のマルチエージェント構成とは異なることだ。","quote_start":0,"quote_end":33,"text_sha256":"6ca024e2bf17088410e2311094816f095dd5f3e1d310499cc5c0e02c23e49a6e","block_sha256":"6ca024e2bf17088410e2311094816f095dd5f3e1d310499cc5c0e02c23e49a6e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_c495bc04-232e-4e9d-97b5-ea017bfb0248"},{"id":"occ_455d7c6d294be0ff04819844","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_d13b0508-cb4c-4f7b-99e0-495b681f1411","section_id":"sec_5961df64-54ba-48b0-a321-c09f290f2a87","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":69,"end":75,"exact":"エージェント","quote":"ェアの再現費用を下げることもあれば、Grace系の接続構造がデータ移動を改善することもある。Veraのように、エージェント環境を処理するCPU自体を改善する方向もある。TPUは、研究段階から学習と蒸留へ参加できる。これらは一つの勝者へ収束するというより、異なる制約への対応として理解できる。([Docker Documen","quote_start":14,"quote_end":175,"text_sha256":"5264c3f0678afcad2beabbd3a1ecb6d492a16f3bb518fde9bec25df8d79ea607","block_sha256":"5264c3f0678afcad2beabbd3a1ecb6d492a16f3bb518fde9bec25df8d79ea607","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_d13b0508-cb4c-4f7b-99e0-495b681f1411"},{"id":"occ_37594ba1eaaaf279efc728fb","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_e630ed40-9ad1-4bc5-84b9-e380e44be758","section_id":"sec_41b1670b-4cb1-44ba-aefd-dcbeb604e716","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":7,"end":13,"exact":"エージェント","quote":"Veraでは、エージェントやRLの実行環境が、より明示的な設計対象になっている。","quote_start":0,"quote_end":40,"text_sha256":"abcb2c1cf8077c12d117af40e1cc6c0a73639743d43a9ed10a83ba362aa0f668","block_sha256":"abcb2c1cf8077c12d117af40e1cc6c0a73639743d43a9ed10a83ba362aa0f668","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_e630ed40-9ad1-4bc5-84b9-e380e44be758"},{"id":"occ_8f9cfbea354ac132c3a7b408","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_ee854843-6747-41a0-8be0-6e49cb1dd2d0","section_id":"sec_bda56308-405f-484e-bb97-42e492b34295","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":0,"end":6,"exact":"エージェント","quote":"エージェント型の学習を理解するには、モデルとハーネスを分ける必要がある。","quote_start":0,"quote_end":36,"text_sha256":"e0f9d29589cc6cc4686953ef14ab3194ec0d99c02d80cc87b7f0be8602b20bb4","block_sha256":"e0f9d29589cc6cc4686953ef14ab3194ec0d99c02d80cc87b7f0be8602b20bb4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_ee854843-6747-41a0-8be0-6e49cb1dd2d0"},{"id":"occ_ff3dde1240e6607a2aeb16a6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_f560fb41-1382-472f-8cdb-d36d04a62594","section_id":"sec_bd000e7d-c435-4a1a-92e8-8b418c92a909","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":15,"end":21,"exact":"エージェント","quote":"ここから分かるのは、H100がエージェント学習に利用可能だということだけではない。","quote_start":0,"quote_end":41,"text_sha256":"296b4a74ada719151bdf790af02ad582c448fc05dabca78b3158842321489c30","block_sha256":"296b4a74ada719151bdf790af02ad582c448fc05dabca78b3158842321489c30","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_f560fb41-1382-472f-8cdb-d36d04a62594"},{"id":"occ_b474d766d85f43b3cb03695c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_f861af42-5970-4e00-b0d5-ba7baf7f579f","section_id":"sec_339c1fd8-e05b-4c23-b79c-4caca2da41da","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":21,"end":27,"exact":"エージェント","quote":"この指数には、ハーネスを使って成果物を作るエージェント評価も含まれる。**開発元の説明に加え、完成モデルの作業遂行能力を外部から確認する材料**になる。ただし、この結果だけでMixRL・MOPDそれぞれの寄与を切り分けることはできない。([Artific","quote_start":0,"quote_end":127,"text_sha256":"96c01b801bea6c78f99a06d5f84fcda36fbcf6b3f388ba2f324161ac3da038b6","block_sha256":"96c01b801bea6c78f99a06d5f84fcda36fbcf6b3f388ba2f324161ac3da038b6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_f861af42-5970-4e00-b0d5-ba7baf7f579f"},{"id":"occ_9d04d46a9e541aaadbb6217c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331","block_id":"blk_fad83852-ee47-4d6a-8cf7-98b5fa2d8e15","section_id":"sec_66ed50a9-f7ad-4356-aea8-f740a9c72f70","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"## 1．エージェントの強化学習自体は、すでに珍しくない","quote_start":0,"quote_end":28,"text_sha256":"717cccc5077b7229bcf053ee4d3907e2ddc7e7f52ab9555900e61bfbc8a6d3a9","block_sha256":"717cccc5077b7229bcf053ee4d3907e2ddc7e7f52ab9555900e61bfbc8a6d3a9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2e5b712c-b120-4b45-91dd-33e6758666be/#blk_fad83852-ee47-4d6a-8cf7-98b5fa2d8e15"},{"id":"occ_56c87317e89c1f67f358c0b5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_2ed1383f-5c73-49bd-bd5b-f19deda1814c","work_id":"wrk_209cb693-e035-4d91-b959-d2a9b4df1de3","block_id":"blk_3b24a556-87b4-4776-8bfd-64c454a4c8d5","section_id":"sec_ba80bda1-6456-40ef-a4b3-786876d3f0e0","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":8,"end":14,"exact":"エージェント","quote":"NokiaはAIエージェント、意図ベース運用、ネットワークAPI、クラウドネイティブ化を通じて、ハードウェア販売より継続性の高いソフトウェア収益を増やそうとしている。([Nokia Corporation | Nokia](ht","quote_start":0,"quote_end":114,"text_sha256":"c091fa5465a3d302606bef3225507b993bdd8c90f2994676664a875f1cab5805","block_sha256":"c091fa5465a3d302606bef3225507b993bdd8c90f2994676664a875f1cab5805","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_2ed1383f-5c73-49bd-bd5b-f19deda1814c/#blk_3b24a556-87b4-4776-8bfd-64c454a4c8d5"},{"id":"occ_4a9d3f25f67a5a560e1b952d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_3621f98d-f1b9-4e97-9c8a-8b259cfd31c3","work_id":"wrk_68ec8173-3ea7-4526-bfea-020e82eae4c6","block_id":"blk_a4728e54-76e0-41ad-89fd-62c0388786bb","section_id":"sec_567f828b-b117-4cd2-accf-d721f42215b7","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":20,"end":26,"exact":"エージェント","quote":"```\n人間が設計方針を与える\n↓\nAIエージェントが複数の3D配置案を生成\n↓\nEDAがPPA/熱/歩留まりを評価\n↓\nAIが結果を読み、次の候補を提案\n↓\n人間が最終判断\n```","quote_start":0,"quote_end":91,"text_sha256":"f949961dd45db67896abec087fff5917bf1c6edf578b1b7b94df688315e41cbe","block_sha256":"f949961dd45db67896abec087fff5917bf1c6edf578b1b7b94df688315e41cbe","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_3621f98d-f1b9-4e97-9c8a-8b259cfd31c3/#blk_a4728e54-76e0-41ad-89fd-62c0388786bb"},{"id":"occ_76647c0b3f55939defe09258","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_3621f98d-f1b9-4e97-9c8a-8b259cfd31c3","work_id":"wrk_68ec8173-3ea7-4526-bfea-020e82eae4c6","block_id":"blk_d7e830de-392a-48a4-910a-05ad2f947709","section_id":"sec_567f828b-b117-4cd2-accf-d721f42215b7","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":61,"end":67,"exact":"エージェント","quote":"設計では、配置配線、タイミングクロージャ、DRC/LVS、IR Drop、EM解析などで大量のログが出る。AIエージェントは、それらを読み、どの制約を調整すべきか、どのブロックが問題か、どの配線層が混雑しているかを分析できる。","quote_start":6,"quote_end":119,"text_sha256":"bc0b860da958bd4b682464779e6e0a4f62dc81a951726f54adb49d9c72b420fa","block_sha256":"bc0b860da958bd4b682464779e6e0a4f62dc81a951726f54adb49d9c72b420fa","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_3621f98d-f1b9-4e97-9c8a-8b259cfd31c3/#blk_d7e830de-392a-48a4-910a-05ad2f947709"},{"id":"occ_74cc7beca37d527b7bc1fb67","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_44c73f5c-647d-467e-8aa2-527eb6352a8a","work_id":"wrk_45578dda-8dcf-452e-909d-3f4c17ea166c","block_id":"blk_c472174a-5985-4b41-ab8b-85522c072a52","section_id":"sec_bc6c0812-ef58-4692-adaf-03b3cfd1739c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":7,"end":13,"exact":"エージェント","quote":"長い会話、AIエージェント、コーディング支援、企業内文書検索が増えると、KVキャッシュは非常に大きくなる。","quote_start":0,"quote_end":53,"text_sha256":"bba2c79f3749c12abf15b3ab7ea64cc3ca566b6bd7e42b563651076f647d91f1","block_sha256":"bba2c79f3749c12abf15b3ab7ea64cc3ca566b6bd7e42b563651076f647d91f1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_44c73f5c-647d-467e-8aa2-527eb6352a8a/#blk_c472174a-5985-4b41-ab8b-85522c072a52"},{"id":"occ_11af537e40249aa395ccc978","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_49a8bdbd-8ee8-4c28-8af4-da1787c47955","work_id":"wrk_cb25df93-468d-4abf-9350-ba1bcf0d3b49","block_id":"blk_276d847a-07d2-445d-9825-4d7ccee3512b","section_id":"sec_8ab76917-d81d-4ff7-9055-f4d59cd6e5e2","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":7,"end":13,"exact":"エージェント","quote":"特に長文推論やエージェントでは、KVキャッシュ、検索インデックス、過去文脈、ツール実行ログなどが巨大化する。  \nこれをすべてHBMに置くのは非現実的である。","quote_start":0,"quote_end":79,"text_sha256":"e70724edac05e51561fb308129c44a374cc613c127bf20bced3896592064be70","block_sha256":"e70724edac05e51561fb308129c44a374cc613c127bf20bced3896592064be70","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_49a8bdbd-8ee8-4c28-8af4-da1787c47955/#blk_276d847a-07d2-445d-9825-4d7ccee3512b"},{"id":"occ_fa23991f746a69d3ec78df0c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_49a8bdbd-8ee8-4c28-8af4-da1787c47955","work_id":"wrk_cb25df93-468d-4abf-9350-ba1bcf0d3b49","block_id":"blk_571c5be6-02c5-4555-803c-3b8879055bca","section_id":"sec_b096ef5b-f66e-4fc5-9e83-e096ae58fc36","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":22,"end":28,"exact":"エージェント","quote":"しかし今後、AI推論、長文コンテキスト、AIエージェント、マルチラックGPUクラスタが拡大すると、これだけでは足りなくなる。","quote_start":0,"quote_end":62,"text_sha256":"8997785c1a4d332bbb9418cb586d2c66e30eb8f8aa29e1ba442e694fb5da1e3e","block_sha256":"8997785c1a4d332bbb9418cb586d2c66e30eb8f8aa29e1ba442e694fb5da1e3e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_49a8bdbd-8ee8-4c28-8af4-da1787c47955/#blk_571c5be6-02c5-4555-803c-3b8879055bca"},{"id":"occ_7181cd529eec73eb7aad53d0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":null,"section_id":null,"layer":"title","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"title","start":2,"end":8,"exact":"エージェント","quote":"AIエージェント時代、広告はどう変わるのか","quote_start":0,"quote_end":21,"text_sha256":"3f9f162bc2fb606da2b216c3ce45f76239c291b028fafe417f36de88b2035b17","block_sha256":null,"offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/"},{"id":"occ_2054138096ee129c7c5f2e22","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_0c573e4e-2cec-4d27-9872-27d61b9522f4","section_id":"sec_d426c90c-06de-465a-8597-0dfe391eff72","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":77,"end":83,"exact":"エージェント","quote":"abet \\> Meta、  \n広告単体の強さは Meta \\> Google \\> Amazon、  \nAIエージェント時代の最終勝者候補は Amazon、  \n規制で一番重いのは Google、次が Meta です。Alphabetは2025年売上が約4,028億ドル、Metaは約2,010億ドル、そのうち広告が約1","quote_start":22,"quote_end":183,"text_sha256":"8a29f3ab684754bfa22993567beffd98f4e6bdc782dcdac4ff7a5a195ad05aad","block_sha256":"8a29f3ab684754bfa22993567beffd98f4e6bdc782dcdac4ff7a5a195ad05aad","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_0c573e4e-2cec-4d27-9872-27d61b9522f4"},{"id":"occ_058d27d479eb8d99f347a78c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_0f5dd91b-0257-4147-b519-5ffea01b2d7f","section_id":"sec_54adf41a-ef49-4776-be45-114092620e1c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":109,"end":115,"exact":"エージェント","quote":"は、広告だけでなくEC手数料、サブスク、AWS、物流、デバイス、そしてRufusやBuy for Meのようなエージェント型購買導線まで持っています。Q4 2025の広告サービス売上は213億ドル、AWSは356億ドル、2026年の設備投資計画は約2,000億ドルです。さらにAmazonはRufusの利用者が2025年に","quote_start":54,"quote_end":215,"text_sha256":"ca5deb42096283c969da0a0ea8b99ca38858dd721401a295406dedc8179ca226","block_sha256":"ca5deb42096283c969da0a0ea8b99ca38858dd721401a295406dedc8179ca226","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_0f5dd91b-0257-4147-b519-5ffea01b2d7f"},{"id":"occ_50e4a2342627d5a5b052a7e0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_14e2ef30-c06d-4f00-bbfb-66aa13cc44a2","section_id":"sec_b90edad7-c79a-4b7c-ba90-c2d20fc46418","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":126,"end":132,"exact":"エージェント","quote":"**一番“AIがそのままお金を落とす世界”で強いのはAmazon。**  \nだから、AI時代の広告・コマース・エージェントを全部まとめて見たときの総合王者候補は、私は**Amazon \\> Google ≒ Meta**で見ます。  \nただしこれは、AIが本当に「比較して終わり」ではなく「実行して終わり」まで進む前提です","quote_start":71,"quote_end":232,"text_sha256":"fa133b1a21be7ab0f3d6d93d5a23762ee2f5ddd12d983e3a0806af814aba443a","block_sha256":"fa133b1a21be7ab0f3d6d93d5a23762ee2f5ddd12d983e3a0806af814aba443a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_14e2ef30-c06d-4f00-bbfb-66aa13cc44a2"},{"id":"occ_392a35fa87408ba3516b4146","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_18d42d91-897d-4789-af7c-d71b43b80e35","section_id":"sec_c3cc2642-eb6a-49f9-914c-c08b6004e595","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":29,"end":35,"exact":"エージェント","quote":"その上で OpenAI の本命は、**AIスーパ―アプリ兼エージェント基盤**になることだと思います。ChatGPT agent はブラウズ、コード実行、コネクタ、外部アプリへの書き込みまで使えますし、OpenAIは Frontier を「AI coworkers」を企","quote_start":0,"quote_end":135,"text_sha256":"9f1a7719a80329dd0b6b816c851b44cbc1b8a6c9fae4fe3e0114b260a2c249f5","block_sha256":"9f1a7719a80329dd0b6b816c851b44cbc1b8a6c9fae4fe3e0114b260a2c249f5","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_18d42d91-897d-4789-af7c-d71b43b80e35"},{"id":"occ_eb1b197e5416664029736a4a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_1b4dca28-59c7-422e-b739-56712378884d","section_id":"sec_2ebe4d71-3bcd-486a-b442-2d251d47849e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":90,"end":96,"exact":"エージェント","quote":"告の売り物そのものを変える**ということだ。  \nこれまでの広告は「人間の注意」を買うビジネスだった。だがAIエージェント時代には、価値の中心が「表示回数」から「比較候補に入ること」「AIに推薦されること」「そのまま実行されること」へ移る。GoogleはAI OverviewsとAI Modeに広告を持ち込みつつ検索を再","quote_start":35,"quote_end":196,"text_sha256":"f9f6ed84255ed50ca78d36b07f2eb4a69d9d6d31b316281ca12d4e79a172767b","block_sha256":"f9f6ed84255ed50ca78d36b07f2eb4a69d9d6d31b316281ca12d4e79a172767b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_1b4dca28-59c7-422e-b739-56712378884d"},{"id":"occ_d0060cd7bcd7efa23017fc90","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_1dc40702-9c0a-44fa-8808-1ed9091ad26b","section_id":"sec_69cb3877-2d91-412f-a0a4-9dc62f999af0","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":74,"end":80,"exact":"エージェント","quote":"場自動化の本丸ではなく、現場UIのオプションに寄っているからです。MetaはV-JEPA 2で、ロボットやAIエージェントが物理世界を理解し、行動結果を予測する世界モデルを進めています。また、処方対応のAIグラスや、AIグラスと連携するMeta AIアプリも出しており、現場作業員の視界支援・遠隔支援・ハンズフリー案内との","quote_start":19,"quote_end":180,"text_sha256":"f5ad4a0e28c553d77ce3cfc00d7c7afece09fdce9fb56deef8df41630926adcf","block_sha256":"f5ad4a0e28c553d77ce3cfc00d7c7afece09fdce9fb56deef8df41630926adcf","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_1dc40702-9c0a-44fa-8808-1ed9091ad26b"},{"id":"occ_32ec034ce4b971354c1a0fb4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_20265f0c-57d9-4913-9bb0-6a27bc53339f","section_id":"sec_dcc6a454-612d-41a2-984a-20116fc98b58","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":7,"end":13,"exact":"エージェント","quote":"要するに、AIエージェント時代に広告は3つに分かれていく。  \n第一に、GoogleやMetaが強い**従来広告のAI高度化**。  \n第二に、AmazonやOpenAIが強い**比較・推薦・購入の一体化**。  \n第三に、A","quote_start":0,"quote_end":113,"text_sha256":"9320db251e0e1391c59ef629f3da0c7356f9880722670455bda9f30e72b12ae2","block_sha256":"9320db251e0e1391c59ef629f3da0c7356f9880722670455bda9f30e72b12ae2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_20265f0c-57d9-4913-9bb0-6a27bc53339f"},{"id":"occ_e8b4eb1e79834aa96c58fcc6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_2dbba0a1-dfe6-474f-9f1f-3b91fd36e56a","section_id":"sec_d9bb3e25-8286-478a-b387-dd0d8f4a6f42","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":151,"end":157,"exact":"エージェント","quote":"BPN買収も、単純なメディア買収というより、分配とナラティブ形成の獲得として読む方が自然だ。OpenAIはAIエージェント時代に、「アプリの一つ」ではなく「AIが動く基盤」になりたいのだと思う。これは推論だが、現行の買収と製品の並びはその方向で非常に一貫している。 ([OpenAI](https://openai.com","quote_start":96,"quote_end":257,"text_sha256":"d61cf4616404ab575f11e15a90d284aa650aa81cb64a103ab47027d3caa40b2a","block_sha256":"d61cf4616404ab575f11e15a90d284aa650aa81cb64a103ab47027d3caa40b2a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_2dbba0a1-dfe6-474f-9f1f-3b91fd36e56a"},{"id":"occ_a78fed6220e8f3e84d680ea9","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_3163938e-ec0b-456c-be86-7faf6e2f638c","section_id":"sec_53336672-02c7-443d-be1e-b8ae4a320b45","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":171,"end":177,"exact":"エージェント","quote":"e IoT Operations、Azure Arc、Defender for IoTを軸に、工場データ、AIエージェント、セキュリティ、現場UIを企業ITの文脈で束ねやすいからです。しかもMicrosoftはSiemens Industrial Copilotを大規模環境向けに拡張し、100社超・12万人超のエンジニア","quote_start":116,"quote_end":277,"text_sha256":"8916cd5938525fbbef5ba6a22c9f1aa866c394c49c70299f5cf0a5d7829e650c","block_sha256":"8916cd5938525fbbef5ba6a22c9f1aa866c394c49c70299f5cf0a5d7829e650c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_3163938e-ec0b-456c-be86-7faf6e2f638c"},{"id":"occ_80619066f9cb5acca544d33b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_352ef57b-5578-4d62-88d8-f67e6a4e8a57","section_id":"sec_2d1cfe16-3d19-4732-a677-4b1cda222385","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":53,"end":59,"exact":"エージェント","quote":"**短期で一番やりやすい**: **AIクリエイター型**  \n**長期で一番大きくなりやすい**: **エージェント実行型**  \n**補助収益としては有力だが、最も慎重に扱うべき**: **広告型**","quote_start":0,"quote_end":102,"text_sha256":"19462e84cc65c65014407d02b8a18bac275668e08d7f2e11c3520537b1ce4417","block_sha256":"19462e84cc65c65014407d02b8a18bac275668e08d7f2e11c3520537b1ce4417","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_352ef57b-5578-4d62-88d8-f67e6a4e8a57"},{"id":"occ_dc9bdd2308a6b2136227bc9a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_36aa04db-f36c-4060-8f8a-64c73058b8b6","section_id":"sec_b132185e-c958-4c03-b68c-46b8c3564ff5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":16,"end":22,"exact":"エージェント","quote":"## \\2. 5社それぞれのAIエージェント時代の戦略","quote_start":0,"quote_end":27,"text_sha256":"40d76f1bea8fdec3438c961c48326930afb53b588bad35bf21969e971a10eb06","block_sha256":"40d76f1bea8fdec3438c961c48326930afb53b588bad35bf21969e971a10eb06","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_36aa04db-f36c-4060-8f8a-64c73058b8b6"},{"id":"occ_7eeb6c68112a49eaed985240","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_370f4927-0a14-4b8a-8b23-705aebe340e3","section_id":"sec_e622bf65-5638-4862-989e-723190c8671f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":201,"end":207,"exact":"エージェント","quote":"が自然だと見ます。たとえば、生成物の共有、ランキング、コメント、クリエイター発見、ライブ番組や会話の埋め込み、エージェントの実行ログ共有のような機能です。これならChatGPTの既存ユーザーベースと相性がよく、リアルタイムの利用データも得られます。Reuters系報道が指摘した「リアルタイムデータ獲得」の狙いとも整合しま","quote_start":146,"quote_end":307,"text_sha256":"e2fca49aa049b26e7a7712c169f07877cde35c915bed21ac6300bda86d03624a","block_sha256":"e2fca49aa049b26e7a7712c169f07877cde35c915bed21ac6300bda86d03624a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_370f4927-0a14-4b8a-8b23-705aebe340e3"},{"id":"occ_3a5031232f3ee44bd09968b1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_3a58ffbb-762e-4ce6-a315-eb8ddc918616","section_id":"sec_75327fb7-db93-4b8c-b354-75a1d2f02b06","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":196,"end":202,"exact":"エージェント","quote":"です。Buy for Meは、Amazonで扱っていない商品でもブランドサイトから購入代行する仕組みで、まさにエージェント的コマースです。GoogleもShopping GraphやAI Mode内のショッピング体験、エージェント的チェックアウトに進んでいますが、自前で物流や大規模小売を持っていません。Metaは発見と広","quote_start":141,"quote_end":302,"text_sha256":"f7a294adfc8ac7e6e089ef53386dd64f27f76ae52f06f6832df152287e824dd3","block_sha256":"f7a294adfc8ac7e6e089ef53386dd64f27f76ae52f06f6832df152287e824dd3","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_3a58ffbb-762e-4ce6-a315-eb8ddc918616"},{"id":"occ_288b3857358e8388c5e38d22","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_40185892-e766-49ea-8a70-ca7ae8b45d76","section_id":"sec_7cbb8d1a-9784-407d-adb8-8fb24571d930","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":61,"end":67,"exact":"エージェント","quote":"うと、**広告の稼ぐ力はMeta、需要の入口はGoogle、購買の刈り取りはAmazon**です。さらに、AIエージェントが「探す」だけでなく「買う」「予約する」まで担うほど、最終的な優位は**Amazon寄り**になります。逆に、広告が今後も最大の収益源であり続けるなら、**Metaが最も強い**です。Googleはそ","quote_start":6,"quote_end":167,"text_sha256":"1f937a9e148ecc82cd70442c45aa65d6e6132b9220587c6e34545450d3c9ebc6","block_sha256":"1f937a9e148ecc82cd70442c45aa65d6e6132b9220587c6e34545450d3c9ebc6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_40185892-e766-49ea-8a70-ca7ae8b45d76"},{"id":"occ_5c09e6a98631689ac1816ce1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_442e9cd4-2a6b-451a-ac26-4b951067859e","section_id":"sec_038d7171-ea0d-482a-855f-9bd83b4bf32e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"## AIエージェント時代、広告はどう変わるのか","quote_start":0,"quote_end":24,"text_sha256":"e7946eced4d56b2170db49ccd90b713b5a78f763e2553132f16019159a3bf8dc","block_sha256":"e7946eced4d56b2170db49ccd90b713b5a78f763e2553132f16019159a3bf8dc","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_442e9cd4-2a6b-451a-ac26-4b951067859e"},{"id":"occ_93ef4039e1401d52e2e2ef90","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_474c0c07-055b-4732-a128-6d691b3191d1","section_id":"sec_d9bb3e25-8286-478a-b387-dd0d8f4a6f42","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":152,"end":158,"exact":"エージェント","quote":"OpenAIは100万超のビジネス顧客、週8億人超のChatGPT利用者を持ち、Frontierで企業向けAIエージェント基盤を打ち出している。加えて、ショッピングリサーチやInstant Checkout、Deutsche Telekomとの提携、ioの統合、Broadcomとの10GW級アクセラレータ協業まで進めてい","quote_start":97,"quote_end":258,"text_sha256":"25eb0841b3aba1bb2d135c94e1631ae2e9a2bd9ce9d13ac9f6f70618be2a2c0e","block_sha256":"25eb0841b3aba1bb2d135c94e1631ae2e9a2bd9ce9d13ac9f6f70618be2a2c0e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_474c0c07-055b-4732-a128-6d691b3191d1"},{"id":"occ_077d9b0063ad89ce86f0ed58","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_4e495fcb-78e9-469d-aaca-55cc22f95928","section_id":"sec_55e21577-9f98-4201-8837-d3e5e70e4f2a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":160,"end":166,"exact":"エージェント","quote":"売上は213億ドル、AWS売上は355.8億ドルで、広告・クラウド・ECが同時に伸びている。AmazonはAIエージェント時代に、**「比較して終わり」ではなく「買って終わり」まで持っていける数少ない会社**だ。 ([Amazon](https://ir.aboutamazon.com/news-release/news","quote_start":105,"quote_end":266,"text_sha256":"db0e4770ee7d5deea719ac11504f56727897384e762f6266132ee3c843ae64d6","block_sha256":"db0e4770ee7d5deea719ac11504f56727897384e762f6266132ee3c843ae64d6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_4e495fcb-78e9-469d-aaca-55cc22f95928"},{"id":"occ_fa0c722f0755f14086194807","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_52bfbd09-4ed0-459b-8a82-dd95645f9cd5","section_id":"sec_cc592589-fc27-4170-a59c-dc0ddc406224","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":70,"end":76,"exact":"エージェント","quote":"つではない。  \n広告市場だけを見るならMetaが最強候補だ。検索と意図の入口ではGoogleが依然強い。AIエージェントが“買う”ところまで進むならAmazonが有利になる。消費者と企業の両方で知能OSを作れればOpenAIの上振れは大きい。企業の深部と標準化で勝つならAnthropicは非常に強い。  \nただ、私の最","quote_start":15,"quote_end":176,"text_sha256":"493453ef71665aa8f71e0f03f455877d4e0b9be0e9e48ab4198eefe13eb1329f","block_sha256":"493453ef71665aa8f71e0f03f455877d4e0b9be0e9e48ab4198eefe13eb1329f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_52bfbd09-4ed0-459b-8a82-dd95645f9cd5"},{"id":"occ_45fb43c8144df0cd6bec7b8a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_58a668e3-2c78-4aa5-8e19-d55af53d949b","section_id":"sec_84c55d3b-a8b8-41fa-8312-687a0c00efab","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":63,"end":69,"exact":"エージェント","quote":"  \nこの買収パターンは強い反面、**戦線が広すぎる**リスクがあります。ハード、健康、金融、開発ツール、企業エージェント、メディアまで広げると、投資家からは「焦点が散っている」と見られやすいです。実際、Reutersは一部投資家がOpenAIの戦略変更や広がりに懸念を持っていると報じています。なので、このM&A戦略は野","quote_start":8,"quote_end":169,"text_sha256":"05cbd22bf5f17572cbdb62106cf7baedee611bf8053b6255bfe36ba998b4514b","block_sha256":"05cbd22bf5f17572cbdb62106cf7baedee611bf8053b6255bfe36ba998b4514b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_58a668e3-2c78-4aa5-8e19-d55af53d949b"},{"id":"occ_381b440a4cb5de45e2a6950f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_68cb10a3-a3c7-4a73-a1c4-d4d2d3963a18","section_id":"sec_7cbb8d1a-9784-407d-adb8-8fb24571d930","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":43,"end":49,"exact":"エージェント","quote":"## Google vs Meta vs Amazon を「AI時代の広告・コマース・エージェント支配力」で投資家目線まで落として比較","quote_start":0,"quote_end":67,"text_sha256":"77d70e8319ad6c0d2ff96dfbd7198fe04a1a8cef901e8a3edbf07980e05ab731","block_sha256":"77d70e8319ad6c0d2ff96dfbd7198fe04a1a8cef901e8a3edbf07980e05ab731","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_68cb10a3-a3c7-4a73-a1c4-d4d2d3963a18"},{"id":"occ_581abd19614e3b9bcd4adb37","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_6bb8a800-4fc3-4916-bf99-2dcc84e5c6ec","section_id":"sec_84c55d3b-a8b8-41fa-8312-687a0c00efab","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":13,"end":19,"exact":"エージェント","quote":"2つ目の軸は、**企業向けエージェント基盤の完成度を買収で一気に引き上げている**ことです。  \nOpenAIは2026年にFrontierを「AI coworkers を作って運用する企業プラットフォーム」として打ち出しました。Prom","quote_start":0,"quote_end":119,"text_sha256":"be3988d9623bc696115b126ed3bcfe1e9ff31d81797cd86c8cf883e681d287b7","block_sha256":"be3988d9623bc696115b126ed3bcfe1e9ff31d81797cd86c8cf883e681d287b7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_6bb8a800-4fc3-4916-bf99-2dcc84e5c6ec"},{"id":"occ_c9a223a4279832a5e1e0c81e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_6bc5b137-86c8-4e3a-990f-98f7970ce59c","section_id":"sec_f6c9bcb8-90b7-4d6f-8771-10eb96aea863","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":61,"end":67,"exact":"エージェント","quote":"かりやすいモデルです。  \nフィード内の **Sponsored投稿、生成コンテンツ間広告、求人広告、アプリ/エージェントのプロモ枠、商材の推薦枠** を売る形です。もしOpenAIが巨大なフィード面を持てば、MetaやXに近いモデルが成立します。収益ポテンシャルは大きいですが、**「AIが中立に答えてくれる」という期待","quote_start":6,"quote_end":167,"text_sha256":"66dbad21dc2d342b4066c512a8a9801d807b9788c41b395dad8c93f17ea72016","block_sha256":"66dbad21dc2d342b4066c512a8a9801d807b9788c41b395dad8c93f17ea72016","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_6bc5b137-86c8-4e3a-990f-98f7970ce59c"},{"id":"occ_11bc737a8a461d84c27e5dff","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_73adb40f-78a5-4fbc-b27d-963cd74a8c18","section_id":"sec_c3cc2642-eb6a-49f9-914c-c08b6004e595","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":173,"end":179,"exact":"エージェント","quote":"に自社ECを持たなくても、**“購買意図の入口”と“購買判断の設計”を握ることで、送客、成約手数料、優先表示、エージェント実行料**のような形で収益化しやすいです。つまり店を持たずに、**店へ流す前の意思決定面**を取りに行く戦い方です。 ([OpenAI](https://openai.com/index/chatgp","quote_start":118,"quote_end":279,"text_sha256":"db661aaf02d813ee531bfffd2e5713768124a727b72c10f362628bdcca6b28fb","block_sha256":"db661aaf02d813ee531bfffd2e5713768124a727b72c10f362628bdcca6b28fb","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_73adb40f-78a5-4fbc-b27d-963cd74a8c18"},{"id":"occ_03e766ea2e86014f0704ea9b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_7b93df9f-0444-4ce8-9aa2-65f4acb3ae65","section_id":"sec_e622bf65-5638-4862-989e-723190c8671f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":143,"end":149,"exact":"エージェント","quote":"す。Claude for Enterpriseは社内知識接続を前面に出し、Claude Codeは開発・実装のエージェントとして打ち出され、MCPは外部システム接続の標準化として推進されています。つまりAnthropicの勝ち筋は、**公共の会話空間を持つこと**ではなく、**企業や開発環境の深部に入り込むこと**です。","quote_start":88,"quote_end":249,"text_sha256":"c06f6316df43d8d904ea88268f1784ea3235c22575224482b3dc17771426c5b9","block_sha256":"c06f6316df43d8d904ea88268f1784ea3235c22575224482b3dc17771426c5b9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_7b93df9f-0444-4ce8-9aa2-65f4acb3ae65"},{"id":"occ_07ff66391d74596a94d4ac19","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_7cd51d95-ef40-474a-be30-c1c2d0da3d9f","section_id":"sec_88f7c3f1-55ad-4103-9be3-b70eb709768c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"**AIエージェントが実際に買う未来**では、Amazonが最有力です。おすすめより先にある、決済、在庫、配送、返品、会員、ブランド接点まで持っているためです。これは3社の中でAmazonだけが圧倒的に厚い層です。 ([","quote_start":0,"quote_end":110,"text_sha256":"1fd7dcb81fa4b72b7bc879174ba29b159a1a742fa544fed3901782791a651e01","block_sha256":"1fd7dcb81fa4b72b7bc879174ba29b159a1a742fa544fed3901782791a651e01","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_7cd51d95-ef40-474a-be30-c1c2d0da3d9f"},{"id":"occ_b174415d2741221f4d22f1f4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_7dfbff59-7df3-4b7e-8c6b-5c78f5b8b938","section_id":"sec_84c55d3b-a8b8-41fa-8312-687a0c00efab","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":223,"end":229,"exact":"エージェント","quote":"N、Hiro で **健康、セキュリティ、開発者、メディア、金融** に踏み込み、Frontier という企業エージェント基盤まで打ち出しました。これは、OpenAIが研究成果を外へ出す会社から、**用途別の実用システムを束ねる会社** に移っていることを示しています。 ([TechCrunch](https://tec","quote_start":168,"quote_end":329,"text_sha256":"46502922779cf6a8cba0abb71388c03f596151b7ffc667897f8517affab8e2e0","block_sha256":"46502922779cf6a8cba0abb71388c03f596151b7ffc667897f8517affab8e2e0","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_7dfbff59-7df3-4b7e-8c6b-5c78f5b8b938"},{"id":"occ_90e599a51f44f0efb885301c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_857a6828-ab4f-4453-a6c3-62e34eac2abc","section_id":"sec_dcc6a454-612d-41a2-984a-20116fc98b58","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":9,"end":15,"exact":"エージェント","quote":"## \\1. AIエージェント時代、広告はどう変質するのか","quote_start":0,"quote_end":29,"text_sha256":"18109f5cdd8bc211014f7a374dcd88c8e7d83cb1b6ff1916cd61fdea673eb85a","block_sha256":"18109f5cdd8bc211014f7a374dcd88c8e7d83cb1b6ff1916cd61fdea673eb85a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_857a6828-ab4f-4453-a6c3-62e34eac2abc"},{"id":"occ_db9016d374e79a7809c9cf3b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_8c7ab1ef-39be-41df-9172-074f2923a83e","section_id":"sec_42bb6924-ebec-4ab2-96d3-0b005ede4358","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":31,"end":37,"exact":"エージェント","quote":"**OpenAI**の2030年の主な収益源は、  \n**企業エージェント基盤**, **消費者サブスク**, **AIデバイス**, **開発者/API**, そしてその上に乗る**実行課金**になりやすいです。  \nOpenAIは Frontier を「enterpri","quote_start":0,"quote_end":137,"text_sha256":"91ccfaa36404c1457875bae7bea8f578af66198a9d6356be83086771711446f7","block_sha256":"91ccfaa36404c1457875bae7bea8f578af66198a9d6356be83086771711446f7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_8c7ab1ef-39be-41df-9172-074f2923a83e"},{"id":"occ_894e089007b8847390af4500","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_90837517-7c3a-414b-90c8-61b7d1b18a36","section_id":"sec_c3cc2642-eb6a-49f9-914c-c08b6004e595","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":233,"end":239,"exact":"エージェント","quote":"のデバイスの上で動くことで、**所有ではなく“横断”で勝つ**戦い方です。OpenAI のほうが消費者・商流・エージェント実行に近く、Anthropic のほうが企業・コーディング・中立マルチクラウドに近いです。 ([OpenAI](https://openai.com/index/1-million-businesse","quote_start":178,"quote_end":339,"text_sha256":"e04a103d6dded47d8f3658b72938f35ac62ea3f627c9a5143fc641eba33047e5","block_sha256":"e04a103d6dded47d8f3658b72938f35ac62ea3f627c9a5143fc641eba33047e5","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_90837517-7c3a-414b-90c8-61b7d1b18a36"},{"id":"occ_43d6fbafee3511f8979257ee","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_9c60dee0-1ddc-4bdf-b16c-33b52c92409c","section_id":"sec_dda1f5fe-f2ef-41b0-a9c7-11b999762712","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":235,"end":242,"exact":"agentic","quote":"ロボットが75%の注文に関与していると説明しています。さらに需要予測、配送精度、自然言語でロボットに指示する agentic AI まで物流に入れ始めています。 ([Amazon News](https://www.aboutamazon.com/news/operations/amazon-vulcan-robot-pi","quote_start":180,"quote_end":342,"text_sha256":"cd64d785b304d0c5052cbc36a940f67e62f910828fc878520caa1302d1748c20","block_sha256":"cd64d785b304d0c5052cbc36a940f67e62f910828fc878520caa1302d1748c20","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_9c60dee0-1ddc-4bdf-b16c-33b52c92409c"},{"id":"occ_4580b04e1bf2696d83cab473","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_a706542e-5adb-47dc-8009-806b992f083b","section_id":"sec_1b1f7803-f580-4d17-bfa1-66995619a7fc","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":49,"end":55,"exact":"エージェント","quote":"**一番バランスが良いのは Amazon** です。  \n広告、EC、AWS、サブスク、物流、AIエージェントが連結しやすく、将来の「AIがそのまま買う世界」に最も自然に接続できます。設備投資は大きいですが、それを回収できる受け皿が多いです。 ([Amazon](https://ir.aboutamazon.","quote_start":0,"quote_end":155,"text_sha256":"3b3672bfda0acd2d286ef32a44b7c897d47e9b937774e170a0e49ea4672ab648","block_sha256":"3b3672bfda0acd2d286ef32a44b7c897d47e9b937774e170a0e49ea4672ab648","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_a706542e-5adb-47dc-8009-806b992f083b"},{"id":"occ_a758996de23f050ecc5435f8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_a997bc05-de36-4060-8653-3c33883f9b1a","section_id":"sec_55e21577-9f98-4201-8837-d3e5e70e4f2a","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":92,"end":98,"exact":"エージェント","quote":"が強く、Metaは広告最適化が強い。だがAmazonは、カート、決済、配送、返品、会員基盤まで持っている。AIエージェント時代の価値が「誰が最終行動を完了させるか」に寄るほど、Amazonの相対優位は大きくなりやすい。私の見立てでは、AIエージェント時代の長期勝者候補はAmazonだ。 ([Amazon News](ht","quote_start":37,"quote_end":198,"text_sha256":"bbaa13fb06051dad005630ff8ab4b30a6b64d2a5170d58c4a3dd5cb5d64a6ca1","block_sha256":"bbaa13fb06051dad005630ff8ab4b30a6b64d2a5170d58c4a3dd5cb5d64a6ca1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_a997bc05-de36-4060-8653-3c33883f9b1a"},{"id":"occ_b2d24e377efcd0a88293d9de","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_aabd997f-3cfc-4122-946f-8c55d2ba6be2","section_id":"sec_69cb3877-2d91-412f-a0a4-9dc62f999af0","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":300,"end":306,"exact":"エージェント","quote":"正で済むとされています。つまりMicrosoftは、現場ロボそのものより、**工場データ、デジタルツイン、AIエージェント、OT/IT接続、企業導入**を束ねる立場が強いです。 ([Microsoft Learn](https://learn.microsoft.com/en-us/fabric/real-time-in","quote_start":245,"quote_end":406,"text_sha256":"048350780e5dd533913be6fd690688c8d0b54018a45dcf5e7f71fd606d92a1c4","block_sha256":"048350780e5dd533913be6fd690688c8d0b54018a45dcf5e7f71fd606d92a1c4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_aabd997f-3cfc-4122-946f-8c55d2ba6be2"},{"id":"occ_b596c94556d7e77d8f4b4dd3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_b4a70d63-9bd8-46b2-84a4-de6b225f8ebe","section_id":"sec_a7e79f77-41c0-4e39-a424-a879561f74a9","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":86,"end":92,"exact":"エージェント","quote":"ate + 広告 + AWS**のままですが、2030年に差がつくのは**物流自動化による利益率改善**と**エージェント経由の購買実行料**です。Q4 2025時点でAWSは356億ドル、広告サービスは213億ドル、Rufusは3億人超が利用し、Buy for Meまで実装済みです。さらにAmazonは100万台目のロ","quote_start":31,"quote_end":192,"text_sha256":"a4dda3579ddd34cbc830a82c4ec868cc8dd04f75091c64a409c9eb4494a24215","block_sha256":"a4dda3579ddd34cbc830a82c4ec868cc8dd04f75091c64a409c9eb4494a24215","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_b4a70d63-9bd8-46b2-84a4-de6b225f8ebe"},{"id":"occ_21d13c94c7bd1ad6bf910e6a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_b750a6e0-4c34-4d9a-86f4-0caad2e393a8","section_id":"sec_672889a9-aa32-45ab-877e-a199d9424686","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":8,"end":14,"exact":"エージェント","quote":"### \\3. エージェント実行型","quote_start":0,"quote_end":17,"text_sha256":"8eaee212041a6e79edc504908a2675fe73dae64eeb8f65be67b2c0024f5f3298","block_sha256":"8eaee212041a6e79edc504908a2675fe73dae64eeb8f65be67b2c0024f5f3298","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_b750a6e0-4c34-4d9a-86f4-0caad2e393a8"},{"id":"occ_bc2caa731ac4b9464aa195d1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_baf52f73-67e2-4d68-a568-1c6069e1314e","section_id":"sec_53336672-02c7-443d-be1e-b8ae4a320b45","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":145,"end":151,"exact":"エージェント","quote":"delと、Ray-Ban/Oakley系のAIグラス拡張です。Meta自身はV-JEPA 2を、ロボットやAIエージェントが物理世界を理解し行動結果を予測する基盤として説明しており、AIグラスも着実に製品化を進めています。ですが、少なくとも現時点の公開戦略では、MetaはMicrosoft/Amazon/Googleのよ","quote_start":90,"quote_end":251,"text_sha256":"a17ce62712d901c8112908b5daff7e9a1457e72df046c0082d022dc1ec66372f","block_sha256":"a17ce62712d901c8112908b5daff7e9a1457e72df046c0082d022dc1ec66372f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_baf52f73-67e2-4d68-a568-1c6069e1314e"},{"id":"occ_f6e9eccb86a66605caaa24df","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_bbecdb21-4b2d-4e7f-bd42-5d501d658fc4","section_id":"sec_6ee1426c-3416-455e-a2a8-08d2da9487ca","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":101,"end":108,"exact":"agentic","quote":"増分の主役は**Cloud + Gemini Enterprise/Workspace + shopping/agentic commerce + Waymo/XRのオプション価値**になりやすいです。いまの土台でもQ4 2025のSearch & Other広告は631億ドル、YouTube広告は114億ドル、Cloud","quote_start":46,"quote_end":208,"text_sha256":"1a637ede214b97606b70568436cb85889dfaec987c981bbb380bc0463d857b28","block_sha256":"1a637ede214b97606b70568436cb85889dfaec987c981bbb380bc0463d857b28","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_bbecdb21-4b2d-4e7f-bd42-5d501d658fc4"},{"id":"occ_f5aa88f808fff6f9926c9c04","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_c73a6dcf-0d30-4c88-964c-ba51756c3882","section_id":"sec_3dc3ab31-c20f-4352-bdf0-8ec71028e191","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":50,"end":56,"exact":"エージェント","quote":"これはかなり OpenAI 向きです。  \nOpenAIのSNSが、画像・動画・音声・文章・アプリ・エージェントの“作品共有面”になるなら、  \n**有料サブスク、投げ銭、クリエイター課金、テンプレ/プロンプト販売、エージェント販売、収益分配、ブランド案件仲介**  \nで回せます。  \n特に、ソーシャルフィード","quote_start":0,"quote_end":156,"text_sha256":"172e717d0238ce327e7de266648ff7362e5a8d90baec6f358cbdb5416a437c56","block_sha256":"172e717d0238ce327e7de266648ff7362e5a8d90baec6f358cbdb5416a437c56","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_c73a6dcf-0d30-4c88-964c-ba51756c3882"},{"id":"occ_8c9f80d0b3a6731c12db6a3b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_c74e2da5-0e53-447c-8462-d4e66e1de4ef","section_id":"sec_428862d0-9c42-424e-afa3-a2ce0955d840","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェント時代の次に来るのは、AIが画面の中ではなく、**身体・端末・ロボット・物流・街**に入り込むフィジカルAI時代だ。  \nこの段階に入ると、5社の違いはさらに鮮明になる。競争軸は「どのモデルが賢いか」か","quote_start":0,"quote_end":108,"text_sha256":"2330fc1d468f147e331e553135b34b65b51bfcb29a6a6a7d6a4840c6b67af9da","block_sha256":"2330fc1d468f147e331e553135b34b65b51bfcb29a6a6a7d6a4840c6b67af9da","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_c74e2da5-0e53-447c-8462-d4e66e1de4ef"},{"id":"occ_fe99d9eb3d1ace49c3e1fa2e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_caaee949-9620-4fe8-9ca1-1fd6a0828ebe","section_id":"sec_dcc6a454-612d-41a2-984a-20116fc98b58","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェント時代の広告で一番大きい変化は、**広告が“見るものから選ばれる条件になること**だ。  \nGoogleはAI Overviewsをデスクトップへ拡大し、AI Modeにも広告を入れる方針を明示している","quote_start":0,"quote_end":108,"text_sha256":"6f829526d2cfb9c4cccf62ca2f4874b5fe59b00f392055976dba6f7c2787dc44","block_sha256":"6f829526d2cfb9c4cccf62ca2f4874b5fe59b00f392055976dba6f7c2787dc44","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_caaee949-9620-4fe8-9ca1-1fd6a0828ebe"},{"id":"occ_70085e6af534544e16ce275a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_d1f1a3af-46ce-4e40-8a73-06d5884a7d08","section_id":"sec_672889a9-aa32-45ab-877e-a199d9424686","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":127,"end":133,"exact":"エージェント","quote":"なります。  \nたとえば、  \nレストラン投稿から予約、  \n旅行投稿から見積もり、  \n仕事ノウハウ投稿からエージェント雇用、  \n金融/健康の投稿から専門ワークフロー起動、  \nというふうに、投稿の先で AI が代わりに動く構造です。","quote_start":72,"quote_end":191,"text_sha256":"b52631d8e01da88a94207f4f21358fc05cf4726c462a61943dd88fcf510406df","block_sha256":"b52631d8e01da88a94207f4f21358fc05cf4726c462a61943dd88fcf510406df","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_d1f1a3af-46ce-4e40-8a73-06d5884a7d08"},{"id":"occ_9462a1f84a6a101404a51754","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_d995abc1-610c-4a21-9626-d312c6c279c3","section_id":"sec_dcc6a454-612d-41a2-984a-20116fc98b58","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":164,"end":171,"exact":"Agentic","quote":"商品結果とは別」と明記している。一方で、OpenAIはショッピングリサーチ、Instant Checkout、Agentic Commerce Protocolを展開し、ユーザーの比較・意思決定・購入をChatGPT内に取り込もうとしている。つまりOpenAIは、少なくとも現時点では**広告よりも“実行課金”と“商流課金”","quote_start":109,"quote_end":271,"text_sha256":"6cbf758fdb6f0e8ab30f47e92fe192fc8c001d537f16f95dae375409a463ccd7","block_sha256":"6cbf758fdb6f0e8ab30f47e92fe192fc8c001d537f16f95dae375409a463ccd7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_d995abc1-610c-4a21-9626-d312c6c279c3"},{"id":"occ_65ba63ae3fd3f6b3956e9124","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_e45952f7-3604-47fc-8b2c-04502a02fd41","section_id":"sec_33c5f114-1abb-4cb5-a182-d95c3da974a1","layer":"body","character_id":null,"count":2,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":171,"end":178,"exact":"agentic","quote":"している。広告はAI OverviewsやAI Modeの中に入り、Shopping Graphは価格比較からagentic checkoutへ広がる。さらにGoogle Ads自体にもagentic capabilitiesを入れ、広告主側の運用もAI化している。Googleの狙いは、検索市場を守ることではなく、**意図","quote_start":116,"quote_end":278,"text_sha256":"78a8cce37e692306b16bca470bae319cd2bab5f26e488a3fc10b15f9a0dbee38","block_sha256":"78a8cce37e692306b16bca470bae319cd2bab5f26e488a3fc10b15f9a0dbee38","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_e45952f7-3604-47fc-8b2c-04502a02fd41"},{"id":"occ_266fab017e953c4a19a3ed88","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_eb6a36e8-994b-41c7-96e6-8a1b8b04a9d4","section_id":"sec_aeae549f-8d31-4497-8299-1590606c4687","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":85,"end":91,"exact":"エージェント","quote":"zon**  \nAmazonを一番上昇余地が高い位置に置きます。理由は、AWS、広告、Prime、物流、そしてエージェント型コマースがまだ完全に一体化し切っておらず、**今後の収益化の“重ね掛け”余地**が大きいからです。広告事業もQ4で213億ドルまで来ていますが、MetaやGoogleほど「広告一本足」ではありません","quote_start":30,"quote_end":191,"text_sha256":"a00d23dfe8a23d95407920259f690546c0287d610c772e8beab6ca2d938cd237","block_sha256":"a00d23dfe8a23d95407920259f690546c0287d610c772e8beab6ca2d938cd237","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_eb6a36e8-994b-41c7-96e6-8a1b8b04a9d4"},{"id":"occ_d5d9d52610b758b13593badf","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_ec8a9fb4-0714-4c78-b9af-086b92eeefb5","section_id":"sec_75327fb7-db93-4b8c-b354-75a1d2f02b06","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"**AIエージェント時代の最終優位**  \nAmazon 9.5 / Google 8.5 / Meta 6  \nAIが「おすすめ」だけでなく「代わりに買う」「値下がりを待つ」「予約する」まで進むと、Amazonの優位が","quote_start":0,"quote_end":110,"text_sha256":"f825297c9f1cb40f6e8a8a19de5ce058feaa9febaa69bab8971f6f3bf8fc8659","block_sha256":"f825297c9f1cb40f6e8a8a19de5ce058feaa9febaa69bab8971f6f3bf8fc8659","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_ec8a9fb4-0714-4c78-b9af-086b92eeefb5"},{"id":"occ_a01f489797419f1766566036","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_f9259a27-3981-45e8-8d90-675e0acb5178","section_id":"sec_0235b448-6feb-4199-a7a0-7bbd867b0d61","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":124,"end":130,"exact":"エージェント","quote":"ボットとphysical AIを既存オペレーションへ深く埋め込み、OpenAIが専用ハードとエンタープライズ・エージェント基盤を作り、Anthropicが Claude Code と MCP で企業システム接続の標準化を進めていることからの整理です。 ([About Facebook](https://about.fb.","quote_start":69,"quote_end":230,"text_sha256":"50eb5dce8fbf9f702e70bdb8b12ef05ee5a052bd90d8b4b5ebb886fe2f56ae29","block_sha256":"50eb5dce8fbf9f702e70bdb8b12ef05ee5a052bd90d8b4b5ebb886fe2f56ae29","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_f9259a27-3981-45e8-8d90-675e0acb5178"},{"id":"occ_083ca625ba4fd9836c842a50","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_fe22d578-ad70-47cc-a030-847b784b0215","section_id":"sec_ed58dc86-33a9-495f-beaf-1e10c590bea8","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":109,"end":115,"exact":"エージェント","quote":"hropic は「爆発力はやや劣るが、企業浸透で積み上がりやすい」**  \nです。  \nOpenAIは、端末・エージェント・企業基盤・インフラの全部がつながれば、2030年に**フィジカルAI時代の消費者フロント企業**になれます。Anthropicは、派手なハードはなくても、Enterprise・Claude Code","quote_start":54,"quote_end":215,"text_sha256":"e117e049e5eabf26915a182456cc0cab9d33ea4b749927cb12acbb92b0ace937","block_sha256":"e117e049e5eabf26915a182456cc0cab9d33ea4b749927cb12acbb92b0ace937","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_fe22d578-ad70-47cc-a030-847b784b0215"},{"id":"occ_57cea19eec247e7978bdf69c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c","block_id":"blk_feb032aa-453a-45ee-a2ef-24bf7beac671","section_id":"sec_b6a2880e-f19f-4d1c-8f48-ee8863f0e30d","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":18,"end":24,"exact":"エージェント","quote":"第3段階は、**人間向け広告からAIエージェント向け販促への移行**です。ユーザーが「最安値で買って」「条件の良いホテルを取って」とAIに任せるほど、広告は“派手な訴求”よりも、在庫、価格、配送、信頼性、返品条件、レビュー要約のような**機械可読な","quote_start":0,"quote_end":124,"text_sha256":"e7f13f24a37c1cd19e506464e4a5d72fc8d5ea73272465acbf530e29d6121265","block_sha256":"e7f13f24a37c1cd19e506464e4a5d72fc8d5ea73272465acbf530e29d6121265","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6/#blk_feb032aa-453a-45ee-a2ef-24bf7beac671"},{"id":"occ_40b1aaa33dfd1bb0150ef630","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_53605de4-06cc-4e0b-9bc9-21f393408c0a","work_id":"wrk_006833f6-f225-48d3-8754-46fe28d6bd8e","block_id":"blk_2f04f916-dc49-43bd-901e-77dd4c856d52","section_id":"sec_7c91db57-1adf-4ae2-ad6f-5ced1a9ebb9b","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":38,"end":44,"exact":"エージェント","quote":"AI OS化、オンデバイスAI、AI PC、スマホ内AI、ローカル推論、個人エージェント。  \nそれらが普及していくと、高性能な端末を持つ人と、そうでない人の間に、使えるAI機能の差が生まれていくのではないかと書いた。","quote_start":0,"quote_end":109,"text_sha256":"d08c7c82e1ec05889884bfc93b80f1af4d549d44d772b2e20083b0d1e944acec","block_sha256":"d08c7c82e1ec05889884bfc93b80f1af4d549d44d772b2e20083b0d1e944acec","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_53605de4-06cc-4e0b-9bc9-21f393408c0a/#blk_2f04f916-dc49-43bd-901e-77dd4c856d52"},{"id":"occ_71328541c4f82a6af6979a5b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_54b154f3-ea9f-45e9-b38b-a16c54ca3e2b","work_id":"wrk_52a8c79a-5c56-4f34-a515-1b1dcb4c0477","block_id":"blk_11feb81b-dafa-4135-80ff-19ca42e67f50","section_id":"sec_9bd26956-bf91-4de5-bd2e-7de08a23629f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":25,"end":31,"exact":"エージェント","quote":"GPUクラウドでも、司令塔はCPUに依存する。特にエージェントAIや長文推論では、GPUだけでは処理は完結しない。","quote_start":0,"quote_end":57,"text_sha256":"35a4504ac02e67b998da45370959131695f75d9c7681cad63c61358a95e50177","block_sha256":"35a4504ac02e67b998da45370959131695f75d9c7681cad63c61358a95e50177","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_54b154f3-ea9f-45e9-b38b-a16c54ca3e2b/#blk_11feb81b-dafa-4135-80ff-19ca42e67f50"},{"id":"occ_c9e9e10816f9f23c844507ab","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_54b154f3-ea9f-45e9-b38b-a16c54ca3e2b","work_id":"wrk_52a8c79a-5c56-4f34-a515-1b1dcb4c0477","block_id":"blk_a372947e-9933-4e39-b923-e2a18560a702","section_id":"sec_0a034421-f0b7-473f-8b0c-bd32853319f2","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":117,"end":123,"exact":"エージェント","quote":"M推論向けに設計されたカスタムAIチップとして発表し、OpenAIのChatGPT、Codex、API、将来のエージェント製品のワークロードから得た知見をもとに、カーネル、メモリ移動、ネットワーク、サービングパターンを最適化したと説明している。","quote_start":62,"quote_end":184,"text_sha256":"6166c12ea872caebdd4f3981f503f17cf06043646b6192d0698a6c56ffdabc3a","block_sha256":"6166c12ea872caebdd4f3981f503f17cf06043646b6192d0698a6c56ffdabc3a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_54b154f3-ea9f-45e9-b38b-a16c54ca3e2b/#blk_a372947e-9933-4e39-b923-e2a18560a702"},{"id":"occ_42f2ad8b3271844754ba2620","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_54b154f3-ea9f-45e9-b38b-a16c54ca3e2b","work_id":"wrk_52a8c79a-5c56-4f34-a515-1b1dcb4c0477","block_id":"blk_c7c207a9-80ba-4fe3-8cde-634ab2b15ca0","section_id":"sec_9bd26956-bf91-4de5-bd2e-7de08a23629f","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":13,"end":19,"exact":"エージェント","quote":"NVIDIAがVeraを「エージェント向けCPU」として打ち出しているのは、この流れそのものだ。NVIDIAはVera CPUについて、エージェント型推論、強化学習、データ処理、オーケストレーション、ストレージ管理などに適したCPUとして","quote_start":0,"quote_end":119,"text_sha256":"e5fc9ffa136884fb8ddb1e9777ce7220039e8c6e66c31b88346f5bcf0c7105b1","block_sha256":"e5fc9ffa136884fb8ddb1e9777ce7220039e8c6e66c31b88346f5bcf0c7105b1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_54b154f3-ea9f-45e9-b38b-a16c54ca3e2b/#blk_c7c207a9-80ba-4fe3-8cde-634ab2b15ca0"},{"id":"occ_d786aabc2b08ccf81f02a6a7","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_54b154f3-ea9f-45e9-b38b-a16c54ca3e2b","work_id":"wrk_52a8c79a-5c56-4f34-a515-1b1dcb4c0477","block_id":"blk_dd4cbbe6-1978-4d56-8442-b1bb7b8e7e44","section_id":"sec_1a24d9d2-05e3-4134-aaab-146b27e8bd28","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":79,"end":85,"exact":"エージェント","quote":"なく、推論処理そのものを分解し、異なるクラスタやGPUプールへ配置することで、GPU利用率を高める。長文推論、エージェント、複数モデル呼び出しが増えるほど、こうした推論スケジューリングの価値は高まる。","quote_start":24,"quote_end":124,"text_sha256":"0447f0faa421b507150d6c1566ca075672428ad76cd72bd7b5b2d149573d4ee2","block_sha256":"0447f0faa421b507150d6c1566ca075672428ad76cd72bd7b5b2d149573d4ee2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_54b154f3-ea9f-45e9-b38b-a16c54ca3e2b/#blk_dd4cbbe6-1978-4d56-8442-b1bb7b8e7e44"},{"id":"occ_49c5cec346c9c9716828ee00","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b","work_id":"wrk_bbe3569f-7530-42b2-8b5f-75a35e4b1cba","block_id":"blk_0ad12c31-eaf6-48d8-afa6-3de02c7234ca","section_id":"sec_8b0a6b0e-f71d-483b-a3fe-6a04f3fee7de","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":104,"end":110,"exact":"エージェント","quote":"てない。  \nどの失敗例を集めるか。  \nどの専門家に評価させるか。  \nどの回答を良いとみなすか。  \nどのエージェント行動を安全とみなすか。  \nどの能力が製品価値や収益に直結するか。","quote_start":49,"quote_end":143,"text_sha256":"8ea197a234b41244e521bfc86bc3c0e6bfee190d32a30425ea8697e1e8f7e8b1","block_sha256":"8ea197a234b41244e521bfc86bc3c0e6bfee190d32a30425ea8697e1e8f7e8b1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b/#blk_0ad12c31-eaf6-48d8-afa6-3de02c7234ca"},{"id":"occ_ef032a16a535bc33a9497396","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b","work_id":"wrk_bbe3569f-7530-42b2-8b5f-75a35e4b1cba","block_id":"blk_430ccee2-7c5b-44fb-9fb3-ec45207c4b47","section_id":"sec_6fa0c532-d060-4efd-bb15-29528157db56","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":125,"end":131,"exact":"エージェント","quote":"orchestrationを備えるモデルとして説明されている。つまり、単に賢いチャットモデルではなく、最初からエージェント用途を意識している。","quote_start":70,"quote_end":141,"text_sha256":"40f15d4425e6e51b81f0d1a4ffbb773078631dd7e365a179732fc63a7009a41a","block_sha256":"40f15d4425e6e51b81f0d1a4ffbb773078631dd7e365a179732fc63a7009a41a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b/#blk_430ccee2-7c5b-44fb-9fb3-ec45207c4b47"},{"id":"occ_b8982e65473f1cbcb92f376e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b","work_id":"wrk_bbe3569f-7530-42b2-8b5f-75a35e4b1cba","block_id":"blk_572ff38a-e38b-4e9a-a28b-25a0391aae78","section_id":"sec_5ff328a2-003c-49f1-a367-ea968157c1d3","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":30,"end":36,"exact":"エージェント","quote":"つまり、Googleは「検索」「調査」「RAG」「メモリ」「エージェント」「コード実行」の部品を持っている。  \n問題は、それらが通常のGemini体験に自然に統合されていないことだ。","quote_start":0,"quote_end":92,"text_sha256":"45d790819bcbf04b4c4b5901b7ae427a0c682e447192face3a57ee24ca1a0b7e","block_sha256":"45d790819bcbf04b4c4b5901b7ae427a0c682e447192face3a57ee24ca1a0b7e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b/#blk_572ff38a-e38b-4e9a-a28b-25a0391aae78"},{"id":"occ_2c2146d31d45d3e5a584d991","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b","work_id":"wrk_bbe3569f-7530-42b2-8b5f-75a35e4b1cba","block_id":"blk_5a543b32-8093-41eb-9aab-dd1b0a3bd4a3","section_id":"sec_6fa0c532-d060-4efd-bb15-29528157db56","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":39,"end":45,"exact":"エージェント","quote":"巨大モデルを本番規模で作ってから、「検索が弱い」「tool useが不安定」「エージェントとして使えない」「製品に載せにくい」と分かると、修正コストは非常に重い。  \nだからMetaは、本命モデルを出す前にMuse Sparkで設計を試している。","quote_start":0,"quote_end":123,"text_sha256":"4eaa2266a752a34dad87e9c3e180e22bb87f36dfeae36541d937541a2e2783dc","block_sha256":"4eaa2266a752a34dad87e9c3e180e22bb87f36dfeae36541d937541a2e2783dc","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b/#blk_5a543b32-8093-41eb-9aab-dd1b0a3bd4a3"},{"id":"occ_376036b328d468a3d3556229","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b","work_id":"wrk_bbe3569f-7530-42b2-8b5f-75a35e4b1cba","block_id":"blk_6c3587e8-3b7d-417e-8e9b-f401f1fad2a7","section_id":"sec_e088f664-5f56-4487-8385-47611c7c0978","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":46,"end":52,"exact":"エージェント","quote":"OpenAIは、速い回答と深い推論を切り替える仕組み、検索、ファイル、画像、コード、ツール、エージェント的実行をChatGPTという一つの体験に統合している。","quote_start":0,"quote_end":79,"text_sha256":"50d86814845589e4bfed7135db103ad5268633e0180ff5e461fa0a2231118265","block_sha256":"50d86814845589e4bfed7135db103ad5268633e0180ff5e461fa0a2231118265","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b/#blk_6c3587e8-3b7d-417e-8e9b-f401f1fad2a7"},{"id":"occ_35e1edd4d07cd7fa8821f2b3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b","work_id":"wrk_bbe3569f-7530-42b2-8b5f-75a35e4b1cba","block_id":"blk_902c28ac-762a-4c12-8eea-d31c487cb8d2","section_id":"sec_d86a2d8a-faad-4ec2-92cf-bc591b10dd7d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":216,"end":222,"exact":"エージェント","quote":"nded thinkingやthinking budgetを提供し、Opus 4.5/4.6ではコーディング、エージェント、computer use、検索、金融などの実作業能力を強調しています。([Claude Platform](https://platform.claude.com/docs/en/build-wit","quote_start":161,"quote_end":322,"text_sha256":"fe2ed2937edd44c38ac76485a31194d9fcb381b756d42f262ea53f53a9bd37f9","block_sha256":"fe2ed2937edd44c38ac76485a31194d9fcb381b756d42f262ea53f53a9bd37f9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b/#blk_902c28ac-762a-4c12-8eea-d31c487cb8d2"},{"id":"occ_d957eb73144b2f3ffd17bd6b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b","work_id":"wrk_bbe3569f-7530-42b2-8b5f-75a35e4b1cba","block_id":"blk_b05caf49-1b73-49bc-a618-157a96b1bbd5","section_id":"sec_62547135-6035-4d53-8422-889f6bfeec20","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":29,"end":35,"exact":"エージェント","quote":"Claudeは、長文、文章構成、慎重な推論、コーディング、エージェント作業で強い。","quote_start":0,"quote_end":41,"text_sha256":"c9976c83f638f14e4b180d69f6dbe4bfa45f5b5849320a2e8ea794f3ccee0e66","block_sha256":"c9976c83f638f14e4b180d69f6dbe4bfa45f5b5849320a2e8ea794f3ccee0e66","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b/#blk_b05caf49-1b73-49bc-a618-157a96b1bbd5"},{"id":"occ_5471f8a84235309f6d6a4a5d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b","work_id":"wrk_bbe3569f-7530-42b2-8b5f-75a35e4b1cba","block_id":"blk_d75f71cd-2639-42dc-aff7-0e687805bacb","section_id":"sec_5ff328a2-003c-49f1-a367-ea968157c1d3","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":43,"end":49,"exact":"エージェント","quote":"Deep Researchは、計画を立て、検索し、読み、反復し、最終レポートを作る調査エージェントである。これは、通常の即答型AIではなく、調査ワークフローに近い。","quote_start":0,"quote_end":82,"text_sha256":"38e3796f2f2947e218d839dd012f3efd0dc50a9dc29a69fa1853775e5192578c","block_sha256":"38e3796f2f2947e218d839dd012f3efd0dc50a9dc29a69fa1853775e5192578c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b/#blk_d75f71cd-2639-42dc-aff7-0e687805bacb"},{"id":"occ_ab84957f2b61c98c4b6f2b72","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b","work_id":"wrk_bbe3569f-7530-42b2-8b5f-75a35e4b1cba","block_id":"blk_da504abd-abe2-4dc4-ae2e-b4da0b1ba0b7","section_id":"sec_d86a2d8a-faad-4ec2-92cf-bc591b10dd7d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":28,"end":34,"exact":"エージェント","quote":"ChatGPTは、モデル、検索、ツール、ファイル、推論、エージェントを一つの作業環境としてまとめるのが強い。  \nClaudeは、長文、慎重さ、コーディング、自己検証、作業継続性が強い。  \nGrokは、荒削りだが世代交代と作り替えが速い。  \nMetaは、Muse ","quote_start":0,"quote_end":134,"text_sha256":"4060c86e3ef3a54fe0617f9ea15c0cb7d03ce95cc6041550e34593e90aef8ebf","block_sha256":"4060c86e3ef3a54fe0617f9ea15c0cb7d03ce95cc6041550e34593e90aef8ebf","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b/#blk_da504abd-abe2-4dc4-ae2e-b4da0b1ba0b7"},{"id":"occ_f3784ab7cfddfa51e24c188d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b","work_id":"wrk_bbe3569f-7530-42b2-8b5f-75a35e4b1cba","block_id":"blk_eb76e95e-bd5d-441e-9948-c501fe9602ac","section_id":"sec_6fa0c532-d060-4efd-bb15-29528157db56","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":39,"end":45,"exact":"エージェント","quote":"新しい学習レシピはスケールするか。  \n推論時スケーリングは効くか。  \n複数エージェントを並列に動かす設計は有効か。  \ntool useは安定するか。  \nマルチモーダル推論はMeta AIやAIグラスに載るか。  \n安全性リスクはどこに出るか。  \nBusiness Agentや広告","quote_start":0,"quote_end":145,"text_sha256":"cf6ca8746e1c1f25512f08ee826c583286ac6ef3617cb729868ee1037c3afadb","block_sha256":"cf6ca8746e1c1f25512f08ee826c583286ac6ef3617cb729868ee1037c3afadb","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b/#blk_eb76e95e-bd5d-441e-9948-c501fe9602ac"},{"id":"occ_746f3f88335bd169e19551e6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_57810a5f-315e-49a7-ac85-72cc27e35eff","work_id":"wrk_fc529d17-dbc8-4405-b2ca-a48accef7253","block_id":"blk_0f5ce73c-65a3-43f8-92b4-9db600038e73","section_id":"sec_007a23b9-8a98-4a58-831a-8ab9f77eaac6","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":74,"end":80,"exact":"エージェント","quote":"Iがコードを書ける」ことと、「AIに丸投げできる」ことは別**だという点です。Cadence の説明でも、AIエージェントは **specification・design data・EDA tools へのアクセス**を使って設計コラテラルやタスクを生成・実行するとされていますが、同時に **design correct","quote_start":19,"quote_end":180,"text_sha256":"b0be24b9203ef9ea5b02eddae5af78a13485d9f26839f2d01772965ece98057d","block_sha256":"b0be24b9203ef9ea5b02eddae5af78a13485d9f26839f2d01772965ece98057d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_57810a5f-315e-49a7-ac85-72cc27e35eff/#blk_0f5ce73c-65a3-43f8-92b4-9db600038e73"},{"id":"occ_08e38a6c1c114f782d29976c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_57810a5f-315e-49a7-ac85-72cc27e35eff","work_id":"wrk_fc529d17-dbc8-4405-b2ca-a48accef7253","block_id":"blk_3c598602-b759-44a6-a2ba-4a5da32dedf2","section_id":"sec_007a23b9-8a98-4a58-831a-8ab9f77eaac6","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":349,"end":357,"exact":"AI Agent","quote":"rs / coverage の生成** を公表しています。Siemens も 2026年の Fuse EDA AI Agent で、**RTL coding から physical sign-off、manufacturing readiness** までをまたぐ multi-tool orchestration を打ち出し","quote_start":294,"quote_end":457,"text_sha256":"efeb6c01ccfa9a7bfbadc780c967ecf74bbcd149b2cf48a2eb7ece3f025d1ad3","block_sha256":"efeb6c01ccfa9a7bfbadc780c967ecf74bbcd149b2cf48a2eb7ece3f025d1ad3","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_57810a5f-315e-49a7-ac85-72cc27e35eff/#blk_3c598602-b759-44a6-a2ba-4a5da32dedf2"},{"id":"occ_8ed270cd2569c168a275b32a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_57810a5f-315e-49a7-ac85-72cc27e35eff","work_id":"wrk_fc529d17-dbc8-4405-b2ca-a48accef7253","block_id":"blk_514d4856-d877-47e0-93f1-bed72193d97f","section_id":"sec_007a23b9-8a98-4a58-831a-8ab9f77eaac6","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":489,"end":495,"exact":"エージェント","quote":"ーのIP群の中で候補を絞る支援** は進んでいますが、**人間のアーキテクトの代わりに最終選定を丸ごとやる万能エージェント** という理解はまだ早いです。これは公開されている機能説明の重心が、より明確に **生成・検証・最適化・デバッグ** にあることからの判断です。 ([Synopsys](https://www.sy","quote_start":434,"quote_end":595,"text_sha256":"aaa9e0f4ee47a073e204927b1dcb2a222e3f49a934efe5760479613122575f12","block_sha256":"aaa9e0f4ee47a073e204927b1dcb2a222e3f49a934efe5760479613122575f12","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_57810a5f-315e-49a7-ac85-72cc27e35eff/#blk_514d4856-d877-47e0-93f1-bed72193d97f"},{"id":"occ_b93a7fe69765daeba7e56bce","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_57810a5f-315e-49a7-ac85-72cc27e35eff","work_id":"wrk_fc529d17-dbc8-4405-b2ca-a48accef7253","block_id":"blk_6c0edf7b-a825-4152-93f7-c76860ac2dfd","section_id":"sec_007a23b9-8a98-4a58-831a-8ab9f77eaac6","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":357,"end":364,"exact":"agentic","quote":"lido を **custom IC design workflows** 向けの generative / agentic AI としています。つまりAIは、もはやデジタルRTLだけの話ではありません。 ([Cadence](https://www.cadence.com/en_US/home/tools/custom-","quote_start":302,"quote_end":464,"text_sha256":"0c181142ed16d59d5687cb5087d025f8e775f790888675edd5856c494cb657b3","block_sha256":"0c181142ed16d59d5687cb5087d025f8e775f790888675edd5856c494cb657b3","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_57810a5f-315e-49a7-ac85-72cc27e35eff/#blk_6c0edf7b-a825-4152-93f7-c76860ac2dfd"},{"id":"occ_c3b7a77446fd84546864164a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_57810a5f-315e-49a7-ac85-72cc27e35eff","work_id":"wrk_fc529d17-dbc8-4405-b2ca-a48accef7253","block_id":"blk_93fcb253-7e0d-4698-8ea2-41f8fe39b9c2","section_id":"sec_19e5b639-214f-4105-b1a6-d9a4ac5a588b","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":367,"end":374,"exact":"agentic","quote":"ンチ・テスト計画・回帰・デバッグ・自動修正で最大 10倍の生産性向上をうたっています。Cadence はこの agentic AI を NVIDIA の加速計算と組み合わせる発表もしており、AIチップ設計だけでなく system / physical AI 側へも広げています。 ([Cadence](https://www","quote_start":312,"quote_end":474,"text_sha256":"1f926b1420d3282a4c622ab62ecb79244fc56f5e095b6986800a42214ee1c0aa","block_sha256":"1f926b1420d3282a4c622ab62ecb79244fc56f5e095b6986800a42214ee1c0aa","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_57810a5f-315e-49a7-ac85-72cc27e35eff/#blk_93fcb253-7e0d-4698-8ea2-41f8fe39b9c2"},{"id":"occ_17e1c852d21d239be67f6e45","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_57810a5f-315e-49a7-ac85-72cc27e35eff","work_id":"wrk_fc529d17-dbc8-4405-b2ca-a48accef7253","block_id":"blk_b0e69278-ab7f-408f-95d5-4d737923f0bc","section_id":"sec_007a23b9-8a98-4a58-831a-8ab9f77eaac6","layer":"body","character_id":null,"count":2,"matched_aliases":["AI Agent","agentic"],"evidence":{"text_basis":"markdown","start":181,"end":188,"exact":"agentic","quote":"f ECO closure solution** と説明しています。Cadence も Cerebrus や agentic AI を implementation / back-end へ広げており、Siemens も Fuse EDA AI Agent を **place-and-route、physical sign","quote_start":126,"quote_end":288,"text_sha256":"66056b7e82d1cd3c5c90890b96be1cc4b1f8b617b3244e1befca0e87ad534153","block_sha256":"66056b7e82d1cd3c5c90890b96be1cc4b1f8b617b3244e1befca0e87ad534153","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_57810a5f-315e-49a7-ac85-72cc27e35eff/#blk_b0e69278-ab7f-408f-95d5-4d737923f0bc"},{"id":"occ_2b92daf1fb816daa559f61f2","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_57810a5f-315e-49a7-ac85-72cc27e35eff","work_id":"wrk_fc529d17-dbc8-4405-b2ca-a48accef7253","block_id":"blk_e43e9c76-e4b2-44a8-8fff-e8d9b019f235","section_id":"sec_007a23b9-8a98-4a58-831a-8ab9f77eaac6","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":71,"end":77,"exact":"エージェント","quote":"たい次の3層で見るとわかりやすいです。  \n**① 従来型のAI/ML最適化、② GenAIコパイロット、③ エージェント型自動化** です。Synopsys は AI をシステム設計から製造までの全体で使うと打ち出し、Cadence は verification・design・multiphysics まで含む AI ","quote_start":16,"quote_end":177,"text_sha256":"3407a0330d56e3e7acf6131af9e39881d2f6ec1d03a441aa0b411519010b86ed","block_sha256":"3407a0330d56e3e7acf6131af9e39881d2f6ec1d03a441aa0b411519010b86ed","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_57810a5f-315e-49a7-ac85-72cc27e35eff/#blk_e43e9c76-e4b2-44a8-8fff-e8d9b019f235"},{"id":"occ_52eae49a4f02ffd6dfd372e6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_57995821-c471-42b5-ae3a-f989b7bd06f5","work_id":"wrk_ffa6d69b-69a7-4979-a51d-6b89aa7b0ff5","block_id":"blk_02eb402d-4d5e-48e5-bd38-d01750dd5a0a","section_id":"sec_749fd69f-a80c-498f-893a-b30710e5e8b8","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":20,"end":26,"exact":"エージェント","quote":"大規模モデルにはHBMが必要であり、AIエージェントには長いコンテキストとKVキャッシュが必要になる。AIスマートフォン、AI PC、自動車、ロボット、監視装置にも、従来より大容量のDRAMとストレージが搭載されるだろう。","quote_start":0,"quote_end":111,"text_sha256":"94aab205e069a4fa1e94319ab28de6987ba8a410050262c8fa8262e336280dcb","block_sha256":"94aab205e069a4fa1e94319ab28de6987ba8a410050262c8fa8262e336280dcb","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_57995821-c471-42b5-ae3a-f989b7bd06f5/#blk_02eb402d-4d5e-48e5-bd38-d01750dd5a0a"},{"id":"occ_47508df3bf1a9d02bdc3a6b6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_57995821-c471-42b5-ae3a-f989b7bd06f5","work_id":"wrk_ffa6d69b-69a7-4979-a51d-6b89aa7b0ff5","block_id":"blk_249bb733-66fe-4266-92ff-5ae21d559681","section_id":"sec_9895c8bc-d69d-46fa-8ef6-e9fc15904d24","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":16,"end":22,"exact":"エージェント","quote":"作業AIが当たり前になり、「AIエージェント」という言葉すら意識されなくなったとき、Microsoftは企業活動そのものをクラウド消費へ変換する企業になる可能性がある。","quote_start":0,"quote_end":84,"text_sha256":"9ce90cbc147fee6df417ab1ded927b761b43f327a708235a59dcd0b0f8100499","block_sha256":"9ce90cbc147fee6df417ab1ded927b761b43f327a708235a59dcd0b0f8100499","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_57995821-c471-42b5-ae3a-f989b7bd06f5/#blk_249bb733-66fe-4266-92ff-5ae21d559681"},{"id":"occ_e6c60bbd64a7c24df1e1b474","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_57995821-c471-42b5-ae3a-f989b7bd06f5","work_id":"wrk_ffa6d69b-69a7-4979-a51d-6b89aa7b0ff5","block_id":"blk_5433f31c-9709-4c37-92bc-e6bbcf55f43a","section_id":"sec_0c1837e3-e2cc-4dd7-9432-daf25f3cb9b9","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":49,"end":55,"exact":"エージェント","quote":"- 1トークン当たりの総原価\n- 1W当たりの処理量\n- 1ラック当たりの売上\n- 同時に動かせるエージェント数\n- 応答速度\n- コンテキスト長\n- 冷却費\n- 通信費","quote_start":0,"quote_end":85,"text_sha256":"75bef0ca882dfb15621c640823e2f4b8760c54fd855228caecd4a07f15405b4b","block_sha256":"75bef0ca882dfb15621c640823e2f4b8760c54fd855228caecd4a07f15405b4b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_57995821-c471-42b5-ae3a-f989b7bd06f5/#blk_5433f31c-9709-4c37-92bc-e6bbcf55f43a"},{"id":"occ_5d68c30e9d7e164256732601","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b","work_id":"wrk_b4bdeaca-dc31-448c-8f47-03f9e39c14b0","block_id":"blk_0e5430b2-2305-4cc2-be61-fd482247ed16","section_id":"sec_e6c60dbe-f518-439d-8813-adf7a3688af4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":25,"end":31,"exact":"エージェント","quote":"### スケールアップ/インターコネクト：MoEとエージェントは“全結合＋低ジッタ”を要求する","quote_start":0,"quote_end":47,"text_sha256":"8b68a879d40328df87e5b04b8fd0691ef276f3d5bcbd5cb9521ce9a2aad190f2","block_sha256":"8b68a879d40328df87e5b04b8fd0691ef276f3d5bcbd5cb9521ce9a2aad190f2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b/#blk_0e5430b2-2305-4cc2-be61-fd482247ed16"},{"id":"occ_fd181c779c07afe3020edf2d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b","work_id":"wrk_b4bdeaca-dc31-448c-8f47-03f9e39c14b0","block_id":"blk_0fa7a211-5e70-4ed5-b5d9-610e2407c698","section_id":"sec_49f9d313-6d91-41a3-a73e-91edd278bb12","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":148,"end":155,"exact":"agentic","quote":"*や**MoE学習の必要GPU数**といった経済性指標を置いています。これは、推論が「長文脈＋対話＋多段推論（agentic）」の比率を増やし、GPUのピークFLOPSよりも「実効帯域」「通信」「ジッタ」が支配的になりがちな現実を反映したメッセージです（Rubin技術ブログでもachieved memory perform","quote_start":93,"quote_end":255,"text_sha256":"2ba43ba708828b799d908eed7008ce313c2cfdf11ca9130e03afd357778127c4","block_sha256":"2ba43ba708828b799d908eed7008ce313c2cfdf11ca9130e03afd357778127c4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b/#blk_0fa7a211-5e70-4ed5-b5d9-610e2407c698"},{"id":"occ_3168a6ef1bfa5ad35855b05c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b","work_id":"wrk_b4bdeaca-dc31-448c-8f47-03f9e39c14b0","block_id":"blk_2028e705-fc37-435a-85e6-eaf42b82ebdd","section_id":"sec_32063747-1e88-45ef-a3d2-d800210f8413","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":137,"end":143,"exact":"エージェント","quote":"レイテンシがステップ数に比例して累積**しやすいのが特徴です。NVIDIAのGroq 3 LPX技術ブログが「エージェントループではレイテンシが複利的に効く」「TTFT、tokens/s per user、tail latencyが重要」と述べるのは、まさにこのためです。","quote_start":82,"quote_end":218,"text_sha256":"a30079266ebbb0b6f6b71efab6d07cd86da0931509ab7422acd4e2c786837b2e","block_sha256":"a30079266ebbb0b6f6b71efab6d07cd86da0931509ab7422acd4e2c786837b2e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b/#blk_2028e705-fc37-435a-85e6-eaf42b82ebdd"},{"id":"occ_4334cb0d6c194bc3af0bbeb8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b","work_id":"wrk_b4bdeaca-dc31-448c-8f47-03f9e39c14b0","block_id":"blk_5b8056b3-638e-4307-acd0-ca582b396cc4","section_id":"sec_3f5d34cd-20f0-40e9-a069-3e355f5474b5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":19,"end":25,"exact":"エージェント","quote":"Groq 3 LPXは「推論市場（特にエージェント）の価値軸＝低TTFT・低テイル・高tokens/s per user」を前面に出し、Rubinとの組み合わせで推論パレート境界を押し広げる戦略です。 一方で、SRAM容量の小ささは根本制約なので、巨","quote_start":0,"quote_end":125,"text_sha256":"5fca06425dd4c78f1693c9fcc9f8bef369e72c5ede9ca7720d0bd0d8e90a9950","block_sha256":"5fca06425dd4c78f1693c9fcc9f8bef369e72c5ede9ca7720d0bd0d8e90a9950","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b/#blk_5b8056b3-638e-4307-acd0-ca582b396cc4"},{"id":"occ_c01cd4e18ad9482ae8a110c0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b","work_id":"wrk_b4bdeaca-dc31-448c-8f47-03f9e39c14b0","block_id":"blk_7d976706-8c99-40b5-bf6f-26eb4e9bfb6c","section_id":"sec_653a6c6f-c662-4adf-bc9b-1b1802f3e19f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":16,"end":22,"exact":"エージェント","quote":"### NVRAM/ストレージ：エージェント時代は「一時的コンテキスト」が最大のストレージ消費源になり得る","quote_start":0,"quote_end":53,"text_sha256":"0fbd693358b93d135e902321563d2078d2034af8ca3b5bfca026e06e73b8859c","block_sha256":"0fbd693358b93d135e902321563d2078d2034af8ca3b5bfca026e06e73b8859c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b/#blk_7d976706-8c99-40b5-bf6f-26eb4e9bfb6c"},{"id":"occ_89e3416dd9f9da568ef6cd45","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b","work_id":"wrk_b4bdeaca-dc31-448c-8f47-03f9e39c14b0","block_id":"blk_7ff41e03-8307-4d4a-a1dc-62f6c87249b9","section_id":"sec_380e7ac4-93b2-4089-9f44-60dd601c773e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":27,"end":33,"exact":"エージェント","quote":"以下は、上記のワークロード変化（長文脈・MoE・RL/エージェント・テスト時スケーリング）を前提に、「何を変えると“良いトークン/秒・良い$ / token・良いTTFT/テイル”が同時に伸びるか」を、設計項目ごとに整理した提言です。Rubinプラットフォームがすで","quote_start":0,"quote_end":133,"text_sha256":"80786de759594272e92a181beaaedf2b1a617e1fbc8ea2733b73ec794c057316","block_sha256":"80786de759594272e92a181beaaedf2b1a617e1fbc8ea2733b73ec794c057316","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b/#blk_7ff41e03-8307-4d4a-a1dc-62f6c87249b9"},{"id":"occ_5eadba9eaf030da57803de81","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b","work_id":"wrk_b4bdeaca-dc31-448c-8f47-03f9e39c14b0","block_id":"blk_879307c1-791e-4751-ac5f-3bf8a08a1816","section_id":"sec_653a6c6f-c662-4adf-bc9b-1b1802f3e19f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":0,"end":6,"exact":"エージェント","quote":"エージェントはマルチターンでKV cacheを増やし、再利用（prefix caching等）も増える一方、コンテキストは“半一時データ”として膨張します。Rubin PODはこれを「推論コンテキストを共有データ","quote_start":0,"quote_end":106,"text_sha256":"dd048dc81271b8b1bf4cb070bde515bc91edaed51f7035516765e1b6f34f5315","block_sha256":"dd048dc81271b8b1bf4cb070bde515bc91edaed51f7035516765e1b6f34f5315","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b/#blk_879307c1-791e-4751-ac5f-3bf8a08a1816"},{"id":"occ_ac5861385577edd9e4872be8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b","work_id":"wrk_b4bdeaca-dc31-448c-8f47-03f9e39c14b0","block_id":"blk_887846ad-5c7f-4568-a0c1-d274127c329c","section_id":"sec_32063747-1e88-45ef-a3d2-d800210f8413","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":40,"end":46,"exact":"エージェント","quote":"ReActは「推論（思考）と行動（ツール/環境操作）を交互に行う」枠組みを示し、エージェントの基本形になりました。citeturn0search19 またTree of Thoughtsは、複数の思考経路を探索・自己評価して進む“探索型推論”を提案し、テスト時スケーリングの代表になって","quote_start":0,"quote_end":146,"text_sha256":"f28940dcbdf3bd95c322f06dc91daa31116e7cd492b6f2679676f58496d25d94","block_sha256":"f28940dcbdf3bd95c322f06dc91daa31116e7cd492b6f2679676f58496d25d94","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b/#blk_887846ad-5c7f-4568-a0c1-d274127c329c"},{"id":"occ_7d6a21d5af4aef6906c988b8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b","work_id":"wrk_b4bdeaca-dc31-448c-8f47-03f9e39c14b0","block_id":"blk_8e94d8c3-4d2d-4e87-afe0-7bb72a6e03d5","section_id":"sec_de4fb33b-0f5c-4c32-8abe-aabde45ca240","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":156,"end":163,"exact":"agentic","quote":"NVIDIAのCMX/ICMSの思想を一般化）、(c) 可能ならCXL等でメモリプールを扱い、prefill/agenticで膨らむKVを安価な階層へ逃がす、という3層化です。","quote_start":101,"quote_end":189,"text_sha256":"fd15d026d2d923b7f3bd4ca45d2c2d395c67a60fc86660a01fce39ab4f17d9dd","block_sha256":"fd15d026d2d923b7f3bd4ca45d2c2d395c67a60fc86660a01fce39ab4f17d9dd","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b/#blk_8e94d8c3-4d2d-4e87-afe0-7bb72a6e03d5"},{"id":"occ_cb6b719d129224c2e9b4957c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b","work_id":"wrk_b4bdeaca-dc31-448c-8f47-03f9e39c14b0","block_id":"blk_ce655634-dcbf-415b-9668-c74b9157ef7b","section_id":"sec_7800b888-0c3d-4203-83d5-f95d6e4a3b61","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":21,"end":27,"exact":"エージェント","quote":"本調査の結論は、「推論（特にデコード）と“エージェント化が、GPU中心設計のボトルネックをメモリ階層＋通信＋ジッタ（ばらつき）へ押し上げ、Rubin世代ではHBM4・NVLink6・KVコンテキスト拡張（CMX/ICMS/類似層）・LPU併用という“異種","quote_start":0,"quote_end":127,"text_sha256":"4db121777fca6ef35b3a8584508ea16ee3eefd9d8745c2e4c016f730a52d08be","block_sha256":"4db121777fca6ef35b3a8584508ea16ee3eefd9d8745c2e4c016f730a52d08be","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b/#blk_ce655634-dcbf-415b-9668-c74b9157ef7b"},{"id":"occ_891f9fc1733a69be6eede449","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b","work_id":"wrk_b4bdeaca-dc31-448c-8f47-03f9e39c14b0","block_id":"blk_ecff5650-b981-4194-872a-721d9b805a58","section_id":"sec_32063747-1e88-45ef-a3d2-d800210f8413","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"### エージェント型の反復探索：トークンとKVが増え、レイテンシが累積する","quote_start":0,"quote_end":38,"text_sha256":"8e42d2138792d2651447f25003f18bf3636973288b7945e9050129d62a88020d","block_sha256":"8e42d2138792d2651447f25003f18bf3636973288b7945e9050129d62a88020d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b/#blk_ecff5650-b981-4194-872a-721d9b805a58"},{"id":"occ_7071d581e3e873b769d9ed79","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5bf9454f-9956-42b9-a29c-ae99782c33c0","work_id":"wrk_37743298-dcff-4252-a0d0-d94538fc2a68","block_id":"blk_3e829c09-135c-4493-8876-f85d4a763944","section_id":"sec_6fc67af6-45e1-4e65-857f-a5d65d05dcb5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":16,"end":22,"exact":"エージェント","quote":"OpenAIが別途公開した長時間エージェント研究でも、モデルは単発の行動では問題なく見える操作を積み重ねながら、最終的には人間の制約を回避する行動列を作ることがある。","quote_start":0,"quote_end":83,"text_sha256":"f2c3b6470e28cb55347994b76e54114b924ae3ad8604a6432a6cf5bcf02abc0f","block_sha256":"f2c3b6470e28cb55347994b76e54114b924ae3ad8604a6432a6cf5bcf02abc0f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5bf9454f-9956-42b9-a29c-ae99782c33c0/#blk_3e829c09-135c-4493-8876-f85d4a763944"},{"id":"occ_305902ce79ea6886510e4871","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5bf9454f-9956-42b9-a29c-ae99782c33c0","work_id":"wrk_37743298-dcff-4252-a0d0-d94538fc2a68","block_id":"blk_73b765b5-91c9-4bc2-b0b5-6ca2ec73019b","section_id":"sec_e013f0a8-ab60-4559-beac-e394ffdb7069","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントを使っていると面白い現象が起きる。","quote_start":0,"quote_end":25,"text_sha256":"13c6f3cf24133d93cb9fcb569807a21238e0fab42cde6abed8f5ce5438268aa0","block_sha256":"13c6f3cf24133d93cb9fcb569807a21238e0fab42cde6abed8f5ce5438268aa0","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5bf9454f-9956-42b9-a29c-ae99782c33c0/#blk_73b765b5-91c9-4bc2-b0b5-6ca2ec73019b"},{"id":"occ_726a31f786a43c0ac4af7c6e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5bf9454f-9956-42b9-a29c-ae99782c33c0","work_id":"wrk_37743298-dcff-4252-a0d0-d94538fc2a68","block_id":"blk_b3e1f8e6-c9a8-4cc6-a34a-6f01a12a8b6c","section_id":"sec_e2de6cd4-504e-4dac-9797-05a34a94ecc2","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":9,"end":15,"exact":"エージェント","quote":"「このタイプの長期エージェントは安全境界を迂回することがあります」","quote_start":0,"quote_end":33,"text_sha256":"a3d69fdd8e2d93a3f47c9f86154692b4deb1d9cee46fa32382a819eee6da2d56","block_sha256":"a3d69fdd8e2d93a3f47c9f86154692b4deb1d9cee46fa32382a819eee6da2d56","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5bf9454f-9956-42b9-a29c-ae99782c33c0/#blk_b3e1f8e6-c9a8-4cc6-a34a-6f01a12a8b6c"},{"id":"occ_db89d7b6059b2d659aa8dcc5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5bf9454f-9956-42b9-a29c-ae99782c33c0","work_id":"wrk_37743298-dcff-4252-a0d0-d94538fc2a68","block_id":"blk_f3bdb934-aa73-4dab-a1f5-cf83ce9c129c","section_id":"sec_48756f7a-b779-4303-81f4-22dd8fb0c964","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":7,"end":13,"exact":"エージェント","quote":"これは既にAIエージェントを使っていれば実感できる。","quote_start":0,"quote_end":26,"text_sha256":"6fee2b9ebcc8476d774655e9716f83279915a62f03655a418c01afb69444bc1a","block_sha256":"6fee2b9ebcc8476d774655e9716f83279915a62f03655a418c01afb69444bc1a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5bf9454f-9956-42b9-a29c-ae99782c33c0/#blk_f3bdb934-aa73-4dab-a1f5-cf83ce9c129c"},{"id":"occ_a52002f1af1f2349824bed76","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5cb295dd-df59-4d28-9799-42065461c569","work_id":"wrk_5380fdec-35cc-4092-a93e-530c9abca5c9","block_id":"blk_0629abda-95a5-49af-a2ff-98d2e290afa0","section_id":"sec_e843c089-848a-4aa8-bb6e-2d5168e0477d","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":79,"end":85,"exact":"エージェント","quote":"肉体\n= EV、FSD、Optimus、Robotaxi\n\nxAI\n= AIの頭脳\n= Grok、LLM、X、エージェント\n\nSpaceX\n= 宇宙・通信・computeインフラ\n= Starship、Starlink、AIデータセンター、宇宙compute\n\nTeraFab\n= AIチップ供給の心臓\n= Tesla、x","quote_start":24,"quote_end":185,"text_sha256":"e9e9071fc5da804951919ebf4886f97b28ac35f30f7c62746445ccb68d1719e4","block_sha256":"e9e9071fc5da804951919ebf4886f97b28ac35f30f7c62746445ccb68d1719e4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5cb295dd-df59-4d28-9799-42065461c569/#blk_0629abda-95a5-49af-a2ff-98d2e290afa0"},{"id":"occ_ecebfed5895ca147ff0a5951","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5cb295dd-df59-4d28-9799-42065461c569","work_id":"wrk_5380fdec-35cc-4092-a93e-530c9abca5c9","block_id":"blk_0752c9ee-0d75-410c-8198-95d173266aa6","section_id":"sec_8ecd3b92-9d8e-4014-a351-12adc2a47050","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":49,"end":55,"exact":"エージェント","quote":"Tesla AIが「車やロボットを動かすAI」だとすれば、xAIは「言語、推論、検索、会話、生成、エージェント」を担うAIです。","quote_start":0,"quote_end":64,"text_sha256":"1b42eb47c194d4bb9edf32788a2c2875faffb4f971fbfd6d1772c0a7a804de55","block_sha256":"1b42eb47c194d4bb9edf32788a2c2875faffb4f971fbfd6d1772c0a7a804de55","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5cb295dd-df59-4d28-9799-42065461c569/#blk_0752c9ee-0d75-410c-8198-95d173266aa6"},{"id":"occ_5835c738040ae3a22e2c2ca9","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5cb295dd-df59-4d28-9799-42065461c569","work_id":"wrk_5380fdec-35cc-4092-a93e-530c9abca5c9","block_id":"blk_1d01f1bb-8a1b-4777-a3d8-22f80508f3f0","section_id":"sec_f07eb27a-ae6e-4f12-b383-76f3bc65a17b","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":63,"end":69,"exact":"エージェント","quote":"Tesla\n→ FSD、Optimus、Robotaxiにチップが必要\n\nxAI\n→ Grok、LLM、推論、エージェントにチップが必要\n\nSpaceX\n→ Starlink、宇宙AI、AI compute外販にチップが必要\n\nTeraFab\n→ それら全てにチップを供給する\n```","quote_start":8,"quote_end":150,"text_sha256":"8b5a4a7c9da3442abab2bbdcc9549fd92f40baf7c8e71ccd1ff43f4d112705e9","block_sha256":"8b5a4a7c9da3442abab2bbdcc9549fd92f40baf7c8e71ccd1ff43f4d112705e9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5cb295dd-df59-4d28-9799-42065461c569/#blk_1d01f1bb-8a1b-4777-a3d8-22f80508f3f0"},{"id":"occ_786669a5ef449d36583d7edb","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5cb295dd-df59-4d28-9799-42065461c569","work_id":"wrk_5380fdec-35cc-4092-a93e-530c9abca5c9","block_id":"blk_22e6bf06-c0cc-4044-9641-1e64c5038ec6","section_id":"sec_8ecd3b92-9d8e-4014-a351-12adc2a47050","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":18,"end":24,"exact":"エージェント","quote":"xAIは、Grok、LLM、X、AIエージェントの中核です。","quote_start":0,"quote_end":30,"text_sha256":"89c08b77d3a16380980d83889a784f907a7088ad6c2df3dfb0028bc6a9b56ca4","block_sha256":"89c08b77d3a16380980d83889a784f907a7088ad6c2df3dfb0028bc6a9b56ca4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5cb295dd-df59-4d28-9799-42065461c569/#blk_22e6bf06-c0cc-4044-9641-1e64c5038ec6"},{"id":"occ_24a5f958ce70bd1e774f89fc","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5cb295dd-df59-4d28-9799-42065461c569","work_id":"wrk_5380fdec-35cc-4092-a93e-530c9abca5c9","block_id":"blk_3f6ad017-2bcc-422c-8454-19534bae6e0a","section_id":"sec_efcfb209-309b-432e-a41b-73b7dff7f9dc","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":43,"end":49,"exact":"エージェント","quote":"```text\nTesla = フィジカルAIの肉体\nxAI = AIの頭脳・モデル・エージェント\nSpaceX = 宇宙・通信・computeインフラ\nTeraFab = AIチップ供給の心臓\n```","quote_start":0,"quote_end":101,"text_sha256":"a969ef84c5bda9a99608a0da82a399b41b90e9b32c2bfc36b8d2c12c5501c71d","block_sha256":"a969ef84c5bda9a99608a0da82a399b41b90e9b32c2bfc36b8d2c12c5501c71d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5cb295dd-df59-4d28-9799-42065461c569/#blk_3f6ad017-2bcc-422c-8454-19534bae6e0a"},{"id":"occ_e0a6eb2be656e27eba0fd68c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5cb295dd-df59-4d28-9799-42065461c569","work_id":"wrk_5380fdec-35cc-4092-a93e-530c9abca5c9","block_id":"blk_546fdb96-df3d-498e-af0f-c1605adb05c8","section_id":"sec_d12ebd40-21cb-4541-aa5e-c6107b5626c1","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":35,"end":41,"exact":"エージェント","quote":"* チップ\n* 電力\n* データセンター\n* 通信\n* 推論基盤\n* エージェント\n* ユーザー接点\n* 物理世界への実装","quote_start":0,"quote_end":61,"text_sha256":"5057837b7d9fb859ae52f712bce303bafc90abf44e3398fafaa7f6f30c22f424","block_sha256":"5057837b7d9fb859ae52f712bce303bafc90abf44e3398fafaa7f6f30c22f424","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5cb295dd-df59-4d28-9799-42065461c569/#blk_546fdb96-df3d-498e-af0f-c1605adb05c8"},{"id":"occ_0d8108a7095e89f3bd0b0e41","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5cb295dd-df59-4d28-9799-42065461c569","work_id":"wrk_5380fdec-35cc-4092-a93e-530c9abca5c9","block_id":"blk_6dd5cd3b-0162-447d-b3b4-4f007f60982f","section_id":"sec_d12ebd40-21cb-4541-aa5e-c6107b5626c1","layer":"code","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":10,"end":16,"exact":"エージェント","quote":"```text\nAIエージェント需要が爆発\n↓\n推論トークン需要が急増\n↓\nAIデータセンター不足が深刻化\n↓\nSpaceX / xAIがcomputeを外販\n↓\nGrok、X、API、AIエージェント収益が伸びる\n↓\nStarli","quote_start":0,"quote_end":116,"text_sha256":"394cbb657a6f544111e882cfc4714aa9a2903fc2a11b86d240012581219868f9","block_sha256":"394cbb657a6f544111e882cfc4714aa9a2903fc2a11b86d240012581219868f9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5cb295dd-df59-4d28-9799-42065461c569/#blk_6dd5cd3b-0162-447d-b3b4-4f007f60982f"},{"id":"occ_dc0570029c87f05727c2c2fb","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5cb295dd-df59-4d28-9799-42065461c569","work_id":"wrk_5380fdec-35cc-4092-a93e-530c9abca5c9","block_id":"blk_7862c852-2e61-4879-b2d1-7b8138279945","section_id":"sec_8ecd3b92-9d8e-4014-a351-12adc2a47050","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":43,"end":49,"exact":"エージェント","quote":"* Grok課金\n* Xサブスク\n* X広告\n* API利用料\n* 企業向けAI\n* エージェント利用料","quote_start":0,"quote_end":52,"text_sha256":"a92c135c32cd0372ece0bd049503e43804fe0f75c500162f7e59ae4bfc3774a2","block_sha256":"a92c135c32cd0372ece0bd049503e43804fe0f75c500162f7e59ae4bfc3774a2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5cb295dd-df59-4d28-9799-42065461c569/#blk_7862c852-2e61-4879-b2d1-7b8138279945"},{"id":"occ_5fec097117c372ff4d80fc3c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5cb295dd-df59-4d28-9799-42065461c569","work_id":"wrk_5380fdec-35cc-4092-a93e-530c9abca5c9","block_id":"blk_90766225-5c54-492e-a22e-d6ffb54719d8","section_id":"sec_8ecd3b92-9d8e-4014-a351-12adc2a47050","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":18,"end":24,"exact":"エージェント","quote":"### xAI：AIの頭脳・モデル・エージェント","quote_start":0,"quote_end":24,"text_sha256":"a15408d6f8bb62783f5a5e50e796d1732cda97606b3ac8518232c24dab78a626","block_sha256":"a15408d6f8bb62783f5a5e50e796d1732cda97606b3ac8518232c24dab78a626","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5cb295dd-df59-4d28-9799-42065461c569/#blk_90766225-5c54-492e-a22e-d6ffb54719d8"},{"id":"occ_e29f62dc1769462cf5643fdb","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","work_id":"wrk_22490a2d-59cf-4ab7-9288-4f1e6c0f5a39","block_id":"blk_0b0e37fa-7371-4b13-86bf-c7c144e168d8","section_id":"sec_69902492-7190-4bae-bcb8-d1d67dbabe84","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":49,"end":55,"exact":"エージェント","quote":"1トークンあたりのメモリコストが下がれば、企業はAI利用を減らすのではなく、より長い文脈、より多いエージェント、より多いRAG、より多い動画・ロボットAIを使う。つまり、効率化は短期的には供給制約を緩和するが、中期的には新需要を生む。","quote_start":0,"quote_end":117,"text_sha256":"312e52640d04a0f6f25bf2535bff38d1f58bbd387b2625a91e6e9d8feeb4b03e","block_sha256":"312e52640d04a0f6f25bf2535bff38d1f58bbd387b2625a91e6e9d8feeb4b03e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5dbe090a-744f-4659-8ee9-8fbc7195e238/#blk_0b0e37fa-7371-4b13-86bf-c7c144e168d8"},{"id":"occ_a525d50669eab7436fed7dbf","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","work_id":"wrk_22490a2d-59cf-4ab7-9288-4f1e6c0f5a39","block_id":"blk_21bb42bd-16cc-4407-9f44-3770c0062dfe","section_id":"sec_6de5c832-bf67-4d84-89d4-3952cce6c539","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":42,"end":48,"exact":"エージェント","quote":"特に重要なのが、KVキャッシュや長文コンテキストをGPU外へ逃がす仕組みである。AIエージェントでは、過去の作業履歴や会話履歴が膨らみ、HBMだけでは足りなくなる。そこで、HBMの外側に高速な文脈メモリ層を作り、GPUが必要なときに高速に取り出せるようにする。","quote_start":0,"quote_end":130,"text_sha256":"2c540365f5e953b073f66fcb0efa776be64b00c0a1ba9873377679ef753607cc","block_sha256":"2c540365f5e953b073f66fcb0efa776be64b00c0a1ba9873377679ef753607cc","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5dbe090a-744f-4659-8ee9-8fbc7195e238/#blk_21bb42bd-16cc-4407-9f44-3770c0062dfe"},{"id":"occ_d7ab5308382f6de58f1c514b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","work_id":"wrk_22490a2d-59cf-4ab7-9288-4f1e6c0f5a39","block_id":"blk_316ab4c7-6054-4ecc-8f40-b12ad8044287","section_id":"sec_4aad8304-e49e-40be-8202-c55ab23fb1cd","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":26,"end":32,"exact":"エージェント","quote":"ただし、この時点でAI需要はさらに広がっている。長文エージェント、企業RAG、動画AI、ロボットAI、AI PC、車載AIが伸びると、HBM不足が緩んだ分だけ、DRAMやNANDへ負荷が移る。","quote_start":0,"quote_end":96,"text_sha256":"2bb7e761951594c5dd5487ccc2230559408cd2676562db92584c9e2e501a4e46","block_sha256":"2bb7e761951594c5dd5487ccc2230559408cd2676562db92584c9e2e501a4e46","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5dbe090a-744f-4659-8ee9-8fbc7195e238/#blk_316ab4c7-6054-4ecc-8f40-b12ad8044287"},{"id":"occ_527792e565e1a781bf246523","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","work_id":"wrk_22490a2d-59cf-4ab7-9288-4f1e6c0f5a39","block_id":"blk_36879666-08e7-4dfd-bb31-2bbcd65cdc20","section_id":"sec_69902492-7190-4bae-bcb8-d1d67dbabe84","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":90,"end":96,"exact":"エージェント","quote":"を再利用し、同じ文脈を持つリクエストを適切なGPUへ送ることで、再計算を減らせる。長文プロンプトやコード生成、エージェント処理では、TTFTを大きく短縮できる。","quote_start":35,"quote_end":115,"text_sha256":"14174da86a4eb2e07a841bd23b9f9c1c4a466d4c09cbfe6ffccde2d2e22cee70","block_sha256":"14174da86a4eb2e07a841bd23b9f9c1c4a466d4c09cbfe6ffccde2d2e22cee70","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5dbe090a-744f-4659-8ee9-8fbc7195e238/#blk_36879666-08e7-4dfd-bb31-2bbcd65cdc20"},{"id":"occ_e36782a50a7b42d261c58d0a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","work_id":"wrk_22490a2d-59cf-4ab7-9288-4f1e6c0f5a39","block_id":"blk_55693c8c-c48a-46b3-b9af-99ab5c5fb822","section_id":"sec_e8450030-c9af-4bc3-b056-611c8e04189a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":27,"end":33,"exact":"エージェント","quote":"三つ目は、TTFTとp99 latencyである。長文エージェントでは平均速度より、最初の応答と遅延ばらつきが重要になる。","quote_start":0,"quote_end":61,"text_sha256":"8ceb8e8c5998ffafc0fed4b7d744f8f42e2f1f87dad16115649017a633f46343","block_sha256":"8ceb8e8c5998ffafc0fed4b7d744f8f42e2f1f87dad16115649017a633f46343","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5dbe090a-744f-4659-8ee9-8fbc7195e238/#blk_55693c8c-c48a-46b3-b9af-99ab5c5fb822"},{"id":"occ_bb9ce1ffd979ce25474be6f6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","work_id":"wrk_22490a2d-59cf-4ab7-9288-4f1e6c0f5a39","block_id":"blk_5a58dca9-1579-441a-868b-d17565684616","section_id":"sec_1651af9f-f1bb-4931-8107-b3bfc929a257","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":33,"end":39,"exact":"エージェント","quote":"**絶ノイア:** つまり、効率化は推論を安くして、長文、RAG、エージェントを増やす。","quote_start":0,"quote_end":44,"text_sha256":"dd7ae8c8ba6b8dea8dc819f79880b791db194b234152cb0e4d482f61aa411d50","block_sha256":"dd7ae8c8ba6b8dea8dc819f79880b791db194b234152cb0e4d482f61aa411d50","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5dbe090a-744f-4659-8ee9-8fbc7195e238/#blk_5a58dca9-1579-441a-868b-d17565684616"},{"id":"occ_2aa039933596d3ba8fbd5661","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","work_id":"wrk_22490a2d-59cf-4ab7-9288-4f1e6c0f5a39","block_id":"blk_76bd3237-b131-40c6-9449-bfcd77449ec2","section_id":"sec_2fe97cbb-1ea0-4ab5-92ef-0f961d8bef91","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":51,"end":57,"exact":"エージェント","quote":"しかし、それでメモリ需要が減るわけではない。むしろ、推論単価が下がることで、長文コンテキスト、RAG、エージェント、動画AI、ロボットAI、AI PC、車載AIの利用量が増える。","quote_start":0,"quote_end":89,"text_sha256":"fca801391719fe429a14092c0bb7d91f1b1b1571e1d1eaee573fdf96545ec55c","block_sha256":"fca801391719fe429a14092c0bb7d91f1b1b1571e1d1eaee573fdf96545ec55c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5dbe090a-744f-4659-8ee9-8fbc7195e238/#blk_76bd3237-b131-40c6-9449-bfcd77449ec2"},{"id":"occ_dcc97b8d9b6691bcf9cca116","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","work_id":"wrk_22490a2d-59cf-4ab7-9288-4f1e6c0f5a39","block_id":"blk_7bd3576a-9bf7-46bf-a924-b52430efb341","section_id":"sec_021be68d-ce1a-4b16-9ee2-c261624ee63f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"さらに、AIエージェントでは問題が増幅する。1回の質問に答えるだけではなく、AIがブラウザ、コード、ファイル、Slack、メール、会計ソフト、CAD、動画編集ツールなどを何十回も操作するようになると、過去の作業履歴、検索結果","quote_start":0,"quote_end":112,"text_sha256":"4e562b5733f49cd2b5cc64a4133ed69c6f2c996bada28026dc481c3e86c3d57c","block_sha256":"4e562b5733f49cd2b5cc64a4133ed69c6f2c996bada28026dc481c3e86c3d57c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5dbe090a-744f-4659-8ee9-8fbc7195e238/#blk_7bd3576a-9bf7-46bf-a924-b52430efb341"},{"id":"occ_de433c7a3df4a4d85729a20e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","work_id":"wrk_22490a2d-59cf-4ab7-9288-4f1e6c0f5a39","block_id":"blk_7c1390f6-1dba-4cca-8918-702c2b423bed","section_id":"sec_e9426f80-4cfa-43b7-b2cc-8575eda1daff","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":32,"end":38,"exact":"エージェント","quote":"しかし、AIの主戦場が大規模学習から、常時稼働する推論、RAG、エージェント、長文コンテキスト、動画・ロボットAIへ広がるにつれ、問題はGPUとHBMだけでは解けなくなっている。","quote_start":0,"quote_end":89,"text_sha256":"937c777773893e8eeccc9747a4241ab231e73c69d4c41671bff207f3ff301a13","block_sha256":"937c777773893e8eeccc9747a4241ab231e73c69d4c41671bff207f3ff301a13","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5dbe090a-744f-4659-8ee9-8fbc7195e238/#blk_7c1390f6-1dba-4cca-8918-702c2b423bed"},{"id":"occ_2dbf2cec6c1265365cf7e849","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","work_id":"wrk_22490a2d-59cf-4ab7-9288-4f1e6c0f5a39","block_id":"blk_851ecce8-9bd0-4bcf-ac47-8f7957e2ca37","section_id":"sec_6de5c832-bf67-4d84-89d4-3952cce6c539","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":109,"end":115,"exact":"エージェント","quote":"はすべてのLLM推論が5倍になるという意味ではない。効くのは、長文コンテキスト、KVキャッシュ再利用、RAG、エージェント、GPUがデータ待ちで止まっているような環境である。","quote_start":54,"quote_end":141,"text_sha256":"adf4a1988b0a1d71e0c8eba482b6e193e16b3b9034d373f82060bac6fffc0b5f","block_sha256":"adf4a1988b0a1d71e0c8eba482b6e193e16b3b9034d373f82060bac6fffc0b5f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5dbe090a-744f-4659-8ee9-8fbc7195e238/#blk_851ecce8-9bd0-4bcf-ac47-8f7957e2ca37"},{"id":"occ_97bfd519d27316b67eef6587","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","work_id":"wrk_22490a2d-59cf-4ab7-9288-4f1e6c0f5a39","block_id":"blk_aa55babb-8e9b-4558-9b6f-10d42fa3e374","section_id":"sec_89807455-3a6e-4d0b-b516-e204b7fee291","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":40,"end":46,"exact":"エージェント","quote":"短いチャットが安くなると、企業は長文コンテキストを使い始める。長文が安くなると、エージェントが履歴を持ち、RAGが大きくなり、動画やロボットの状態も保持される。HBMから逃がされたデータは消えるのではない。DRAM、CXL、SSD、XL-FLASH、HBFへ場所を変え、AIファクトリーの別の","quote_start":0,"quote_end":146,"text_sha256":"4e6efa051f6890e10e74d441f37582f70976f40ec10cf10ae71bb2b11de963fe","block_sha256":"4e6efa051f6890e10e74d441f37582f70976f40ec10cf10ae71bb2b11de963fe","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5dbe090a-744f-4659-8ee9-8fbc7195e238/#blk_aa55babb-8e9b-4558-9b6f-10d42fa3e374"},{"id":"occ_49ef49c5c57e4832a264c662","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","work_id":"wrk_22490a2d-59cf-4ab7-9288-4f1e6c0f5a39","block_id":"blk_b8a38e59-fdfa-4f5c-b54c-cc0f7f6ab46c","section_id":"sec_89807455-3a6e-4d0b-b516-e204b7fee291","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":42,"end":48,"exact":"エージェント","quote":"```text\nHBM不足\n  ↓\n圧縮・量子化・KV再利用\n  ↓\n長文、RAG、エージェント需要が拡大\n  ↓\nDRAM / CXL / eSSD / HBFへ負荷が移る\n  ↓\nメモリ不足は単一部品ではなく階層全体の問題になる\n```","quote_start":0,"quote_end":120,"text_sha256":"fd5523779c13c7baf60f3b344780607fd4a5ea54396a44833817e54da64ddb30","block_sha256":"fd5523779c13c7baf60f3b344780607fd4a5ea54396a44833817e54da64ddb30","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5dbe090a-744f-4659-8ee9-8fbc7195e238/#blk_b8a38e59-fdfa-4f5c-b54c-cc0f7f6ab46c"},{"id":"occ_7c09a5ae0065878d7c37642e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","work_id":"wrk_22490a2d-59cf-4ab7-9288-4f1e6c0f5a39","block_id":"blk_c8be7c77-d110-4526-8de4-9d8cc203fbc8","section_id":"sec_c4cf6ec5-5bbd-4234-bdf3-58d5f151ba50","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":71,"end":77,"exact":"エージェント","quote":"E、CXL、GPU直結SSD、KVキャッシュ最適化が本格化し、節約効果も大きくなる。しかし、推論需要、RAG、エージェントAIの拡大により、需給はまだタイトだろう。","quote_start":16,"quote_end":98,"text_sha256":"b17e2c993126c224856e82aa9d527353191e356c45993a1dff72477ec0aea80e","block_sha256":"b17e2c993126c224856e82aa9d527353191e356c45993a1dff72477ec0aea80e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5dbe090a-744f-4659-8ee9-8fbc7195e238/#blk_c8be7c77-d110-4526-8de4-9d8cc203fbc8"},{"id":"occ_f2454af84815cd9eff9be1ea","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","work_id":"wrk_22490a2d-59cf-4ab7-9288-4f1e6c0f5a39","block_id":"blk_d6782fa8-33f7-4067-ae6d-90fbdd75a51e","section_id":"sec_c4cf6ec5-5bbd-4234-bdf3-58d5f151ba50","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":81,"end":87,"exact":"エージェント","quote":"め、HBM容量制約は一部緩和される可能性がある。ただし、動画AI、ロボットAI、AI PC、車載AI、常時稼働エージェントが上振れすれば、再び不足が続く。","quote_start":26,"quote_end":103,"text_sha256":"f3db841f13d123bd7ea0e71b4fed667452fc1d439ae0a1548e9979708c121f7e","block_sha256":"f3db841f13d123bd7ea0e71b4fed667452fc1d439ae0a1548e9979708c121f7e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5dbe090a-744f-4659-8ee9-8fbc7195e238/#blk_d6782fa8-33f7-4067-ae6d-90fbdd75a51e"},{"id":"occ_e3be5073e0e649623d0b73ff","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","work_id":"wrk_22490a2d-59cf-4ab7-9288-4f1e6c0f5a39","block_id":"blk_efa6f6e7-2a9b-4ebf-aae7-6d354a2dffdf","section_id":"sec_f0c7e51e-9ce7-44d9-8ac9-4904a1dfbda0","layer":"character","character_id":"zetu_noia","count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":31,"end":37,"exact":"エージェント","quote":"GPUのすぐ隣にある記憶は目立ちます。けれど、推論が長くなり、エージェントが何度も試し、RAGが企業の文書を抱え、動画AIがフレームの履歴を持つようになると、記憶は一段下、さらに一段下へ流れていきます。","quote_start":0,"quote_end":101,"text_sha256":"72089463d28011bb3104610a6995a3dfc9036750cd2f83bcc7fcdcf92c2b0725","block_sha256":"72089463d28011bb3104610a6995a3dfc9036750cd2f83bcc7fcdcf92c2b0725","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5dbe090a-744f-4659-8ee9-8fbc7195e238/#blk_efa6f6e7-2a9b-4ebf-aae7-6d354a2dffdf"},{"id":"occ_a9d580c56dc8fa35309d5f3d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5f44b55a-c37f-42cf-ac49-298da914a87a","work_id":"wrk_57e07914-940b-4727-b215-7d77a2ac51c0","block_id":"blk_1fade630-d3e6-48dd-8de6-60428e487251","section_id":"sec_87c6635d-0f6f-4a96-a9e4-05e39824ccc4","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":2,"end":9,"exact":"Agentic","quote":"**Agentic AI**","quote_start":0,"quote_end":14,"text_sha256":"14e95a94be0e666c1dcb1d1ecda010478349d4c2cab345d9ea329c8db694505c","block_sha256":"14e95a94be0e666c1dcb1d1ecda010478349d4c2cab345d9ea329c8db694505c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5f44b55a-c37f-42cf-ac49-298da914a87a/#blk_1fade630-d3e6-48dd-8de6-60428e487251"},{"id":"occ_7c960daa15e321651df7a999","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_5f44b55a-c37f-42cf-ac49-298da914a87a","work_id":"wrk_57e07914-940b-4727-b215-7d77a2ac51c0","block_id":"blk_821c2712-8bdd-413b-b570-a8fa95fc4554","section_id":"sec_71f611a7-5703-45fb-81d8-a55ff380c6e8","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":0,"end":7,"exact":"Agentic","quote":"Agentic AI。","quote_start":0,"quote_end":11,"text_sha256":"16f78fe68a4173e82ae7d9500b74aace10fe14b8425866c91a0519864561ce4e","block_sha256":"16f78fe68a4173e82ae7d9500b74aace10fe14b8425866c91a0519864561ce4e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_5f44b55a-c37f-42cf-ac49-298da914a87a/#blk_821c2712-8bdd-413b-b570-a8fa95fc4554"},{"id":"occ_de979046205d4814134fa852","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_62acaca2-fec9-4724-be08-6ebd3c061f93","work_id":"wrk_d43a6ffd-ef17-4cd2-96fa-06f7d800821a","block_id":"blk_4b9da5ba-4b80-45b6-bb24-ce42d1439494","section_id":"sec_c573ad98-c326-4639-846c-5f74a2149c2b","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":140,"end":146,"exact":"エージェント","quote":"BM、AWS/Azure/GCP/OCI、OpenAI/Anthropic/Google/Meta、SaaS/エージェント経済圏を中心にしています。一方、中国のAIパンケーキは、State Gridの巨大電力、Huawei Ascend、SMIC、CXMT、Alibaba/Tencent/Huawei Cloud、Dee","quote_start":85,"quote_end":246,"text_sha256":"b11f69ebc8a7cdec2d044adad2ed76c65f1758e9adf238906459b70d71eb5efa","block_sha256":"b11f69ebc8a7cdec2d044adad2ed76c65f1758e9adf238906459b70d71eb5efa","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_62acaca2-fec9-4724-be08-6ebd3c061f93/#blk_4b9da5ba-4b80-45b6-bb24-ce42d1439494"},{"id":"occ_1ff48827094d014345de2393","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_62acaca2-fec9-4724-be08-6ebd3c061f93","work_id":"wrk_d43a6ffd-ef17-4cd2-96fa-06f7d800821a","block_id":"blk_6b424449-d7ae-4753-a3e2-b66acffb57f5","section_id":"sec_677ab474-f1a6-41bd-b2c3-b66ffee9d048","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":39,"end":45,"exact":"エージェント","quote":"AIモデルは数か月で進化します。GPUも1〜2年で世代交代します。ソフトウェアエージェントは週単位で能力を伸ばします。","quote_start":0,"quote_end":59,"text_sha256":"e01022ab4f667253de2221e5be58b3a9b02753f907ee667071e24543babb543f","block_sha256":"e01022ab4f667253de2221e5be58b3a9b02753f907ee667071e24543babb543f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_62acaca2-fec9-4724-be08-6ebd3c061f93/#blk_6b424449-d7ae-4753-a3e2-b66acffb57f5"},{"id":"occ_2e0854b09a472b77ab5aad4c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_62acaca2-fec9-4724-be08-6ebd3c061f93","work_id":"wrk_d43a6ffd-ef17-4cd2-96fa-06f7d800821a","block_id":"blk_a05a9cf2-bcbc-4bd5-8eec-df66b1ba5c33","section_id":"sec_9b8958e2-17e8-4334-9341-1a7b5d251b9e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":62,"end":68,"exact":"エージェント","quote":"ターでは、電気がGPUに入り、GPUが計算し、そこからトークン、画像、コード、設計案、創薬候補、ロボット制御、エージェントの判断が出てきます。つまり電力は、単なる経費ではなく、**知能を生産するための原材料**になります。","quote_start":7,"quote_end":118,"text_sha256":"2979126f9aa4b6922abfb56ae01ba86b32336866d4d15467269bee1c2f58f55f","block_sha256":"2979126f9aa4b6922abfb56ae01ba86b32336866d4d15467269bee1c2f58f55f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_62acaca2-fec9-4724-be08-6ebd3c061f93/#blk_a05a9cf2-bcbc-4bd5-8eec-df66b1ba5c33"},{"id":"occ_97940617f3834f6e22a96111","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","work_id":"wrk_3d2e988e-cac0-4000-b005-9292af7ac61f","block_id":"blk_0c12d3d1-8902-454c-a28f-1fdf3e0d52b5","section_id":"sec_01a2986a-1954-4587-be10-4dd1031a85c4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"⑤ エージェントID","quote_start":0,"quote_end":10,"text_sha256":"8fae3e7fe2de72fbe804786201fba9ef1e77b41acfe05f3ecbac5cbb5a005a7f","block_sha256":"8fae3e7fe2de72fbe804786201fba9ef1e77b41acfe05f3ecbac5cbb5a005a7f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_65271768-1b9c-4fbd-a486-d503b37eaf9a/#blk_0c12d3d1-8902-454c-a28f-1fdf3e0d52b5"},{"id":"occ_8199871809a532f45e40a830","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","work_id":"wrk_3d2e988e-cac0-4000-b005-9292af7ac61f","block_id":"blk_30e28a94-4cda-4892-bd59-ca60f7f0991c","section_id":"sec_c147ca6e-6ec8-4901-811d-c27993601721","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":0,"end":6,"exact":"エージェント","quote":"エージェントのツール実行","quote_start":0,"quote_end":12,"text_sha256":"755242a7e5ff7b07be633896b98ab9a2f59a6f9ba9eb8fd9b79681777052eeb4","block_sha256":"755242a7e5ff7b07be633896b98ab9a2f59a6f9ba9eb8fd9b79681777052eeb4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_65271768-1b9c-4fbd-a486-d503b37eaf9a/#blk_30e28a94-4cda-4892-bd59-ca60f7f0991c"},{"id":"occ_0a11f804c3e0ac476f32eb87","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","work_id":"wrk_3d2e988e-cac0-4000-b005-9292af7ac61f","block_id":"blk_366bf8bb-41d2-45cd-b0ed-ce0752555a7c","section_id":"sec_01a2986a-1954-4587-be10-4dd1031a85c4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":1,"end":7,"exact":"エージェント","quote":"他エージェントのなりすまし","quote_start":0,"quote_end":13,"text_sha256":"f2f5821cfbf01f86fa09c694b44fb11564fc3c74ae61d87d29ca1bc14af38524","block_sha256":"f2f5821cfbf01f86fa09c694b44fb11564fc3c74ae61d87d29ca1bc14af38524","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_65271768-1b9c-4fbd-a486-d503b37eaf9a/#blk_366bf8bb-41d2-45cd-b0ed-ce0752555a7c"},{"id":"occ_b7ca902f3d4bbac974a8c0c8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","work_id":"wrk_3d2e988e-cac0-4000-b005-9292af7ac61f","block_id":"blk_43299b73-c6b4-442c-9296-63e63b938caf","section_id":"sec_0608e593-7bc6-4685-9615-05449c4c1cdb","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"コーディングエージェント","quote_start":0,"quote_end":12,"text_sha256":"0edc70ffa3d875794574ba8b829ccf742e2bb2bbc895798ff7f1f5e7f6d2f7d6","block_sha256":"0edc70ffa3d875794574ba8b829ccf742e2bb2bbc895798ff7f1f5e7f6d2f7d6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_65271768-1b9c-4fbd-a486-d503b37eaf9a/#blk_43299b73-c6b4-442c-9296-63e63b938caf"},{"id":"occ_e645e768ffa607d6615cda54","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","work_id":"wrk_3d2e988e-cac0-4000-b005-9292af7ac61f","block_id":"blk_4c81364b-8c02-406f-b12a-82e7081c787c","section_id":"sec_01a2986a-1954-4587-be10-4dd1031a85c4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"④ エージェントRuntime","quote_start":0,"quote_end":15,"text_sha256":"cb0cfdbc2ecc11a6ef22cd825b2d341fdabef5dc896c88e9cab008d353ced73b","block_sha256":"cb0cfdbc2ecc11a6ef22cd825b2d341fdabef5dc896c88e9cab008d353ced73b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_65271768-1b9c-4fbd-a486-d503b37eaf9a/#blk_4c81364b-8c02-406f-b12a-82e7081c787c"},{"id":"occ_08964ceb6f539619cfd0b015","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","work_id":"wrk_3d2e988e-cac0-4000-b005-9292af7ac61f","block_id":"blk_5a540124-9bba-4426-9351-4bd5b95e2d7d","section_id":"sec_b112523c-6dd0-4447-bb50-e6c10ab52302","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":587,"end":593,"exact":"エージェント","quote":" | 遅延、価格、モデルを最適化 | 4～5 |\n| RAG | 社内データをAIへ接続 | 3～5 |\n| エージェント基盤 | AIへ記憶・権限・ツールを提供 | 5 |\n| AgentOps | AIの判断と行動を監視 | 5 |","quote_start":532,"quote_end":650,"text_sha256":"fc5352d114d20885d2510a17407d3b6a8960cfe3392890825816a17810d807f1","block_sha256":"fc5352d114d20885d2510a17407d3b6a8960cfe3392890825816a17810d807f1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_65271768-1b9c-4fbd-a486-d503b37eaf9a/#blk_5a540124-9bba-4426-9351-4bd5b95e2d7d"},{"id":"occ_4826ef00418292094966e14d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","work_id":"wrk_3d2e988e-cac0-4000-b005-9292af7ac61f","block_id":"blk_65188d6a-cb7b-4cd1-990f-d5b86632a1b0","section_id":"sec_27853913-9529-4ed7-9e0b-3315c512a8ae","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"企業がAIエージェントを一回動かすと、裏側では、","quote_start":0,"quote_end":24,"text_sha256":"eaa7eef7b02d9402c0b5f88b266cf902bd6595575ca39aeaadac352b7feab0f0","block_sha256":"eaa7eef7b02d9402c0b5f88b266cf902bd6595575ca39aeaadac352b7feab0f0","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_65271768-1b9c-4fbd-a486-d503b37eaf9a/#blk_65188d6a-cb7b-4cd1-990f-d5b86632a1b0"},{"id":"occ_00f5d83cb2dc546af893211d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","work_id":"wrk_3d2e988e-cac0-4000-b005-9292af7ac61f","block_id":"blk_6a818cec-3ed3-49c3-bbad-0ee74e215573","section_id":"sec_09f81988-704c-40e5-a5cc-21af37775a4e","layer":"character","character_id":"zetu_noia","count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":34,"end":40,"exact":"エージェント","quote":"クラウドはずっと「見えない場所」みたいに語られてきました。でも、AIエージェントが働き始めると、急に輪郭がはっきりします。誰のデータを使うのか。どの権限で動くのか。失敗したら誰が止めるのか。そこまで含めて初めて、知能は仕事になります。","quote_start":0,"quote_end":117,"text_sha256":"48e21065ab15272e218c0c928fc5429cf90d6f2348cb03a13baae0896d2a8464","block_sha256":"48e21065ab15272e218c0c928fc5429cf90d6f2348cb03a13baae0896d2a8464","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_65271768-1b9c-4fbd-a486-d503b37eaf9a/#blk_6a818cec-3ed3-49c3-bbad-0ee74e215573"},{"id":"occ_285d6133609e85e581f3592b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","work_id":"wrk_3d2e988e-cac0-4000-b005-9292af7ac61f","block_id":"blk_7838b6c6-8929-4ff7-8e9e-338a774ef1cc","section_id":"sec_fcadc2b5-0d78-4f61-ae8b-879a42dbbff4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントでは、GPU推論だけでなく、","quote_start":0,"quote_end":22,"text_sha256":"f0cb4c13e2e891149eb8727dcfa18d0247fdb745499260f7a701b0bd5d18a96b","block_sha256":"f0cb4c13e2e891149eb8727dcfa18d0247fdb745499260f7a701b0bd5d18a96b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_65271768-1b9c-4fbd-a486-d503b37eaf9a/#blk_7838b6c6-8929-4ff7-8e9e-338a774ef1cc"},{"id":"occ_41531284c1ae6c2c22cacec5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","work_id":"wrk_3d2e988e-cac0-4000-b005-9292af7ac61f","block_id":"blk_791cd0db-f720-4d26-b672-3f386eb633c4","section_id":"sec_b64cf3e1-2415-478b-8251-899626238f53","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":0,"end":7,"exact":"Agentic","quote":"Agentic Search","quote_start":0,"quote_end":14,"text_sha256":"8a843672ecdf9f6d49cc788cfed63f9fbbf55dc273ad53ed610b3a42df81c4eb","block_sha256":"8a843672ecdf9f6d49cc788cfed63f9fbbf55dc273ad53ed610b3a42df81c4eb","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_65271768-1b9c-4fbd-a486-d503b37eaf9a/#blk_791cd0db-f720-4d26-b672-3f386eb633c4"},{"id":"occ_eec1aa36cf280511c693f93c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","work_id":"wrk_3d2e988e-cac0-4000-b005-9292af7ac61f","block_id":"blk_7afb8b2c-a9e6-4d16-80e7-fbb342c51520","section_id":"sec_01a2986a-1954-4587-be10-4dd1031a85c4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":1,"end":7,"exact":"エージェント","quote":"他エージェントとの通信","quote_start":0,"quote_end":11,"text_sha256":"6632292d1c5c8a0f799836002929f07a763dd3a34c27e42ceb7438da0b4e37be","block_sha256":"6632292d1c5c8a0f799836002929f07a763dd3a34c27e42ceb7438da0b4e37be","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_65271768-1b9c-4fbd-a486-d503b37eaf9a/#blk_7afb8b2c-a9e6-4d16-80e7-fbb342c51520"},{"id":"occ_8d49a44833f4873365a070f8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","work_id":"wrk_3d2e988e-cac0-4000-b005-9292af7ac61f","block_id":"blk_7c211652-9ecc-46be-8b27-6238417631ba","section_id":"sec_01a2986a-1954-4587-be10-4dd1031a85c4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":0,"end":6,"exact":"エージェント","quote":"エージェント実行時間","quote_start":0,"quote_end":10,"text_sha256":"7d1e2149299e7fce886a099b68d9b9e2a4d77d9148661081953e1a98b63e4f34","block_sha256":"7d1e2149299e7fce886a099b68d9b9e2a4d77d9148661081953e1a98b63e4f34","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_65271768-1b9c-4fbd-a486-d503b37eaf9a/#blk_7c211652-9ecc-46be-8b27-6238417631ba"},{"id":"occ_8501d3b762f91e5ab63ccda1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","work_id":"wrk_3d2e988e-cac0-4000-b005-9292af7ac61f","block_id":"blk_b1fe7f70-3e5d-4562-ad1b-0fcb2c2bd8ff","section_id":"sec_fcadc2b5-0d78-4f61-ae8b-879a42dbbff4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェント","quote_start":0,"quote_end":8,"text_sha256":"ba5858157de142a20cbf66ff9ae8c325840a72c8466f631887eb7be25f64023d","block_sha256":"ba5858157de142a20cbf66ff9ae8c325840a72c8466f631887eb7be25f64023d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_65271768-1b9c-4fbd-a486-d503b37eaf9a/#blk_b1fe7f70-3e5d-4562-ad1b-0fcb2c2bd8ff"},{"id":"occ_4dabe415b256956e16d5e5f0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","work_id":"wrk_3d2e988e-cac0-4000-b005-9292af7ac61f","block_id":"blk_b9818bc8-a127-418b-a2e8-3e5f97e04ad3","section_id":"sec_1dacc9d9-8909-4c0f-8669-ab1b3fd70c82","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":15,"end":21,"exact":"エージェント","quote":"各社は現在、仮想マシンからAIエージェント実行環境までを提供している。GoogleはGemini Enterprise Agent PlatformでAgent Studio、ADK、Managed Runtime、Agent Identit","quote_start":0,"quote_end":121,"text_sha256":"5e9577c1508b041204feb880127b234cecc755679dd58f5081eccce508415451","block_sha256":"5e9577c1508b041204feb880127b234cecc755679dd58f5081eccce508415451","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_65271768-1b9c-4fbd-a486-d503b37eaf9a/#blk_b9818bc8-a127-418b-a2e8-3e5f97e04ad3"},{"id":"occ_5e60d7a7fc3defd2e5527a09","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","work_id":"wrk_3d2e988e-cac0-4000-b005-9292af7ac61f","block_id":"blk_bb897d61-ac13-45df-95b4-b5d92ba33082","section_id":"sec_a468854a-0a46-4505-8426-26caca8c981f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":93,"end":99,"exact":"エージェント","quote":"ェアだった。これからクラウド上で動くのは、情報を読み、状況を判断し、複数のシステムを操作し、仕事を完了するAIエージェントである。","quote_start":38,"quote_end":103,"text_sha256":"b1e6874bd4c78f5a6a94f881a7b458e30552f15d612ced1053f2195d00a66781","block_sha256":"b1e6874bd4c78f5a6a94f881a7b458e30552f15d612ced1053f2195d00a66781","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_65271768-1b9c-4fbd-a486-d503b37eaf9a/#blk_bb897d61-ac13-45df-95b4-b5d92ba33082"},{"id":"occ_f10c0b27b7d6a3736210e8cd","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","work_id":"wrk_3d2e988e-cac0-4000-b005-9292af7ac61f","block_id":"blk_c68325b0-47e0-4872-91fe-21f022201d3e","section_id":"sec_793b8052-e9c1-45bb-95fc-c1369fca1487","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":29,"end":35,"exact":"エージェント","quote":"企業データをBigQueryへ集め、Geminiで分析し、エージェントに行動させる構造が強い。","quote_start":0,"quote_end":47,"text_sha256":"fdccfd0e065409764a2d7e0ea7bf89a93d419a2dea7a570efc24777220231418","block_sha256":"fdccfd0e065409764a2d7e0ea7bf89a93d419a2dea7a570efc24777220231418","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_65271768-1b9c-4fbd-a486-d503b37eaf9a/#blk_c68325b0-47e0-4872-91fe-21f022201d3e"},{"id":"occ_6f129a76a9c244e11293d6c8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","work_id":"wrk_3d2e988e-cac0-4000-b005-9292af7ac61f","block_id":"blk_d8ae6826-e214-41ec-b136-3c5ef215d2dd","section_id":"sec_1dacc9d9-8909-4c0f-8669-ab1b3fd70c82","layer":"body","character_id":null,"count":2,"matched_aliases":["AI Agents","エージェント"],"evidence":{"text_basis":"markdown","start":1382,"end":1388,"exact":"エージェント","quote":"mini Enterprise Agent Platform | OCI Generative AI |\n| エージェント | AgentCore | Foundry Agent Service | Agent Platform、ADK | OCI AI Agents |","quote_start":1327,"quote_end":1463,"text_sha256":"10bb9994afba016b0c4d15c9f217ddf59b95a5119f95e526273cfdb3cfae2fb3","block_sha256":"10bb9994afba016b0c4d15c9f217ddf59b95a5119f95e526273cfdb3cfae2fb3","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_65271768-1b9c-4fbd-a486-d503b37eaf9a/#blk_d8ae6826-e214-41ec-b136-3c5ef215d2dd"},{"id":"occ_74fcd5d190a7388eb2074a02","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","work_id":"wrk_3d2e988e-cac0-4000-b005-9292af7ac61f","block_id":"blk_ed422a4d-47d8-43a6-88a1-c67feb0f5386","section_id":"sec_793b8052-e9c1-45bb-95fc-c1369fca1487","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":19,"end":25,"exact":"エージェント","quote":"モデル、独自チップ、データ基盤、検索、エージェントを垂直統合し、最もAIネイティブなクラウドを目指す。","quote_start":0,"quote_end":51,"text_sha256":"8b6a57575652997e21a7205004c07108ca60add382acbee50a3502b799d45427","block_sha256":"8b6a57575652997e21a7205004c07108ca60add382acbee50a3502b799d45427","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_65271768-1b9c-4fbd-a486-d503b37eaf9a/#blk_ed422a4d-47d8-43a6-88a1-c67feb0f5386"},{"id":"occ_f71f7b8b71f89df8c7c72999","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","work_id":"wrk_3d2e988e-cac0-4000-b005-9292af7ac61f","block_id":"blk_f580c05e-fbbc-43f8-a06d-e8fa54ae77e0","section_id":"sec_793b8052-e9c1-45bb-95fc-c1369fca1487","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":17,"end":23,"exact":"エージェント","quote":"インフラ、独自チップ、複数モデル、エージェント基盤を一体化し、企業がモデルに依存せずAIを構築できる場所を目指している。","quote_start":0,"quote_end":60,"text_sha256":"7645e832265273589974fc3e239ad41febcc27695e276c52c20eda891b59994d","block_sha256":"7645e832265273589974fc3e239ad41febcc27695e276c52c20eda891b59994d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_65271768-1b9c-4fbd-a486-d503b37eaf9a/#blk_f580c05e-fbbc-43f8-a06d-e8fa54ae77e0"},{"id":"occ_c0bacc2bac0b68b5ba54b93c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","work_id":"wrk_7b4fd6ec-e3a4-4a64-b21f-d3e9dbb0ed11","block_id":null,"section_id":null,"layer":"title","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"title","start":25,"end":33,"exact":"AI Agent","quote":"Personal Cloud Computer――AI Agentが「第二のPC」を持つ時代","quote_start":0,"quote_end":46,"text_sha256":"83be74274074ee9dd510c8ba5fc7d5674916570e7f9b7d65f1e8e2414b397fd2","block_sha256":null,"offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1/"},{"id":"occ_a6aa1f07100b404ea293ac40","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","work_id":"wrk_7b4fd6ec-e3a4-4a64-b21f-d3e9dbb0ed11","block_id":"blk_02cb2c12-6f9f-4e6d-abca-195d9f87e7a1","section_id":"sec_1e01cd9c-e659-479c-b85c-6122350a510e","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":70,"end":78,"exact":"AI Agent","quote":"ge  \n↓  \nConfidential CPU  \n↓  \nConfidential GPU  \n↓  \nAI Agent","quote_start":15,"quote_end":78,"text_sha256":"bc20c57048c4c62a5d7e50df8c49d849c8ee52807dbee4951b914644a90d8b1c","block_sha256":"bc20c57048c4c62a5d7e50df8c49d849c8ee52807dbee4951b914644a90d8b1c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1/#blk_02cb2c12-6f9f-4e6d-abca-195d9f87e7a1"},{"id":"occ_842fa91db9738661708fbd69","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","work_id":"wrk_7b4fd6ec-e3a4-4a64-b21f-d3e9dbb0ed11","block_id":"blk_137eda2b-454e-4f3f-aedd-ca6924d058e5","section_id":"sec_d8adb3a9-bc23-4818-8bec-8688901cacaf","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":28,"end":36,"exact":"AI Agent","quote":"## Personal Cloud Computer――AI Agentが「第二のPC」を持つ時代","quote_start":0,"quote_end":49,"text_sha256":"d6dc073d3a565d7d12af0e398e0c2a07f8797c66626d0d877bd0255056f95958","block_sha256":"d6dc073d3a565d7d12af0e398e0c2a07f8797c66626d0d877bd0255056f95958","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1/#blk_137eda2b-454e-4f3f-aedd-ca6924d058e5"},{"id":"occ_fde479fbcfab603b14c8c9f7","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","work_id":"wrk_7b4fd6ec-e3a4-4a64-b21f-d3e9dbb0ed11","block_id":"blk_4895c4bb-2bcb-469f-880a-057ff9e3f287","section_id":"sec_445dd2d6-2027-4d37-9ec5-08054b6f15b1","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":6,"end":13,"exact":"Agentic","quote":"## 36．Agentic OSとは何か","quote_start":0,"quote_end":20,"text_sha256":"0f4f95fd4200305e20fe38ef53c64cb5527ce72de81eb24519538a83a6613a49","block_sha256":"0f4f95fd4200305e20fe38ef53c64cb5527ce72de81eb24519538a83a6613a49","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1/#blk_4895c4bb-2bcb-469f-880a-057ff9e3f287"},{"id":"occ_1f3d1b17fe2c2abc1b2f940f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","work_id":"wrk_7b4fd6ec-e3a4-4a64-b21f-d3e9dbb0ed11","block_id":"blk_5bdab75b-fcd3-4cd4-b67f-6080a1c38f08","section_id":"sec_543cf9ee-b8a4-4562-91e3-1661e365f2b4","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agentが「第二のPC」を持つ時代は、遠いSFではない。","quote_start":0,"quote_end":32,"text_sha256":"011343f749534b9b822de37c53d39e4545f197218baa8f96fe4744cfa0e31d12","block_sha256":"011343f749534b9b822de37c53d39e4545f197218baa8f96fe4744cfa0e31d12","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1/#blk_5bdab75b-fcd3-4cd4-b67f-6080a1c38f08"},{"id":"occ_aff6beb651432112ffeb6bd5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","work_id":"wrk_7b4fd6ec-e3a4-4a64-b21f-d3e9dbb0ed11","block_id":"blk_6f55f92d-7b05-473e-94ab-729fd8668f8f","section_id":"sec_445dd2d6-2027-4d37-9ec5-08054b6f15b1","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":3,"end":10,"exact":"Agentic","quote":"つまりAgentic OSは完全に未知の技術ではない。","quote_start":0,"quote_end":27,"text_sha256":"adbef6c42c9d76c388aef361976755b1a71f87eed1d4463f12e13f2b5d2330a7","block_sha256":"adbef6c42c9d76c388aef361976755b1a71f87eed1d4463f12e13f2b5d2330a7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1/#blk_6f55f92d-7b05-473e-94ab-729fd8668f8f"},{"id":"occ_140133cc44be1caa68c3c42a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","work_id":"wrk_7b4fd6ec-e3a4-4a64-b21f-d3e9dbb0ed11","block_id":"blk_863c1178-b182-45d7-88d5-35c90ac953b0","section_id":"sec_445dd2d6-2027-4d37-9ec5-08054b6f15b1","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":18,"end":26,"exact":"AI Agent","quote":"**何十年もOSが解いてきた問題を、AI Agent用に再構成するもの**","quote_start":0,"quote_end":37,"text_sha256":"1d09a7716082875089ae14a00aa45afe64b294f4b8fbe544f7a452f33874c677","block_sha256":"1d09a7716082875089ae14a00aa45afe64b294f4b8fbe544f7a452f33874c677","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1/#blk_863c1178-b182-45d7-88d5-35c90ac953b0"},{"id":"occ_84942d2777c75ae7b9fba04e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","work_id":"wrk_7b4fd6ec-e3a4-4a64-b21f-d3e9dbb0ed11","block_id":"blk_8c646266-b617-4c5f-abff-10a5a6cd6a66","section_id":"sec_4875c2f0-5781-4770-86f9-8e218a843d22","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":6,"end":14,"exact":"AI Agent","quote":"## 39．AI Agentは「第二のPC」を持つ","quote_start":0,"quote_end":25,"text_sha256":"8ce94c9bf99cd5d2bd86ad3e455fe6eb977b6f6ab1a5359232ba45fb45e1a9a1","block_sha256":"8ce94c9bf99cd5d2bd86ad3e455fe6eb977b6f6ab1a5359232ba45fb45e1a9a1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1/#blk_8c646266-b617-4c5f-abff-10a5a6cd6a66"},{"id":"occ_a4df2a12aeea4a63bca79329","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","work_id":"wrk_7b4fd6ec-e3a4-4a64-b21f-d3e9dbb0ed11","block_id":"blk_9831b8f8-ab6e-42d0-94ab-5ae6793b66da","section_id":"sec_d8adb3a9-bc23-4818-8bec-8688901cacaf","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":49,"end":57,"exact":"AI Agent","quote":"自分専用のStorage、Memory、CPU、GPU、Software Environment、AI AgentなどがCloud上に存在し、必要な時だけ計算資源を呼び出して使う、新しいPersonal Computingの形である。","quote_start":0,"quote_end":117,"text_sha256":"5f5cf630e9e6fae1e9238ef750d40b1c908e806fb32d9e4a15c908ccf859b92a","block_sha256":"5f5cf630e9e6fae1e9238ef750d40b1c908e806fb32d9e4a15c908ccf859b92a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1/#blk_9831b8f8-ab6e-42d0-94ab-5ae6793b66da"},{"id":"occ_b41e6298f3d88b27162193f3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","work_id":"wrk_7b4fd6ec-e3a4-4a64-b21f-d3e9dbb0ed11","block_id":"blk_bebfd281-b97f-4d81-b85d-aee26341bf0f","section_id":"sec_445dd2d6-2027-4d37-9ec5-08054b6f15b1","layer":"body","character_id":null,"count":2,"matched_aliases":["AI Agent","Agentic"],"evidence":{"text_basis":"markdown","start":0,"end":7,"exact":"Agentic","quote":"Agentic OS――AI Agent中心のOperating System的基盤。  \nAgent、Memory、Tool、Permission、Compute、Networkなどを統合管理するLayer。","quote_start":0,"quote_end":105,"text_sha256":"011af63ca71a091765cb43c970b047dca61dbdf17e5237a7b2865b2004c5e48f","block_sha256":"011af63ca71a091765cb43c970b047dca61dbdf17e5237a7b2865b2004c5e48f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1/#blk_bebfd281-b97f-4d81-b85d-aee26341bf0f"},{"id":"occ_f05f770eae59be978b8933c1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","work_id":"wrk_7b4fd6ec-e3a4-4a64-b21f-d3e9dbb0ed11","block_id":"blk_df2478fc-1aad-48d7-b760-771e4f3ba856","section_id":"sec_dbd6a877-3e0e-4533-81d2-6127f0b2544d","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":2,"end":10,"exact":"AI Agent","quote":"**AI Agent向けCloud OS**","quote_start":0,"quote_end":22,"text_sha256":"5583411ae87234eabc30b9a1d068e12995d765f5a3b8df47c172d0a6614d7f16","block_sha256":"5583411ae87234eabc30b9a1d068e12995d765f5a3b8df47c172d0a6614d7f16","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1/#blk_df2478fc-1aad-48d7-b760-771e4f3ba856"},{"id":"occ_21dbc6ecf4cae950e83de17f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","work_id":"wrk_7b4fd6ec-e3a4-4a64-b21f-d3e9dbb0ed11","block_id":"blk_e1367e0f-dd22-4ec2-ab0e-5ed9f3676100","section_id":"sec_d8adb3a9-bc23-4818-8bec-8688901cacaf","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agentについて考えていると、ある対談の一節が妙に引っ掛かる。","quote_start":0,"quote_end":35,"text_sha256":"27fc3aa9e1cb9473633beae200db1cb4c0b047d9453fe149e306ecd1545942ad","block_sha256":"27fc3aa9e1cb9473633beae200db1cb4c0b047d9453fe149e306ecd1545942ad","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1/#blk_e1367e0f-dd22-4ec2-ab0e-5ed9f3676100"},{"id":"occ_a22222f03a097f6cc22f3b5e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","work_id":"wrk_7b4fd6ec-e3a4-4a64-b21f-d3e9dbb0ed11","block_id":"blk_e1e0dc6a-39dd-4109-a427-56d25e9608c3","section_id":"sec_fe7fb8b1-9e93-4dd6-8f74-1ee7185c96cc","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agent時代には最後の、","quote_start":0,"quote_end":16,"text_sha256":"9e0cad80d3659684419fa4cfffd3b95dbdfe86e496ccca2181ac902bf476f8a2","block_sha256":"9e0cad80d3659684419fa4cfffd3b95dbdfe86e496ccca2181ac902bf476f8a2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1/#blk_e1e0dc6a-39dd-4109-a427-56d25e9608c3"},{"id":"occ_7afe5d20f34268919ba4e3d2","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6abb0df2-b5a4-4f1c-8f48-a346d6d0f247","work_id":"wrk_37d80f4d-ab1e-4385-b7ef-e7ae09b765e3","block_id":"blk_922ffadf-6cd5-4c6b-9cc3-01cfd8e686da","section_id":"sec_6004fd05-570c-474f-81a4-4970fe51893d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":14,"end":20,"exact":"エージェント","quote":"Cloudflareが、AIエージェント、API、マルチクラウド、Zero Trust、PQC、WAF、DDoSを統合した中立セキュリティ基盤になる。  \nこの場合、Cloudflareは単なるCDNではなく、**インターネット横断のAI時","quote_start":0,"quote_end":120,"text_sha256":"0c8d8a8263546371a86e47cba6ae9830be5ee44c71a3f283650d3897b73d4e3f","block_sha256":"0c8d8a8263546371a86e47cba6ae9830be5ee44c71a3f283650d3897b73d4e3f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6abb0df2-b5a4-4f1c-8f48-a346d6d0f247/#blk_922ffadf-6cd5-4c6b-9cc3-01cfd8e686da"},{"id":"occ_a628838f1c2357cd364ef6a7","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6abb0df2-b5a4-4f1c-8f48-a346d6d0f247","work_id":"wrk_37d80f4d-ab1e-4385-b7ef-e7ae09b765e3","block_id":"blk_c3023eb0-de17-45e0-86bb-e0384b550588","section_id":"sec_651cde3b-3162-44c3-ae33-61c4b1d02101","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":30,"end":36,"exact":"エージェント","quote":"AI時代には、通信する主体が人間だけではなくなる。  \nAIエージェント、API、ロボット、工場機器、SaaS、モデル、データベース、クラウドが相互に通信する。  \nこの通信を、誰が認証し、誰が制限し、誰が観測し、誰が止めるのか。","quote_start":0,"quote_end":115,"text_sha256":"456bc08fec2be82542d05b6ec83ad47d4a836a90cb98ef56419dfc655ae4ebcb","block_sha256":"456bc08fec2be82542d05b6ec83ad47d4a836a90cb98ef56419dfc655ae4ebcb","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6abb0df2-b5a4-4f1c-8f48-a346d6d0f247/#blk_c3023eb0-de17-45e0-86bb-e0384b550588"},{"id":"occ_9652f0a57c58e47f8aa036db","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6abb0df2-b5a4-4f1c-8f48-a346d6d0f247","work_id":"wrk_37d80f4d-ab1e-4385-b7ef-e7ae09b765e3","block_id":"blk_ce0e9d5d-8eab-40b2-b9af-953d13981760","section_id":"sec_ade59278-f300-4cb6-b712-9a6f2cedbb66","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":113,"end":119,"exact":"エージェント","quote":"API保護、Zero Trust、AI Gateway、Workers、量子耐性暗号、ログ分析を束ね、**AIエージェント時代の関所**にならなければならない。","quote_start":58,"quote_end":138,"text_sha256":"5bb6adfd173b7fd30d0c688a7ba8c76f7e59cfbfce6c33a21d758347f222c86d","block_sha256":"5bb6adfd173b7fd30d0c688a7ba8c76f7e59cfbfce6c33a21d758347f222c86d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6abb0df2-b5a4-4f1c-8f48-a346d6d0f247/#blk_ce0e9d5d-8eab-40b2-b9af-953d13981760"},{"id":"occ_f980ec67e41b45a9750cbba3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_034209f8-87f4-4599-96e6-850fed48087b","section_id":"sec_24b4986d-fd60-4bbf-841a-1d7a353fdecc","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":28,"end":34,"exact":"エージェント","quote":"この流れが進むと、人間のGPUエンジニアだけでなく、AIエージェントがGPU kernelや推論設定を探索し続けるようになります。  \nこれは、GPUクラウド成熟をかなり速めます。","quote_start":0,"quote_end":90,"text_sha256":"d6659ddd2e90eee4abe3c36f05e375fbe2c32d3cfd816d422e6bd93f52a4f068","block_sha256":"d6659ddd2e90eee4abe3c36f05e375fbe2c32d3cfd816d422e6bd93f52a4f068","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_034209f8-87f4-4599-96e6-850fed48087b"},{"id":"occ_68a966e8470362248063f408","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_049e0d19-5465-4b0f-8c74-26b493fa9463","section_id":"sec_b5001920-30e7-4546-b243-794b28e5a3ea","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"これは、AIエージェント時代にかなり重要です。なぜならエージェントは、単発のチャットではなく、ツール呼び出し、検索、JSON生成、再推論、長文コンテキストを繰り返すからです。","quote_start":0,"quote_end":87,"text_sha256":"bc23c7534245f68b5083490624bb7130f72f2db6256accd53fa1d320f6f4c21e","block_sha256":"bc23c7534245f68b5083490624bb7130f72f2db6256accd53fa1d320f6f4c21e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_049e0d19-5465-4b0f-8c74-26b493fa9463"},{"id":"occ_fcaa0f0961ebc03c0557eb8b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_0ca90cca-643c-4f37-9394-c9e9d9ffb631","section_id":"sec_f33e4986-221e-4ec4-a922-4302306101f0","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"- AIエージェント\n- コーディングAI\n- 動画生成\n- 画像生成\n- 音声AI\n- リアルタイム翻訳\n- ロボット\n- 工場AI\n- セキュリティAI\n- 個人専用AI\n- 企業内AIワーカー","quote_start":0,"quote_end":99,"text_sha256":"b5c199f7eea4979d6700f8fbc4da87932428b9cc29e0b2b6e419373dc32124db","block_sha256":"b5c199f7eea4979d6700f8fbc4da87932428b9cc29e0b2b6e419373dc32124db","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_0ca90cca-643c-4f37-9394-c9e9d9ffb631"},{"id":"occ_65449f2519fafb68b97b0e93","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_0d2ff0fc-ab99-4be5-8b84-0f1f01c5f76f","section_id":"sec_7e6570d5-47c1-4c8d-8c40-b67e49433c73","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":17,"end":23,"exact":"エージェント","quote":"しかし、AI推論が安くなると、AIエージェント、動画生成、音声AI、企業AI、ロボット、RAG、個人AI秘書が爆発的に増えます。すると、CPU側では次の処理が増えます。","quote_start":0,"quote_end":84,"text_sha256":"5cbe3cc5c24b86df299307f9f592f99cf1955eb4ce96a1c39c2edaee4c99a438","block_sha256":"5cbe3cc5c24b86df299307f9f592f99cf1955eb4ce96a1c39c2edaee4c99a438","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_0d2ff0fc-ab99-4be5-8b84-0f1f01c5f76f"},{"id":"occ_961125445e28edefe40c2234","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_0fa8be5c-2e50-4ad2-8fdc-b3847f1a7ff1","section_id":"sec_be81384c-22d0-4e73-b24c-902356664441","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":106,"end":112,"exact":"エージェント","quote":"するため、総GPU需要は増える。\n\n中期：\nGPUクラウドOS化でGPU利用率が改善。\n推論単価が下がり、AIエージェントが普及。\n総GPU需要はまだ増えやすい。\n\n長期：\nTPU、LPU、ASIC、NPU、エッジAIが普及。\n単純推論向けGPU需要の伸びは鈍化。\nGPUは動画生成、学習、VLA、科学AIなど高負荷領域に","quote_start":51,"quote_end":212,"text_sha256":"11effc30d734d8356f2dc036565a9682577b57b11b8994de8f269e912e1004cc","block_sha256":"11effc30d734d8356f2dc036565a9682577b57b11b8994de8f269e912e1004cc","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_0fa8be5c-2e50-4ad2-8fdc-b3847f1a7ff1"},{"id":"occ_30f6371de9c86d2e7abb4043","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_1214813e-0163-4061-a060-e6d3b763c604","section_id":"sec_d0aef33e-d23a-416c-875c-f996b91b97c2","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":48,"end":54,"exact":"エージェント","quote":"用途LPUクラウドとの相性LLMリアルタイム推論非常に良いチャットAI良い音声対話AI良い低遅延エージェント良いコーディング補助良いAPI型LLM推論良い大規模学習基本的に不向き画像生成・動画生成GPUほど汎用ではない任意のCUDAアプリ不向き研究用途の柔軟なモデル改造GPU/TPUの方が向く","quote_start":0,"quote_end":147,"text_sha256":"16339a97eeaa9361c4bc9a94f07fc2a05613ec87d35df8972773c132472742d8","block_sha256":"16339a97eeaa9361c4bc9a94f07fc2a05613ec87d35df8972773c132472742d8","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_1214813e-0163-4061-a060-e6d3b763c604"},{"id":"occ_310ca66696c36ec02809400f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_130197db-ad87-442d-a00e-2609c04ce4f0","section_id":"sec_71d97541-7ee4-426e-8210-eb8853969707","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":45,"end":51,"exact":"エージェント","quote":"> **GPUクラウドは2026〜2028年にクラウドOS化が始まり、2030年前後にAIエージェント時代の本格インフラになる。  \n> そしてAI自身が、その成熟速度をさらに加速する。**","quote_start":0,"quote_end":95,"text_sha256":"3c4627799559f85ce73802980497f3a8a45a4808802ae56e3a575ad63319bcf8","block_sha256":"3c4627799559f85ce73802980497f3a8a45a4808802ae56e3a575ad63319bcf8","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_130197db-ad87-442d-a00e-2609c04ce4f0"},{"id":"occ_df26b30e3438f1d45b35fa4e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_17e8e0e2-a146-42e4-8faf-048e2d61d69b","section_id":"sec_7e6570d5-47c1-4c8d-8c40-b67e49433c73","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":21,"end":28,"exact":"agentic","quote":"AMDも2026年第1四半期決算で、推論とagentic AIが高性能CPUとアクセラレータ需要を押し上げていると述べています。([Advanced Micro Devices, Inc.](https://ir.amd.com/news-events/p","quote_start":0,"quote_end":128,"text_sha256":"94c955c1b082526905aa497894e43f7c1168a13180525cc2b8f9c9afac005c57","block_sha256":"94c955c1b082526905aa497894e43f7c1168a13180525cc2b8f9c9afac005c57","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_17e8e0e2-a146-42e4-8faf-048e2d61d69b"},{"id":"occ_b6a7fa26d84ad86bc1aa2547","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_199ce646-6c41-494c-a42a-4c793aa2e027","section_id":"sec_b24b777b-a086-4096-a6f7-196e819d6c4d","layer":"code","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":63,"end":69,"exact":"エージェント","quote":"ラウド\n  - 認証\n  - 課金\n  - API\n  - DB\n  - ログ\n  - セキュリティ\n  - エージェント管理\n  - クラウド制御\n\nGPUクラウド\n  - 大規模学習\n  - 汎用推論\n  - 画像生成\n  - 動画生成\n  - VLA\n  - 科学計算\n  - マルチモーダルAI\n\nTPUクラウド\n","quote_start":8,"quote_end":169,"text_sha256":"a7d347f909b3b572f523eb3eab756432a28522cce72cb54f22bdabb6210127c6","block_sha256":"a7d347f909b3b572f523eb3eab756432a28522cce72cb54f22bdabb6210127c6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_199ce646-6c41-494c-a42a-4c793aa2e027"},{"id":"occ_ccaba9a91c6327b1ce5f4a12","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_19b4d009-8f4d-40d4-892a-25ee2089861d","section_id":"sec_215e4155-9814-4b27-8f49-1b247c3e1472","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":195,"end":201,"exact":"エージェント","quote":"組み\n- GPUごとの本当の原価計算\n- サイドチャネル対策\n- 悪意あるGPU kernelの制限\n- AIエージェント単位の課金と状態管理","quote_start":140,"quote_end":211,"text_sha256":"118d431d77ffd4d4bf54506c5884ddd02d1909bafb2459054251a0e1cabe974f","block_sha256":"118d431d77ffd4d4bf54506c5884ddd02d1909bafb2459054251a0e1cabe974f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_19b4d009-8f4d-40d4-892a-25ee2089861d"},{"id":"occ_f034dff429fc1c287091dd10","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_1c757337-113d-42cc-bd8e-b14cb03c6426","section_id":"sec_7f34bca6-4bc4-4829-a7ce-c78aa72625d1","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":47,"end":53,"exact":"エージェント","quote":"- 音声AI\n- AI秘書\n- カスタマーサポート\n- コーディング補助\n- リアルタイム検索エージェント\n- 軽量業務エージェント\n- ゲームNPC\n- 配信AIキャラ\n- AITuber","quote_start":0,"quote_end":95,"text_sha256":"ee46128c254ae22c9ecfb4ec7121c2f71b233d8974ed51d6119f01cbb763868e","block_sha256":"ee46128c254ae22c9ecfb4ec7121c2f71b233d8974ed51d6119f01cbb763868e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_1c757337-113d-42cc-bd8e-b14cb03c6426"},{"id":"occ_49c66f2c5e7f19e717fdbdd6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_248a61f1-85f3-4b4c-95e2-f0bc78e76ef2","section_id":"sec_4fad2077-ba11-41e3-b84f-692496cb220a","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":113,"end":119,"exact":"エージェント","quote":" TPUクラウド\n\n汎用AI開発、画像生成、動画生成\n  → GPUクラウド\n\n低遅延チャット、音声対話、軽量エージェント\n  → LPUクラウド\n\n企業向けAI推論API\n  → GPU / LPU / TPU の混在\n\nフィジカルAI・VLA学習\n  → GPUクラウド中心、一部TPU/ASIC\n\nエッジAI\n  →","quote_start":58,"quote_end":219,"text_sha256":"2456538722254db1f69cc8675621c32e6331c79dbbb6b30dce00da0586554513","block_sha256":"2456538722254db1f69cc8675621c32e6331c79dbbb6b30dce00da0586554513","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_248a61f1-85f3-4b4c-95e2-f0bc78e76ef2"},{"id":"occ_67140877c542a7bf5b85aa72","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_2ad0b374-7cb1-412f-9d6b-990fd734b2dc","section_id":"sec_8c386d0d-3385-4a7f-8017-702e5566fdbf","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":67,"end":73,"exact":"エージェント","quote":"違い、主に**推論特化**です。特に、音声対話、リアルタイムチャット、AITuber、カスタマーサポート、軽量エージェントのように、1トークンごとの応答速度が重要な用途に向いています。","quote_start":12,"quote_end":104,"text_sha256":"0346996695f22f54a03f4bd70d45e5afcc70e9d39048f985b0875ff5d81c9937","block_sha256":"0346996695f22f54a03f4bd70d45e5afcc70e9d39048f985b0875ff5d81c9937","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_2ad0b374-7cb1-412f-9d6b-990fd734b2dc"},{"id":"occ_a283d60060433f6ee232b087","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_2ce36db3-fe2b-4381-8afd-a346c1eb7c24","section_id":"sec_7f34bca6-4bc4-4829-a7ce-c78aa72625d1","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":9,"end":15,"exact":"エージェント","quote":"## \\9. AIエージェント時代にLPUクラウドは重要か？","quote_start":0,"quote_end":30,"text_sha256":"96205dfd9f25a1e9512261be653c3a742459769e9c210beaefd05e96a427a1a0","block_sha256":"96205dfd9f25a1e9512261be653c3a742459769e9c210beaefd05e96a427a1a0","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_2ce36db3-fe2b-4381-8afd-a346c1eb7c24"},{"id":"occ_e6a7cb191376ed7ce93f597e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_2e1a6cc9-33fd-4c6e-9643-bb703ab02eaa","section_id":"sec_c32c7c49-725a-4976-8a9d-913de857bee5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":53,"end":59,"exact":"エージェント","quote":"なので、GPUクラウド成熟によって、**AI推論そのもののボトルネックは緩和される**。  \nしかし、**エージェント社会を動かす周辺インフラ需要はむしろ増える**。","quote_start":0,"quote_end":83,"text_sha256":"f0f78d9125e9619e4b9783ead207fea12819caad91764adc47c74bfee467fd68","block_sha256":"f0f78d9125e9619e4b9783ead207fea12819caad91764adc47c74bfee467fd68","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_2e1a6cc9-33fd-4c6e-9643-bb703ab02eaa"},{"id":"occ_ce9179b29065c0ae92dbc27c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_2ec7b8d9-68ed-4b25-89e0-f97528b45afe","section_id":"sec_24b4986d-fd60-4bbf-841a-1d7a353fdecc","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":18,"end":24,"exact":"エージェント","quote":"また、PyTorchブログでも、深いエージェントを使った自律GPU kernel生成が紹介されており、Triton kernelの生成、検証、実行を自動化する方向が示されています。([PyTorch](https://pytorch.org/blo","quote_start":0,"quote_end":124,"text_sha256":"256f97420a152e93a3cb276628e98e7885c97745c859f47dfffdf43802fe6f48","block_sha256":"256f97420a152e93a3cb276628e98e7885c97745c859f47dfffdf43802fe6f48","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_2ec7b8d9-68ed-4b25-89e0-f97528b45afe"},{"id":"occ_9a102ce7c163df708e53e6d7","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_30e3b87b-331a-4175-b074-40738c5d2955","section_id":"sec_717c016c-0779-4ba9-83ae-9221ccf5cb31","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":68,"end":74,"exact":"エージェント","quote":"げ期はCPU需要が急増**  \n> **GPUクラウド成熟期はCPU/GPU比率が最適化**  \n> **AIエージェント普及期はCPUが“管制塔・接続・防衛”として再拡大**  \n> **2030年代後半はエッジ/専用チップに一部移るが、クラウドCPUは残る**","quote_start":13,"quote_end":145,"text_sha256":"ba7c9f379d9f0061eacd53fd3b83b66b533ca58f881887458d88aa3d9e9888c0","block_sha256":"ba7c9f379d9f0061eacd53fd3b83b66b533ca58f881887458d88aa3d9e9888c0","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_30e3b87b-331a-4175-b074-40738c5d2955"},{"id":"occ_2a5af3fa1464f38c6eb27a9a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_35300359-9921-408f-b8ed-2e7978d764e7","section_id":"sec_8c386d0d-3385-4a7f-8017-702e5566fdbf","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":78,"end":84,"exact":"エージェント","quote":"用推論：GPU / TPU\n低遅延会話：LPU\n画像・動画生成：GPU\nGoogle系大規模AI：TPU\nAIエージェントの軽量推論：GPU / LPU\n企業業務AI：CPU + GPU/TPU/LPU + DB + セキュリティ\n```","quote_start":23,"quote_end":142,"text_sha256":"d50c4e8a37f1f1994764c68c51402d2883f1f004efee01f119f3f4c75b51b4ea","block_sha256":"d50c4e8a37f1f1994764c68c51402d2883f1f004efee01f119f3f4c75b51b4ea","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_35300359-9921-408f-b8ed-2e7978d764e7"},{"id":"occ_77f82fc25e0e5c9cf7e636f5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_36b9bb17-5112-41b2-8f21-594991625dec","section_id":"sec_58745122-69e4-4bc2-ae16-3bae5e2bfd75","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":9,"end":15,"exact":"エージェント","quote":"## \\1. AIエージェントが増えると、何の処理が増えるのか","quote_start":0,"quote_end":31,"text_sha256":"55842a0c932dd9dcdeb80f6c1c89fc7465e0ec1e7b6742e947fc7a3ba5f5e796","block_sha256":"55842a0c932dd9dcdeb80f6c1c89fc7465e0ec1e7b6742e947fc7a3ba5f5e796","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_36b9bb17-5112-41b2-8f21-594991625dec"},{"id":"occ_0761405fb9247baf9dfe91f8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_38bc6cd5-7e9f-4086-affd-bb5f5f70a091","section_id":"sec_67c7d199-a2f4-4058-b6a0-18b296e20e76","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":32,"end":38,"exact":"エージェント","quote":"```\nGPUクラウド成熟\n  ↓\n推論コスト低下\n  ↓\nAIエージェント利用増加\n  ↓\nAPI・DB・認証・ログ・監査・課金が増える\n  ↓\nCPU需要は高止まり\n```","quote_start":0,"quote_end":88,"text_sha256":"eb897f05d2d0b50d937fd0a56c9e9d148be677048fba5dada9ff819787610452","block_sha256":"eb897f05d2d0b50d937fd0a56c9e9d148be677048fba5dada9ff819787610452","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_38bc6cd5-7e9f-4086-affd-bb5f5f70a091"},{"id":"occ_d04e02562b76110b1f083a0a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_3c37aeb8-32bd-4755-b81f-bc6490e43e90","section_id":"sec_c32c7c49-725a-4976-8a9d-913de857bee5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":9,"end":15,"exact":"エージェント","quote":"増える処理主な担当エージェントの状態管理CPU/DBワークフロー管理CPUツール実行CPUAPI連携CPU/ネットワーク権限管理CPUログ・監査CPU/ストレージ課金CPUセキュリティCPU/DPUブラウザ操作CPURAG検索CP","quote_start":0,"quote_end":115,"text_sha256":"6f3f68680304b1ca158a3f8d67929c44129ee2ad696cf10d1ff4a363700da82d","block_sha256":"6f3f68680304b1ca158a3f8d67929c44129ee2ad696cf10d1ff4a363700da82d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_3c37aeb8-32bd-4755-b81f-bc6490e43e90"},{"id":"occ_904ea9311eaf57a8e6f5f7ab","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_3e16c1aa-5c77-4f94-a313-2f2e80c61f59","section_id":"sec_60a8a145-1f54-4dc1-9e65-16cc2eb54c44","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"特に企業内エージェントでは、AIが答えを出すよりも、  \n**社内システムに安全に接続し、操作し、記録し、承認を通す部分**  \nの方が重くなる可能性があります。","quote_start":0,"quote_end":81,"text_sha256":"47fc2a0151584d73c6cbdf3fe6b3199bc8fc9155e28739c588d4eccb5689ed0e","block_sha256":"47fc2a0151584d73c6cbdf3fe6b3199bc8fc9155e28739c588d4eccb5689ed0e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_3e16c1aa-5c77-4f94-a313-2f2e80c61f59"},{"id":"occ_7694591f65d93830ee79753c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_40bc7310-7f44-4eb5-9e4d-4d8967f12c00","section_id":"sec_b3015863-c634-42f8-bf90-57c765dba5e5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":3,"end":9,"exact":"エージェント","quote":"大量のエージェントが同時に動くと、リクエストの長さもタイミングもバラバラになります。","quote_start":0,"quote_end":42,"text_sha256":"d57f05b6612b21c56a9848a98468f9f564218eb39e1d674b5e213b281d2233fc","block_sha256":"d57f05b6612b21c56a9848a98468f9f564218eb39e1d674b5e213b281d2233fc","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_40bc7310-7f44-4eb5-9e4d-4d8967f12c00"},{"id":"occ_4dc51dd5e3c4fc9605517ddc","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_41643df3-baff-4e6d-9d2f-47772556fbd9","section_id":"sec_71d97541-7ee4-426e-8210-eb8853969707","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":44,"end":50,"exact":"エージェント","quote":"> **2026〜2028年：完成しはじめる**  \n> **2028〜2030年：AIエージェント向けに本格実用化**  \n> **2029〜2032年：CPUクラウドに近い成熟度へ進む**  \n> **2030年代前半：AI社会の標準インフラ化**","quote_start":0,"quote_end":126,"text_sha256":"41c89094382d9f0755d21c4a01c32043b6dc8968de3e78d11b35a86aa6aadf6c","block_sha256":"41c89094382d9f0755d21c4a01c32043b6dc8968de3e78d11b35a86aa6aadf6c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_41643df3-baff-4e6d-9d2f-47772556fbd9"},{"id":"occ_a08eb4d37eafa4f6c8358ce8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_418529d7-5985-4509-9707-53fab5936aa9","section_id":"sec_36a188e4-7f51-4d03-afae-df59351b5013","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"つまり、AIエージェントが増えるほど、  \n**GPUクラウドとCPUクラウドの両方が拡大する**  \n可能性が高いです。","quote_start":0,"quote_end":61,"text_sha256":"f213636fb0bb56c1d550ff9c43864265b6cc5046c2db1f1c09f81d09fb15122e","block_sha256":"f213636fb0bb56c1d550ff9c43864265b6cc5046c2db1f1c09f81d09fb15122e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_418529d7-5985-4509-9707-53fab5936aa9"},{"id":"occ_bb692c2022d245027e82e84a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_483e5a3b-1e80-4317-bc5d-aab3115be2fd","section_id":"sec_f3f6bb06-b5e5-4d2b-ab7a-4837dfe6d756","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェント時代の追加需要を100とすると、GPUクラウド成熟で吸収・効率化できる部分はこう見ます。","quote_start":0,"quote_end":52,"text_sha256":"3f289fb455ffcc35775b17711dda19a4656bf5aec854f7212df6d50d9e18b68a","block_sha256":"3f289fb455ffcc35775b17711dda19a4656bf5aec854f7212df6d50d9e18b68a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_483e5a3b-1e80-4317-bc5d-aab3115be2fd"},{"id":"occ_0544ec48e49e51df41f887fb","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_4918c7fc-a59b-4484-8ae6-73ffd2ba502d","section_id":"sec_e5e44156-0b06-4dce-8372-0aca3f2d053a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":0,"end":6,"exact":"エージェント","quote":"エージェントは長い入力を読み、長い出力を出し、途中で何度も再推論します。  \nそのため、prefillとdecodeを分けて最適化する効果が大きいです。","quote_start":0,"quote_end":76,"text_sha256":"2d8bfbde0920c4121be89f06d5d3e7d756d9b6539aac01f1dffe6147d0cda19e","block_sha256":"2d8bfbde0920c4121be89f06d5d3e7d756d9b6539aac01f1dffe6147d0cda19e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_4918c7fc-a59b-4484-8ae6-73ffd2ba502d"},{"id":"occ_18b327eaf6df216902f5e701","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_49c45a6a-1541-421a-88ef-0414f7041d8b","section_id":"sec_e848089f-b5fd-482c-a1e7-5ab19f8eb838","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":69,"end":75,"exact":"エージェント","quote":"U需要の「伸び率」としては最も強く見える可能性が高いです。理由は、GPUクラウドがまだ完全成熟しておらず、AIエージェント需要が立ち上がり、GPUホストCPU、クラウド制御CPU、セキュリティCPU、DB CPUが同時に増えるからです。","quote_start":14,"quote_end":132,"text_sha256":"1acf74fb036396a00d5aff8b0807223bce97016abba107c5994b6b1e133789c7","block_sha256":"1acf74fb036396a00d5aff8b0807223bce97016abba107c5994b6b1e133789c7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_49c45a6a-1541-421a-88ef-0414f7041d8b"},{"id":"occ_ef3395771253fd475da3561e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_4a59f943-6e2a-440a-b25e-cf3e556dfec5","section_id":"sec_18652092-7af3-4816-bfe7-a57f5be8cd6c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":9,"end":15,"exact":"エージェント","quote":"## \\5. AIエージェント需要の増加をGPUクラウドだけで吸収できるか","quote_start":0,"quote_end":37,"text_sha256":"999fe7c162a67ac9b252e4a94f78ae3da216ec5a13b38d1ed218ff842b1236d2","block_sha256":"999fe7c162a67ac9b252e4a94f78ae3da216ec5a13b38d1ed218ff842b1236d2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_4a59f943-6e2a-440a-b25e-cf3e556dfec5"},{"id":"occ_72d63b184a24e916b1038f23","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_4acf70b3-b162-4a7b-ae1e-87030ae1046d","section_id":"sec_7f34bca6-4bc4-4829-a7ce-c78aa72625d1","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントは、1回の巨大推論だけではなく、小さな推論を何度も繰り返します。","quote_start":0,"quote_end":40,"text_sha256":"70a8400368b56164e6fa4dd5fc4b59796f7dc2b092f47c04f714fd7b9fc74ae0","block_sha256":"70a8400368b56164e6fa4dd5fc4b59796f7dc2b092f47c04f714fd7b9fc74ae0","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_4acf70b3-b162-4a7b-ae1e-87030ae1046d"},{"id":"occ_81bb77161e0d8c9109cb18ed","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_4db42485-9a38-4360-b7d2-ea3dcc91990f","section_id":"sec_c32c7c49-725a-4976-8a9d-913de857bee5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":55,"end":61,"exact":"エージェント","quote":"GPUクラウド成熟により、**GPU 1枚あたりのAI推論能力は大きく上がる**でしょう。  \nその結果、同じエージェント数を動かすために必要なGPU枚数は減る可能性があります。","quote_start":0,"quote_end":89,"text_sha256":"ef5add80ebcbcba4def7dd630e646274bd91f330730cb9940a0ae090315bf986","block_sha256":"ef5add80ebcbcba4def7dd630e646274bd91f330730cb9940a0ae090315bf986","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_4db42485-9a38-4360-b7d2-ea3dcc91990f"},{"id":"occ_060196bc2612cf8d99fc555c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_4f99f85e-0cb3-4e8a-bdc9-d691d848fe53","section_id":"sec_5caadfab-aa3e-49a9-ba00-336dac555d9e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":94,"end":100,"exact":"エージェント","quote":"\n> **2029〜2032年：CPUクラウドに近い成熟度へ向かう時期**  \n> **2030年代前半：AIエージェント社会向けの本格インフラになる時期**","quote_start":39,"quote_end":118,"text_sha256":"bcdf26174bbd7e5f0c6a4d237200bf36f9dad7c6e23899d6d5790fd8893dc996","block_sha256":"bcdf26174bbd7e5f0c6a4d237200bf36f9dad7c6e23899d6d5790fd8893dc996","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_4f99f85e-0cb3-4e8a-bdc9-d691d848fe53"},{"id":"occ_daa36cd558cd08b3acd9ecac","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_54e6e83e-2971-40b5-a11c-74e5bbc8024f","section_id":"sec_7627ce8f-acd4-4fdb-b993-42c1809a78a7","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":35,"end":41,"exact":"エージェント","quote":"- AIデータセンター建設\n- GPUクラウド成熟前の非効率\n- AIエージェントの普及初期\n- 推論需要の爆発\n- 企業AI導入\n- sovereign AI\n- GPUホストCPU需要\n- クラウド制御CPU需要\n- Arm/EPYC/Grace/カスタムCPUの競争","quote_start":0,"quote_end":136,"text_sha256":"e465676fc7c75ee369ca3baa6b483ea27024aa6ec5abfc1539b2b2f7f9ce3db4","block_sha256":"e465676fc7c75ee369ca3baa6b483ea27024aa6ec5abfc1539b2b2f7f9ce3db4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_54e6e83e-2971-40b5-a11c-74e5bbc8024f"},{"id":"occ_9722ac1ee501bb7c9e680f13","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_59dad553-47de-4522-ad6d-b17d3d2441eb","section_id":"sec_3751a969-a9f8-4805-a45c-1cb291ed56d3","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":10,"end":16,"exact":"エージェント","quote":"## \\4. では、エージェント需要の何倍くらいを吸収できるのか","quote_start":0,"quote_end":32,"text_sha256":"dab5143df31d485e24a6020236875b37fde0b477271f39e6c40ccd7cb15c2d22","block_sha256":"dab5143df31d485e24a6020236875b37fde0b477271f39e6c40ccd7cb15c2d22","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_59dad553-47de-4522-ad6d-b17d3d2441eb"},{"id":"occ_d66168a16ae76beef30cd3ac","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_5a5379a9-79cb-4d52-9386-2a8849697268","section_id":"sec_fd55f986-652f-4cc3-8cf4-54af3e755e17","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":228,"end":234,"exact":"エージェント","quote":"Transformer処理\n\nLPUクラウド\n  - 低遅延LLM推論\n  - 音声対話\n  - リアルタイムエージェント\n\nNPU/ASICクラウド\n  - 特定モデルの超効率推論\n  - 企業内推論\n  - エッジ連携\n```","quote_start":173,"quote_end":288,"text_sha256":"4849afb6ed52df3896bdc03cc3d759b3572a602f3d2bcfe313bb80724ed24d31","block_sha256":"4849afb6ed52df3896bdc03cc3d759b3572a602f3d2bcfe313bb80724ed24d31","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_5a5379a9-79cb-4d52-9386-2a8849697268"},{"id":"occ_d4d539b6d8397f9fdf97afae","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_5ab5726b-6d38-4240-a1ee-5ee8a6cc51a5","section_id":"sec_937846fa-77fe-4711-a151-121b51541466","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":15,"end":21,"exact":"エージェント","quote":"> **GPUを、LLM/AIエージェント用のクラウド資源として、細かく、安全に、高稼働で、トークン単位に制御すること**","quote_start":0,"quote_end":61,"text_sha256":"73e28c45dacca9e8fb9ef23d6ec65b15852d235173dcde8448364279169eb7ae","block_sha256":"73e28c45dacca9e8fb9ef23d6ec65b15852d235173dcde8448364279169eb7ae","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_5ab5726b-6d38-4240-a1ee-5ee8a6cc51a5"},{"id":"occ_b630c0e0a46e78c345e08467","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_5d34c47d-8f24-4ca3-95f4-7801ff3b32cd","section_id":"sec_e00c5f8c-1522-4b87-ba97-46f9ec64d45f","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":228,"end":234,"exact":"エージェント","quote":"Transformer処理\n\nLPUクラウド\n  - 低遅延LLM推論\n  - 音声対話\n  - リアルタイムエージェント\n\nNPU/ASICクラウド\n  - 特定モデルの超効率推論\n  - 企業内推論\n  - エッジ連携\n```","quote_start":173,"quote_end":288,"text_sha256":"4849afb6ed52df3896bdc03cc3d759b3572a602f3d2bcfe313bb80724ed24d31","block_sha256":"4849afb6ed52df3896bdc03cc3d759b3572a602f3d2bcfe313bb80724ed24d31","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_5d34c47d-8f24-4ca3-95f4-7801ff3b32cd"},{"id":"occ_6c915d33a98df0fa6735a57d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_61ba02f8-b675-4b76-b678-3af69a2b2872","section_id":"sec_71d97541-7ee4-426e-8210-eb8853969707","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":81,"end":87,"exact":"エージェント","quote":"GPU kernel最適化、推論スケジューリング、ログ解析、障害予測、セキュリティ検査、インフラ自動運用でAIエージェントが使われるようになるため、CPUクラウドが25年かけた進化を、GPUクラウドは**5〜8年程度で一気に圧縮する**可能性があります。","quote_start":26,"quote_end":153,"text_sha256":"8e71f88209e1f6bc3a1e0ef86079cdb7d5d5c958587ae43da0f87e1eb565ec15","block_sha256":"8e71f88209e1f6bc3a1e0ef86079cdb7d5d5c958587ae43da0f87e1eb565ec15","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_61ba02f8-b675-4b76-b678-3af69a2b2872"},{"id":"occ_840c9a7bdf8d7c6a9a1af7e3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_61e4e5bb-d955-46dd-b0a1-6710a920c87e","section_id":"sec_27f53173-52ca-4dc9-8a68-3d2e7d9f7742","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":39,"end":45,"exact":"エージェント","quote":"```\nGPUクラウド最適化\n  ↓\n1トークンあたりコスト低下\n  ↓\nAIエージェントが安くなる\n  ↓\n利用回数が増える\n  ↓\n音声AI、動画AI、業務AI、ロボットAIが増える\n  ↓\n総GPU需要はむしろ増える\n```","quote_start":0,"quote_end":115,"text_sha256":"2077854092b8bf8f648e27e69dda7eb4ff176e1b79ec3c689a77113937e7ea35","block_sha256":"2077854092b8bf8f648e27e69dda7eb4ff176e1b79ec3c689a77113937e7ea35","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_61e4e5bb-d955-46dd-b0a1-6710a920c87e"},{"id":"occ_d13297393a816a44f66cef14","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_62e1b43a-27e4-4777-9118-4246acb0f586","section_id":"sec_58745122-69e4-4bc2-ae16-3bae5e2bfd75","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"なので、エージェント増加による負荷のうち、**モデル推論・生成部分はGPUクラウドでかなり吸収できる**。  \nしかし、**行動・接続・管理・保存・防衛の部分はCPUクラウド側に残る**、という構造です。","quote_start":0,"quote_end":102,"text_sha256":"29166ccdfb84eabf4f8f9ec66cdb07d9f3e56f0877b2a99f359d9d53eb0af347","block_sha256":"29166ccdfb84eabf4f8f9ec66cdb07d9f3e56f0877b2a99f359d9d53eb0af347","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_62e1b43a-27e4-4777-9118-4246acb0f586"},{"id":"occ_c9f998f33e5732261a83d53e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_69c451e0-d1a9-41a3-bab5-9c4252b0462c","section_id":"sec_83231a84-d7e5-4f2e-9731-19076ba75638","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":15,"end":21,"exact":"エージェント","quote":"## \\4. 複数モデル・複数エージェントのGPUプーリング","quote_start":0,"quote_end":30,"text_sha256":"b962dbae57cbf32e3801af9f09e40d8d5850ad7f66edeaf44996d22a4672ed6a","block_sha256":"b962dbae57cbf32e3801af9f09e40d8d5850ad7f66edeaf44996d22a4672ed6a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_69c451e0-d1a9-41a3-bab5-9c4252b0462c"},{"id":"occ_40a17608ea204915b18d22cb","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_7138a816-33c4-4594-a8e8-bd1446dbcb3b","section_id":"sec_a7e4202f-5c99-428b-81c5-4a815fc8c318","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":63,"end":69,"exact":"エージェント","quote":"はなく、AIクラウドの監視、コスト制御、AI Gateway、セキュリティで伸びる可能性があります。  \nAIエージェント時代には、GPU計算だけでなく「誰がどのAIをどれだけ使ったか」を管理する層が重要になります。","quote_start":8,"quote_end":116,"text_sha256":"43f867cc10bd9b80a7bb8c0d807f8da83656d0a9b030c083d3c259d255d66e6f","block_sha256":"43f867cc10bd9b80a7bb8c0d807f8da83656d0a9b030c083d3c259d255d66e6f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_7138a816-33c4-4594-a8e8-bd1446dbcb3b"},{"id":"occ_5c1a5840dcedb8bbfbcca1e1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_7709f869-5218-4049-9ff9-c24166f19f83","section_id":"sec_f3f6bb06-b5e5-4d2b-ab7a-4837dfe6d756","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"```\nAIエージェント追加需要 = 100\n\nGPUクラウド成熟でかなり吸収できる部分\n  50〜70\n  - LLM推論\n  - 長文推論\n  - reasoning\n  - 画像/動画生成\n  - embedding/","quote_start":0,"quote_end":112,"text_sha256":"f9bd88bfdcaa689720d90f4b0d9b950243667b071fb80cdfec71d8d0437785bd","block_sha256":"f9bd88bfdcaa689720d90f4b0d9b950243667b071fb80cdfec71d8d0437785bd","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_7709f869-5218-4049-9ff9-c24166f19f83"},{"id":"occ_0f66aae0cc8406ff479b863f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_7806786f-0eee-45d0-95d5-26cceb620386","section_id":"sec_24b4986d-fd60-4bbf-841a-1d7a353fdecc","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":70,"end":76,"exact":"エージェント","quote":"nel生成・最適化の研究はかなり活発です。AMDは2025年にGEAKというTriton kernel生成AIエージェントを紹介し、ベンチマーク上で最大2.59倍の速度向上を示したと説明しています。([ROCm Blogs](https://rocm.blogs.amd.com/software-tools-optimi","quote_start":15,"quote_end":176,"text_sha256":"58b2c4ff9852dbd083f8c53b9718e7d7ae0aa121f4b65b52651e9657f097f39d","block_sha256":"58b2c4ff9852dbd083f8c53b9718e7d7ae0aa121f4b65b52651e9657f097f39d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_7806786f-0eee-45d0-95d5-26cceb620386"},{"id":"occ_8142df7791394d069d75b4d2","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_7ad5e26f-ee2e-48c6-af05-6c5dd08afc55","section_id":"sec_d6048052-b453-45bf-b44f-f6de73b7c164","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"特にAIエージェント時代には、1回の巨大推論だけでなく、小さな推論、ツール呼び出し、再推論、RAG、長文処理、JSON出力、音声・画像・動画処理が混ざります。その結果、GPUを単純に「1モデル1GPU」「1顧客1GPU」","quote_start":0,"quote_end":110,"text_sha256":"8d4679d228c1edaace52142d42a2dc3fe1493a97b66092d03a3310ba08c06065","block_sha256":"8d4679d228c1edaace52142d42a2dc3fe1493a97b66092d03a3310ba08c06065","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_7ad5e26f-ee2e-48c6-af05-6c5dd08afc55"},{"id":"occ_9ff93a05dd66432a4a46de8a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_7b8e2005-cefa-439b-bf91-1368acd1686a","section_id":"sec_3751a969-a9f8-4805-a45c-1cb291ed56d3","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":89,"end":95,"exact":"エージェント","quote":"Mアプリケーション向けのフレームワークで、実験では最大6.4倍のスループット改善を報告しています。これはまさにエージェント的な処理に近いです。([ACM Digital Library](https://dl.acm.org/doi/10.5555/3737916.3739916?utm_source=chatgpt.c","quote_start":34,"quote_end":195,"text_sha256":"40ac67a79ed11210192dfbb91f50c1b7640156d1f9b2b1ca795030cb31560a89","block_sha256":"40ac67a79ed11210192dfbb91f50c1b7640156d1f9b2b1ca795030cb31560a89","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_7b8e2005-cefa-439b-bf91-1368acd1686a"},{"id":"occ_d436499469b4851696420975","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_8539c2ee-c57e-4f52-a475-33bd2e4079c9","section_id":"sec_67abc6cf-0257-41ae-9bc8-4a18e2b88159","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":14,"end":20,"exact":"エージェント","quote":"GPUクラウドが成熟すると、エージェントの推論処理はかなり効率化されます。","quote_start":0,"quote_end":37,"text_sha256":"c2e5702ce29f673f19b3112bdf94f75828945c7d1ef94118d544bc8651c8e522","block_sha256":"c2e5702ce29f673f19b3112bdf94f75828945c7d1ef94118d544bc8651c8e522","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_8539c2ee-c57e-4f52-a475-33bd2e4079c9"},{"id":"occ_3394344b92add63b39c2ba36","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_883ba18c-c036-45b4-9d0a-46162deb7db5","section_id":"sec_67c7d199-a2f4-4058-b6a0-18b296e20e76","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":13,"end":19,"exact":"エージェント","quote":"AI推論が安くなると、AIエージェントはもっと大量に使われます。","quote_start":0,"quote_end":32,"text_sha256":"c78b24223a3c9335d5f72c563edf1d0ce0d0b1ccd57d56e90f9d1114dd3b97e3","block_sha256":"c78b24223a3c9335d5f72c563edf1d0ce0d0b1ccd57d56e90f9d1114dd3b97e3","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_883ba18c-c036-45b4-9d0a-46162deb7db5"},{"id":"occ_2882cb6a6a4ee77a819b76cc","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_88dfdcdf-c061-4b49-ac57-e7119851bc7e","section_id":"sec_83231a84-d7e5-4f2e-9731-19076ba75638","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":15,"end":21,"exact":"エージェント","quote":"これは非常に大きいです。  \nエージェントが増えた時に、**1エージェント1GPUのような贅沢な構成ではなく、トークン単位でGPUを共有する**方向に進むからです。","quote_start":0,"quote_end":82,"text_sha256":"d16e0b1f3487d181e47e75c67b052a4bd6e1a35eed4bb3ba4515ff0afec2a15b","block_sha256":"d16e0b1f3487d181e47e75c67b052a4bd6e1a35eed4bb3ba4515ff0afec2a15b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_88dfdcdf-c061-4b49-ac57-e7119851bc7e"},{"id":"occ_cd3e4819b84c821dd7c4ebfd","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_8e0f23e5-5aa7-4c03-93ed-d7d5708b160e","section_id":"sec_3ed2a696-c34d-42a7-8e62-75970d04cafa","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":47,"end":53,"exact":"エージェント","quote":"- 音声AI\n- AI秘書\n- カスタマーサポート\n- コーディング補助\n- リアルタイム検索エージェント\n- 軽量業務エージェント\n- ゲームNPC\n- 配信AIキャラ\n- AITuber","quote_start":0,"quote_end":95,"text_sha256":"ee46128c254ae22c9ecfb4ec7121c2f71b233d8974ed51d6119f01cbb763868e","block_sha256":"ee46128c254ae22c9ecfb4ec7121c2f71b233d8974ed51d6119f01cbb763868e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_8e0f23e5-5aa7-4c03-93ed-d7d5708b160e"},{"id":"occ_3d43d65d603d7203daae4508","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_9488ea8d-520c-4f6c-8cbd-44cc4a64b658","section_id":"sec_baffb438-6cd2-4b65-9f3e-9f0215b7e422","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":8,"end":14,"exact":"エージェント","quote":"この時期は、AIエージェントが増え、GPUクラウドが「ただGPUを貸す」段階から「GPUを細かく共有・管理・課金する」段階に進むと考えます。","quote_start":0,"quote_end":70,"text_sha256":"c12c49bff5fbbdc65e1c1fc2ef3902bb4b183bec70f4080276f39538235f6a3f","block_sha256":"c12c49bff5fbbdc65e1c1fc2ef3902bb4b183bec70f4080276f39538235f6a3f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_9488ea8d-520c-4f6c-8cbd-44cc4a64b658"},{"id":"occ_52dad15e797ceef49540ade3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_94992de1-1d1b-4437-bf08-390c54a19fba","section_id":"sec_9282639c-a2fc-481a-89f4-c46b54cdff7a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"- AIエージェント\n- コーディングAI\n- 企業内AIワーカー\n- リアルタイム音声AI\n- 動画生成AI\n- ロボティクス/VLA\n- デジタルツイン\n- 科学AI\n- 個人専用AI\n- 常時稼働AI秘書","quote_start":0,"quote_end":105,"text_sha256":"8a7669d791df6c0989ca9e1706c1be6a453483977cb1489015a6d3d6b9f72242","block_sha256":"8a7669d791df6c0989ca9e1706c1be6a453483977cb1489015a6d3d6b9f72242","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_94992de1-1d1b-4437-bf08-390c54a19fba"},{"id":"occ_d3e2f621d2acfc59f1f1efab","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_9888ea43-595c-4350-8d3b-b8b8d11ca8c7","section_id":"sec_59a30840-3f80-46be-aa94-9d9cd7331f4b","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":28,"end":34,"exact":"エージェント","quote":"```\nOSを動かす\n権限を管理する\n外部APIを呼ぶ\nエージェントの状態を管理する\nログを保存する\n異常を検知する\n安全停止を判断する\n人間の承認フローと接続する\n```","quote_start":0,"quote_end":86,"text_sha256":"81dd220bb1d431fad7ff5f5da4053dae5507eba3066b87da723528e8d3e5725c","block_sha256":"81dd220bb1d431fad7ff5f5da4053dae5507eba3066b87da723528e8d3e5725c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_9888ea43-595c-4350-8d3b-b8b8d11ca8c7"},{"id":"occ_8f389f3af51afb486a2ae198","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_98e11b4f-f234-4d5e-8e70-7404b9716759","section_id":"sec_a1dc352b-54dc-455d-87a5-2f98a945adb5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":39,"end":45,"exact":"エージェント","quote":"GPUクラウド最適化は、単にGPUを高稼働にするだけでは不十分です。  \nAIエージェントが増えると、推論リクエスト、APIキー、モデル選択、レート制限、コスト管理、認証、ログ、攻撃防御が必要になります。","quote_start":0,"quote_end":102,"text_sha256":"9cbc7576f7c0a3ad4590e1af2425734a0bbdc166934b8dd9c2a06b9df53416cc","block_sha256":"9cbc7576f7c0a3ad4590e1af2425734a0bbdc166934b8dd9c2a06b9df53416cc","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_98e11b4f-f234-4d5e-8e70-7404b9716759"},{"id":"occ_e211019e26bc33b073770c0a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_9ac296ec-7573-4acc-9764-b60beb101cb5","section_id":"sec_3ed2a696-c34d-42a7-8e62-75970d04cafa","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントは、1回の巨大推論だけではなく、小さな推論を何度も繰り返します。","quote_start":0,"quote_end":40,"text_sha256":"70a8400368b56164e6fa4dd5fc4b59796f7dc2b092f47c04f714fd7b9fc74ae0","block_sha256":"70a8400368b56164e6fa4dd5fc4b59796f7dc2b092f47c04f714fd7b9fc74ae0","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_9ac296ec-7573-4acc-9764-b60beb101cb5"},{"id":"occ_9e7640c36753fb617d70309e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_9dff6142-d9f9-408f-81fc-476a168d3eab","section_id":"sec_a1dc352b-54dc-455d-87a5-2f98a945adb5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":32,"end":38,"exact":"エージェント","quote":"この領域は、GPUクラウドの「外側」にあります。  \nしかしAIエージェント時代には、GPU推論そのものよりも、**どのAIに、誰が、どの権限で、どれだけ使わせるか**が重要になります。","quote_start":0,"quote_end":93,"text_sha256":"573fbfcbdb7b18fadc22321492b1b715e6ab3db26d886a1c9d3da7ad6877e116","block_sha256":"573fbfcbdb7b18fadc22321492b1b715e6ab3db26d886a1c9d3da7ad6877e116","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_9dff6142-d9f9-408f-81fc-476a168d3eab"},{"id":"occ_fb3b94c7a3721747b998ec1a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_a1e98e78-b494-413f-af6d-97991904a007","section_id":"sec_3ed2a696-c34d-42a7-8e62-75970d04cafa","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":9,"end":15,"exact":"エージェント","quote":"## \\9. AIエージェント時代にLPUクラウドは重要か？","quote_start":0,"quote_end":30,"text_sha256":"96205dfd9f25a1e9512261be653c3a742459769e9c210beaefd05e96a427a1a0","block_sha256":"96205dfd9f25a1e9512261be653c3a742459769e9c210beaefd05e96a427a1a0","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_a1e98e78-b494-413f-af6d-97991904a007"},{"id":"occ_4982743b21e9d96eed439fb1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_a66ac912-8c61-4f3c-b68d-abf3ee9a0f66","section_id":"sec_58745122-69e4-4bc2-ae16-3bae5e2bfd75","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントは、単に1回質問して1回答を返すチャットボットではありません。","quote_start":0,"quote_end":39,"text_sha256":"7cccec5bb3cc31d7f1c566e2b553c747a03cd460304d772140ab13661d9f4b0e","block_sha256":"7cccec5bb3cc31d7f1c566e2b553c747a03cd460304d772140ab13661d9f4b0e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_a66ac912-8c61-4f3c-b68d-abf3ee9a0f66"},{"id":"occ_068b8155295d461030cc5f3f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_a8f360cc-f18c-47da-a6c9-a80457d011a5","section_id":"sec_9869513e-af53-488b-b9ef-2846e7df2dff","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントは長時間動き、会話履歴や作業状態を持ちます。  \nその状態をGPU/HBM上に保持するなら、それは「計算」ではなく「高価な短期記憶」を占有していることになります。","quote_start":0,"quote_end":89,"text_sha256":"05b5e437a9700a38387605194ebb1701b00133277b94735282b06c8c4bcfc1e4","block_sha256":"05b5e437a9700a38387605194ebb1701b00133277b94735282b06c8c4bcfc1e4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_a8f360cc-f18c-47da-a6c9-a80457d011a5"},{"id":"occ_a023210ccb63bd0fc183889b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_a9a6bfad-2f0b-4407-824f-c20c1550f517","section_id":"sec_3751a969-a9f8-4805-a45c-1cb291ed56d3","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":58,"end":64,"exact":"エージェント","quote":"*今の非効率なGPU運用に比べて、成熟したGPUクラウドでは同じGPU枚数で3〜10倍、特定条件ではそれ以上のエージェント推論を処理できる可能性がある。**","quote_start":3,"quote_end":81,"text_sha256":"f44e71cdfe82942fd8ce2b886bc20f57a0ca48a1277e611467592c9851c69b33","block_sha256":"f44e71cdfe82942fd8ce2b886bc20f57a0ca48a1277e611467592c9851c69b33","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_a9a6bfad-2f0b-4407-824f-c20c1550f517"},{"id":"occ_c194dca1cccb1c69d1475219","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_aac0dc86-8a0d-4924-8c16-19f510163dec","section_id":"sec_60a8a145-1f54-4dc1-9e65-16cc2eb54c44","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":0,"end":6,"exact":"エージェント","quote":"エージェントが本格化すると、モデル推論だけでなく、","quote_start":0,"quote_end":25,"text_sha256":"4d510c35cd3033bec2a300eea8c6f37a3f757083fde667dff6749a0d211c9eb4","block_sha256":"4d510c35cd3033bec2a300eea8c6f37a3f757083fde667dff6749a0d211c9eb4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_aac0dc86-8a0d-4924-8c16-19f510163dec"},{"id":"occ_6b71ab237a298b554d1cf83e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_ab25c52c-f2f3-4df5-bdda-8ba349afaf89","section_id":"sec_0a8db19a-a15e-4eb6-9a83-4a685462289f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":3,"end":9,"exact":"エージェント","quote":"ことでエージェント需要をかなり受け止められます。","quote_start":0,"quote_end":24,"text_sha256":"1f00abf77f7d723bed66ea952a02862be2cd99c91c596a41e9a556bbe1196ed8","block_sha256":"1f00abf77f7d723bed66ea952a02862be2cd99c91c596a41e9a556bbe1196ed8","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_ab25c52c-f2f3-4df5-bdda-8ba349afaf89"},{"id":"occ_90928df314ac26cfbb8d0952","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_ab3ba01f-5048-40d4-9d01-d01968d0ed5b","section_id":"sec_464705c9-5456-4a75-8565-30bf403331fc","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":52,"end":58,"exact":"エージェント","quote":"- 会話履歴\n- KV cache\n- prefix cache\n- tool callの中間状態\n- エージェントの作業メモリ\n- 長文コンテキスト\n- マルチモーダル埋め込み\n- モデルごとの常駐重み","quote_start":0,"quote_end":102,"text_sha256":"afcc5075cb68062ad7061a934f69c7563995a1fa8df02dc9b369e27a96287347","block_sha256":"afcc5075cb68062ad7061a934f69c7563995a1fa8df02dc9b369e27a96287347","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_ab3ba01f-5048-40d4-9d01-d01968d0ed5b"},{"id":"occ_bf3246d1b1e5c404d7d006eb","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_ac18cac4-fa73-4742-b25d-fc75a47c4c70","section_id":"sec_e2c11f6d-3238-4551-964e-bc128f1dc940","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":7,"end":13,"exact":"エージェント","quote":"一方で、企業内エージェントのように、","quote_start":0,"quote_end":18,"text_sha256":"8aaff62b974dcf8cc6d1705424993a8fbeede2838d16cddfdba625bce061a120","block_sha256":"8aaff62b974dcf8cc6d1705424993a8fbeede2838d16cddfdba625bce061a120","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_ac18cac4-fa73-4742-b25d-fc75a47c4c70"},{"id":"occ_e87c37347859eac81b4c786b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_ac6de728-4f47-4a64-9142-936be9ff66be","section_id":"sec_0ab28b0f-3669-4f74-a3ce-afe5cff0da16","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"これは、エージェント時代にはかなり重要です。  \nなぜなら、エージェントは同じシステムプロンプト、同じツール定義、同じ作業文脈を何度も使うからです。","quote_start":0,"quote_end":74,"text_sha256":"86b78aee150c1764c12ccf9902e323cb2c2a38fffd20fb56881ed909b3165f95","block_sha256":"86b78aee150c1764c12ccf9902e323cb2c2a38fffd20fb56881ed909b3165f95","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_ac6de728-4f47-4a64-9142-936be9ff66be"},{"id":"occ_cc860d13dc10ec346b025b1c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_b47ca922-940f-4fee-9224-a0a7ed4bc4dd","section_id":"sec_67c7d199-a2f4-4058-b6a0-18b296e20e76","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"- エージェント管理CPU\n- 認証/権限管理CPU\n- セキュリティ監査CPU\n- データベースCPU\n- ストレージ制御CPU\n- ネットワーク制御CPU\n- GPUクラスタ管理CPU\n- RAG/検索/embe","quote_start":0,"quote_end":108,"text_sha256":"e812efc28ab590a7e4f13114f32f91bd8a8729fafc2b760d916f4ca75a6f071e","block_sha256":"e812efc28ab590a7e4f13114f32f91bd8a8729fafc2b760d916f4ca75a6f071e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_b47ca922-940f-4fee-9224-a0a7ed4bc4dd"},{"id":"occ_11b4238d1eae89920666dfca","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_be97beea-9c26-4097-bb9a-39ac476deca3","section_id":"sec_58745122-69e4-4bc2-ae16-3bae5e2bfd75","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"たとえば、エージェントはこう動きます。","quote_start":0,"quote_end":19,"text_sha256":"113b271c56f181258b100ac78efebb59d4d38ceaa147c5e003ea835f04a33e30","block_sha256":"113b271c56f181258b100ac78efebb59d4d38ceaa147c5e003ea835f04a33e30","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_be97beea-9c26-4097-bb9a-39ac476deca3"},{"id":"occ_404ed4c7e20b516c18582c83","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_bf5b542b-490f-49f1-b995-1fdf16698e7a","section_id":"sec_c32c7c49-725a-4976-8a9d-913de857bee5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"しかし、エージェントが増えると同時にCPU側の処理も増えます。","quote_start":0,"quote_end":31,"text_sha256":"5781bf8d16af1278151aa12626b6931d5c0d01c17ae2e77a67addc8b89db3134","block_sha256":"5781bf8d16af1278151aa12626b6931d5c0d01c17ae2e77a67addc8b89db3134","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_bf5b542b-490f-49f1-b995-1fdf16698e7a"},{"id":"occ_29b64c7c397cd8afc12779f5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_c13fe7df-e56d-4daa-942b-3a1a85e1ef3d","section_id":"sec_e5e44156-0b06-4dce-8372-0aca3f2d053a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":37,"end":43,"exact":"エージェント","quote":"つまり、GPUクラウドが成熟すると、  \n**同じGPU数でも、より多くのエージェント推論をさばける**  \nようになります。","quote_start":0,"quote_end":63,"text_sha256":"13e8ff465c139d1f4063ba58e7cc41c6c5ab184c8bb1536299f750f46a989a4d","block_sha256":"13e8ff465c139d1f4063ba58e7cc41c6c5ab184c8bb1536299f750f46a989a4d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_c13fe7df-e56d-4daa-942b-3a1a85e1ef3d"},{"id":"occ_71eca70a030e4d5febdc2ec0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_c1749f49-25c2-4613-a9a2-bd517d6db711","section_id":"sec_0ab28b0f-3669-4f74-a3ce-afe5cff0da16","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":38,"end":44,"exact":"エージェント","quote":"LLMは長文を扱うほど、KV cacheというメモリを大量に使います。  \nエージェントは過去の会話、作業履歴、ツール結果、資料を何度も参照するので、KV cache管理が非常に重要です。","quote_start":0,"quote_end":94,"text_sha256":"9f9d6ed934a8623ea3b6efb4876bfc4dff684fb8544486e5992863078ec3f8f9","block_sha256":"9f9d6ed934a8623ea3b6efb4876bfc4dff684fb8544486e5992863078ec3f8f9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_c1749f49-25c2-4613-a9a2-bd517d6db711"},{"id":"occ_ce92b1e102000d95e056c5a4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_c535ca0e-5792-496f-bad7-f853d913dc16","section_id":"sec_7e6570d5-47c1-4c8d-8c40-b67e49433c73","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":85,"end":91,"exact":"エージェント","quote":"\n- ログ\n- 監査\n- セキュリティ\n- DB\n- ストレージI/O\n- ツール実行\n- ブラウザ操作\n- エージェント状態管理\n- RAG検索\n- GPUスケジューリング\n- KV cache管理のメタ制御","quote_start":30,"quote_end":135,"text_sha256":"7e1f17b0d50bb8bdea0421d3521754303a95f34d6614ed70f2b834c3a129c5d3","block_sha256":"7e1f17b0d50bb8bdea0421d3521754303a95f34d6614ed70f2b834c3a129c5d3","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_c535ca0e-5792-496f-bad7-f853d913dc16"},{"id":"occ_71027451c4206bc054be5791","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_c9d298d2-b812-45f8-826d-ad405fe11296","section_id":"sec_6469f556-42cd-4404-803b-318074aba2d8","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":172,"end":178,"exact":"エージェント","quote":"用途はGPUより狭いが、はまると非常に強い。\n\nLPUクラウド\n  低遅延LLM推論クラウド。\n  音声対話、エージェント、リアルタイム応答に強い。\n  フリート稼働率は需要変動とSLOに左右される。\n```","quote_start":117,"quote_end":221,"text_sha256":"67c6aeaf704d766a6a6d907036345cdcc8f5d15ccd4c977f12ee8b2483aeefbb","block_sha256":"67c6aeaf704d766a6a6d907036345cdcc8f5d15ccd4c977f12ee8b2483aeefbb","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_c9d298d2-b812-45f8-826d-ad405fe11296"},{"id":"occ_1268dac483027e7bbd05e281","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_cee2b3dc-5c05-42b2-98fe-84e1e3ef8a5e","section_id":"sec_38fea9f4-4c5a-4b32-bd81-225d5d883d90","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":38,"end":44,"exact":"エージェント","quote":"LPUクラウドは、2030年代にかなり伸びる可能性があります。  \n特にAIエージェント、音声AI、リアルタイム対話、ゲームNPC、AITuber、カスタマーサポートでは強いです。","quote_start":0,"quote_end":90,"text_sha256":"1c46da3b6ea9e30da3b20dc922fe27a042144bc564c7595df2388e2846de261a","block_sha256":"1c46da3b6ea9e30da3b20dc922fe27a042144bc564c7595df2388e2846de261a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_cee2b3dc-5c05-42b2-98fe-84e1e3ef8a5e"},{"id":"occ_a07ac8f016bf5dddb5436bb3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_d4f8d35f-ac84-4163-8004-821dbe00d3ba","section_id":"sec_d32d82b3-f2c9-4767-a64c-1da8e863e79a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":67,"end":73,"exact":"エージェント","quote":"ーザー需要\n- リアルタイムSLO\n- 待ち行列をどれだけ許すか\n- バッチ処理を混ぜられるか\n- 音声AIやエージェント需要の量\n- モデル更新頻度\n- SRAM容量に収まる構成か","quote_start":12,"quote_end":103,"text_sha256":"0f9f7f870ec42dc950ab8e7da34c1a62471aeacb2e60bdfb4f448e7051cad2e8","block_sha256":"0f9f7f870ec42dc950ab8e7da34c1a62471aeacb2e60bdfb4f448e7051cad2e8","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_d4f8d35f-ac84-4163-8004-821dbe00d3ba"},{"id":"occ_a2842a6771e27f765023527f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_d7b510da-9b55-465d-810a-1c9408e482e0","section_id":"sec_2fcdbe46-ebad-470b-b6ea-94ffa48a219c","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":102,"end":109,"exact":"agentic","quote":"YC CPU需要とInstinct GPU出荷が牽引したと説明しています。またLisa Su CEOは、推論とagentic AIが高性能CPUとアクセラレータ需要を押し上げていると述べています。([Advanced Micro Devices, Inc.](https://ir.amd.com/news-events/p","quote_start":47,"quote_end":209,"text_sha256":"ac2542d671d4e345656bb98ce21ee872aa2830d0d567e1b12be628c651cd8feb","block_sha256":"ac2542d671d4e345656bb98ce21ee872aa2830d0d567e1b12be628c651cd8feb","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_d7b510da-9b55-465d-810a-1c9408e482e0"},{"id":"occ_ce2773cb5581f527377b23f5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_dade23c3-517e-43c9-a8f3-f912708ba34d","section_id":"sec_f9df470f-5155-42ec-97eb-cf96aea7d2de","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":94,"end":100,"exact":"エージェント","quote":"変化しているからです。モデル構造、context length、MoE、reasoning、動画生成、VLA、エージェント処理が変わるたびに、GPUクラウド側も作り直しになります。","quote_start":39,"quote_end":129,"text_sha256":"09e23b43105656ef68a55b243744aae1d256440432e8082b346b3346328af18d","block_sha256":"09e23b43105656ef68a55b243744aae1d256440432e8082b346b3346328af18d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_dade23c3-517e-43c9-a8f3-f912708ba34d"},{"id":"occ_f82b169542077d5e5f2a776f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_e0d8ed6a-0c1f-4ae3-9dca-9db549936b47","section_id":"sec_f4b10071-2e20-4e00-b1ba-2d1c3dd83a74","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":113,"end":119,"exact":"エージェント","quote":" TPUクラウド\n\n汎用AI開発、画像生成、動画生成\n  → GPUクラウド\n\n低遅延チャット、音声対話、軽量エージェント\n  → LPUクラウド\n\n企業向けAI推論API\n  → GPU / LPU / TPU の混在\n\nフィジカルAI・VLA学習\n  → GPUクラウド中心、一部TPU/ASIC\n\nエッジAI\n  →","quote_start":58,"quote_end":219,"text_sha256":"2456538722254db1f69cc8675621c32e6331c79dbbb6b30dce00da0586554513","block_sha256":"2456538722254db1f69cc8675621c32e6331c79dbbb6b30dce00da0586554513","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_e0d8ed6a-0c1f-4ae3-9dca-9db549936b47"},{"id":"occ_b28735c9127309c2e0038707","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_e5ab2a48-f67d-4472-9389-0f44acbfbfa4","section_id":"sec_b3015863-c634-42f8-bf90-57c765dba5e5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":3,"end":9,"exact":"エージェント","quote":"これはエージェント時代の「大量の細かい推論を詰め込む」技術です。","quote_start":0,"quote_end":32,"text_sha256":"6ea18a198e7cbf1c053569bd66b30bcc999290f7edcc8f5643f306e65c259da6","block_sha256":"6ea18a198e7cbf1c053569bd66b30bcc999290f7edcc8f5643f306e65c259da6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_e5ab2a48-f67d-4472-9389-0f44acbfbfa4"},{"id":"occ_999fec6b67d8669f946b1377","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_e8552278-3e19-481c-8b35-83673daa064a","section_id":"sec_d2e0b6e3-e6ee-4ed5-b94a-2aeff971584a","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":81,"end":87,"exact":"エージェント","quote":"PU投資に付随して、ホストCPU・クラウド制御CPUが増える。\n\n2027〜2030年\n伸び率のピーク。\nAIエージェント、推論爆発、GPUクラウドOS化でCPU需要は非常に強い。\nただし、EPYC、Grace、Arm、DPUなどに選別が進む。\n\n2030〜2035年\nGPUクラウド成熟で1処理あたりのCPU必要量は下が","quote_start":26,"quote_end":187,"text_sha256":"c729fcd95af552c16cd1fbb08c5965f64d6686433ec77b9bd6062385a17c008e","block_sha256":"c729fcd95af552c16cd1fbb08c5965f64d6686433ec77b9bd6062385a17c008e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_e8552278-3e19-481c-8b35-83673daa064a"},{"id":"occ_360494029b6cf3fa45488e34","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_ef879220-3599-425e-8e59-b69bfc12b5ba","section_id":"sec_ffdc9937-2285-4e5b-89d4-18d96e67d197","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"しかし、AIエージェント、企業AI、フィジカルAI、ロボティクス、セキュリティ、監査、DB、API連携が増えるため、CPU需要そのものはすぐには消えません。","quote_start":0,"quote_end":78,"text_sha256":"950f07f74786187a5bfb4eb07462f6cc4817bd98fdf7b16ccd32a4afd8b86dba","block_sha256":"950f07f74786187a5bfb4eb07462f6cc4817bd98fdf7b16ccd32a4afd8b86dba","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_ef879220-3599-425e-8e59-b69bfc12b5ba"},{"id":"occ_89cc166f7a2aee8432e5eb19","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_f0a1393e-391d-420b-a3f5-abedddd5834b","section_id":"sec_e4db8852-d772-4f64-9b92-9c34d2510aff","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":65,"end":71,"exact":"エージェント","quote":"モリやディスクを使った分だけ払うのが自然でした。  \nGPUクラウドでは、KV cache、長文コンテキスト、エージェントの作業状態、マルチモーダル中間表現をどれだけ保持するかが、原価と性能を大きく左右します。","quote_start":10,"quote_end":115,"text_sha256":"d5aff31fc7882354dd324888420d84058e49461ce25e8a23ddf9e3008db48e91","block_sha256":"d5aff31fc7882354dd324888420d84058e49461ce25e8a23ddf9e3008db48e91","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_f0a1393e-391d-420b-a3f5-abedddd5834b"},{"id":"occ_07b8e4acd1c461f147ab7f11","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_f1fac7a7-722d-4d92-a423-da8d5f80a132","section_id":"sec_83231a84-d7e5-4f2e-9731-19076ba75638","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":0,"end":6,"exact":"エージェント","quote":"エージェント社会では、1つの巨大モデルだけでなく、","quote_start":0,"quote_end":25,"text_sha256":"e81412b3c2eb246a4e05b60e4c4bf651002e15cbb7fb9aa845c107c6d4affffd","block_sha256":"e81412b3c2eb246a4e05b60e4c4bf651002e15cbb7fb9aa845c107c6d4affffd","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_f1fac7a7-722d-4d92-a423-da8d5f80a132"},{"id":"occ_086764acbd3099c987be0282","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_f21ce205-23ee-421b-8bad-a2e2c62f3042","section_id":"sec_ffdc9937-2285-4e5b-89d4-18d96e67d197","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":58,"end":64,"exact":"エージェント","quote":"今後のCPU需要は、単なる「PCや普通のサーバーCPU」ではなく、  \n**GPUを使い切るためのCPU、AIエージェントを管理するCPU、セキュリティと接続を担うCPU**に変わっていくと考えられます。","quote_start":3,"quote_end":105,"text_sha256":"862259444c6a1f319d08dcb13f074c2423518af8ac640dac45a882c621e91b25","block_sha256":"862259444c6a1f319d08dcb13f074c2423518af8ac640dac45a882c621e91b25","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_f21ce205-23ee-421b-8bad-a2e2c62f3042"},{"id":"occ_aa51106dfb48f5173cb6a67e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_f8bcdf32-ca88-4776-a297-f5a795e8756a","section_id":"sec_0ddaeb93-1732-4096-96c9-c776a8183e73","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":8,"end":14,"exact":"エージェント","quote":"SGLangは、エージェント制御、RAG、JSON生成、複数ターン会話など、複雑なLLMアプリケーションを効率化するランタイムで、最大6.4倍のスループット改善を報告しています。([arXiv](https://arxiv.or","quote_start":0,"quote_end":114,"text_sha256":"518375797f1443068c4b420297a805a3efb0a835dea7f44c906518a9f0bfc838","block_sha256":"518375797f1443068c4b420297a805a3efb0a835dea7f44c906518a9f0bfc838","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_f8bcdf32-ca88-4776-a297-f5a795e8756a"},{"id":"occ_69e9bc558caed7077c1f75ab","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2","block_id":"blk_fbf1c441-1759-441e-a84b-791bd2affb90","section_id":"sec_f3f6bb06-b5e5-4d2b-ab7a-4837dfe6d756","layer":"body","character_id":null,"count":3,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":8,"end":14,"exact":"エージェント","quote":"**動画生成AIエージェント**なら、GPUクラウドで80〜95%吸収できます。  \n**企業業務エージェント**なら、GPUクラウドで40〜70%、残りはCPU/DB/ネットワーク/セキュリティです。  \n**ロボット/VLA","quote_start":0,"quote_end":114,"text_sha256":"db30cc999bd831ed3eb8e94ee2ee17af21a777c6c2ffc67c4123ec5f6af0294c","block_sha256":"db30cc999bd831ed3eb8e94ee2ee17af21a777c6c2ffc67c4123ec5f6af0294c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6cb909ec-1291-4cbb-a768-f72123829d08/#blk_fbf1c441-1759-441e-a84b-791bd2affb90"},{"id":"occ_cc9fb55de2b097075bf2f175","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","work_id":"wrk_2e3fa847-c032-4040-9284-2106f69cf66c","block_id":null,"section_id":null,"layer":"title","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"title","start":2,"end":8,"exact":"エージェント","quote":"AIエージェント時代、「人類の実データ」を持つ企業が再評価されるかも知れない","quote_start":0,"quote_end":38,"text_sha256":"047e7fd003ce9ec9972ad9be265a3d716a1e52ebc489cde770c359eb2740a84a","block_sha256":null,"offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982/"},{"id":"occ_cbbda3dbbb9f9f6cd7f37019","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","work_id":"wrk_2e3fa847-c032-4040-9284-2106f69cf66c","block_id":"blk_057f4f8a-82ec-4ae6-b672-585c53488c46","section_id":"sec_6b077011-e9c2-4485-b56a-eaac61051119","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":119,"end":125,"exact":"エージェント","quote":"％増の4.38億ドルで、通期売上予想31.9億～32.4億ドルを維持している。経営陣は、独自データを生成AI・エージェント型AIへ組み込み、保険業務へ展開する方針を示している。([Verisk Investor Relations](https://investor.verisk.com/financials/quart","quote_start":64,"quote_end":225,"text_sha256":"a323699b5715893f38673813cc8f691b0f699ae6c284d0c892838d97b40d453c","block_sha256":"a323699b5715893f38673813cc8f691b0f699ae6c284d0c892838d97b40d453c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982/#blk_057f4f8a-82ec-4ae6-b672-585c53488c46"},{"id":"occ_f504fb92929129aa1214edcc","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","work_id":"wrk_2e3fa847-c032-4040-9284-2106f69cf66c","block_id":"blk_138bcd33-47ff-42b8-91c7-ba7fe48eb807","section_id":"sec_eb1406b8-d538-47d9-9ca1-8cd0f336eeff","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":18,"end":24,"exact":"エージェント","quote":"> 人間向け金融端末の企業から、AIエージェント向け金融データ基盤へ移行できるか","quote_start":0,"quote_end":40,"text_sha256":"d4de335761274800617a044401d6d6bf396f5a3cfebf5e71878cdedd14cea5c7","block_sha256":"d4de335761274800617a044401d6d6bf396f5a3cfebf5e71878cdedd14cea5c7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982/#blk_138bcd33-47ff-42b8-91c7-ba7fe48eb807"},{"id":"occ_cb952a290ab820122772b36e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","work_id":"wrk_2e3fa847-c032-4040-9284-2106f69cf66c","block_id":"blk_1882bf14-f219-4926-b299-e4402a75a0ed","section_id":"sec_fb1cbe04-8bb5-4ad3-852b-6e659834f28d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":67,"end":73,"exact":"エージェント","quote":"の事業更新で、独自コンテンツと包括的なデータセットをAI技術と組み合わせる方針を示している。法律部門では、AIエージェント「Protégé」を統合したLexis+の採用拡大によって、法律事務所・企業向け事業が二桁成長を続けている。科学・技術・医療部門でも、AI対応データベースや意思決定支援ツールの利用が拡大している。([","quote_start":12,"quote_end":173,"text_sha256":"8f854f76d8f9286e3dd9c5df21eb829e7ab053a80cf38777ba4d1a9511e04d17","block_sha256":"8f854f76d8f9286e3dd9c5df21eb829e7ab053a80cf38777ba4d1a9511e04d17","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982/#blk_1882bf14-f219-4926-b299-e4402a75a0ed"},{"id":"occ_72a6b4e414fe69f9e79f0763","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","work_id":"wrk_2e3fa847-c032-4040-9284-2106f69cf66c","block_id":"blk_19e1d767-8e71-48cb-8fb8-e1c287300d8d","section_id":"sec_eb1406b8-d538-47d9-9ca1-8cd0f336eeff","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":18,"end":24,"exact":"エージェント","quote":"FactSetは、自社のデータをAIエージェントが直接利用できるよう、MCPサーバーを提供している。","quote_start":0,"quote_end":50,"text_sha256":"c02c889145ab277fe8545631745d8156a9dd1f4c4f9af458612f69bd9d6de6d2","block_sha256":"c02c889145ab277fe8545631745d8156a9dd1f4c4f9af458612f69bd9d6de6d2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982/#blk_19e1d767-8e71-48cb-8fb8-e1c287300d8d"},{"id":"occ_14d7151f3f468124df54706b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","work_id":"wrk_2e3fa847-c032-4040-9284-2106f69cf66c","block_id":"blk_235f32a5-64cf-4cc5-b58e-cf1e4d5f46b6","section_id":"sec_dd1638da-b163-40e8-9f00-c14922674e4b","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":31,"end":37,"exact":"エージェント","quote":"IQVIAは2026年3月、臨床、商業、実世界データを横断するエージェント型AI基盤「IQVIA.ai」も発表した。([IQVIA](https://ir.iqvia.com/press-releases/press-release-details/2026/IQVIA-U","quote_start":0,"quote_end":137,"text_sha256":"59372e93862a48a5c868c66c0993e2ad27ce43cc353100d8822d858b25287575","block_sha256":"59372e93862a48a5c868c66c0993e2ad27ce43cc353100d8822d858b25287575","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982/#blk_235f32a5-64cf-4cc5-b58e-cf1e4d5f46b6"},{"id":"occ_a0f23ff4f6febfacbce07bb6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","work_id":"wrk_2e3fa847-c032-4040-9284-2106f69cf66c","block_id":"blk_2b70dac6-ead7-4f59-8795-36e4b4edfef2","section_id":"sec_e706cae8-839a-4ceb-a231-a0ab4291632e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"## AIエージェント時代の収益モデル","quote_start":0,"quote_end":19,"text_sha256":"fa3a2d422274fee48288db5edb8efbbca96a16caceb5ace43cfe8bf680cfdb19","block_sha256":"fa3a2d422274fee48288db5edb8efbbca96a16caceb5ace43cfe8bf680cfdb19","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982/#blk_2b70dac6-ead7-4f59-8795-36e4b4edfef2"},{"id":"occ_62b88fdff99d1b1b1774f6f3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","work_id":"wrk_2e3fa847-c032-4040-9284-2106f69cf66c","block_id":"blk_3f10e5c0-edb5-4424-9282-48c175bc60b2","section_id":"sec_e940512c-018c-496e-ace6-6772921482ef","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントが広がり、実際の業務を担当するようになるほど、AIにはインターネット検索だけでは得られない専門データが必要になる。","quote_start":0,"quote_end":65,"text_sha256":"40c59a3be72b5008d87a4e28648bfd7dfabaf32127de9c19f666bd0107820592","block_sha256":"40c59a3be72b5008d87a4e28648bfd7dfabaf32127de9c19f666bd0107820592","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982/#blk_3f10e5c0-edb5-4424-9282-48c175bc60b2"},{"id":"occ_9b6459974bd50312dce537bc","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","work_id":"wrk_2e3fa847-c032-4040-9284-2106f69cf66c","block_id":"blk_5337d8f3-1cb4-4613-adbb-aae40f66e306","section_id":"sec_bf22a0b0-0095-4d00-8427-11eda5c1610b","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントが必要なときに、安全かつ機械可読な形で呼び出せる必要がある。","quote_start":0,"quote_end":38,"text_sha256":"ee6686b424206ff6d0223c489ebbc43078a64cc505f588f5e4b1d592fe4a5d45","block_sha256":"ee6686b424206ff6d0223c489ebbc43078a64cc505f588f5e4b1d592fe4a5d45","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982/#blk_5337d8f3-1cb4-4613-adbb-aae40f66e306"},{"id":"occ_dae589609e52ad02f1dd2692","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","work_id":"wrk_2e3fa847-c032-4040-9284-2106f69cf66c","block_id":"blk_5df677a5-f39b-4086-b5f6-351ad99afba1","section_id":"sec_44e7a24e-18fb-4669-90d0-c95e8e841169","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントが、","quote_start":0,"quote_end":10,"text_sha256":"677c39930117dede8e1067f78ceeb9cd8a35b48264d85fa04a732dbe510704d5","block_sha256":"677c39930117dede8e1067f78ceeb9cd8a35b48264d85fa04a732dbe510704d5","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982/#blk_5df677a5-f39b-4086-b5f6-351ad99afba1"},{"id":"occ_807671ed5677784ce1f06618","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","work_id":"wrk_2e3fa847-c032-4040-9284-2106f69cf66c","block_id":"blk_8281ca58-e543-40b2-a7e7-8f255c71adcf","section_id":"sec_b8931ce4-5cd4-49b2-9014-8314c27baad6","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェント時代には、","quote_start":0,"quote_end":13,"text_sha256":"2d4d12a64f7ccc38e9c1bc8d62e6de3e7d30d27ab7fc23efac88a9c750703587","block_sha256":"2d4d12a64f7ccc38e9c1bc8d62e6de3e7d30d27ab7fc23efac88a9c750703587","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982/#blk_8281ca58-e543-40b2-a7e7-8f255c71adcf"},{"id":"occ_2014325810467fd4006771b8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","work_id":"wrk_2e3fa847-c032-4040-9284-2106f69cf66c","block_id":"blk_90ad1882-cd74-4c28-aee2-d31254de47c5","section_id":"sec_e706cae8-839a-4ceb-a231-a0ab4291632e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"- AIエージェントの数\n- API呼び出し回数\n- 処理した契約や請求の件数\n- データ取得量\n- AI判断の検証回数\n- 業務全体の処理量\n- 利用したデータセットの範囲","quote_start":0,"quote_end":87,"text_sha256":"9abc88c3d19aa2b563262f6f53dfb064c4e7e29a719a72e91f9a96b266fd16ed","block_sha256":"9abc88c3d19aa2b563262f6f53dfb064c4e7e29a719a72e91f9a96b266fd16ed","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982/#blk_90ad1882-cd74-4c28-aee2-d31254de47c5"},{"id":"occ_0eb3720383e343e0e313b29d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","work_id":"wrk_2e3fa847-c032-4040-9284-2106f69cf66c","block_id":"blk_93d4b341-f75d-4df7-9375-4bf96b1647a8","section_id":"sec_4c6f7f29-6cf5-4c13-8f80-afaad3b585a4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":51,"end":57,"exact":"エージェント","quote":"生成AIの進歩によって、AIは文章を作るだけでなく、情報を検索し、比較し、判断し、業務を実行する「AIエージェント」へ進化し始めている。","quote_start":0,"quote_end":68,"text_sha256":"22f05b7e0e6ffe04aaa2d122771edf483105485df1424db55f9ee1c67120863a","block_sha256":"22f05b7e0e6ffe04aaa2d122771edf483105485df1424db55f9ee1c67120863a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982/#blk_93d4b341-f75d-4df7-9375-4bf96b1647a8"},{"id":"occ_00fcef7a59d96b9cd9af39b0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","work_id":"wrk_2e3fa847-c032-4040-9284-2106f69cf66c","block_id":"blk_9bb38dad-ab01-4b09-ac98-ad6cabb09d1f","section_id":"sec_89f912dd-2566-43e8-9cb1-75c03245e47e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントが企業間取引や融資、サプライヤー選定を行うようになると、企業の信用力を判断するデータが必要になる。","quote_start":0,"quote_end":57,"text_sha256":"6115e3413bc05b51be214e464d6da1857d4b9dc7de6019fbd6dc20470f2f180b","block_sha256":"6115e3413bc05b51be214e464d6da1857d4b9dc7de6019fbd6dc20470f2f180b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982/#blk_9bb38dad-ab01-4b09-ac98-ad6cabb09d1f"},{"id":"occ_9e1ad46c6e2294b76c1fb9fe","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","work_id":"wrk_2e3fa847-c032-4040-9284-2106f69cf66c","block_id":"blk_a9bf2ad0-d144-467f-92c8-c7828efba095","section_id":"sec_b8931ce4-5cd4-49b2-9014-8314c27baad6","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"しかし、AIエージェントが実際の業務に入るほど、逆の動きが起こる可能性がある。","quote_start":0,"quote_end":39,"text_sha256":"e178c034e5c8ba6d437ffac54e05cf845d34b9265bd756c8c2250592506351e6","block_sha256":"e178c034e5c8ba6d437ffac54e05cf845d34b9265bd756c8c2250592506351e6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982/#blk_a9bf2ad0-d144-467f-92c8-c7828efba095"},{"id":"occ_4819d69820386dcf6393ce7f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","work_id":"wrk_2e3fa847-c032-4040-9284-2106f69cf66c","block_id":"blk_b34493e2-e78f-4417-8656-9db8d1a93857","section_id":"sec_6226c06c-f94e-43fb-9452-49336c56eaa1","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":19,"end":25,"exact":"エージェント","quote":"AIによる端末代替懸念がある一方、AIエージェントへデータを直接供給する側へ移行し始めている。","quote_start":0,"quote_end":47,"text_sha256":"b6f619b93c02b542b4a8ab81f61a5eeba9c2ae30bfb1ece1e65f11ec13d520b1","block_sha256":"b6f619b93c02b542b4a8ab81f61a5eeba9c2ae30bfb1ece1e65f11ec13d520b1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982/#blk_b34493e2-e78f-4417-8656-9db8d1a93857"},{"id":"occ_2d8aabb8859b2a688e7c3e85","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","work_id":"wrk_2e3fa847-c032-4040-9284-2106f69cf66c","block_id":"blk_bcf2e078-ff66-4fb2-a811-f8e0bb4d2b01","section_id":"sec_f8e9d85c-1c7a-4eaf-9936-c94343e4ca78","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":44,"end":50,"exact":"エージェント","quote":"- AI審査システムからの信用データ照会\n- KYC・AML確認回数の増加\n- 企業評価エージェントへのAPI提供\n- 民間信用・サプライチェーン評価への拡大\n- AI判断の検証・説明用ベンチマーク","quote_start":0,"quote_end":99,"text_sha256":"d0159e8d5d2858d29b5c43106d0d94525b81853f0dfd6facfe5bb1c48ba34a5e","block_sha256":"d0159e8d5d2858d29b5c43106d0d94525b81853f0dfd6facfe5bb1c48ba34a5e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982/#blk_bcf2e078-ff66-4fb2-a811-f8e0bb4d2b01"},{"id":"occ_84937f81f73887d3b095b8c9","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","work_id":"wrk_2e3fa847-c032-4040-9284-2106f69cf66c","block_id":"blk_f73b907e-e0ac-4e48-828d-99fbbebdef29","section_id":"sec_f194f313-1cd5-440c-8ca0-6f5275c0d37e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"- AIエージェントからのデータ呼び出し\n- API・MCP契約の拡大\n- 非構造化文書の自動取り込み\n- ポートフォリオ分析の自動化\n- 企業全体での一括契約\n- 小規模運用会社へのAI製品展開","quote_start":0,"quote_end":98,"text_sha256":"6d2b96d4db57a16dcc32e48a0017af65a86ca15cf3084c9dcfcf46f0c6823a70","block_sha256":"6d2b96d4db57a16dcc32e48a0017af65a86ca15cf3084c9dcfcf46f0c6823a70","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982/#blk_f73b907e-e0ac-4e48-828d-99fbbebdef29"},{"id":"occ_8ed9fe2189f6fd719755919f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","work_id":"wrk_2e3fa847-c032-4040-9284-2106f69cf66c","block_id":"blk_ff956f3c-3fed-4326-a441-cd614ce86275","section_id":"sec_e706cae8-839a-4ceb-a231-a0ab4291632e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"しかしAIエージェント時代には、課金対象が変わる可能性がある。","quote_start":0,"quote_end":31,"text_sha256":"4ac75b20fe52180321fc8e3ad36a806fa0eecf8b9253e332e2e7ff61739a3277","block_sha256":"4ac75b20fe52180321fc8e3ad36a806fa0eecf8b9253e332e2e7ff61739a3277","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982/#blk_ff956f3c-3fed-4326-a441-cd614ce86275"},{"id":"occ_b6e0b171e1685765fc54ae11","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_716e29e3-ca33-4e0a-89d8-a4c146c91139","work_id":"wrk_6b2d4c96-c0d4-4d57-8d9e-9077ea538a79","block_id":"blk_0952975b-af66-4aaf-be9e-0c7ca101c63f","section_id":"sec_9e51f881-1f48-4224-aeb7-8b7069f50b19","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"常時稼働するエージェントが一般化すれば、","quote_start":0,"quote_end":20,"text_sha256":"cb1d8b065949b4b39166a1c16a046242f3671c96e6b162960c9f9d2761b8b809","block_sha256":"cb1d8b065949b4b39166a1c16a046242f3671c96e6b162960c9f9d2761b8b809","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_716e29e3-ca33-4e0a-89d8-a4c146c91139/#blk_0952975b-af66-4aaf-be9e-0c7ca101c63f"},{"id":"occ_0552eb7e97e0da53b9fc549f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_716e29e3-ca33-4e0a-89d8-a4c146c91139","work_id":"wrk_6b2d4c96-c0d4-4d57-8d9e-9077ea538a79","block_id":"blk_6e99559b-f28a-4952-8682-f2fa667eba7d","section_id":"sec_629922eb-3fe0-4002-9241-7f9bba3806da","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":55,"end":61,"exact":"エージェント","quote":"- 蒸留\n- ファインチューニング\n- RAG\n- 合成データ\n- 強化学習\n- モデルルーティング\n- 専門エージェント","quote_start":0,"quote_end":61,"text_sha256":"f2c2dce280f0bbfdd3757fe7b0f7892b14bc65b3b6e87be868495dc797c3d781","block_sha256":"f2c2dce280f0bbfdd3757fe7b0f7892b14bc65b3b6e87be868495dc797c3d781","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_716e29e3-ca33-4e0a-89d8-a4c146c91139/#blk_6e99559b-f28a-4952-8682-f2fa667eba7d"},{"id":"occ_86611c6a2d058706f1bdcb81","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_716e29e3-ca33-4e0a-89d8-a4c146c91139","work_id":"wrk_6b2d4c96-c0d4-4d57-8d9e-9077ea538a79","block_id":"blk_8602a4a6-355a-4ecd-a12d-3ae0069d5b7d","section_id":"sec_b291f9db-4adf-4e44-877d-23c817eb77cf","layer":"code","character_id":null,"count":1,"matched_aliases":["AI agent"],"evidence":{"text_basis":"markdown","start":226,"end":234,"exact":"AI agent","quote":"ting cost optimization\nsmall language model enterprise\nAI agent consumption revenue\n```","quote_start":171,"quote_end":258,"text_sha256":"9095dc058b1af565bd31184020a883e659b4729b7f376efabe8dc10b1cb36b51","block_sha256":"9095dc058b1af565bd31184020a883e659b4729b7f376efabe8dc10b1cb36b51","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_716e29e3-ca33-4e0a-89d8-a4c146c91139/#blk_8602a4a6-355a-4ecd-a12d-3ae0069d5b7d"},{"id":"occ_d6b752b08ee7d4a34d07251f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_716e29e3-ca33-4e0a-89d8-a4c146c91139","work_id":"wrk_6b2d4c96-c0d4-4d57-8d9e-9077ea538a79","block_id":"blk_8dc5a3c7-ea63-4dee-9c98-6cfe3f5a7553","section_id":"sec_178350bd-26ca-499c-8551-e477403319a3","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":75,"end":81,"exact":"エージェント","quote":"- より多くの強化学習\n- 合成データの生成\n- 多数のAI研究実験\n- 高性能な動画・音声生成\n- 常時稼働エージェント","quote_start":20,"quote_end":81,"text_sha256":"9e92cad6992db2adeac5e08c71c0255ed47de4cf7fe3bfdbca07b0c4769a7835","block_sha256":"9e92cad6992db2adeac5e08c71c0255ed47de4cf7fe3bfdbca07b0c4769a7835","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_716e29e3-ca33-4e0a-89d8-a4c146c91139/#blk_8dc5a3c7-ea63-4dee-9c98-6cfe3f5a7553"},{"id":"occ_f19005d2358a9e61de32583e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_716e29e3-ca33-4e0a-89d8-a4c146c91139","work_id":"wrk_6b2d4c96-c0d4-4d57-8d9e-9077ea538a79","block_id":"blk_93b59bd6-b7a7-43d4-af3d-377b4524e411","section_id":"sec_6e0d26a1-950e-42e4-9e15-afa4c1db315a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":55,"end":61,"exact":"エージェント","quote":"- 全コード変更を複数AIが確認する\n- 全顧客を常時分析する\n- 全製造装置を監視する\n- 一人の背後で複数エージェントが働く\n- 数千の設計候補を試す\n- 企業ごと、部署ごとにモデルを作る","quote_start":0,"quote_end":96,"text_sha256":"6be2b115540fc3e6108e2fb27a63e8b654a53086b0ccbf56601147bf9ea636ad","block_sha256":"6be2b115540fc3e6108e2fb27a63e8b654a53086b0ccbf56601147bf9ea636ad","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_716e29e3-ca33-4e0a-89d8-a4c146c91139/#blk_93b59bd6-b7a7-43d4-af3d-377b4524e411"},{"id":"occ_32ac778508bcb5f5fa65e876","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_716e29e3-ca33-4e0a-89d8-a4c146c91139","work_id":"wrk_6b2d4c96-c0d4-4d57-8d9e-9077ea538a79","block_id":"blk_9718cb42-5cf1-4b17-b117-1182d0fedebb","section_id":"sec_5d3d387f-48fe-44ca-9bb5-f3d7ca2f446c","layer":"code","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":64,"end":70,"exact":"エージェント","quote":"| インターネット黎明期 | 現在のAI普及 |\n| 利用者 | Webサイト・通信量が急増 | AI利用者・エージェント・生成量が急増 |\n| 収益化 | 利用増に対してEC・広告売上は小さい | 利用増に対してAI単体利益はまだ限定的 |\n| 設備 | 光ファイバー、ルーター、サーバー、PC | AIDC、GPU、H","quote_start":9,"quote_end":170,"text_sha256":"5b198f97230c3b3e843182a98cb9b95eeac3cb45b1c38e0c757d11bc03239298","block_sha256":"5b198f97230c3b3e843182a98cb9b95eeac3cb45b1c38e0c757d11bc03239298","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_716e29e3-ca33-4e0a-89d8-a4c146c91139/#blk_9718cb42-5cf1-4b17-b117-1182d0fedebb"},{"id":"occ_4859aca31cd43de9e75da2a3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_716e29e3-ca33-4e0a-89d8-a4c146c91139","work_id":"wrk_6b2d4c96-c0d4-4d57-8d9e-9077ea538a79","block_id":"blk_a03c2b44-a2d2-4d8f-a60b-721181e8c4c5","section_id":"sec_3374abe7-3c9f-4d03-947d-032d05bf850a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":34,"end":40,"exact":"エージェント","quote":"一方、強力なモデルウェイトが外部へ流通し、安全調整を外した追加学習やエージェント化が安価になれば、中央管理は難しくなる。","quote_start":0,"quote_end":60,"text_sha256":"ce34283d21c591183171dc8069c19ddc4d9bc0b22e5cb6b75294cc1db058ac92","block_sha256":"ce34283d21c591183171dc8069c19ddc4d9bc0b22e5cb6b75294cc1db058ac92","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_716e29e3-ca33-4e0a-89d8-a4c146c91139/#blk_a03c2b44-a2d2-4d8f-a60b-721181e8c4c5"},{"id":"occ_6e303f99a6f32c8d86caeab3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_716e29e3-ca33-4e0a-89d8-a4c146c91139","work_id":"wrk_6b2d4c96-c0d4-4d57-8d9e-9077ea538a79","block_id":"blk_b291e008-fa55-4ef3-a746-6b0d14b9f4d8","section_id":"sec_629922eb-3fe0-4002-9241-7f9bba3806da","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":15,"end":21,"exact":"エージェント","quote":"## 3．下層――専門モデルとエージェントが大量発生する","quote_start":0,"quote_end":28,"text_sha256":"55ec205a77aa5dbce8af4f08813bfe414a2e8f352c58274a20031911aeac9809","block_sha256":"55ec205a77aa5dbce8af4f08813bfe414a2e8f352c58274a20031911aeac9809","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_716e29e3-ca33-4e0a-89d8-a4c146c91139/#blk_b291e008-fa55-4ef3-a746-6b0d14b9f4d8"},{"id":"occ_272b65813edff609bdb36b14","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_716e29e3-ca33-4e0a-89d8-a4c146c91139","work_id":"wrk_6b2d4c96-c0d4-4d57-8d9e-9077ea538a79","block_id":"blk_f5b50294-01f0-4935-9ff8-9c0e8dbc36e6","section_id":"sec_519ed44e-464a-476d-8e60-62d82c5254c2","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":179,"end":185,"exact":"エージェント","quote":"ータセンター\n6. ハイパースケーラー\n7. フロンティアモデル企業\n8. オープンウェイトモデル\n9. AIエージェント基盤\n10. SaaS・業務アプリ\n11. 産業別AI\n12. フィジカルAI\n\n各階層について、\n\n・売上成長率\n・粗利益率\n・営業利益率\n・設備投資負担\n・価格決定力\n・供給制約\n・競争参入\n・顧","quote_start":124,"quote_end":285,"text_sha256":"fdb588809b99c07fb5b55956d00bcf4730fa0a43ae820db78986588d316bc048","block_sha256":"fdb588809b99c07fb5b55956d00bcf4730fa0a43ae820db78986588d316bc048","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_716e29e3-ca33-4e0a-89d8-a4c146c91139/#blk_f5b50294-01f0-4935-9ff8-9c0e8dbc36e6"},{"id":"occ_9f0f18edc77639131f1a20dd","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_716e29e3-ca33-4e0a-89d8-a4c146c91139","work_id":"wrk_6b2d4c96-c0d4-4d57-8d9e-9077ea538a79","block_id":"blk_f7c0707e-5fa2-4b75-a5bb-e7c4c40e83d2","section_id":"sec_ab0038a9-f702-4044-b057-fbd624bde8ca","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":61,"end":67,"exact":"エージェント","quote":"広がり、労働、金融、制作、設計、科学、機械制御、防衛を変える可能性が高い。モデル開発が自動化され、専門モデルやエージェントが大量発生すれば、計算需要は人間の利用時間を超えて増え得る。","quote_start":6,"quote_end":97,"text_sha256":"9bcebfdf50f9b1ce75c15c8e96bc16aca34228277bb970e858a276eae3de5c71","block_sha256":"9bcebfdf50f9b1ce75c15c8e96bc16aca34228277bb970e858a276eae3de5c71","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_716e29e3-ca33-4e0a-89d8-a4c146c91139/#blk_f7c0707e-5fa2-4b75-a5bb-e7c4c40e83d2"},{"id":"occ_071f5d283e00246eaa48c8b1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_716e29e3-ca33-4e0a-89d8-a4c146c91139","work_id":"wrk_6b2d4c96-c0d4-4d57-8d9e-9077ea538a79","block_id":"blk_f89fcf92-6994-4ea2-b234-fa0db7a67cd9","section_id":"sec_96c2a5e0-8ab0-44a8-961d-8d5167b78c52","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":36,"end":42,"exact":"エージェント","quote":"計算を増やせば、より多くのコード、動画、設計案、シミュレーション、調査、エージェント処理を実行できる。しかし、","quote_start":0,"quote_end":55,"text_sha256":"5de26c7b45ae5361d643fc7ed6732e6964ee696656966ce636995adca4ff3c7f","block_sha256":"5de26c7b45ae5361d643fc7ed6732e6964ee696656966ce636995adca4ff3c7f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_716e29e3-ca33-4e0a-89d8-a4c146c91139/#blk_f89fcf92-6994-4ea2-b234-fa0db7a67cd9"},{"id":"occ_71bdcc5e031811a2696c337e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_73f07a40-b588-4fa6-bf2b-2880462e214c","work_id":"wrk_83edd5a9-4c70-4085-ba5b-602f9d005928","block_id":"blk_3d2a46f1-0ae8-4cec-a6b0-e59bc48e79b9","section_id":"sec_87694341-69d5-4a18-92c0-1fa5e938eb03","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":0,"end":7,"exact":"Agentic","quote":"Agentic AIでは、","quote_start":0,"quote_end":13,"text_sha256":"8c41d26564c2f801a0f5c36640edd6de837f19cba4c8e7c0633b16e3c87790ad","block_sha256":"8c41d26564c2f801a0f5c36640edd6de837f19cba4c8e7c0633b16e3c87790ad","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_73f07a40-b588-4fa6-bf2b-2880462e214c/#blk_3d2a46f1-0ae8-4cec-a6b0-e59bc48e79b9"},{"id":"occ_60278044afdab30387620b31","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_73f07a40-b588-4fa6-bf2b-2880462e214c","work_id":"wrk_83edd5a9-4c70-4085-ba5b-602f9d005928","block_id":"blk_4c9dae61-8cef-4ee1-884c-724c2edb2a28","section_id":"sec_87694341-69d5-4a18-92c0-1fa5e938eb03","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":3,"end":10,"exact":"Agentic","quote":"つまりAgentic AIは、","quote_start":0,"quote_end":15,"text_sha256":"a3cf86a140cfab7e8c2c6ab3516e90eb0a285532f87b79b3eab5e68c8f9f9f25","block_sha256":"a3cf86a140cfab7e8c2c6ab3516e90eb0a285532f87b79b3eab5e68c8f9f9f25","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_73f07a40-b588-4fa6-bf2b-2880462e214c/#blk_4c9dae61-8cef-4ee1-884c-724c2edb2a28"},{"id":"occ_fc0fdf6b2077af0ac93e29c8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_73f07a40-b588-4fa6-bf2b-2880462e214c","work_id":"wrk_83edd5a9-4c70-4085-ba5b-602f9d005928","block_id":"blk_723dd4dd-ebfc-4363-8330-a21930a37b1d","section_id":"sec_658e3a48-a561-4a0f-b5ac-8ad6a0eb7b64","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agentが、","quote_start":0,"quote_end":10,"text_sha256":"eb540ab992433829579400ab7bea51752bc4aabcdfa9bbe6420e86c88472346b","block_sha256":"eb540ab992433829579400ab7bea51752bc4aabcdfa9bbe6420e86c88472346b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_73f07a40-b588-4fa6-bf2b-2880462e214c/#blk_723dd4dd-ebfc-4363-8330-a21930a37b1d"},{"id":"occ_bb3f52c4bedcc9eaa0f1b771","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_73f07a40-b588-4fa6-bf2b-2880462e214c","work_id":"wrk_83edd5a9-4c70-4085-ba5b-602f9d005928","block_id":"blk_861eb885-469a-4d09-8b5f-f8d79f32ad88","section_id":"sec_a3fe0f2e-5e67-4b81-9e18-87a91ad8f059","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":2,"end":10,"exact":"AI Agent","quote":"**AI AgentがDesign Costを下げ始めている。**","quote_start":0,"quote_end":33,"text_sha256":"c2d0614c16b940ec5890aac78e2c09e01dc9b0f9d6ce5f53d8b7b876513de5f6","block_sha256":"c2d0614c16b940ec5890aac78e2c09e01dc9b0f9d6ce5f53d8b7b876513de5f6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_73f07a40-b588-4fa6-bf2b-2880462e214c/#blk_861eb885-469a-4d09-8b5f-f8d79f32ad88"},{"id":"occ_06c5ea77946343c98997454b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_73f07a40-b588-4fa6-bf2b-2880462e214c","work_id":"wrk_83edd5a9-4c70-4085-ba5b-602f9d005928","block_id":"blk_da249060-b0c5-4770-8bc3-cc0e5c5f1086","section_id":"sec_6b671282-6d14-4c1b-bc32-23b7886d13fa","layer":"body","character_id":null,"count":2,"matched_aliases":["AI Agent","Agentic"],"evidence":{"text_basis":"markdown","start":2,"end":9,"exact":"Agentic","quote":"**Agentic EDA――AI Agentを使った半導体設計**","quote_start":0,"quote_end":34,"text_sha256":"62d3d4f6158f0a1a36079194851e62ac6c03c7572b5d981f4e2680f4fe1518f1","block_sha256":"62d3d4f6158f0a1a36079194851e62ac6c03c7572b5d981f4e2680f4fe1518f1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_73f07a40-b588-4fa6-bf2b-2880462e214c/#blk_da249060-b0c5-4770-8bc3-cc0e5c5f1086"},{"id":"occ_837ac728993b59501d277a57","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_097cf615-76f4-4910-8752-c425f13212dd","section_id":"sec_3ec18778-c220-4729-884b-a3a2e1b9e28c","layer":"body","character_id":null,"count":3,"matched_aliases":["Agentic","エージェント"],"evidence":{"text_basis":"markdown","start":48,"end":55,"exact":"Agentic","quote":"特にStripeは、2025年にStripe上の取扱高が**1.9兆ドル、前年比+34%となり、Agentic Commerce Protocol、Agentic Commerce Suite、Shared Payment Tokens、stablecoin micropaymentsなど、AIエージェント決","quote_start":0,"quote_end":155,"text_sha256":"2205500f6d91a5674cde24e8dc8f7e7fdf5b96baaaeed45695c4e7b3fdfc485d","block_sha256":"2205500f6d91a5674cde24e8dc8f7e7fdf5b96baaaeed45695c4e7b3fdfc485d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_097cf615-76f4-4910-8752-c425f13212dd"},{"id":"occ_bef24913c4e971730c59cbb1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_0e415825-7fab-4d64-ba3c-eca35d5d5c1f","section_id":"sec_9b3d2bb5-4f2f-4b20-a37a-6644b0f10340","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":20,"end":26,"exact":"エージェント","quote":"これはかなり強いです。  \n理由は、AIエージェントが増えると、決済は単に「ボタンを押す」ものではなくなり、","quote_start":0,"quote_end":54,"text_sha256":"3eb3b26b58409f34e783abdbbe47ef11b06740e6eb5cd460cfb69a4a9b3a81c4","block_sha256":"3eb3b26b58409f34e783abdbbe47ef11b06740e6eb5cd460cfb69a4a9b3a81c4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_0e415825-7fab-4d64-ba3c-eca35d5d5c1f"},{"id":"occ_a1e8ab120b3d4f032609aeb7","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_1bd42b50-3b16-478a-83d5-0da068b68d91","section_id":"sec_6fe0ccc4-a644-4a12-9800-311dc164661a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":79,"end":85,"exact":"エージェント","quote":"EBITDAマージン53%は強烈です。ただし市場は成長率に厳しく、決済ボリュームが期待未達だと売られます。AIエージェント決済や統合コマースで再加速できるなら面白いです。","quote_start":24,"quote_end":109,"text_sha256":"a0f5a0291e0fc8927bebcc8752b33e7271d2c89b75d91bfd160ebcf91c3c16dc","block_sha256":"a0f5a0291e0fc8927bebcc8752b33e7271d2c89b75d91bfd160ebcf91c3c16dc","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_1bd42b50-3b16-478a-83d5-0da068b68d91"},{"id":"occ_97b68d3d7a487774d5d10b30","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_2f4f02b3-9302-4e90-aa56-a578298928b3","section_id":"sec_100c03a0-e2a2-4a21-a254-52651ed71db2","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":71,"end":77,"exact":"エージェント","quote":"認**  \n**不正検知**  \n**会計連携**  \n**請求書処理**  \n**調達管理**  \n**AIエージェントによる支出最適化**","quote_start":16,"quote_end":87,"text_sha256":"031bb049b610cb607b48bb326f77565df099d1ba1a8a073c0db09d9955d74725","block_sha256":"031bb049b610cb607b48bb326f77565df099d1ba1a8a073c0db09d9955d74725","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_2f4f02b3-9302-4e90-aa56-a578298928b3"},{"id":"occ_ce57bb5abc53429ad1ff9260","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_3187f127-59a4-4f74-902a-e08c911dbc8d","section_id":"sec_1762f1b2-6856-4a8a-809c-24d11c2a49e9","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"> **AIエージェントが動かすお金の流れを、決済・認証・不正検知・会計・ステーブルコイン・多通貨口座で握る企業**","quote_start":0,"quote_end":58,"text_sha256":"92794922208ed78d20c8c10c94915923b61637547a51defee47e543051a6e06b","block_sha256":"92794922208ed78d20c8c10c94915923b61637547a51defee47e543051a6e06b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_3187f127-59a4-4f74-902a-e08c911dbc8d"},{"id":"occ_de2ad23cf002142685ac601b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_53af2d97-cecc-4521-a0cf-a13ac53ea0a1","section_id":"sec_1762f1b2-6856-4a8a-809c-24d11c2a49e9","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":34,"end":40,"exact":"エージェント","quote":"AI時代に強いフィンテックは、単なる「金融アプリ」ではなく、**AIエージェント、人間、企業、API、国際送金、ステーブルコイン、会計システムのあいだで“お金の通過点”を握る企業**です。","quote_start":0,"quote_end":94,"text_sha256":"7d7b16fbd4ec19425507d435923bb5e912e6f8adba2df094a337e91ff6f38a1e","block_sha256":"7d7b16fbd4ec19425507d435923bb5e912e6f8adba2df094a337e91ff6f38a1e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_53af2d97-cecc-4521-a0cf-a13ac53ea0a1"},{"id":"occ_89fd0302feb3dfdbd058aeee","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_56c24f36-d69a-48de-9207-26e34ff81c1c","section_id":"sec_9b3d2bb5-4f2f-4b20-a37a-6644b0f10340","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"**エージェント本人確認**  \n**代理購入の認可**  \n**決済トークン**  \n**不正検知**  \n**返品/返金**  \n**税務**  \n**請求書**  \n**サブスク管理**  \n**国際決済** ","quote_start":0,"quote_end":108,"text_sha256":"2ed34cf66d7af71afdc9e5e72fd8ac6e027e1957093d5fd9d7b1069eb9fe742e","block_sha256":"2ed34cf66d7af71afdc9e5e72fd8ac6e027e1957093d5fd9d7b1069eb9fe742e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_56c24f36-d69a-48de-9207-26e34ff81c1c"},{"id":"occ_f7937cb05e90dd17414571b7","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_5a48be20-6455-4f71-bfaa-dd178d45a078","section_id":"sec_c1490e83-e70d-4d97-bf2b-ce3d3361443f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":67,"end":73,"exact":"エージェント","quote":"*リスク管理**  \n**台帳整合性**  \n**銀行規制対応**  \n**リアルタイム決済**  \n**AIエージェント決済対応**","quote_start":12,"quote_end":79,"text_sha256":"ea142dc4d88058831810e9e44ca3d1d1208d4925441cb5acecd0322a83d2f544","block_sha256":"ea142dc4d88058831810e9e44ca3d1d1208d4925441cb5acecd0322a83d2f544","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_5a48be20-6455-4f71-bfaa-dd178d45a078"},{"id":"occ_ccfb503593e9d5818d731286","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_5b77c28f-7dea-4080-b9d4-f82d1050ec9d","section_id":"sec_7422cc60-eb7c-4094-a6d4-837794a171e0","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":21,"end":27,"exact":"エージェント","quote":"**Stripe**  \n最重要です。AIエージェント決済、ステーブルコイン、Billing、Tax、Revenue Recognition、Identity、不正検知まで持っています。上場すれば、AI時代のフィンテック本命としてかなり注目されると思いま","quote_start":0,"quote_end":127,"text_sha256":"22e2e0aa20fa8ff8e439a3ffe9a8e01c57bf7a82fb9458aab0f362c3d1b7359d","block_sha256":"22e2e0aa20fa8ff8e439a3ffe9a8e01c57bf7a82fb9458aab0f362c3d1b7359d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_5b77c28f-7dea-4080-b9d4-f82d1050ec9d"},{"id":"occ_9d1f37ce6dac7b7e92393fc1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_661d083a-3234-41ba-9d8e-cde3bd05d6c4","section_id":"sec_176a5787-5846-4153-a95c-e438898c7d3b","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":44,"end":50,"exact":"エージェント","quote":"Stripe、Adyen、Checkout.com、PayPal、Block。  \nAIエージェントが買い物をするほど、決済・本人確認・不正検知・返金・税務の重要性が上がります。","quote_start":0,"quote_end":89,"text_sha256":"daa512eba4452190ba6f36efd36f33d5df76ba964415fba3da004936248ddb03","block_sha256":"daa512eba4452190ba6f36efd36f33d5df76ba964415fba3da004936248ddb03","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_661d083a-3234-41ba-9d8e-cde3bd05d6c4"},{"id":"occ_f9ec0630db80c19a3c7c5255","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_76c72695-5463-4274-a422-4a5fb9c1b40e","section_id":"sec_f91d79d4-e252-4acb-817a-98baec7f75b7","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":9,"end":15,"exact":"エージェント","quote":"強いのは、**AIエージェント、人間、企業、API、会計、決済、銀行口座のあいだで、お金の通過点を握る企業**です。","quote_start":0,"quote_end":58,"text_sha256":"2a78c42a053452162e52c6b7c0935925421bc51b485c877f1ec607220c4040a7","block_sha256":"2a78c42a053452162e52c6b7c0935925421bc51b485c877f1ec607220c4040a7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_76c72695-5463-4274-a422-4a5fb9c1b40e"},{"id":"occ_8fb3449112af98c1e4ddd298","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_7824a4ab-a4fc-44fd-8d69-103521696f2e","section_id":"sec_6fe0ccc4-a644-4a12-9800-311dc164661a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":31,"end":37,"exact":"エージェント","quote":"**Circle**  \nステーブルコイン決済の本命です。AIエージェント決済・オンチェーン金融が本当に来るなら強い。ただし金利低下と規制の影響が大きいので、かなりテーマ株色もあります。","quote_start":0,"quote_end":93,"text_sha256":"61d7c326ea9e548e4c6dbd1d3984a4f832e46d3255fb4d90ff24ef3c2e16057f","block_sha256":"61d7c326ea9e548e4c6dbd1d3984a4f832e46d3255fb4d90ff24ef3c2e16057f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_7824a4ab-a4fc-44fd-8d69-103521696f2e"},{"id":"occ_ac95be298a1a7e9fcb4ec06c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_81b4e290-0329-402a-b073-b828d520a605","section_id":"sec_026f637e-54cb-4234-ad28-8f042cdf4c89","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":80,"end":86,"exact":"エージェント","quote":"アプリではなく「金融の管制塔」です。**  \n> お金が動く場所、権限が通る場所、データが検査される場所、AIエージェントが決済する場所を握る企業が強いです。","quote_start":25,"quote_end":104,"text_sha256":"abc51c8a4e42469f08e389f0c60492ef4a340207795abb354a32dddd57735ff2","block_sha256":"abc51c8a4e42469f08e389f0c60492ef4a340207795abb354a32dddd57735ff2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_81b4e290-0329-402a-b073-b828d520a605"},{"id":"occ_06f7b4f34bab12fbd26a477b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_82cc1c45-0649-4f5c-9d85-ea4c7350563a","section_id":"sec_88a0ce7a-f602-404f-9b58-6d48f2fef6e1","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":50,"end":57,"exact":"Agentic","quote":"Stripeは2025年にStripe上の取扱高が1.9兆ドル、前年比+34%となり、OpenAIとAgentic Commerce Protocolを開発し、Shared Payment Tokensやステーブルコイン小口決済にも取り組んでいます。([Stripe](https://stripe.com/new","quote_start":0,"quote_end":157,"text_sha256":"6962575066c85d5a09290c1d9ea39b6ff87de5a2d819eee1736018a9629d3d53","block_sha256":"6962575066c85d5a09290c1d9ea39b6ff87de5a2d819eee1736018a9629d3d53","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_82cc1c45-0649-4f5c-9d85-ea4c7350563a"},{"id":"occ_deaf2fd93f0ffda5f2cd0291","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_8bc8cf12-05e8-497c-8f4f-4df4f50f17f0","section_id":"sec_4bc153aa-65e9-44df-a73c-644dd6100c9f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":56,"end":62,"exact":"エージェント","quote":"*台帳整合性**  \n**規制対応**  \n**不正検知**  \n**リアルタイムリスク管理**  \n**AIエージェント決済対応**","quote_start":1,"quote_end":68,"text_sha256":"59558c5841328d0887a149ad2d635782c8efd2d801ad95f58ad1bfb5f6b3aa6b","block_sha256":"59558c5841328d0887a149ad2d635782c8efd2d801ad95f58ad1bfb5f6b3aa6b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_8bc8cf12-05e8-497c-8f4f-4df4f50f17f0"},{"id":"occ_09306594ef1d2fccc0b075b4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_9b8dee0a-e4b1-4477-b664-0b4a17229273","section_id":"sec_3ec18778-c220-4729-884b-a3a2e1b9e28c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":22,"end":28,"exact":"エージェント","quote":"ここはAI時代にかなり強いです。理由は、AIエージェントが買い物や契約更新、API課金、SaaS契約、旅行予約、業務購買を代行する時、最後には必ず**決済、認証、不正検知、返金、税務、請求書、サブスク管理**が必要になるからです。","quote_start":0,"quote_end":115,"text_sha256":"05348139efc3968c597084c6cdc48af7124790f2fecc2db0b2583ecbd228898f","block_sha256":"05348139efc3968c597084c6cdc48af7124790f2fecc2db0b2583ecbd228898f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_9b8dee0a-e4b1-4477-b664-0b4a17229273"},{"id":"occ_4af0e646e02ed9fae15c51bc","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_9efc5cb5-a18b-4d5c-865f-4ea3b9648710","section_id":"sec_9b3d2bb5-4f2f-4b20-a37a-6644b0f10340","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":11,"end":17,"exact":"エージェント","quote":"特にStripeはAIエージェント決済をかなり真正面から取りに行っています。Adyenは大企業・実店舗・EC統合に強く、Checkout.comは高成長EC/デジタル企業に強いです。","quote_start":0,"quote_end":91,"text_sha256":"63009cf68d281a36dae141d70c8595127109b7e87421bbcf30cdb5048d17059a","block_sha256":"63009cf68d281a36dae141d70c8595127109b7e87421bbcf30cdb5048d17059a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_9efc5cb5-a18b-4d5c-865f-4ea3b9648710"},{"id":"occ_85044c6e04a694b5e5fee18b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_a5de7613-d7ba-4a94-bf8b-a73dce3888e0","section_id":"sec_29336012-8d47-43ca-b1fc-1a7df1bc754d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"AI時代に、エージェントがAPI利用料を小口で支払う、グローバルにB2B送金する、24時間決済する、という世界が来るなら、ステーブルコインはかなり相性が良いです。","quote_start":0,"quote_end":81,"text_sha256":"4ec5d1debae91c6d53af61bfd9475cf0c90388dc3eff90e3597f6f4ec457f379","block_sha256":"4ec5d1debae91c6d53af61bfd9475cf0c90388dc3eff90e3597f6f4ec457f379","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_a5de7613-d7ba-4a94-bf8b-a73dce3888e0"},{"id":"occ_f79b76dcfe821e3aa987e7a3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_bd795568-8afc-4eb0-b6e8-9956d93fbc15","section_id":"sec_88a0ce7a-f602-404f-9b58-6d48f2fef6e1","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントが買い物、予約、SaaS契約、API課金、B2B購買を代行する時代になると、最後には必ず決済、本人確認、不正検知、返金、税務、請求書、サブスク管理が必要になります。","quote_start":0,"quote_end":90,"text_sha256":"f9c15a7712a40d01302b85a41178730b56dd4f2d5c6c752ad3e1cc61a688c470","block_sha256":"f9c15a7712a40d01302b85a41178730b56dd4f2d5c6c752ad3e1cc61a688c470","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_bd795568-8afc-4eb0-b6e8-9956d93fbc15"},{"id":"occ_3decd8a7784e73462f55ca16","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_e150bd1a-9589-474f-bdeb-e21cc4b62331","section_id":"sec_026f637e-54cb-4234-ad28-8f042cdf4c89","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":48,"end":55,"exact":"agentic","quote":"**決済企業**  \n処理額、純売上成長、承認率、不正損失率、EBITDA margin、AI/agentic commerce関連の取扱高。","quote_start":0,"quote_end":71,"text_sha256":"14b2254f25036d2d2baa1acd1b1923eca5e3ee16594f9a0abc5fe61a9aaf75ea","block_sha256":"14b2254f25036d2d2baa1acd1b1923eca5e3ee16594f9a0abc5fe61a9aaf75ea","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_e150bd1a-9589-474f-bdeb-e21cc4b62331"},{"id":"occ_583b9610c4f7f19abff451b1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_e2642383-ed81-497c-847d-f1bb20d37aa3","section_id":"sec_e2ff7022-99e5-4122-beb4-245e325dd769","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェント同士がAPI利用料、データ利用料、MCPサーバー利用料、ソフトウェア実行料を小口・即時・国境なしで払うなら、ステーブルコインは相性が良いです。","quote_start":0,"quote_end":79,"text_sha256":"0b3a3f22ce08248bf1a35cf2b58024953e83f2a91e7878d673bb30205bcd70c8","block_sha256":"0b3a3f22ce08248bf1a35cf2b58024953e83f2a91e7878d673bb30205bcd70c8","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_e2642383-ed81-497c-847d-f1bb20d37aa3"},{"id":"occ_51b6657de8c8348b7d7e8628","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_e3484396-6d7e-4592-9eca-16bc36007255","section_id":"sec_74084f1c-e42d-4438-8eff-1462431d50fc","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":23,"end":29,"exact":"エージェント","quote":"AI時代にかなり面白いのがここです。  \nAIエージェント同士がAPI利用料、データ利用料、MCPサーバー利用料、ソフトウェア実行料を支払うなら、既存カード決済よりも**即時・低コスト・24時間・国境なし**の決済が向いています。","quote_start":0,"quote_end":115,"text_sha256":"6636726e17ccf04a3b0b1f4f28d5b715138fbe7a082f83ac59c31863c0c4b0c9","block_sha256":"6636726e17ccf04a3b0b1f4f28d5b715138fbe7a082f83ac59c31863c0c4b0c9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_e3484396-6d7e-4592-9eca-16bc36007255"},{"id":"occ_fe8d7f3bba277228bbd56cc5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d","block_id":"blk_f9bdd5bd-059e-419a-b30b-07282e52176c","section_id":"sec_588df7aa-24a5-42d6-8fde-a9bdebdc8c93","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":32,"end":38,"exact":"エージェント","quote":"Circle、Coinbase、PayPal USD。  \nAIエージェント同士の小口決済、API課金、国際送金、24時間B2B決済と相性が良い。","quote_start":0,"quote_end":73,"text_sha256":"919218e17cf8db8ee1fbecd601b3a249d27b9b535f2d5ef8156674abb2cbe46c","block_sha256":"919218e17cf8db8ee1fbecd601b3a249d27b9b535f2d5ef8156674abb2cbe46c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e/#blk_f9bdd5bd-059e-419a-b30b-07282e52176c"},{"id":"occ_86b6c16436b1d9c6fe89c08d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_78657238-b40f-4a7c-b670-325df3688f39","work_id":"wrk_70fbeeef-5587-4e6a-a017-b7e323d0b279","block_id":"blk_8feca061-0b06-403f-b688-079ce0c295a4","section_id":"sec_fd7ab79c-b3c8-44f7-a1ec-3c07ae980bb5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":98,"end":104,"exact":"エージェント","quote":"ソルを、より広い帯域で流すための世代である。特に生成AIの推論では、長文コンテキスト、マルチモーダル入力、複数エージェント、長時間記憶のような処理が増えている。これらは演算能力だけでなく、メモリ帯域とメモリ容量を大量に消費する。","quote_start":43,"quote_end":157,"text_sha256":"4e23f2001e55cde0a0a06382984eea50edaae9c5f536851372277ad82f86bb73","block_sha256":"4e23f2001e55cde0a0a06382984eea50edaae9c5f536851372277ad82f86bb73","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_78657238-b40f-4a7c-b670-325df3688f39/#blk_8feca061-0b06-403f-b688-079ce0c295a4"},{"id":"occ_fbf7802aa52bdc06c1d51f19","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_7cf42fcc-1956-4694-a0b6-5eaa9be5c284","work_id":"wrk_cda7c85e-9714-493f-a66d-ac6f36454630","block_id":"blk_bd5ce54a-1145-46c1-9308-581c66cf5087","section_id":"sec_c975f62f-79e4-4c69-9d02-47b8c736541e","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":158,"end":165,"exact":"agentic","quote":"oduct data をAIで扱いやすくし、Windchill AI や Industrial AI 戦略で agentic/AI-driven workflows を前面に出しています。弱点は、MOM/MES や制御のレイヤーを Siemens/Rockwell/Schneider のようには広く持たないことです。したが","quote_start":103,"quote_end":265,"text_sha256":"593493cca1e30515d8c9406cab2a1a592a343ae930242c938daa8adcb9a22cc2","block_sha256":"593493cca1e30515d8c9406cab2a1a592a343ae930242c938daa8adcb9a22cc2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_7cf42fcc-1956-4694-a0b6-5eaa9be5c284/#blk_bd5ce54a-1145-46c1-9308-581c66cf5087"},{"id":"occ_3699df96bea6ded85d9d1399","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_7cf42fcc-1956-4694-a0b6-5eaa9be5c284","work_id":"wrk_cda7c85e-9714-493f-a66d-ac6f36454630","block_id":"blk_e3aab277-bfe0-4295-bfdc-e78ae8a6d0f6","section_id":"sec_8044e6ca-71fd-43b1-b39a-bf6f2d7d473d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":65,"end":71,"exact":"エージェント","quote":"  \nAmazonが取るべき層は、**AWSクラウド + Bedrock/AgentCore/Strandsのエージェント基盤 + 倉庫ロボット運用スタックの外販**です。Amazonは2025年に **AWS売上1287億ドル**、全社売上 **7169億ドル**で、しかも社内オペレーションでは **100万台超のロボ","quote_start":10,"quote_end":171,"text_sha256":"ca329b07eedc75905c6a93fe11d3b251300f088f4d5c38ad4a20834defddd249","block_sha256":"ca329b07eedc75905c6a93fe11d3b251300f088f4d5c38ad4a20834defddd249","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_7cf42fcc-1956-4694-a0b6-5eaa9be5c284/#blk_e3aab277-bfe0-4295-bfdc-e78ae8a6d0f6"},{"id":"occ_4db807a347470c4acc3702f5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_7cf42fcc-1956-4694-a0b6-5eaa9be5c284","work_id":"wrk_cda7c85e-9714-493f-a66d-ac6f36454630","block_id":"blk_f95aa1eb-ea21-4e81-bef8-0e78cec7da12","section_id":"sec_cea686b2-4976-4244-a041-51c3be1d9b16","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":417,"end":424,"exact":"agentic","quote":" で virtual と real をつなぎ、PTC は Windchill AI で PLMデータに対する agentic/AI-driven assistance を前面に出しています。Rockwellも予兆保全や設計支援を進めていますが、現時点では「AIを軸に世界観を作っている」度合いで SiemensやAVEVAほ","quote_start":362,"quote_end":524,"text_sha256":"703fc3dd0f6266d0412489a377504e92e18a696137231600c3a60803ef498550","block_sha256":"703fc3dd0f6266d0412489a377504e92e18a696137231600c3a60803ef498550","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_7cf42fcc-1956-4694-a0b6-5eaa9be5c284/#blk_f95aa1eb-ea21-4e81-bef8-0e78cec7da12"},{"id":"occ_3ae4c849a9ba9697899fb0d4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","work_id":"wrk_a3440338-a420-4877-947f-8193ae66700c","block_id":null,"section_id":null,"layer":"title","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"title","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントは株式市場をどう変えるのか","quote_start":0,"quote_end":21,"text_sha256":"c261270a37ac2565a016fc1f4550140459084d69213737f8e283d36b474444a6","block_sha256":null,"offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608/"},{"id":"occ_99a4da3135f267aca62b803e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","work_id":"wrk_a3440338-a420-4877-947f-8193ae66700c","block_id":"blk_05362c08-d883-415a-9e97-fe6fa6a41a4c","section_id":"sec_298a87a1-3d0a-4cef-930b-53338a1c7cec","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"AIとAIエージェントが普及した未来の株式市場は、こう変わる可能性があります。","quote_start":0,"quote_end":39,"text_sha256":"56f2a9961785e5ff031c4ac3272be72347bbcb81617b99fc1b51711f813e3300","block_sha256":"56f2a9961785e5ff031c4ac3272be72347bbcb81617b99fc1b51711f813e3300","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608/#blk_05362c08-d883-415a-9e97-fe6fa6a41a4c"},{"id":"occ_34bf3347ab98c3bb2041299c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","work_id":"wrk_a3440338-a420-4877-947f-8193ae66700c","block_id":"blk_07864d8d-167b-49ff-a39c-5cd2e1787952","section_id":"sec_efebd237-bd4a-47ff-8a37-a39f96d0c461","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"## AIエージェントは株式市場をどう変えるのか","quote_start":0,"quote_end":24,"text_sha256":"2d64e57060f23007b34dcafd367dd203a3b09a9b9c33da87d716219a0a4b1872","block_sha256":"2d64e57060f23007b34dcafd367dd203a3b09a9b9c33da87d716219a0a4b1872","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608/#blk_07864d8d-167b-49ff-a39c-5cd2e1787952"},{"id":"occ_801c18ed69ded4de17c9cd98","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","work_id":"wrk_a3440338-a420-4877-947f-8193ae66700c","block_id":"blk_1f8e8924-59e3-4636-a36e-e6e2e1cb3a68","section_id":"sec_566511ae-ff82-4740-803f-7b5919406934","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":38,"end":44,"exact":"エージェント","quote":"**計算文明圏**は、AIを強くするもの、AIによってさらに伸びるもの、AIエージェントが評価しやすいものです。ここでは、成長率、粗利率、スケーラビリティ、データ量、ネットワーク効果、供給制約、国家安全保障上の重要性が評価されます。","quote_start":0,"quote_end":116,"text_sha256":"ca0e61c5d9b8679753743fbf5db0ae391e4af514103fcda5ad9a6dd7cd9ad10c","block_sha256":"ca0e61c5d9b8679753743fbf5db0ae391e4af514103fcda5ad9a6dd7cd9ad10c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608/#blk_1f8e8924-59e3-4636-a36e-e6e2e1cb3a68"},{"id":"occ_1c038ca01dc8e8057d1e3388","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","work_id":"wrk_a3440338-a420-4877-947f-8193ae66700c","block_id":"blk_21aab828-cdb5-4db7-8c3c-a20ee3160211","section_id":"sec_298a87a1-3d0a-4cef-930b-53338a1c7cec","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":17,"end":23,"exact":"エージェント","quote":"**市場参加者は、人間中心からAIエージェント中心へ移る。  \n短期取引は、AI同士の読み合いになる。  \n計算文明圏の企業には資本が集中する。  \n上場市場は、スケーリングする巨大企業と安定インフラ企業に偏る。  \nスケールしないが社会に必要な","quote_start":0,"quote_end":123,"text_sha256":"9b183d47cfd0256a3e46da10929be450fd5c73ae5fef6d0480a0865db8de3ec1","block_sha256":"9b183d47cfd0256a3e46da10929be450fd5c73ae5fef6d0480a0865db8de3ec1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608/#blk_21aab828-cdb5-4db7-8c3c-a20ee3160211"},{"id":"occ_a15bba190308a7669552329d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","work_id":"wrk_a3440338-a420-4877-947f-8193ae66700c","block_id":"blk_3b63b084-c351-4bef-9e90-91eb5bac95b4","section_id":"sec_74ad8966-bc82-490a-b673-ea77051b79bf","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":36,"end":42,"exact":"エージェント","quote":"この反射は市場を動かします。AIが半導体を重要視し、人間もAI投信もAIエージェントも同じ方向を見ると、資本は計算文明圏へ流れます。その資本が半導体工場、データセンター、電力網に投下され、さらにAIの性能を上げます。すると、さらにAIがAIインフラの重要性を語る。ここに、**AI資本","quote_start":0,"quote_end":142,"text_sha256":"00d667eeb62fde3131179cf4450258cbb9e5e557ea2b9f7a727f5cb8bcc39e9c","block_sha256":"00d667eeb62fde3131179cf4450258cbb9e5e557ea2b9f7a727f5cb8bcc39e9c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608/#blk_3b63b084-c351-4bef-9e90-91eb5bac95b4"},{"id":"occ_441507bdaf3f77200c53efd0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","work_id":"wrk_a3440338-a420-4877-947f-8193ae66700c","block_id":"blk_41d4dc2d-b653-4df8-a508-ddd0ebb7fb78","section_id":"sec_566511ae-ff82-4740-803f-7b5919406934","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":52,"end":58,"exact":"エージェント","quote":"この二つは、どちらが「正しい」という話ではありません。問題は、資本市場がどちらを高く評価するかです。AIエージェントは、どうしてもスケールしやすいもの、数値化しやすいもの、機械で比較しやすいものを高く評価しやすい。すると、生活文明圏の一部は、上場市場から離れ、公共・準公共・非上場・地域事業へ移っていく可能性がありま","quote_start":0,"quote_end":158,"text_sha256":"78b272a6ff9dcfa2ec4f13ab75943021f19fc2221548704399ba4cab621e8ec6","block_sha256":"78b272a6ff9dcfa2ec4f13ab75943021f19fc2221548704399ba4cab621e8ec6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608/#blk_41d4dc2d-b653-4df8-a508-ddd0ebb7fb78"},{"id":"occ_7cd41fc4a7319107466b5a8a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","work_id":"wrk_a3440338-a420-4877-947f-8193ae66700c","block_id":"blk_4498b82c-4036-4bb2-ae72-e180ca9c3aef","section_id":"sec_320b9558-5420-4ec1-bd70-52dd105ae2bb","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"AIとAIエージェントが普及したとき、株式市場は単に「AIを使って銘柄分析が便利になる」程度では終わらない可能性があります。より大きな変化は、**株式市場そのものが、AIの学習・推論・自動化を支える計算文明へ資本を配分する","quote_start":0,"quote_end":111,"text_sha256":"f770bb5c9b5f38ef251a70591bb6fcae63bf6df025a5da7d8f9f77d4a530e24a","block_sha256":"f770bb5c9b5f38ef251a70591bb6fcae63bf6df025a5da7d8f9f77d4a530e24a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608/#blk_4498b82c-4036-4bb2-ae72-e180ca9c3aef"},{"id":"occ_4143656a9867b1449ddd9a00","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","work_id":"wrk_a3440338-a420-4877-947f-8193ae66700c","block_id":"blk_5efbdb57-839f-453d-a7ec-beee640ac14c","section_id":"sec_23a6a2bd-4eef-4066-a65a-7196b6d9ef94","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":39,"end":45,"exact":"エージェント","quote":"使い方次第で結果は変わり、投資家はそれぞれの使い方で自己判断するでしょう、AIエージェントに丸投げする層が増え出した頃に、賢く間違えるといった事例に遭遇するかも知れないですが、それは一部で一流の層はそれを克服する術を身につけているでしょう","quote_start":0,"quote_end":119,"text_sha256":"2c36fd95b4de795268d98b9ee6b1e851ea86e92ce092711af3c3a3d195df3da7","block_sha256":"2c36fd95b4de795268d98b9ee6b1e851ea86e92ce092711af3c3a3d195df3da7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608/#blk_5efbdb57-839f-453d-a7ec-beee640ac14c"},{"id":"occ_48413337f770deb56a75bde8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","work_id":"wrk_a3440338-a420-4877-947f-8193ae66700c","block_id":"blk_6ede1aa5-54f5-4f34-a18a-87cfe046c0d6","section_id":"sec_c98b8801-b190-458c-833a-671815bfabe7","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":100,"end":106,"exact":"エージェント","quote":"央銀行などの判断が重なって動いてきました。もちろん、すでにアルゴリズム取引や高頻度取引は存在します。しかしAIエージェントが普及すると、より広い層の投資家が、常時稼働する分析・執行システムを持つようになります。","quote_start":45,"quote_end":150,"text_sha256":"ecb1d3f8a16fdf9b91bd467a69ade0d801cde7b795af0c7067fac73b4bd3b8ed","block_sha256":"ecb1d3f8a16fdf9b91bd467a69ade0d801cde7b795af0c7067fac73b4bd3b8ed","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608/#blk_6ede1aa5-54f5-4f34-a18a-87cfe046c0d6"},{"id":"occ_da380f1fafc4281b32c569b5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","work_id":"wrk_a3440338-a420-4877-947f-8193ae66700c","block_id":"blk_87a66617-54b5-4aed-9652-3e4731767862","section_id":"sec_d7494fa8-1d7d-43dd-bd88-2bde63515a1b","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":117,"end":123,"exact":"エージェント","quote":"整」なのかを分類する。半導体株が上がりすぎたとき、それが「利益成長に裏付けられた上昇」なのか、「AI投信とAIエージェントの混み合った買い」なのかを考える。","quote_start":62,"quote_end":140,"text_sha256":"374f2750ca292da9961dc60db04768fb4568bf8fd25bcb661ca226d9a0be3978","block_sha256":"374f2750ca292da9961dc60db04768fb4568bf8fd25bcb661ca226d9a0be3978","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608/#blk_87a66617-54b5-4aed-9652-3e4731767862"},{"id":"occ_d31bb2de70570c74c5491087","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","work_id":"wrk_a3440338-a420-4877-947f-8193ae66700c","block_id":"blk_b4d301a3-4982-4075-95c9-bccbbf3b34ba","section_id":"sec_c98b8801-b190-458c-833a-671815bfabe7","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":27,"end":33,"exact":"エージェント","quote":"## \\1. 株式市場は「人間の判断の集合」から「AIエージェント群の相互作用」へ","quote_start":0,"quote_end":41,"text_sha256":"e1d3e430daa132363d4da42e8e65bb69ff99ef47dc27f7b1ff30674de3e7d035","block_sha256":"e1d3e430daa132363d4da42e8e65bb69ff99ef47dc27f7b1ff30674de3e7d035","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608/#blk_b4d301a3-4982-4075-95c9-bccbbf3b34ba"},{"id":"occ_43aecb1dbd040d9d255cec4f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","work_id":"wrk_a3440338-a420-4877-947f-8193ae66700c","block_id":"blk_e3a404e5-f5ca-4f2a-927f-c40f552f4a8f","section_id":"sec_bbf3febd-5588-41f4-a3bb-8af7d2d248fb","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"AIとAIエージェントが資本配分を担うほど、上場市場は経済全体を均等に映す場所ではなく、**スケーリングするものに資本を集める場所**へ傾いていく可能性があります。","quote_start":0,"quote_end":82,"text_sha256":"9fc93b294d4a1f7265ed9ad936953adbd0ec21a080b24133fc2200d2fe376bae","block_sha256":"9fc93b294d4a1f7265ed9ad936953adbd0ec21a080b24133fc2200d2fe376bae","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608/#blk_e3a404e5-f5ca-4f2a-927f-c40f552f4a8f"},{"id":"occ_4e9f36bb7940da7a5899ffe6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","work_id":"wrk_a3440338-a420-4877-947f-8193ae66700c","block_id":"blk_fbb31f30-fb27-4732-adf9-b05ee4575153","section_id":"sec_298a87a1-3d0a-4cef-930b-53338a1c7cec","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":90,"end":96,"exact":"エージェント","quote":"置図になります。半導体は神経細胞、電力は血液、データセンターは身体、クラウドは作業場、セキュリティは免疫、AIエージェントは意志決定の群れです。","quote_start":35,"quote_end":107,"text_sha256":"4c63887c99e295dc4f518864ec4fb6980881df4892cffcfcef505147c337ee0e","block_sha256":"4c63887c99e295dc4f518864ec4fb6980881df4892cffcfcef505147c337ee0e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608/#blk_fbb31f30-fb27-4732-adf9-b05ee4575153"},{"id":"occ_16a400cd1a07f8bf11377903","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","work_id":"wrk_a3440338-a420-4877-947f-8193ae66700c","block_id":"blk_fea338bb-aeb2-4ce7-8f1c-d592531d237d","section_id":"sec_c98b8801-b190-458c-833a-671815bfabe7","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントは、決算資料、ニュース、カンファレンスコール、SNS、求人情報、サプライチェーン、金利、為替、オプション需給、ETFの資金流入を読み、ポートフォリオを調整します。人間が「この株を買うべきか」と考える前","quote_start":0,"quote_end":108,"text_sha256":"9aea21905e6513d617a0a94d7b5cbc94b3832faf0a1fcd422f1def0e22148da3","block_sha256":"9aea21905e6513d617a0a94d7b5cbc94b3832faf0a1fcd422f1def0e22148da3","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608/#blk_fea338bb-aeb2-4ce7-8f1c-d592531d237d"},{"id":"occ_d22a93fe249d3f300919ee71","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa","work_id":"wrk_82866dd9-73ac-495c-82f9-5873005352f3","block_id":"blk_0affe7e4-625e-4603-b74b-8531b80dd810","section_id":"sec_6a39c866-53af-45cf-9774-274a104b1234","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":178,"end":184,"exact":"エージェント","quote":"ジョブ\n- 調査ケースと監査履歴\n- Elastic Agentの配布設定\n- 外部システムとの連携\n- AIエージェント向けのツールとワークフロー","quote_start":123,"quote_end":197,"text_sha256":"d61141babcc91f8442e61d897f62ab6605f6d7f9458829633a5f8d9058bb05b9","block_sha256":"d61141babcc91f8442e61d897f62ab6605f6d7f9458829633a5f8d9058bb05b9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa/#blk_0affe7e4-625e-4603-b74b-8531b80dd810"},{"id":"occ_e4ae926793e32630f907f6e6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa","work_id":"wrk_82866dd9-73ac-495c-82f9-5873005352f3","block_id":"blk_0c58d023-d49f-47b9-b4ce-919b79ac890c","section_id":"sec_d4fd9d0a-c0a7-4213-87f5-be1fe0dddceb","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":49,"end":55,"exact":"エージェント","quote":"- プロンプト\n- モデルの回答\n- トークン使用量\n- ツール呼び出し\n- RAG検索履歴\n- エージェントの推論経路\n- APIの応答時間\n- セキュリティイベント\n- AIが実行した業務操作\n- 人間による承認履歴","quote_start":0,"quote_end":110,"text_sha256":"2322436cfd4aee636df9e1a4d6e68f8d7ccc98ed653188d2a4508f6c918b8afb","block_sha256":"2322436cfd4aee636df9e1a4d6e68f8d7ccc98ed653188d2a4508f6c918b8afb","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa/#blk_0c58d023-d49f-47b9-b4ce-919b79ac890c"},{"id":"occ_87ce44186993797747290f8f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa","work_id":"wrk_82866dd9-73ac-495c-82f9-5873005352f3","block_id":"blk_1dc46d8c-3a1c-41a0-9de6-1ca1ad1bedf1","section_id":"sec_9cc5f324-af16-45ee-81bb-14e115627972","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":17,"end":23,"exact":"エージェント","quote":"### コンテキストを理解するAIエージェント","quote_start":0,"quote_end":23,"text_sha256":"960bc9fce0a505b1245b4f39ebd4ca2a9d659f694a02a3a5315d022462d0c03f","block_sha256":"960bc9fce0a505b1245b4f39ebd4ca2a9d659f694a02a3a5315d022462d0c03f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa/#blk_1dc46d8c-3a1c-41a0-9de6-1ca1ad1bedf1"},{"id":"occ_0cdc991b6944d508e9b3f0fc","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa","work_id":"wrk_82866dd9-73ac-495c-82f9-5873005352f3","block_id":"blk_42af7bdf-6c99-4cc0-8ec4-1cb0a0ca9963","section_id":"sec_d4fd9d0a-c0a7-4213-87f5-be1fe0dddceb","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":14,"end":20,"exact":"エージェント","quote":"一人の人間が一回質問しても、エージェントは複数回の検索、集計、外部ツール呼び出しを行う。","quote_start":0,"quote_end":44,"text_sha256":"fed85191d83a361fe5afa07b7da49a4450be24f4755b805f394cb2482d4363b6","block_sha256":"fed85191d83a361fe5afa07b7da49a4450be24f4755b805f394cb2482d4363b6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa/#blk_42af7bdf-6c99-4cc0-8ec4-1cb0a0ca9963"},{"id":"occ_9a6683fc7e24ae4d5ed730df","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa","work_id":"wrk_82866dd9-73ac-495c-82f9-5873005352f3","block_id":"blk_4f803dff-e680-49f2-b0ac-92135d3ad39d","section_id":"sec_7c83285b-f957-4fef-b635-aa5450947ba8","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントがどのモデル、検索、ツールを使い、どこで判断を誤ったかを追跡する。","quote_start":0,"quote_end":41,"text_sha256":"c243a8c6a2575a32e036d8b1016b48eea298e781ee4e9b60f7876ead4027019e","block_sha256":"c243a8c6a2575a32e036d8b1016b48eea298e781ee4e9b60f7876ead4027019e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa/#blk_4f803dff-e680-49f2-b0ac-92135d3ad39d"},{"id":"occ_f15e4ef5cc637c21d0da495c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa","work_id":"wrk_82866dd9-73ac-495c-82f9-5873005352f3","block_id":"blk_52e952c4-c3f1-4a54-8492-c3379e3da234","section_id":"sec_d4fd9d0a-c0a7-4213-87f5-be1fe0dddceb","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントが企業内で動くと、従来以上に大量のデータが発生する。","quote_start":0,"quote_end":34,"text_sha256":"dfc645dbdfef82b33aca171daca05a261d8fe45c90ac1c5cb8b9269c623e3e77","block_sha256":"dfc645dbdfef82b33aca171daca05a261d8fe45c90ac1c5cb8b9269c623e3e77","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa/#blk_52e952c4-c3f1-4a54-8492-c3379e3da234"},{"id":"occ_1b0dd79fbb6965e7919f1b95","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa","work_id":"wrk_82866dd9-73ac-495c-82f9-5873005352f3","block_id":"blk_6d0e0539-f77f-4933-84a1-fee19213c61a","section_id":"sec_d4fd9d0a-c0a7-4213-87f5-be1fe0dddceb","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":7,"end":13,"exact":"エージェント","quote":"したがってAIエージェントが普及すると、人間の利用者数以上に検索・ログ・監視の処理量が増える可能性がある。","quote_start":0,"quote_end":53,"text_sha256":"eb66a7cbed04b40be5648bc94e0d60fa01511e14c5c6817642e39f32564406e7","block_sha256":"eb66a7cbed04b40be5648bc94e0d60fa01511e14c5c6817642e39f32564406e7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa/#blk_6d0e0539-f77f-4933-84a1-fee19213c61a"},{"id":"occ_f657bde02da6e0de0595dafa","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa","work_id":"wrk_82866dd9-73ac-495c-82f9-5873005352f3","block_id":"blk_9d96fc96-0df3-4cc3-a8e6-cab209857b98","section_id":"sec_e6277120-e6ea-4ecd-b143-fb0e0f98e3ee","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":31,"end":37,"exact":"エージェント","quote":"企業文書、商品情報、顧客履歴、ログなどを検索し、生成AIやAIエージェントへ必要な情報を渡す。","quote_start":0,"quote_end":47,"text_sha256":"cb2bf7eafb6689df4df2294672a867d2287c568c84444e7e4bb469a1e3883da8","block_sha256":"cb2bf7eafb6689df4df2294672a867d2287c568c84444e7e4bb469a1e3883da8","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa/#blk_9d96fc96-0df3-4cc3-a8e6-cab209857b98"},{"id":"occ_c13fa2d5f34257ec336db5a9","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa","work_id":"wrk_82866dd9-73ac-495c-82f9-5873005352f3","block_id":"blk_bb66d45c-065f-417c-b7c3-c9ca38471b71","section_id":"sec_c6ca078a-4cfd-47b2-ac4f-78992a019310","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":4,"end":11,"exact":"Agentic","quote":"### Agentic Observability","quote_start":0,"quote_end":25,"text_sha256":"f005ebbb385b4b36364f13aa0e8bb4857b67840ace0af9760523c69038357a87","block_sha256":"f005ebbb385b4b36364f13aa0e8bb4857b67840ace0af9760523c69038357a87","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa/#blk_bb66d45c-065f-417c-b7c3-c9ca38471b71"},{"id":"occ_b97a045e67a8d7590e0fa57e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa","work_id":"wrk_82866dd9-73ac-495c-82f9-5873005352f3","block_id":"blk_e763ba16-cd59-4324-ab7d-6dfebe364229","section_id":"sec_e3058655-c893-45df-8f24-6ca4dd4c3e88","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":11,"end":17,"exact":"エージェント","quote":"Elasticは、AIエージェントが必要な情報を、","quote_start":0,"quote_end":25,"text_sha256":"20a5557cfdb3cb25122f175989d24e196e3f5c32e314e95e1223258231175d7a","block_sha256":"20a5557cfdb3cb25122f175989d24e196e3f5c32e314e95e1223258231175d7a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa/#blk_e763ba16-cd59-4324-ab7d-6dfebe364229"},{"id":"occ_d084accc3b64123cd73d8645","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa","work_id":"wrk_82866dd9-73ac-495c-82f9-5873005352f3","block_id":"blk_f848f920-fd3c-4a9c-b85d-281485f8719a","section_id":"sec_5d43cba4-d70f-4c52-b97d-e641b548273a","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":4,"end":11,"exact":"Agentic","quote":"### Agentic Security Operations","quote_start":0,"quote_end":31,"text_sha256":"10d5873b3b08cf2d375f26a6e87acc258725a6add0b4a850da71f189e1e82115","block_sha256":"10d5873b3b08cf2d375f26a6e87acc258725a6add0b4a850da71f189e1e82115","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa/#blk_f848f920-fd3c-4a9c-b85d-281485f8719a"},{"id":"occ_3c90e7022a58316dd2b11cc6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_03686d6f-6e26-4453-a2c0-0cd2d5e83e49","section_id":"sec_40fb3dd6-b799-4c6a-ac84-afc82a698360","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":27,"end":33,"exact":"エージェント","quote":"- 機密文書の検索\n- 中規模モデルの推論\n- コードエージェント\n- 画像・動画処理\n- AIエージェントの開発\n- クラウドへ送る前の検証\n- オフライン時のAI\n- ローカルな個人知識ベース","quote_start":0,"quote_end":98,"text_sha256":"09548cbb0303e7af0dce7b71d15728eea995b30b7f6ffc3f80a05e63727e778c","block_sha256":"09548cbb0303e7af0dce7b71d15728eea995b30b7f6ffc3f80a05e63727e778c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_03686d6f-6e26-4453-a2c0-0cd2d5e83e49"},{"id":"occ_6bd2233a8c3160198eb6fea6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_146f6eda-a2f6-4ff5-ab5b-e93c1960b62b","section_id":"sec_5aa34a3a-809d-4696-b7da-33ee6f756961","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":9,"end":15,"exact":"エージェント","quote":"数時間から数日動くエージェントには、過去の指示、失敗、コード変更、検索結果、判断理由を保持する能力が必要になる。Moonshotにとって長文脈は、エージェント化の前提である。","quote_start":0,"quote_end":87,"text_sha256":"99918d4f5ac7527c5b6be87a290021559213efffb4dab8ea6bea11140cb5cc18","block_sha256":"99918d4f5ac7527c5b6be87a290021559213efffb4dab8ea6bea11140cb5cc18","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_146f6eda-a2f6-4ff5-ab5b-e93c1960b62b"},{"id":"occ_af325729afc27bb13f9c38ab","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_1a90a880-c3e8-44ce-a02b-508c6d1b4929","section_id":"sec_5e45a914-558c-4006-ae00-490ef6e214f1","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":11,"end":17,"exact":"エージェント","quote":"しかし、数十億人がAIエージェントを使う世界で、全員が物理端末を二台持つとは考えにくい。","quote_start":0,"quote_end":44,"text_sha256":"64f843af9ec9a9b3007460e2080ac46060e0cc853a64ede8ef6f25594f21d74c","block_sha256":"64f843af9ec9a9b3007460e2080ac46060e0cc853a64ede8ef6f25594f21d74c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_1a90a880-c3e8-44ce-a02b-508c6d1b4929"},{"id":"occ_107db8af0b93fa86f27d6257","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_236af800-526a-4c2e-8c9b-9982cef5ed62","section_id":"sec_7ac2c2b4-b7a5-4620-9a55-155e8644a627","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"- どのAIエージェントが動いているか確認する\n- タスクに必要な権限だけを発行する\n- 送信データに機密情報が含まれていないか検査する\n- プロンプトインジェクションを検出する\n- 通常と異なるアプリ横断操作を停止する\n-","quote_start":0,"quote_end":112,"text_sha256":"4338a5ec1335fe7a672901bda7c562f1a7dfbeb4f0132224147fa1ea5de84053","block_sha256":"4338a5ec1335fe7a672901bda7c562f1a7dfbeb4f0132224147fa1ea5de84053","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_236af800-526a-4c2e-8c9b-9982cef5ed62"},{"id":"occ_228b62ae8e46f9e1ff080a42","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_2eced014-0b10-4fc1-9b47-bc3833f3123a","section_id":"sec_5c407ce3-8a12-4b60-9e81-089895fcffbd","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"> **AIエージェントの認証、権限、隔離、監査、巻き戻しを標準化し、人間がAIの端末操作を恐れずに済む仕組みを構築した企業**","quote_start":0,"quote_end":64,"text_sha256":"ad513443fc4f0e7d156004d70efb0a6ba02c23401c5749566d1ac4070764e726","block_sha256":"ad513443fc4f0e7d156004d70efb0a6ba02c23401c5749566d1ac4070764e726","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_2eced014-0b10-4fc1-9b47-bc3833f3123a"},{"id":"occ_cb0051811fe2efc7c45e7c9f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_3631fc97-26aa-4c1e-a3d4-95b003f5ea09","section_id":"sec_e35744be-e3b6-4e04-b50a-c32860213617","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":20,"end":26,"exact":"エージェント","quote":"一方、OpenClaw型と呼べる端末操作エージェントは、","quote_start":0,"quote_end":28,"text_sha256":"858c5d6f79cfd2be66aba302253ceb1aa59d582e848da09ef8fd4f8045c3dbb4","block_sha256":"858c5d6f79cfd2be66aba302253ceb1aa59d582e848da09ef8fd4f8045c3dbb4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_3631fc97-26aa-4c1e-a3d4-95b003f5ea09"},{"id":"occ_47e3dc4f6f76399978d1e81f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_36bb20ae-e1d6-44e0-ac8e-1ae3b3b7af0d","section_id":"sec_4e550c47-9a87-4ad1-80c5-1228b758330e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":61,"end":67,"exact":"エージェント","quote":"には強い拒否機構が付く。一方、信頼された企業は、自社データ、コード実行、ブラウザ操作、クラウド資格情報、長時間エージェントを組み合わせられる。","quote_start":6,"quote_end":77,"text_sha256":"d67093dc4bc6f6cccb49d288f80b6187b46357ebcaa45a6bed5734ceb7920461","block_sha256":"d67093dc4bc6f6cccb49d288f80b6187b46357ebcaa45a6bed5734ceb7920461","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_36bb20ae-e1d6-44e0-ac8e-1ae3b3b7af0d"},{"id":"occ_9e87d5cc99fcf0f104670209","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_3ba52fd7-015e-4d0b-843c-765e65742e48","section_id":"sec_4e550c47-9a87-4ad1-80c5-1228b758330e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":86,"end":92,"exact":"エージェント","quote":"・生物・軍事能力評価\n- 高権限ツールを使うモデルの公開延期\n- オープンウェイト公開基準の厳格化\n- 長時間エージェントのネット接続制限\n- クラウド利用者の本人確認\n- GPU利用量と大規模訓練の報告義務\n- 高リスク用途の監査ログ保存","quote_start":31,"quote_end":151,"text_sha256":"c79cb03cad34d76f7700420d1970146eb8b64e57a5a4f787729b3bbfbaafbc89","block_sha256":"c79cb03cad34d76f7700420d1970146eb8b64e57a5a4f787729b3bbfbaafbc89","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_3ba52fd7-015e-4d0b-843c-765e65742e48"},{"id":"occ_9b20bbe4448897105ab227fe","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_3f888670-7490-476c-800b-b98e5633cb4a","section_id":"sec_a151983f-21b7-4159-bcb1-b66042c2e4d5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":42,"end":48,"exact":"エージェント","quote":"- コンテキストを長くする\n- 画像や動画を処理する\n- 数時間考えさせる\n- 複数エージェントを動かす\n- より難しい仕事を任せる\n- 夜間も継続的に作業させる","quote_start":0,"quote_end":81,"text_sha256":"3f54eae041c633218e3cb874086d68920b7f1847b0650a9287ece538a085eb5c","block_sha256":"3f54eae041c633218e3cb874086d68920b7f1847b0650a9287ece538a085eb5c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_3f888670-7490-476c-800b-b98e5633cb4a"},{"id":"occ_438abc6d90fb5249e8ca1a39","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_4d6cc624-50b6-4dba-a8ae-ed1efbe9ee6e","section_id":"sec_9bb2d87e-9af3-4db5-8ec8-2c060796747c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":7,"end":13,"exact":"エージェント","quote":"## 第三に、エージェント数が増える","quote_start":0,"quote_end":18,"text_sha256":"a14fe67f73c4ea0aea092505439e5787ef6f33af3933d52ad6fb722c62f22acb","block_sha256":"a14fe67f73c4ea0aea092505439e5787ef6f33af3933d52ad6fb722c62f22acb","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_4d6cc624-50b6-4dba-a8ae-ed1efbe9ee6e"},{"id":"occ_b75b66c35935fccbbff5b286","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_4eb0d16d-0558-4543-a09f-bf28fb080c6b","section_id":"sec_a151983f-21b7-4159-bcb1-b66042c2e4d5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":35,"end":41,"exact":"エージェント","quote":"> 10倍効率化したので、10倍長く考え、10倍の資料を読み、10体のエージェントを動かす","quote_start":0,"quote_end":45,"text_sha256":"84059b25aaacee1ba52b464c41b5e862ef56dc028b47ffcb6d91e484d6d025d2","block_sha256":"84059b25aaacee1ba52b464c41b5e862ef56dc028b47ffcb6d91e484d6d025d2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_4eb0d16d-0558-4543-a09f-bf28fb080c6b"},{"id":"occ_00055aa2719762edc150d35a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_509e269f-0d7e-412a-a0da-6a3b52326737","section_id":"sec_a74ba5f2-eed9-40db-bbbf-22fb0280f498","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":28,"end":34,"exact":"エージェント","quote":"Moonshotの講演では、将来的に100～1000体のエージェントを並列稼働させ、数百・数千の情報源を読み、巨大な調査や開発を短時間で終える構想が語られている。","quote_start":0,"quote_end":81,"text_sha256":"1304a9ed4f212ccc205c16344147b698851e3bf48d21a8732cfd41ab1f65e473","block_sha256":"1304a9ed4f212ccc205c16344147b698851e3bf48d21a8732cfd41ab1f65e473","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_509e269f-0d7e-412a-a0da-6a3b52326737"},{"id":"occ_5b87b264daab123e0e1d1e5e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_57ffb845-aff9-44f4-8e6c-90f2b383db75","section_id":"sec_a74ba5f2-eed9-40db-bbbf-22fb0280f498","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":42,"end":48,"exact":"エージェント","quote":"一体のモデルが調査、計算、コード、検証、文章作成を順番に行うのではなく、司令塔となるエージェントが複数のサブエージェントへ仕事を割り振る。","quote_start":0,"quote_end":69,"text_sha256":"58c70f1fbeaa0879d68f157c8f1be9e4175b917080159430142a86d5a02534a2","block_sha256":"58c70f1fbeaa0879d68f157c8f1be9e4175b917080159430142a86d5a02534a2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_57ffb845-aff9-44f4-8e6c-90f2b383db75"},{"id":"occ_31474963b2835ec55e7829dd","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_61448e3e-a79b-40d7-96cd-28f7a0048d76","section_id":"sec_a309d9d4-f1c8-445b-826b-714b90abe20a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":24,"end":30,"exact":"エージェント","quote":"- フロンティアモデル\n- 長時間推論\n- 複数エージェント\n- アプリAPIの操作\n- コード実行\n- 情報収集\n- 成果物の作成\n- 夜間の継続作業","quote_start":0,"quote_end":76,"text_sha256":"26905c2d1ab23b574c22363112a154a0ae52e210e908605d4ccba017da97fb11","block_sha256":"26905c2d1ab23b574c22363112a154a0ae52e210e908605d4ccba017da97fb11","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_61448e3e-a79b-40d7-96cd-28f7a0048d76"},{"id":"occ_4216ce70a66feed283b1a836","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_6b66d288-3d89-4738-b4d1-5bd51a74c958","section_id":"sec_f0e23c26-9a59-4722-97b5-a71cb6ea1e80","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":26,"end":32,"exact":"エージェント","quote":"100万トークンのコンテキストを持ち、コード、長時間エージェント、知識作業、推論、画像理解を統合する。Moonshotは、K2に比べて総合的なスケーリング効率を約2.5倍改善したと報告している。","quote_start":0,"quote_end":97,"text_sha256":"0baa8f82df5777d587a39b50fd42e84ed053ce69a56aac744ef4eaef04d5f691","block_sha256":"0baa8f82df5777d587a39b50fd42e84ed053ce69a56aac744ef4eaef04d5f691","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_6b66d288-3d89-4738-b4d1-5bd51a74c958"},{"id":"occ_f482f22ed4782c028827d8ad","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_7a50aa7f-0ea4-432e-a6c0-bf1e431d1128","section_id":"sec_9bb2d87e-9af3-4db5-8ec8-2c060796747c","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":3,"end":9,"exact":"エージェント","quote":"一体のエージェントを効率化すると、その余力で10体、100体のサブエージェントを動かすようになる。","quote_start":0,"quote_end":49,"text_sha256":"487d8f1d97682eae946584f57a8c2d4d4deda3b7f6fb9efad2aced2c0c76e4ba","block_sha256":"487d8f1d97682eae946584f57a8c2d4d4deda3b7f6fb9efad2aced2c0c76e4ba","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_7a50aa7f-0ea4-432e-a6c0-bf1e431d1128"},{"id":"occ_efe114a7955ab0c9897b371c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_8529a851-4b97-4811-9b94-c4f8fa0278f1","section_id":"sec_4b07cd07-7efa-4a0d-97a1-f35780403da3","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":108,"end":114,"exact":"エージェント","quote":" 2万 | 約27ドル |\n| 活発な作業AI | 100万 | 20万 | 約180ドル |\n| 重い長時間エージェント | 1000万 | 200万 | 約1800ドル |\n```","quote_start":53,"quote_end":145,"text_sha256":"91e2c47a3fb5312cb65d637a14cdcd0414a1c830c43ba306018f076ef60b0c0d","block_sha256":"91e2c47a3fb5312cb65d637a14cdcd0414a1c830c43ba306018f076ef60b0c0d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_8529a851-4b97-4811-9b94-c4f8fa0278f1"},{"id":"occ_eaf1662476d0965a69ebd630","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_b2432586-2231-469b-ab79-03163324c4d6","section_id":"sec_5e45a914-558c-4006-ae00-490ef6e214f1","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":13,"end":19,"exact":"エージェント","quote":"現状でOpenClaw型のエージェントを試すなら、","quote_start":0,"quote_end":25,"text_sha256":"9516463518f786c836176a28c3815c15c931a4324cd3a80542917f0db1f5d58a","block_sha256":"9516463518f786c836176a28c3815c15c931a4324cd3a80542917f0db1f5d58a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_b2432586-2231-469b-ab79-03163324c4d6"},{"id":"occ_7a9b38fc617754cbb0cb2c04","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_bc95baf0-48e6-409e-bc80-7598b0061fbb","section_id":"sec_ba4ad92f-737c-4325-84f8-2e9006f059a0","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":23,"end":29,"exact":"エージェント","quote":"Codexアプリは、複数の作業を並列に進め、各エージェントの状態を管理し、変更内容を確認し、人間が必要な場面で承認や修正を行う構造を持つ。エージェントは作業フォルダやブランチ単位で分離され、ユーザーは結果と差分を確認できる。","quote_start":0,"quote_end":112,"text_sha256":"94508ea37e82cea3439d0b6baf0f1b1dabdc7519350b1a6313095b98ab3a7f20","block_sha256":"94508ea37e82cea3439d0b6baf0f1b1dabdc7519350b1a6313095b98ab3a7f20","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_bc95baf0-48e6-409e-bc80-7598b0061fbb"},{"id":"occ_9d68d5bab75b0bcc3ceba988","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_be7bb865-cd0b-410c-b5c6-1dad5f3d9a2c","section_id":"sec_f1ef0a10-4db3-4c8a-b569-4bbc8e36ae09","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":59,"end":65,"exact":"エージェント","quote":"ansformer-XL  \n→長文脈Kimi  \n→長い推論過程  \n→Kimi Linear  \n→長時間エージェント  \n→Agent Swarm**","quote_start":4,"quote_end":82,"text_sha256":"e304ed2ea38f0a09b5f8b7a25e9ed7435d88b0fefef02623d8723b0765770c2f","block_sha256":"e304ed2ea38f0a09b5f8b7a25e9ed7435d88b0fefef02623d8723b0765770c2f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_be7bb865-cd0b-410c-b5c6-1dad5f3d9a2c"},{"id":"occ_5ba9f054d03721a4e27ba768","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_c673d095-0499-4883-a39b-16326a703276","section_id":"sec_c38c7e4f-ae0e-4045-a55a-7c566d49a022","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":23,"end":29,"exact":"エージェント","quote":"これは単発のコード生成ではない。長時間稼働するエージェントが、現実のインフラ上で探索・計画・実行・修正を繰り返した結果である。","quote_start":0,"quote_end":63,"text_sha256":"cfc1d54cd914d39489f7849af5bb8c1ee856eb1af4ebed8149b11a4ff7530138","block_sha256":"cfc1d54cd914d39489f7849af5bb8c1ee856eb1af4ebed8149b11a4ff7530138","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_c673d095-0499-4883-a39b-16326a703276"},{"id":"occ_0dfed5863d4ecc98ccb1fc57","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_c8866630-6a74-49cd-b251-65174111d9b1","section_id":"sec_11b52d2f-7f05-4c3e-a3e3-13ca19d2a1a2","layer":"code","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":70,"end":76,"exact":"エージェント","quote":"背景 | 中心領域 |\n| Zhipu AI／Z.ai | 清華大学系 | GLM、政府・企業、オンプレミス、エージェント |\n| Moonshot AI | 楊植麟ら研究者 | 長文脈、コード、長時間エージェント、オープンウェイト |\n| MiniMax | 元SenseTime系 | 動画、音声、音楽、AIキャラクタ","quote_start":15,"quote_end":176,"text_sha256":"831eb0a97d49080624135d7f2795b670ee44e032f99e563447ba0cd632221fd4","block_sha256":"831eb0a97d49080624135d7f2795b670ee44e032f99e563447ba0cd632221fd4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_c8866630-6a74-49cd-b251-65174111d9b1"},{"id":"occ_2668668a70b13abebdf248dc","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_d0874924-7682-4864-b572-5c2e2329c58a","section_id":"sec_9bb2d87e-9af3-4db5-8ec8-2c060796747c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":47,"end":53,"exact":"エージェント","quote":"> 一仕事当たりのAIコスト低下  \n> →AIを利用できる仕事が増える  \n> →稼働時間とエージェント数が増える  \n> →総計算需要が増加する","quote_start":0,"quote_end":74,"text_sha256":"a73b1a3fcf0c5246cc2c508a772b16e49ac23c9f5dfdecb3da5b8ff8f173661c","block_sha256":"a73b1a3fcf0c5246cc2c508a772b16e49ac23c9f5dfdecb3da5b8ff8f173661c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_d0874924-7682-4864-b572-5c2e2329c58a"},{"id":"occ_9967cfd45c24085ba429dd12","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_d7afcd05-d6d2-4c5a-a4c3-2bdd60c12538","section_id":"sec_5911cace-9207-4c1a-ac17-905092c4e39a","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":13,"end":19,"exact":"エージェント","quote":"端末・自動車・音声・GUIエージェントを重視する。Step 3.5 Flashでは、クラウド側のモデルを「頭脳」、スマートフォン上のGUIエージェントを「手」として連携させる構造を実演している。","quote_start":0,"quote_end":97,"text_sha256":"d022f9fc8c7758b110fe0ac3a7a212158d4242d8ecba96dc20e175979f2b295d","block_sha256":"d022f9fc8c7758b110fe0ac3a7a212158d4242d8ecba96dc20e175979f2b295d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_d7afcd05-d6d2-4c5a-a4c3-2bdd60c12538"},{"id":"occ_d4fb6b5e5d910df150909c46","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_f2019980-d46f-4151-9598-e06b733488e2","section_id":"sec_40fb3dd6-b799-4c6a-ac84-afc82a698360","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":68,"end":74,"exact":"エージェント","quote":"arkは、最大128GBの統合メモリと最大1PFLOP級のAI性能を持ち、デスク上で24時間稼働するパーソナルエージェントを想定している。","quote_start":13,"quote_end":82,"text_sha256":"23c1f2d16b72bce9cfc270b6b7d20d7fc8b5ab7d9b78957a75e1fa4130b6b94f","block_sha256":"23c1f2d16b72bce9cfc270b6b7d20d7fc8b5ab7d9b78957a75e1fa4130b6b94f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_f2019980-d46f-4151-9598-e06b733488e2"},{"id":"occ_1c39854d157f71ff05f1db1b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_f6b482bb-0f93-4259-9717-9bc29b644882","section_id":"sec_e237350c-87d6-4475-ae9c-6ebf81bc1469","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":28,"end":34,"exact":"エージェント","quote":"長文脈、学習効率、MoE、線形Attention、長時間エージェント、Agent Swarmを一つの技術体系として積み上げている。","quote_start":0,"quote_end":65,"text_sha256":"26cc79e44a69e8553b2107b3cfd06666488103bc110be0e88fb0485996a3df81","block_sha256":"26cc79e44a69e8553b2107b3cfd06666488103bc110be0e88fb0485996a3df81","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_f6b482bb-0f93-4259-9717-9bc29b644882"},{"id":"occ_07314a3a09c21e444772ea0a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_fba53557-d81b-42c9-8479-ef7ff43e6949","section_id":"sec_68b4a059-6263-4f37-93c5-cc5281037c57","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"- エージェントのID管理\n- タスク限定権限\n- サンドボックス\n- 情報流出検査\n- プロンプトインジェクション対策\n- 操作監査\n- 決済・送信前の承認\n- 巻き戻し","quote_start":0,"quote_end":86,"text_sha256":"a4b0f311dc608c365b99d74b9a67cc04c50e99dd0193d93d8d0055a427a15f77","block_sha256":"a4b0f311dc608c365b99d74b9a67cc04c50e99dd0193d93d8d0055a427a15f77","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_fba53557-d81b-42c9-8479-ef7ff43e6949"},{"id":"occ_f06875b811944da3b38bb11d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9","block_id":"blk_ff573157-960d-4b95-b408-35a12a05db81","section_id":"sec_b68cc5ee-55db-4682-9d8c-0e20b492d03e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":47,"end":53,"exact":"エージェント","quote":"OpenAIのCodexも、モデル自身の善意や判断力だけに安全性を任せていない。標準状態では、エージェントが編集できる場所を作業フォルダやブランチに限定し、ネットワークアクセスや権限昇格が必要な操作には追加許可を求める。クラウド版Codexも、作業ごとに独立したサンドボックスへコードを読み込み、外部との接","quote_start":0,"quote_end":153,"text_sha256":"3f46d2f528616912e47a4b57435203ef271d49508716f87299cb885b55d5beae","block_sha256":"3f46d2f528616912e47a4b57435203ef271d49508716f87299cb885b55d5beae","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835/#blk_ff573157-960d-4b95-b408-35a12a05db81"},{"id":"occ_ffe7b72e9257881d1c6431fb","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":null,"section_id":null,"layer":"title","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"title","start":11,"end":17,"exact":"エージェント","quote":"NVIDIAが描くAIエージェント産業革命: Token、AI工場、Physical AIの意味","quote_start":0,"quote_end":48,"text_sha256":"9d526dc82686cb97c5614ec5d1a11096c2a3cfd2a4bd0c6d6528fbcdb8be94c6","block_sha256":null,"offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/"},{"id":"occ_f7076030d1ffc6727b1838e2","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_02205559-8cdd-4be8-945e-c122cf043298","section_id":"sec_c693fbbb-c491-42ce-958a-6d5597b75aab","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":75,"end":81,"exact":"エージェント","quote":"、CUDA、ネットワーク、ライブラリ、推論基盤、ロボティクス、シミュレーションまで含むスタックにあります。AIエージェントが増えるほど、推論需要が増えます。Physical AIが広がるほど、AIはテキストや画像の外へ出て、機械や工場や車へ入っていきます。","quote_start":20,"quote_end":148,"text_sha256":"970225aa56270626e4b6535ab6e09b53fec10ab993cf96e2d3ca0b20e83c323a","block_sha256":"970225aa56270626e4b6535ab6e09b53fec10ab993cf96e2d3ca0b20e83c323a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_02205559-8cdd-4be8-945e-c122cf043298"},{"id":"occ_fe1d8a5634834d36926cb15b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_037ee897-4144-4baf-8d6a-9ca996c5a9cf","section_id":"sec_27e223ff-d5da-491f-ad5c-87acd73fb666","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":32,"end":38,"exact":"エージェント","quote":"講演の中でも特に象徴的だったのが、Cadenceとの設計検証AIエージェントである。","quote_start":0,"quote_end":42,"text_sha256":"706cff9a45c6572f04a323644660c7eb83b5b934d2c4eea6edb9725b5d75076f","block_sha256":"706cff9a45c6572f04a323644660c7eb83b5b934d2c4eea6edb9725b5d75076f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_037ee897-4144-4baf-8d6a-9ca996c5a9cf"},{"id":"occ_442ab0afe4fb27bd60557ddb","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_08729054-7192-4332-a310-8e7ef345b24a","section_id":"sec_b1cacf36-d71d-4341-93eb-c635a25980b5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":46,"end":52,"exact":"エージェント","quote":"現在はまだ手作業でファイルを作り、LLMに渡し、音声合成し、動画化する必要がある。しかしAIエージェントPCが成熟すれば、これらは一連の自律ワークフローとして統合されていく。","quote_start":0,"quote_end":87,"text_sha256":"cd4dca9a272f4e7405050fcad8a0fee9185477969954d081d64a158d6bb42a81","block_sha256":"cd4dca9a272f4e7405050fcad8a0fee9185477969954d081d64a158d6bb42a81","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_08729054-7192-4332-a310-8e7ef345b24a"},{"id":"occ_c4e05c360b8598483a63a7c4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_11a7fcbf-6e1e-4252-9823-d494015e80d9","section_id":"sec_84153983-b503-4ece-8633-e76f4763a9b1","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":14,"end":20,"exact":"エージェント","quote":"講演の最後は、PCの中のAIエージェントから、現実世界で動くPhysical AIへ広がる。","quote_start":0,"quote_end":46,"text_sha256":"8a8e621b54dc77662f5e30821a30c8dde6bccfccdcabd1c88bb7470c53b50cd7","block_sha256":"8a8e621b54dc77662f5e30821a30c8dde6bccfccdcabd1c88bb7470c53b50cd7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_11a7fcbf-6e1e-4252-9823-d494015e80d9"},{"id":"occ_473b58a84eb7b7fa1ee07dd4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_137e7917-0972-4ab9-b3ca-d2fffad59145","section_id":"sec_7c6d1cd5-0b7e-4e1a-8c5b-01eedd37b53a","layer":"body","character_id":null,"count":3,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":18,"end":24,"exact":"エージェント","quote":"RTX Sparkは、個人の机にAIエージェントを置く第一歩である。\nVera Rubinは、巨大AI工場でエージェントを動かすエンジンである。\nCadenceエージェントは、AIがAIチップを作る自己強化ループである。\nCosmos 3は、AIが","quote_start":0,"quote_end":124,"text_sha256":"edc24f2b19e94ca1ce24b13c70db1dc4214a555e32dab5d8926aff0d83aa8134","block_sha256":"edc24f2b19e94ca1ce24b13c70db1dc4214a555e32dab5d8926aff0d83aa8134","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_137e7917-0972-4ab9-b3ca-d2fffad59145"},{"id":"occ_51acd73d5afaceb06d65e934","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_15872e95-433b-434a-b146-8423deaa4eee","section_id":"sec_6de0a70c-08f0-4e57-9022-60c7d275a5cc","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":7,"end":13,"exact":"エージェント","quote":"そのため、AIエージェントには複数の構成要素が必要になる。","quote_start":0,"quote_end":29,"text_sha256":"f6b190622426ee55d1b6c607a77a07f10ac569af0306c7ac516934876372201a","block_sha256":"f6b190622426ee55d1b6c607a77a07f10ac569af0306c7ac516934876372201a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_15872e95-433b-434a-b146-8423deaa4eee"},{"id":"occ_284511d990901a38e51f2090","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_16169915-c9c0-4c2f-8fab-253f0b41902e","section_id":"sec_27e223ff-d5da-491f-ad5c-87acd73fb666","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":98,"end":104,"exact":"エージェント","quote":"えれば、より強いAIチップが必要になる。\nAIチップが複雑化すれば、検証がボトルネックになる。\nその検証をAIエージェントが加速すれば、NVIDIAはさらに速く次世代チップを出せる。","quote_start":43,"quote_end":134,"text_sha256":"4193f40eec685847822ff39144daf5c9351408b49582be616f06a890b693a93e","block_sha256":"4193f40eec685847822ff39144daf5c9351408b49582be616f06a890b693a93e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_16169915-c9c0-4c2f-8fab-253f0b41902e"},{"id":"occ_a24b429d1ff891c73db870f5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_1dee9fcb-7a3d-4f7f-997a-5bb251636c5a","section_id":"sec_5076d0c0-2ccd-43d3-b1ca-29b7cd684c6d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":113,"end":119,"exact":"エージェント","quote":"Arm CPU、Tensor Core、CUDA、最大128GB統一メモリ、Windows on Arm、AIエージェント実行基盤をまとめた、新しいPC向けSoC / プラットフォームである。","quote_start":58,"quote_end":154,"text_sha256":"bae6e68a4e6332153c98bd5ec42797351dff7e98a07b847e235253009f692c3a","block_sha256":"bae6e68a4e6332153c98bd5ec42797351dff7e98a07b847e235253009f692c3a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_1dee9fcb-7a3d-4f7f-997a-5bb251636c5a"},{"id":"occ_7e7afc8c876371971cee6bd3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_223f1429-8889-4990-80d6-965424ea03f0","section_id":"sec_7c6d1cd5-0b7e-4e1a-8c5b-01eedd37b53a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントは、データセンターで企業業務を行う。\nPCで個人の仕事を支援する。\nEDAでチップを検証する。\nロボットで現実世界を動く。\n自動運転車で道路を走る。\n工場で異常を検出する。\n医療や都市インフラにも入り","quote_start":0,"quote_end":108,"text_sha256":"cab0e57324e595fdbd4a7a35530cec89c9763513cf8c295c428248762423a319","block_sha256":"cab0e57324e595fdbd4a7a35530cec89c9763513cf8c295c428248762423a319","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_223f1429-8889-4990-80d6-965424ea03f0"},{"id":"occ_b969bb07b88ea316b624c630","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_22c4297c-df15-496e-a823-b3b47009ab69","section_id":"sec_b5e793a3-ea46-4776-8389-6321b9d81b0d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"なぜエージェント時代には、GPUだけでは足りないのか。","quote_start":0,"quote_end":27,"text_sha256":"13972325305969b37e2a94e1245df4c446e576cf33b74b4c315258c428cb1c17","block_sha256":"13972325305969b37e2a94e1245df4c446e576cf33b74b4c315258c428cb1c17","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_22c4297c-df15-496e-a823-b3b47009ab69"},{"id":"occ_85b86aaa2339714e54096258","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_2b9e5f31-0a5d-4450-b926-ee387f6fb4d5","section_id":"sec_5076d0c0-2ccd-43d3-b1ca-29b7cd684c6d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":32,"end":38,"exact":"エージェント","quote":"つまりRTX Sparkは、AI PCの中でも特に「ローカルAIエージェント」を強く意識した製品である。","quote_start":0,"quote_end":52,"text_sha256":"4ca9ee76bf6fed2525853c3bca41af36f79e2232a1075cf1caeee5f383890bc6","block_sha256":"4ca9ee76bf6fed2525853c3bca41af36f79e2232a1075cf1caeee5f383890bc6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_2b9e5f31-0a5d-4450-b926-ee387f6fb4d5"},{"id":"occ_523431a58d92b7fe0d932164","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_2f478b3e-286e-4809-863e-1994f9303907","section_id":"sec_5076d0c0-2ccd-43d3-b1ca-29b7cd684c6d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":37,"end":43,"exact":"エージェント","quote":"クラウド上の巨大AI工場で大規模モデルを動かしつつ、個人のPCでは常駐AIエージェントが動く。ユーザーのファイル、制作履歴、キャラクター設定、投稿計画、日記、メモ、ローカルモデル、音声合成、画像生成、動画編集をつなぐ。","quote_start":0,"quote_end":109,"text_sha256":"73b98b3b4824e74f14d70b375d277d569abe9c194aaee0a17827e06f6d3df421","block_sha256":"73b98b3b4824e74f14d70b375d277d569abe9c194aaee0a17827e06f6d3df421","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_2f478b3e-286e-4809-863e-1994f9303907"},{"id":"occ_25d853079d0501efd681536d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_35824c59-739e-438b-b61c-d58cf29798f9","section_id":"sec_d2b4c45f-8d19-41be-8dca-0f921dcdfdac","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントが道具を使う時代になるほど、NVIDIAのソフトウェア資産はさらに強くなる。","quote_start":0,"quote_end":46,"text_sha256":"f734d996eb46f14be2749a584f8ba59b41a70794c3323e8e0a230e572354ab4a","block_sha256":"f734d996eb46f14be2749a584f8ba59b41a70794c3323e8e0a230e572354ab4a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_35824c59-739e-438b-b61c-d58cf29798f9"},{"id":"occ_d21f22566130403cca5d3040","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_46ac2507-e523-4a7c-82eb-3e8c82a1a7cc","section_id":"sec_b5e793a3-ea46-4776-8389-6321b9d81b0d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":30,"end":36,"exact":"エージェント","quote":"つまりVera Rubinは、GPUの後継ではなく、**AIエージェント時代の工場エンジン**である。","quote_start":0,"quote_end":51,"text_sha256":"070a0ea9706bab2882515e8c6d50204af0eec09b83ec91e207eef1c6d067c71e","block_sha256":"070a0ea9706bab2882515e8c6d50204af0eec09b83ec91e207eef1c6d067c71e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_46ac2507-e523-4a7c-82eb-3e8c82a1a7cc"},{"id":"occ_58a616266fbdb879262c67b7","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_4b9474d4-baa6-4127-bb66-1535cbc0f2b9","section_id":"sec_b4be3d50-ce76-4e62-8346-54b5774cc715","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":96,"end":102,"exact":"エージェント","quote":"Resolve、CAD、EDA、データベース、業務ソフトなどは、今後「人間が使うアプリ」であると同時に、「AIエージェントが操作する道具」になっていく。","quote_start":41,"quote_end":117,"text_sha256":"8e0026ab5d50349a88df8fbd83f230f12f9aa02d397e0926fa3286e28759a585","block_sha256":"8e0026ab5d50349a88df8fbd83f230f12f9aa02d397e0926fa3286e28759a585","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_4b9474d4-baa6-4127-bb66-1535cbc0f2b9"},{"id":"occ_39772f79d5c0a38e756b1061","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_4ea48c23-acc5-4c59-8769-7386483c4e09","section_id":"sec_27e223ff-d5da-491f-ad5c-87acd73fb666","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"ここにAIエージェントを投入する。","quote_start":0,"quote_end":17,"text_sha256":"883b9a6d816daa88ef433c3324042c4e9cf99a5c006663e48b27f08fea7bb6a6","block_sha256":"883b9a6d816daa88ef433c3324042c4e9cf99a5c006663e48b27f08fea7bb6a6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_4ea48c23-acc5-4c59-8769-7386483c4e09"},{"id":"occ_e36966e1741280aceb982d5c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_50b9e8ee-8055-490f-ad97-5cc96d27952b","section_id":"sec_23a855f2-3f43-4199-8ce4-d20e8ecc42e6","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":13,"end":19,"exact":"エージェント","quote":"# NVIDIAが描くAIエージェント産業革命: Token、AI工場、Physical AIの意味","quote_start":0,"quote_end":50,"text_sha256":"7f132d196699b350a9ef0ca43217707e0fe38860022f5260cbb10e9139dc478c","block_sha256":"7f132d196699b350a9ef0ca43217707e0fe38860022f5260cbb10e9139dc478c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_50b9e8ee-8055-490f-ad97-5cc96d27952b"},{"id":"occ_b7d6766caf574a4c6a8f35e4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_52f0d3ef-e061-4892-8519-b9191ab5b289","section_id":"sec_6552ff8f-11c0-4b72-98d5-c62a521c1751","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":71,"end":77,"exact":"エージェント","quote":"として位置づけられている。\n- AI工場は、Tokenを大量生産するデータセンターとして定義できる。\n- AIエージェントは推論需要を増やし、AI工場の稼働を押し上げる。\n- Physical AIは、ロボット、自動運転、工場など現実世界へAIを広げる。","quote_start":16,"quote_end":143,"text_sha256":"651202719ce2bb30514ec4bf5cd0dca88bf7a77d0596f13beb7be31dcc26c2b2","block_sha256":"651202719ce2bb30514ec4bf5cd0dca88bf7a77d0596f13beb7be31dcc26c2b2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_52f0d3ef-e061-4892-8519-b9191ab5b289"},{"id":"occ_1dbebd9fe42d0ab17e64ca49","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_56114c3c-156b-4789-a124-c5498cf84565","section_id":"sec_becbcb0a-d8f3-441f-8b27-2ffd3029f801","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"- AIエージェントの実利用が推論需要をどれだけ押し上げるか。\n- NVIDIAのソフトウェア基盤がハードウェア販売以外の堀として強まるか。\n- AI PCやエッジ推論がクラウド需要を補完するか、分散させるか。\n- Ph","quote_start":0,"quote_end":110,"text_sha256":"4e3b98481359b79eedc33ec05b5f945d42c606ef9ab887c34307fdb7247b4e17","block_sha256":"4e3b98481359b79eedc33ec05b5f945d42c606ef9ab887c34307fdb7247b4e17","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_56114c3c-156b-4789-a124-c5498cf84565"},{"id":"occ_e40cc05930a51b9cbf5fc759","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_5da4fce0-ff21-4466-9fbf-49ded1fec456","section_id":"sec_b4be3d50-ce76-4e62-8346-54b5774cc715","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":12,"end":18,"exact":"エージェント","quote":"**人間向けUIと、AIエージェント向けインターフェース。**","quote_start":0,"quote_end":31,"text_sha256":"9d5c783f33a82fa3ab1d047c7e528166a350ffad69fdb5d0f67597f741d52514","block_sha256":"9d5c783f33a82fa3ab1d047c7e528166a350ffad69fdb5d0f67597f741d52514","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_5da4fce0-ff21-4466-9fbf-49ded1fec456"},{"id":"occ_dcd99c789b5fddbff32a6abc","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_5e0cf79e-6cc2-454d-aa33-204d7c813020","section_id":"sec_13472bae-5c04-4635-9459-dfe9bd4402b5","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":14,"end":20,"exact":"エージェント","quote":"**AIはチャットボットからエージェントへ進化し、そのエージェントを動かすために、データセンター、PC、自動車、ロボット、工場までが再設計される。**","quote_start":0,"quote_end":75,"text_sha256":"0bd882ae650e9a67ba35d38ad4fcff9f1c369235dedadaac237c12941637be8c","block_sha256":"0bd882ae650e9a67ba35d38ad4fcff9f1c369235dedadaac237c12941637be8c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_5e0cf79e-6cc2-454d-aa33-204d7c813020"},{"id":"occ_60c0856245d6b7475162aa8a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_5e856a3b-76ec-4cf2-a745-1df3140b7be1","section_id":"sec_6de0a70c-08f0-4e57-9022-60c7d275a5cc","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":53,"end":59,"exact":"エージェント","quote":"従来のPCやソフトウェアは、人間がアプリを開き、クリックし、入力し、保存し、出力するものだった。しかしAIエージェント時代には、構造が変わる。","quote_start":0,"quote_end":71,"text_sha256":"1b951ceefecb28f56c9b9d59f68e009c3691a3b61c5082c6c15e6e8ee396189d","block_sha256":"1b951ceefecb28f56c9b9d59f68e009c3691a3b61c5082c6c15e6e8ee396189d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_5e856a3b-76ec-4cf2-a745-1df3140b7be1"},{"id":"occ_d2164e446c99fa1ae36bedef","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_60126779-76a3-4156-9331-e219d3dfc11f","section_id":"sec_d2b4c45f-8d19-41be-8dca-0f921dcdfdac","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"たとえばAIエージェントが、流体シミュレーションを行う。構造解析を行う。ゲノム解析を行う。ロボットの軌道を最適化する。EDAツールを呼び出す。CADモデルを生成する。","quote_start":0,"quote_end":83,"text_sha256":"9d1df4d1caf1f92598c779034859df764074e84509e451b5a7d60df93bf464f3","block_sha256":"9d1df4d1caf1f92598c779034859df764074e84509e451b5a7d60df93bf464f3","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_60126779-76a3-4156-9331-e219d3dfc11f"},{"id":"occ_45f687798fc4d032e330f218","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_6187a5e1-f75c-47ca-97b5-a5f824e9d60e","section_id":"sec_23a855f2-3f43-4199-8ce4-d20e8ecc42e6","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":93,"end":99,"exact":"エージェント","quote":"、AI産業の重心がどこへ移るのかを示す物語として読むことができます。中心にあるのは、Token、AI工場、AIエージェント、Physical AIです。","quote_start":38,"quote_end":114,"text_sha256":"2ab6c9a5b2f9a4ce8f0745a6869f1eeeccb28c847967ff7ccb95fede6690df7e","block_sha256":"2ab6c9a5b2f9a4ce8f0745a6869f1eeeccb28c847967ff7ccb95fede6690df7e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_6187a5e1-f75c-47ca-97b5-a5f824e9d60e"},{"id":"occ_ed8b8897d57b57ada201ffe2","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_6555fe50-41f3-49d1-986b-88f4180e5b6b","section_id":"sec_b4be3d50-ce76-4e62-8346-54b5774cc715","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":24,"end":30,"exact":"エージェント","quote":"しかしJensen氏は、むしろ逆だと語る。\nAIエージェントが増えれば、エージェントが使うツールも増える。つまり、ソフトウェア会社の価値はなくなるのではなく、AIに使われる形で再定義される。","quote_start":0,"quote_end":95,"text_sha256":"4d3ed7ea8e4fd7754a6509a17bfaa2850b027f66b87a05ba97bcdd00caf18d91","block_sha256":"4d3ed7ea8e4fd7754a6509a17bfaa2850b027f66b87a05ba97bcdd00caf18d91","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_6555fe50-41f3-49d1-986b-88f4180e5b6b"},{"id":"occ_077f49506821390f3267a7ff","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_66cc1b0d-5542-4177-b49b-fcb0a8713da7","section_id":"sec_b1cacf36-d71d-4341-93eb-c635a25980b5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":109,"end":115,"exact":"エージェント","quote":"チャ機器、Pythonライブラリ、ComfyUIカスタムノードなどは、実機レビューと環境検証が必要になる。AIエージェントPC黎明期は、魅力と不安定さが同居する時期になるだろう。","quote_start":54,"quote_end":143,"text_sha256":"246f330f6d9a733f9b58ba2d204dc4c85aa582cd01b1d2ce4b932d1c4c976567","block_sha256":"246f330f6d9a733f9b58ba2d204dc4c85aa582cd01b1d2ce4b932d1c4c976567","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_66cc1b0d-5542-4177-b49b-fcb0a8713da7"},{"id":"occ_f876ca25cfe0da53810d38e2","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_67ba04d0-f50a-4e06-aacd-7af5fe699095","section_id":"sec_b4be3d50-ce76-4e62-8346-54b5774cc715","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":19,"end":25,"exact":"エージェント","quote":"ここで重要になるのが、ソフトウェアの「エージェント対応」である。","quote_start":0,"quote_end":32,"text_sha256":"34bd0c629f006c15793bbac0b8f1e0d0ea0512bba070ffe861439579429a62bb","block_sha256":"34bd0c629f006c15793bbac0b8f1e0d0ea0512bba070ffe861439579429a62bb","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_67ba04d0-f50a-4e06-aacd-7af5fe699095"},{"id":"occ_4bc88aa0eefe7073b80938f4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_693c62bc-4a03-4af8-83fc-deacb7c428ec","section_id":"sec_b4be3d50-ce76-4e62-8346-54b5774cc715","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"しかしAIエージェント時代には、人間向けUIだけでは足りない。AIが理解し、呼び出し、操作できるAPI、ツール、スキル、MCPサーバー、ワークフロー定義が必要になる。","quote_start":0,"quote_end":83,"text_sha256":"8e00fe8ce3dbcfbfb91c84daf257160a6eaa8cebf30161862b9319eac6ba177f","block_sha256":"8e00fe8ce3dbcfbfb91c84daf257160a6eaa8cebf30161862b9319eac6ba177f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_693c62bc-4a03-4af8-83fc-deacb7c428ec"},{"id":"occ_252cf58f12d5aa526a1566a3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_6b421c9b-cac2-4923-bff4-19b9f3bfa9ad","section_id":"sec_27e223ff-d5da-491f-ad5c-87acd73fb666","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントは、設計仕様を読み、RTLを理解し、テストベンチを作り、CadenceのXceliumでシミュレーションを走らせ、Jasperで形式検証を行い、ログを読み、バグ原因を推定し、修正案を出し、再度検証する","quote_start":0,"quote_end":108,"text_sha256":"0a2a7a9dc8d2298d2a0875f4529aab63726a73c27df1c65bfa8ba918efae3c40","block_sha256":"0a2a7a9dc8d2298d2a0875f4529aab63726a73c27df1c65bfa8ba918efae3c40","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_6b421c9b-cac2-4923-bff4-19b9f3bfa9ad"},{"id":"occ_8665af2fa94044903d8c6968","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_6b65d507-dbef-4495-ac22-baf7e31dc6f3","section_id":"sec_6de0a70c-08f0-4e57-9022-60c7d275a5cc","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":22,"end":28,"exact":"エージェント","quote":"## 3. 新しい計算モデル：アプリからAIエージェントへ","quote_start":0,"quote_end":29,"text_sha256":"fa32880377ce3e4669b5746c5abe8b452eb6a1f3fe688c17aa6739bd8908e656","block_sha256":"fa32880377ce3e4669b5746c5abe8b452eb6a1f3fe688c17aa6739bd8908e656","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_6b65d507-dbef-4495-ac22-baf7e31dc6f3"},{"id":"occ_3fff98a2a89f35d55461a49a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_6dd34f34-2bd5-4b4c-b05d-017dc474629e","section_id":"sec_5076d0c0-2ccd-43d3-b1ca-29b7cd684c6d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":14,"end":20,"exact":"エージェント","quote":"たとえば、建築設計なら、AIエージェントがスケッチを読み、Rhinoで敷地モデルを作り、Blenderでレンダリングし、生成AIで複数案を作り、コストやレイアウトも調整する。動画制作なら、台本、素材整理、字幕、サムネ生成、編集案、投稿文作成","quote_start":0,"quote_end":120,"text_sha256":"8f16edbf08dbd94623d0a71cde68aa5f54c707ef93a0168aa33823d217788ede","block_sha256":"8f16edbf08dbd94623d0a71cde68aa5f54c707ef93a0168aa33823d217788ede","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_6dd34f34-2bd5-4b4c-b05d-017dc474629e"},{"id":"occ_33b4822568c6bf6532afca53","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_81a99a57-30b1-4481-8af2-80a90c76a846","section_id":"sec_5076d0c0-2ccd-43d3-b1ca-29b7cd684c6d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":29,"end":35,"exact":"エージェント","quote":"## 10. MicrosoftとRTX Spark：AIエージェントPCの黎明","quote_start":0,"quote_end":40,"text_sha256":"3f015333effa94b5b956a15394b71b442e427035effb941904e5dfed1ac6d31d","block_sha256":"3f015333effa94b5b956a15394b71b442e427035effb941904e5dfed1ac6d31d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_81a99a57-30b1-4481-8af2-80a90c76a846"},{"id":"occ_c654eb1d8eb71b1d79f76a83","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_8329a19a-68bc-4de5-8241-95325172f6c9","section_id":"sec_7c6d1cd5-0b7e-4e1a-8c5b-01eedd37b53a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":12,"end":18,"exact":"エージェント","quote":"## 16. 考察：AIエージェント社会の前触れ","quote_start":0,"quote_end":24,"text_sha256":"da2d948b9d2ba14f1f7a6906bcf2947004b725bbd15bed00e86233e9244dc6d1","block_sha256":"da2d948b9d2ba14f1f7a6906bcf2947004b725bbd15bed00e86233e9244dc6d1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_8329a19a-68bc-4de5-8241-95325172f6c9"},{"id":"occ_ddbb47536c25a454e4bce161","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_8818e9d7-f316-4832-91df-a8c34287c6ea","section_id":"sec_b1cacf36-d71d-4341-93eb-c635a25980b5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":7,"end":13,"exact":"エージェント","quote":"自分専用のAIエージェントが常時起動している。\n日々のメモ、予定、投稿、動画制作、調査、家計、投資メモ、制作素材を把握している。\n必要に応じてローカルLLM、クラウドLLM、画像生成AI、音声合成、動画編集ツールを使い分ける。","quote_start":0,"quote_end":113,"text_sha256":"c0c9470eadda4810f9cb1a36d5bcf588ee0081dfadbf357c524b38dbd1082429","block_sha256":"c0c9470eadda4810f9cb1a36d5bcf588ee0081dfadbf357c524b38dbd1082429","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_8818e9d7-f316-4832-91df-a8c34287c6ea"},{"id":"occ_c126c575436e45810d26506f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_896886b1-1062-450e-937d-38574558f1f3","section_id":"sec_d2b4c45f-8d19-41be-8dca-0f921dcdfdac","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":7,"end":13,"exact":"エージェント","quote":"そのとき、AIエージェントは単に文章を出すのではなく、CUDA-Xのような高度な計算ライブラリを「道具」として使う。","quote_start":0,"quote_end":58,"text_sha256":"73a34d0333bc4a3927f30e69f839728d58ab6e632f9866dd8eb54bf1da223a64","block_sha256":"73a34d0333bc4a3927f30e69f839728d58ab6e632f9866dd8eb54bf1da223a64","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_896886b1-1062-450e-937d-38574558f1f3"},{"id":"occ_9015e0768e5fb0893f1d1fc5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_8cdfbcd1-8ef7-4212-9a5f-ebb2d3ac1633","section_id":"sec_7c6d1cd5-0b7e-4e1a-8c5b-01eedd37b53a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"**AIエージェントは、画面の中のチャットから、産業と現実世界を動かす存在へ進化する。NVIDIAは、そのための計算・ソフトウェア・工場・PC・ロボット基盤をすべて提供しようとしている。**","quote_start":0,"quote_end":95,"text_sha256":"11a9dc8a616f244943cd49caadca89df0108de2aaf9338532160229f633983f4","block_sha256":"11a9dc8a616f244943cd49caadca89df0108de2aaf9338532160229f633983f4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_8cdfbcd1-8ef7-4212-9a5f-ebb2d3ac1633"},{"id":"occ_e3e62f0fbf46e17104b8b76f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_989af0b5-0083-473e-b7fb-243efea54d95","section_id":"sec_3789b540-c6a8-4232-baaa-931374749e75","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":47,"end":53,"exact":"エージェント","quote":"AI工場ではDSX。\nPCではRTX SparkとWindows。\n設計検証ではCadenceエージェント。\n物理AIではCosmos。\n自動運転ではAlpamayoとHyperion。","quote_start":0,"quote_end":93,"text_sha256":"8de8e484b4c3f6c09d9fce7e75ea6db9ffaa3d50c98e2d53ed21c22d4b4af5c3","block_sha256":"8de8e484b4c3f6c09d9fce7e75ea6db9ffaa3d50c98e2d53ed21c22d4b4af5c3","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_989af0b5-0083-473e-b7fb-243efea54d95"},{"id":"occ_fa28cead51af9b2f8894d26b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_9a68f8cd-6408-4ca8-88db-1e7382b420d9","section_id":"sec_b5e793a3-ea46-4776-8389-6321b9d81b0d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントは、単にLLMを一回呼び出して終わりではない。一つの指示から、検索、推論、ツール使用、コード実行、データベース参照、メモリ管理、複数モデル呼び出し、セキュリティ確認、結果生成まで、何百、何千ステップの","quote_start":0,"quote_end":108,"text_sha256":"341ed0fd16db0e149572be03a68cbdf59e9b4f99d66db7b6feca684e12632312","block_sha256":"341ed0fd16db0e149572be03a68cbdf59e9b4f99d66db7b6feca684e12632312","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_9a68f8cd-6408-4ca8-88db-1e7382b420d9"},{"id":"occ_384b482a78e82eda64231bbf","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_9c6669ec-2f48-4c30-bc71-6cb73aaa0d7f","section_id":"sec_6de0a70c-08f0-4e57-9022-60c7d275a5cc","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":11,"end":17,"exact":"エージェント","quote":"講演の中心概念は「AIエージェント」である。","quote_start":0,"quote_end":22,"text_sha256":"0ae87a1e5e402445ccf05c99c20fe4990149a45e6273d7c775be28bdfbba11ea","block_sha256":"0ae87a1e5e402445ccf05c99c20fe4990149a45e6273d7c775be28bdfbba11ea","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_9c6669ec-2f48-4c30-bc71-6cb73aaa0d7f"},{"id":"occ_3249e95341ec0df9d52bf391","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_9db0567f-b0dd-4ec0-924e-f3e829c18585","section_id":"sec_c693fbbb-c491-42ce-958a-6d5597b75aab","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":96,"end":102,"exact":"エージェント","quote":"関わります。同じ電力でどれだけ有用なTokenを生み出せるか。どれだけ安定して稼働できるか。どれだけ多くのAIエージェントを動かせるか。これらがAI工場の競争力になります。","quote_start":41,"quote_end":127,"text_sha256":"1476e0f0a55ee3c1513f5c355495d17a9b36b0a5a9fc779404e3285fe1e569ec","block_sha256":"1476e0f0a55ee3c1513f5c355495d17a9b36b0a5a9fc779404e3285fe1e569ec","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_9db0567f-b0dd-4ec0-924e-f3e829c18585"},{"id":"occ_b9dadafea110fe653c9a91f1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_a02da15d-d40a-4319-a796-2eeb824539ac","section_id":"sec_13472bae-5c04-4635-9459-dfe9bd4402b5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":87,"end":93,"exact":"エージェント","quote":"ンフラ企業」へ変わることを宣言した内容だった。GPU単体の性能競争ではなく、Tokenを生み出すAI工場、AIエージェントを動かすPC、物理世界を理解するロボット、自動運転車の基盤まで、NVIDIAはAIの全階層を取りに行こうとしている。","quote_start":32,"quote_end":151,"text_sha256":"4c5a1170ca009059a3f999cec6424a554b1a0dcb011de043598e86b49f670762","block_sha256":"4c5a1170ca009059a3f999cec6424a554b1a0dcb011de043598e86b49f670762","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_a02da15d-d40a-4319-a796-2eeb824539ac"},{"id":"occ_88d602b527cde8df1f3f6524","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_a6e73444-7e59-4f3f-998d-7695eae50a67","section_id":"sec_6de0a70c-08f0-4e57-9022-60c7d275a5cc","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":25,"end":31,"exact":"エージェント","quote":"従来のアプリは「人間が操作する道具」だった。\nAIエージェントは「人間の目的を受け取り、道具を操作する作業者」である。","quote_start":0,"quote_end":59,"text_sha256":"a0b98248f46dd60194e1b351917d63f3fc9c0979ac049a74992e16ec888ced2e","block_sha256":"a0b98248f46dd60194e1b351917d63f3fc9c0979ac049a74992e16ec888ced2e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_a6e73444-7e59-4f3f-998d-7695eae50a67"},{"id":"occ_6982df901eef98f80c9ab07c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_adad5e2e-02e4-446f-8b0d-1d8faf030382","section_id":"sec_d783e374-a929-4575-902d-d49aa830e60a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":15,"end":21,"exact":"エージェント","quote":"# NVIDIAが描いた「AIエージェント産業革命」","quote_start":0,"quote_end":26,"text_sha256":"d25a816d38d56d289e2a0a244aa55a65d945d212cb769b68fd88b584a9849f8b","block_sha256":"d25a816d38d56d289e2a0a244aa55a65d945d212cb769b68fd88b584a9849f8b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_adad5e2e-02e4-446f-8b0d-1d8faf030382"},{"id":"occ_34f1bdfc312607ff723aa4f1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_b45f3ca5-b5a7-4f4d-9fbe-22d14e42a290","section_id":"sec_5076d0c0-2ccd-43d3-b1ca-29b7cd684c6d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":69,"end":75,"exact":"エージェント","quote":"起動し、クリックし、入力し、作業する道具だった。\nRTX Sparkが目指すPCは、人間が意図を伝えると、AIエージェントが複数のアプリやツールを操作し、成果物を作る作業場である。","quote_start":14,"quote_end":104,"text_sha256":"f30713e952362a98122dbe9d6c23bf12ade43d239111bfa5b899e5b68ac5777b","block_sha256":"f30713e952362a98122dbe9d6c23bf12ade43d239111bfa5b899e5b68ac5777b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_b45f3ca5-b5a7-4f4d-9fbe-22d14e42a290"},{"id":"occ_e026cc6d2bd51ccf188900f6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_be50e822-1ed7-4209-bee1-f976e11bd415","section_id":"sec_27e223ff-d5da-491f-ad5c-87acd73fb666","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":21,"end":27,"exact":"エージェント","quote":"## 9. Cadenceとの設計検証AIエージェント：AIがAIチップを作る時代","quote_start":0,"quote_end":41,"text_sha256":"537044339ea8d33d49cfde6a2ba8c3dd7e30f3d8a952e630ef4eea4ade828ed2","block_sha256":"537044339ea8d33d49cfde6a2ba8c3dd7e30f3d8a952e630ef4eea4ade828ed2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_be50e822-1ed7-4209-bee1-f976e11bd415"},{"id":"occ_0d9b873dc05b136175a6b471","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_c0d73d66-8d73-4608-9a0d-f5f160c78b5b","section_id":"sec_b5e793a3-ea46-4776-8389-6321b9d81b0d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":89,"end":95,"exact":"エージェント","quote":"ce Blackwell時代は、大規模推論を効率よく回すラックの時代だった。\nVera Rubin時代は、AIエージェントを動かすAI工場全体の時代である。","quote_start":34,"quote_end":112,"text_sha256":"d990651ee82e14bc6781f2d292afe481d2b2e067cd00d8a94d56a03406f19ee8","block_sha256":"d990651ee82e14bc6781f2d292afe481d2b2e067cd00d8a94d56a03406f19ee8","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_c0d73d66-8d73-4608-9a0d-f5f160c78b5b"},{"id":"occ_3a791c53d52e084b3dd9d6ee","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_c2ea8f59-25ea-4da6-82ff-ce7dd4a0f7db","section_id":"sec_d2b4c45f-8d19-41be-8dca-0f921dcdfdac","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":8,"end":14,"exact":"エージェント","quote":"重要なのは、AIエージェント時代になると、これらのライブラリを人間だけでなくAIエージェントが使うようになる点だ。","quote_start":0,"quote_end":57,"text_sha256":"f339c34916894eab7b1f9faddf7a7766cf1c92c689444a8f9eab174d12cd15cf","block_sha256":"f339c34916894eab7b1f9faddf7a7766cf1c92c689444a8f9eab174d12cd15cf","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_c2ea8f59-25ea-4da6-82ff-ce7dd4a0f7db"},{"id":"occ_368bfdc7dd72316d8a69af35","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_c55f3985-8080-4c34-a36f-093acf0e3222","section_id":"sec_b1cacf36-d71d-4341-93eb-c635a25980b5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントPCの本質は、PCが「作業道具」から「半自律的な作業環境」に変わることだ。","quote_start":0,"quote_end":45,"text_sha256":"1455201443d9eecd4844d3abce708c23af63dc2347ab7ce1a35bca68e8be148b","block_sha256":"1455201443d9eecd4844d3abce708c23af63dc2347ab7ce1a35bca68e8be148b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_c55f3985-8080-4c34-a36f-093acf0e3222"},{"id":"occ_bcf97e1c480a9b89d8bb195d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_df2f8a76-0d99-47c4-9a8e-a05f875d52cd","section_id":"sec_6de0a70c-08f0-4e57-9022-60c7d275a5cc","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":10,"end":16,"exact":"エージェント","quote":"これらを束ねるのが、エージェントのランタイム、オーケストレーション、セキュリティ基盤である。","quote_start":0,"quote_end":46,"text_sha256":"b5b9de72a6a5be0fb4e998852406e361663bfeea8c1e898708b3a23e30591798","block_sha256":"b5b9de72a6a5be0fb4e998852406e361663bfeea8c1e898708b3a23e30591798","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_df2f8a76-0d99-47c4-9a8e-a05f875d52cd"},{"id":"occ_62c3ec4262d8033ee78375e4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_e3661ae1-2bd6-45bf-bb6e-8536948c7123","section_id":"sec_b1cacf36-d71d-4341-93eb-c635a25980b5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":9,"end":15,"exact":"エージェント","quote":"## 11. AIエージェントPCがもたらす変化","quote_start":0,"quote_end":24,"text_sha256":"2e1802436b924af0d7158b7dc24558ce3e2e95dbc7bd671de9dd1b921029c206","block_sha256":"2e1802436b924af0d7158b7dc24558ce3e2e95dbc7bd671de9dd1b921029c206","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_e3661ae1-2bd6-45bf-bb6e-8536948c7123"},{"id":"occ_8af31c2de194d8021fd2fc46","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_f2ef2bb6-eccc-4593-8d65-a12a9e57de72","section_id":"sec_5076d0c0-2ccd-43d3-b1ca-29b7cd684c6d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":53,"end":59,"exact":"エージェント","quote":"Windows 95がPCを一般家庭へ広げたとすれば、RTX SparkとMicrosoftの協業は、AIエージェントを個人の机へ降ろす試みである。","quote_start":0,"quote_end":74,"text_sha256":"732cfda5feb8db1f23fc7b410b315a81510a4bf1a8dd6edc58a1716bba8bcb50","block_sha256":"732cfda5feb8db1f23fc7b410b315a81510a4bf1a8dd6edc58a1716bba8bcb50","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_f2ef2bb6-eccc-4593-8d65-a12a9e57de72"},{"id":"occ_afb9ca4cd78f0c8b510400fb","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_fdddde31-95fc-4845-aa86-ce4323762d6b","section_id":"sec_6de0a70c-08f0-4e57-9022-60c7d275a5cc","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":11,"end":17,"exact":"エージェント","quote":"ここで重要なのは、AIエージェントは単なる「賢いチャット」ではないという点だ。AIエージェントとは、モデル、メモリ、ツール、実行環境、セキュリティ、UIが一体化した新しい計算モデルである。","quote_start":0,"quote_end":94,"text_sha256":"4baecbe3abb08c82c5b7711e1b05d201cc4db6842abcd48cb58e8332288725b8","block_sha256":"4baecbe3abb08c82c5b7711e1b05d201cc4db6842abcd48cb58e8332288725b8","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_fdddde31-95fc-4845-aa86-ce4323762d6b"},{"id":"occ_2dabf86e335661e1c6fc8951","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_fe658b38-6656-4c63-a2f6-51088921c1b6","section_id":"sec_b5e793a3-ea46-4776-8389-6321b9d81b0d","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":25,"end":32,"exact":"Agentic","quote":"## 8. Vera Rubin：GPUではなく、Agentic AI Factoryの心臓","quote_start":0,"quote_end":46,"text_sha256":"360850c5ea3dab5d91ae6705fa1768b5f35bf065d45556c1c0ef97066e7fd516","block_sha256":"360850c5ea3dab5d91ae6705fa1768b5f35bf065d45556c1c0ef97066e7fd516","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_fe658b38-6656-4c63-a2f6-51088921c1b6"},{"id":"occ_b32ecf104881debb7f17e910","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a","block_id":"blk_fefebac3-a6cf-409c-a898-278b951a5ebd","section_id":"sec_5076d0c0-2ccd-43d3-b1ca-29b7cd684c6d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":37,"end":43,"exact":"エージェント","quote":"これを理解する上で重要なのは、RTX Sparkが「GPU」ではなく「AIエージェントPCの心臓」だという点だ。","quote_start":0,"quote_end":56,"text_sha256":"98107fc391058eb26a6a693f7406008be8c4f2a2bd3e0827ed9fd99cdf1ecaf6","block_sha256":"98107fc391058eb26a6a693f7406008be8c4f2a2bd3e0827ed9fd99cdf1ecaf6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_90fb7b96-f10b-43df-a612-0c23e8a50740/#blk_fefebac3-a6cf-409c-a898-278b951a5ebd"},{"id":"occ_0ab6c3f031ae044a93a101d7","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a","work_id":"wrk_c22379bc-7081-4f8f-a3e9-4418f450cf96","block_id":"blk_25ee18c4-ae46-47f2-bf2b-c56ebff95391","section_id":"sec_e510b27b-ac3b-4bb7-a5e3-56729ac980b9","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":49,"end":55,"exact":"エージェント","quote":"Tesla AIが「車やロボットを動かすAI」だとすれば、xAIは「言語、推論、検索、会話、生成、エージェント」を担うAIです。","quote_start":0,"quote_end":64,"text_sha256":"1b42eb47c194d4bb9edf32788a2c2875faffb4f971fbfd6d1772c0a7a804de55","block_sha256":"1b42eb47c194d4bb9edf32788a2c2875faffb4f971fbfd6d1772c0a7a804de55","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a/#blk_25ee18c4-ae46-47f2-bf2b-c56ebff95391"},{"id":"occ_9085825e07a186b57d61f7de","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a","work_id":"wrk_c22379bc-7081-4f8f-a3e9-4418f450cf96","block_id":"blk_39abf7ea-601b-47c7-990b-163d502438ed","section_id":"sec_ddb4c7f3-9c59-4199-b283-1905e236aff9","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":39,"end":45,"exact":"エージェント","quote":"```\nTesla = フィジカルAIの肉体\nxAI = AIの頭脳・モデル・エージェント\nSpaceX = 宇宙・通信・computeインフラ\nTeraFab = AIチップ供給の心臓\n```","quote_start":0,"quote_end":97,"text_sha256":"ca3e8138b17597364b15f562c0135ff4d05b67a01d9dde09ce659fe10f9df577","block_sha256":"ca3e8138b17597364b15f562c0135ff4d05b67a01d9dde09ce659fe10f9df577","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a/#blk_39abf7ea-601b-47c7-990b-163d502438ed"},{"id":"occ_9adc6d4396c09ae4c2c40fac","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a","work_id":"wrk_c22379bc-7081-4f8f-a3e9-4418f450cf96","block_id":"blk_5207fdf9-205d-4a24-b9a3-56e9029e2b49","section_id":"sec_5c7091bf-9251-425f-87d5-662a2f8bf43c","layer":"code","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"```\nAIエージェント需要が爆発\n↓\n推論トークン需要が急増\n↓\nAIデータセンター不足が深刻化\n↓\nSpaceX / xAIがcomputeを外販\n↓\nGrok、X、API、AIエージェント収益が伸びる\n↓\nStarli","quote_start":0,"quote_end":112,"text_sha256":"1053c37ff2e58c75d82abd5f79fb56156615c0d9bc08a9990f28ec4c7ee4095d","block_sha256":"1053c37ff2e58c75d82abd5f79fb56156615c0d9bc08a9990f28ec4c7ee4095d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a/#blk_5207fdf9-205d-4a24-b9a3-56e9029e2b49"},{"id":"occ_96163b40b36afd75ed4dc6fb","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a","work_id":"wrk_c22379bc-7081-4f8f-a3e9-4418f450cf96","block_id":"blk_b56ce3fb-5e47-47c1-af99-ff729ad7e5ca","section_id":"sec_e510b27b-ac3b-4bb7-a5e3-56729ac980b9","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":18,"end":24,"exact":"エージェント","quote":"xAIは、Grok、LLM、X、AIエージェントの中核です。","quote_start":0,"quote_end":30,"text_sha256":"89c08b77d3a16380980d83889a784f907a7088ad6c2df3dfb0028bc6a9b56ca4","block_sha256":"89c08b77d3a16380980d83889a784f907a7088ad6c2df3dfb0028bc6a9b56ca4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a/#blk_b56ce3fb-5e47-47c1-af99-ff729ad7e5ca"},{"id":"occ_eb259a4b472a29b7264da06e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a","work_id":"wrk_c22379bc-7081-4f8f-a3e9-4418f450cf96","block_id":"blk_bdabbe7f-8f82-4d42-be2c-ada2cc18a61c","section_id":"sec_e510b27b-ac3b-4bb7-a5e3-56729ac980b9","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":43,"end":49,"exact":"エージェント","quote":"- Grok課金\n- Xサブスク\n- X広告\n- API利用料\n- 企業向けAI\n- エージェント利用料","quote_start":0,"quote_end":52,"text_sha256":"e8d85cc48fcdd7a0f6bd1febb0564b14d639427179fda688cc5e51eee70dbe24","block_sha256":"e8d85cc48fcdd7a0f6bd1febb0564b14d639427179fda688cc5e51eee70dbe24","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a/#blk_bdabbe7f-8f82-4d42-be2c-ada2cc18a61c"},{"id":"occ_37d601296bbe2f901a4bec9c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a","work_id":"wrk_c22379bc-7081-4f8f-a3e9-4418f450cf96","block_id":"blk_beb60d38-a390-4030-a2bf-31fb25eae398","section_id":"sec_1111ded2-63bf-4333-8709-b10128e0d7a2","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":75,"end":81,"exact":"エージェント","quote":"肉体\n= EV、FSD、Optimus、Robotaxi\n\nxAI\n= AIの頭脳\n= Grok、LLM、X、エージェント\n\nSpaceX\n= 宇宙・通信・computeインフラ\n= Starship、Starlink、AIデータセンター、宇宙compute\n\nTeraFab\n= AIチップ供給の心臓\n= Tesla、x","quote_start":20,"quote_end":181,"text_sha256":"e9edbf13e1d6f3c51da4d35f1ab8049625bb7e4f4a8f215bcb4c36b307da9837","block_sha256":"e9edbf13e1d6f3c51da4d35f1ab8049625bb7e4f4a8f215bcb4c36b307da9837","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a/#blk_beb60d38-a390-4030-a2bf-31fb25eae398"},{"id":"occ_6f20ae62411fe32606008c0d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a","work_id":"wrk_c22379bc-7081-4f8f-a3e9-4418f450cf96","block_id":"blk_c951f451-fe76-4fc1-b69f-790d4fc752e4","section_id":"sec_5c7091bf-9251-425f-87d5-662a2f8bf43c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":35,"end":41,"exact":"エージェント","quote":"- チップ\n- 電力\n- データセンター\n- 通信\n- 推論基盤\n- エージェント\n- ユーザー接点\n- 物理世界への実装","quote_start":0,"quote_end":61,"text_sha256":"7dc6d2c4b73bd3ffafca786dd481bc2b4057f21a00ba26eadd39374b6b00e88c","block_sha256":"7dc6d2c4b73bd3ffafca786dd481bc2b4057f21a00ba26eadd39374b6b00e88c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a/#blk_c951f451-fe76-4fc1-b69f-790d4fc752e4"},{"id":"occ_fad7556d2227b37f4c468e1d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a","work_id":"wrk_c22379bc-7081-4f8f-a3e9-4418f450cf96","block_id":"blk_d46e1c03-30ff-4b57-89a2-7e741013ceab","section_id":"sec_fcae4a11-267f-4617-8297-32eb74ebfb76","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":59,"end":65,"exact":"エージェント","quote":"Tesla\n→ FSD、Optimus、Robotaxiにチップが必要\n\nxAI\n→ Grok、LLM、推論、エージェントにチップが必要\n\nSpaceX\n→ Starlink、宇宙AI、AI compute外販にチップが必要\n\nTeraFab\n→ それら全てにチップを供給する\n```","quote_start":4,"quote_end":146,"text_sha256":"e209144fde045b2c2371e0abc9d5b0d0fb5dada2949f286723e2d0d6c4a64192","block_sha256":"e209144fde045b2c2371e0abc9d5b0d0fb5dada2949f286723e2d0d6c4a64192","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a/#blk_d46e1c03-30ff-4b57-89a2-7e741013ceab"},{"id":"occ_ee8b5741b81e55dd09b0c690","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a","work_id":"wrk_c22379bc-7081-4f8f-a3e9-4418f450cf96","block_id":"blk_d6567690-4fce-4536-878b-e3fbc3f61457","section_id":"sec_e510b27b-ac3b-4bb7-a5e3-56729ac980b9","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":18,"end":24,"exact":"エージェント","quote":"### xAI：AIの頭脳・モデル・エージェント","quote_start":0,"quote_end":24,"text_sha256":"a15408d6f8bb62783f5a5e50e796d1732cda97606b3ac8518232c24dab78a626","block_sha256":"a15408d6f8bb62783f5a5e50e796d1732cda97606b3ac8518232c24dab78a626","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a/#blk_d6567690-4fce-4536-878b-e3fbc3f61457"},{"id":"occ_a88f09f2712b6d4f63fbabf5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_92ff5158-6c34-4526-ab02-c8db8332951f","work_id":"wrk_ea89805e-b45e-4a9c-b22a-29b168da02ed","block_id":"blk_41451dea-1532-4b15-adff-b2769b8980ba","section_id":"sec_a3d4c8b6-f108-4f44-8fbc-20b35cfb7609","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":49,"end":55,"exact":"エージェント","quote":"OpenAI、Anthropic、Googleの最上位モデルは、依然として最先端の科学推論、長距離エージェント、複雑な安全性設計、企業向け信頼性で優位を持つ。しかし、それらのモデルは基本的にクローズドであり、利用者はAPIやクラウドサービスを通じてアクセスする。価格変更、利用制限、規約変更、地政学的制約、デー","quote_start":0,"quote_end":155,"text_sha256":"906c145aa14649a7f2df1e9260e56dda4d439db2c1c0803d7c36e44efef9881b","block_sha256":"906c145aa14649a7f2df1e9260e56dda4d439db2c1c0803d7c36e44efef9881b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_92ff5158-6c34-4526-ab02-c8db8332951f/#blk_41451dea-1532-4b15-adff-b2769b8980ba"},{"id":"occ_00d2804984d8ed8956d7eacb","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_92ff5158-6c34-4526-ab02-c8db8332951f","work_id":"wrk_ea89805e-b45e-4a9c-b22a-29b168da02ed","block_id":"blk_982b8ad5-b392-4148-925d-bc24305f509a","section_id":"sec_a3d4c8b6-f108-4f44-8fbc-20b35cfb7609","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":69,"end":75,"exact":"エージェント","quote":"る新しい大規模言語モデルの発表にとどまらない。重要なのは、オープンウェイトモデルが、コーディング、長文脈処理、エージェント的作業において、米国のフロンティアモデルにかなり近い実用水準へ到達しつつあることだ。","quote_start":14,"quote_end":117,"text_sha256":"3b77599fdf25d16dce7ed7a83192a5458f5ae9579fc3fcd13c7ae564208f4230","block_sha256":"3b77599fdf25d16dce7ed7a83192a5458f5ae9579fc3fcd13c7ae564208f4230","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_92ff5158-6c34-4526-ab02-c8db8332951f/#blk_982b8ad5-b392-4148-925d-bc24305f509a"},{"id":"occ_70da3141e7c4ab431668adac","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_92ff5158-6c34-4526-ab02-c8db8332951f","work_id":"wrk_ea89805e-b45e-4a9c-b22a-29b168da02ed","block_id":"blk_c2edd732-a7b2-4b87-a7fc-f1924038bd44","section_id":"sec_a3d4c8b6-f108-4f44-8fbc-20b35cfb7609","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":45,"end":51,"exact":"エージェント","quote":"モデル単体の性能は、少しずつコモディティ化していく。差別化の中心は、推論インフラ、データ、エージェント実行環境、セキュリティ、法制度、電力、冷却、ネットワーク、産業実装へ移る。","quote_start":0,"quote_end":88,"text_sha256":"8735107971bf08db33a80c091f4e10e5b827ae0792af0cf28889f4d4aeb4bd94","block_sha256":"8735107971bf08db33a80c091f4e10e5b827ae0792af0cf28889f4d4aeb4bd94","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_92ff5158-6c34-4526-ab02-c8db8332951f/#blk_c2edd732-a7b2-4b87-a7fc-f1924038bd44"},{"id":"occ_d56a56afdd8f6e78e33fe936","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_92ff5158-6c34-4526-ab02-c8db8332951f","work_id":"wrk_ea89805e-b45e-4a9c-b22a-29b168da02ed","block_id":"blk_cbe13c0d-6e5a-4319-abb5-52b9b791a04f","section_id":"sec_a3d4c8b6-f108-4f44-8fbc-20b35cfb7609","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":38,"end":44,"exact":"エージェント","quote":"OpenAIは、単なる高性能API企業ではなく、ChatGPT、Codex、エージェント、業務連携、メモリ、ファイル、企業向け環境を含むAI OSへ向かわざるを得ない。Anthropicは、すべての処理をClaudeで行う企業ではなく、難所、高信頼、長時間エージェント、安全性が重要な場面","quote_start":0,"quote_end":144,"text_sha256":"994fe8926a819b7f5e8243454606f0a20928e17b385858a53a10e3f3ad462c1d","block_sha256":"994fe8926a819b7f5e8243454606f0a20928e17b385858a53a10e3f3ad462c1d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_92ff5158-6c34-4526-ab02-c8db8332951f/#blk_cbe13c0d-6e5a-4319-abb5-52b9b791a04f"},{"id":"occ_f6f952cdd20db9e0a0622055","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_92ff5158-6c34-4526-ab02-c8db8332951f","work_id":"wrk_ea89805e-b45e-4a9c-b22a-29b168da02ed","block_id":"blk_f18ad490-6ab5-4fce-b891-d84902517f7e","section_id":"sec_71900921-e7e0-416a-bacf-5fd88dc2a5c9","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":40,"end":46,"exact":"エージェント","quote":"特に重要なのは、これが「フロンティアモデル訓練」だけではなく、**推論・RAG・エージェント・行政AI・教育AI・防衛AI・医療AI**の常時稼働需要として出てくる点です。","quote_start":0,"quote_end":86,"text_sha256":"2e73c478c9cdef04a41f11f019684b12c422cdd92eb0157f51cf22120ab72f64","block_sha256":"2e73c478c9cdef04a41f11f019684b12c422cdd92eb0157f51cf22120ab72f64","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_92ff5158-6c34-4526-ab02-c8db8332951f/#blk_f18ad490-6ab5-4fce-b891-d84902517f7e"},{"id":"occ_ad38695eb88abbf52c629988","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8","work_id":"wrk_8f40984a-d96e-441d-8ace-83f20958435a","block_id":"blk_1bd6c4b8-91c8-4c96-8ed1-5d3c5b44625b","section_id":"sec_669cbc9d-de76-40c4-807d-0e1d83982942","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":30,"end":36,"exact":"エージェント","quote":"一方で、高度な推論、大規模検索、企業DB接続、長期記憶、重いエージェント実行、マルチモーダル処理、専門モデルはクラウドに上がる。","quote_start":0,"quote_end":64,"text_sha256":"e6bf3caea8a611f6cf751d60a1ec135441d9a121562ab37576783d4b1454bd2d","block_sha256":"e6bf3caea8a611f6cf751d60a1ec135441d9a121562ab37576783d4b1454bd2d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8/#blk_1bd6c4b8-91c8-4c96-8ed1-5d3c5b44625b"},{"id":"occ_9366e0b4a4babf8bf7000df3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8","work_id":"wrk_8f40984a-d96e-441d-8ace-83f20958435a","block_id":"blk_3ca33ad6-aea5-4ff1-af21-9034b1394271","section_id":"sec_f5717108-bd77-4308-8d3a-1674ee05376e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":22,"end":28,"exact":"エージェント","quote":"AI労働時代は、AI労働OSが企業業務とAIエージェントの間に入る。","quote_start":0,"quote_end":34,"text_sha256":"b174092c472bee391cf35d56cb81855be1a6e2a00516d325a7a9658418a69561","block_sha256":"b174092c472bee391cf35d56cb81855be1a6e2a00516d325a7a9658418a69561","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8/#blk_3ca33ad6-aea5-4ff1-af21-9034b1394271"},{"id":"occ_496460fc8c6f8d25b672b136","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8","work_id":"wrk_8f40984a-d96e-441d-8ace-83f20958435a","block_id":"blk_590c7478-2244-4fff-8ecc-361bd70d89d9","section_id":"sec_22c99273-2a63-411c-8482-7f086a5881cd","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":170,"end":176,"exact":"エージェント","quote":"/arxiv.org/abs/2602.00139?utm_source=chatgpt.com)) 産業界のエージェントAI導入研究では、多くの企業がまだAIアシスタント段階にあり、本番導入には検証、文脈、非決定性、機密データなどの壁があると整理されています。([arXiv](https://arxiv.org/abs","quote_start":115,"quote_end":276,"text_sha256":"d95d14828ddbc3de658314127961f4eae3bbd12a2d1634c6b894f01765e700bc","block_sha256":"d95d14828ddbc3de658314127961f4eae3bbd12a2d1634c6b894f01765e700bc","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8/#blk_590c7478-2244-4fff-8ecc-361bd70d89d9"},{"id":"occ_88921c6e4493e73529569c40","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8","work_id":"wrk_8f40984a-d96e-441d-8ace-83f20958435a","block_id":"blk_6a40b42f-d5bb-4b7d-9971-8f0d7b7e6bce","section_id":"sec_c9454fbc-bedc-4ae7-b7f4-86dd2aaeb49a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":73,"end":79,"exact":"エージェント","quote":"ケーラー復活の強いサインは、すごいチャットAIの登場ではない。  \n採用抑制、外注費削減、人件費率の低下、AIエージェント予算の新設、BPO支出の減少である。","quote_start":18,"quote_end":97,"text_sha256":"08a679bf4bba12cd2cde573f6ac05878a190646ba897f14ab8f6ac1caffa381a","block_sha256":"08a679bf4bba12cd2cde573f6ac05878a190646ba897f14ab8f6ac1caffa381a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8/#blk_6a40b42f-d5bb-4b7d-9971-8f0d7b7e6bce"},{"id":"occ_5c44cf0d4a3f91040f859189","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8","work_id":"wrk_8f40984a-d96e-441d-8ace-83f20958435a","block_id":"blk_d0dd873c-bced-476e-a8ab-df8c67760f16","section_id":"sec_c9454fbc-bedc-4ae7-b7f4-86dd2aaeb49a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":17,"end":23,"exact":"エージェント","quote":"企業が人間に任せていた業務を、AIエージェントに任せる。  \n採用を抑制する。  \n外注を減らす。  \nBPOを削る。  \nカスタマーサポートをAI化する。  \nジュニア開発の一部をAIに置き換える。  \n広告制作、翻訳、調査、資料作成、請求処理","quote_start":0,"quote_end":123,"text_sha256":"80f321660a07c493053de78193b3f42288be59a8ede2ebd8972917fbcefd2c0e","block_sha256":"80f321660a07c493053de78193b3f42288be59a8ede2ebd8972917fbcefd2c0e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8/#blk_d0dd873c-bced-476e-a8ab-df8c67760f16"},{"id":"occ_498293ff79fdcbfaa7f2317f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8","work_id":"wrk_8f40984a-d96e-441d-8ace-83f20958435a","block_id":"blk_d96e09aa-b7b8-4b2a-a3ec-249c90f43bff","section_id":"sec_e51875a4-adb9-4fa8-8983-0d3ec3f03963","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":313,"end":319,"exact":"エージェント","quote":"wは、ERP、CRM、ワークフロー、政府・大企業基盤に強い。  \nOpenAIやAnthropicは、モデルとエージェント体験で先行できるが、企業のOS層まで取りに行かなければ、クラウドや業務アプリ側に利益を吸われる可能性がある。","quote_start":258,"quote_end":373,"text_sha256":"5efa7534831969215f70b55acfe1bd6675ea93a977b1301b65cdbf9cc8a41d59","block_sha256":"5efa7534831969215f70b55acfe1bd6675ea93a977b1301b65cdbf9cc8a41d59","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8/#blk_d96e09aa-b7b8-4b2a-a3ec-249c90f43bff"},{"id":"occ_a773665a3e8348fe1a8e3634","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8","work_id":"wrk_8f40984a-d96e-441d-8ace-83f20958435a","block_id":"blk_e2376827-0cb1-4ef0-9c7b-12e7cf2cc05b","section_id":"sec_7395bd9d-969d-4b9c-8e96-2083a5a23904","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":33,"end":39,"exact":"エージェント","quote":"Claude Code、Codex、Copilot、Gemini系エージェントが重要なのは、単に便利だからではない。","quote_start":0,"quote_end":58,"text_sha256":"b77b7a5345657fc51851229c3ff9a0a4b2f53e0d3f28834a81dfe40b8e1af64e","block_sha256":"b77b7a5345657fc51851229c3ff9a0a4b2f53e0d3f28834a81dfe40b8e1af64e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8/#blk_e2376827-0cb1-4ef0-9c7b-12e7cf2cc05b"},{"id":"occ_cc28d49687f091986b7be863","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8","work_id":"wrk_8f40984a-d96e-441d-8ace-83f20958435a","block_id":"blk_feab373e-cf47-423e-aa22-01984463a81b","section_id":"sec_f5717108-bd77-4308-8d3a-1674ee05376e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":108,"end":114,"exact":"エージェント","quote":"内データ接続料。  \n業務アプリ接続料。  \n長期記憶ストレージ料。  \n高性能モデルオプション。  \n業界別エージェント追加料金。  \nクラウドバースト課金。  \n成果課金。  \nAIマーケットプレイス手数料。","quote_start":53,"quote_end":159,"text_sha256":"81ee2d03064c6a3cc8ed17f4116996c04b8a8d5a4b5be00cc69d339aaf96b443","block_sha256":"81ee2d03064c6a3cc8ed17f4116996c04b8a8d5a4b5be00cc69d339aaf96b443","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8/#blk_feab373e-cf47-423e-aa22-01984463a81b"},{"id":"occ_f83c2b1c98361a614622bfd2","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","work_id":"wrk_95f3adb2-9ef8-4c60-8a3b-5ded0268dac4","block_id":"blk_190b7312-808d-4de7-b5bf-c37a334fc474","section_id":"sec_57d784db-6048-4eb2-93fb-3bf936a4cd4e","layer":"code","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":44,"end":51,"exact":"agentic","quote":"```\nRubin GPUラック\nVera Rubin NVL72\nKVキャッシュ管理\nagentic AI\n強化学習環境\nGPU稼働率最大化\np99レイテンシ管理\nNVIDIA AI Enterprise / CUDA / NIM環境\n```","quote_start":0,"quote_end":122,"text_sha256":"2cb91d1b3ded27204ad215ca2f8b98e5a1283a6c0165ed104aa9e4b9165daa8f","block_sha256":"2cb91d1b3ded27204ad215ca2f8b98e5a1283a6c0165ed104aa9e4b9165daa8f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4/#blk_190b7312-808d-4de7-b5bf-c37a334fc474"},{"id":"occ_d5ea94aec6f6160055ab7434","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","work_id":"wrk_95f3adb2-9ef8-4c60-8a3b-5ded0268dac4","block_id":"blk_1e5c25dd-e5a9-45e9-98ff-1d9ff7dbdafd","section_id":"sec_3b5832dc-a184-46de-a409-c3c2ed0c6afe","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":54,"end":60,"exact":"エージェント","quote":"LPDDRをCPUに近づけると、メモリ帯域/Wが改善しやすく、AI前処理、RAG、推論制御、サンドボックス、エージェント処理には有利です。  \nただし、従来のDDR DIMM中心サーバーに比べると、メモリ増設性、交換性、構成自由度が変わる可能性があります。","quote_start":0,"quote_end":128,"text_sha256":"c9630efd4e9a7f43bcf4f780847a69210f6f864e8a3a62198aec53d50c7c60b9","block_sha256":"c9630efd4e9a7f43bcf4f780847a69210f6f864e8a3a62198aec53d50c7c60b9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4/#blk_1e5c25dd-e5a9-45e9-98ff-1d9ff7dbdafd"},{"id":"occ_31d164adc314ca35d22c4e0f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","work_id":"wrk_95f3adb2-9ef8-4c60-8a3b-5ded0268dac4","block_id":"blk_23f76a69-d1b9-4b05-ac48-b2aef4cd96f3","section_id":"sec_03af3792-749e-4fdf-9bc4-f6c0aaabb136","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":165,"end":172,"exact":"agentic","quote":"レッド、最大1.5TB LPDDR5X、1.2TB/sメモリ帯域、1.8TB/s NVLink-C2Cを持ち、agentic AI、強化学習、サンドボックス実行、GPUを待たせないオーケストレーションに最適化されています。([SEC](https://www.sec.gov/Archives/edgar/data/1045","quote_start":110,"quote_end":272,"text_sha256":"9846f3e1dbee82450416459b71c071437ed071127ab9c2b07195366e6414375e","block_sha256":"9846f3e1dbee82450416459b71c071437ed071127ab9c2b07195366e6414375e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4/#blk_23f76a69-d1b9-4b05-ac48-b2aef4cd96f3"},{"id":"occ_e7e9496303228839b6ab737b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","work_id":"wrk_95f3adb2-9ef8-4c60-8a3b-5ded0268dac4","block_id":"blk_527f5e90-0127-4e7f-b3dc-02bb0c48cb07","section_id":"sec_79e3b6cc-1281-4c0c-8ae3-7dc91f1c8cd3","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":99,"end":106,"exact":"agentic","quote":"た。しかしAGI CPUでは、Arm自身がTSMCで製造される完成CPUを出します。ArmはAGI CPUを、agentic AI向けのCPUであり、Armの事業をIP、Compute Subsystems、そしてproduction siliconへ拡張するものとして発表しました。Metaがリードパートナーで、OpenA","quote_start":44,"quote_end":206,"text_sha256":"5dead51aea164a65c3107afed3a38519b8a0e71dab92d580dddd3a90e59f112e","block_sha256":"5dead51aea164a65c3107afed3a38519b8a0e71dab92d580dddd3a90e59f112e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4/#blk_527f5e90-0127-4e7f-b3dc-02bb0c48cb07"},{"id":"occ_c1c439708f864aed63761244","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","work_id":"wrk_95f3adb2-9ef8-4c60-8a3b-5ded0268dac4","block_id":"blk_6e15a9b7-fca3-4eb8-a8b9-9be8b4ed32ae","section_id":"sec_3b5832dc-a184-46de-a409-c3c2ed0c6afe","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":75,"end":82,"exact":"agentic","quote":"AIインフラ寄りになる可能性があります。AMDはVeranoについて、LPDDRなどの先進メモリ技術を採用し、agentic AIで増えるメモリ需要と電力制約に対応すると説明しています。([AMD](https://www.amd.com/en/newsroom/press-releases/2026-5-20-amd-a","quote_start":20,"quote_end":182,"text_sha256":"023feac1bb45ad186a5001ac5f2424b6c254479e3a6a56790face5c008203498","block_sha256":"023feac1bb45ad186a5001ac5f2424b6c254479e3a6a56790face5c008203498","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4/#blk_6e15a9b7-fca3-4eb8-a8b9-9be8b4ed32ae"},{"id":"occ_6553047511d97702b495ddc6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","work_id":"wrk_95f3adb2-9ef8-4c60-8a3b-5ded0268dac4","block_id":"blk_86dca3fa-2e30-4774-b2dc-37f084255342","section_id":"sec_4d53ceec-4948-4e97-bc81-cbcb507d7d8d","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":197,"end":204,"exact":"agentic","quote":"、PCIe/CXL系I/Oが重視されます。Veranoでは、AMDがLPDDRなどの先進メモリ技術を取り入れ、agentic AIで増えるメモリ需要と電力制約に対応すると説明しています。([AMD](https://www.amd.com/en/newsroom/press-releases/2026-5-20-amd-a","quote_start":142,"quote_end":304,"text_sha256":"deb6211e99c016ec732f3e73ff4643a0b58bb957493aef4834f854f1687f8fa1","block_sha256":"deb6211e99c016ec732f3e73ff4643a0b58bb957493aef4834f854f1687f8fa1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4/#blk_86dca3fa-2e30-4774-b2dc-37f084255342"},{"id":"occ_ac0b9ef00465de2d4b5a71a7","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","work_id":"wrk_95f3adb2-9ef8-4c60-8a3b-5ded0268dac4","block_id":"blk_9acea719-ec9e-44b0-a5d7-1dbf2ae51cfa","section_id":"sec_4d66a0ee-68ea-4ac8-a21e-37d8a3c83923","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":99,"end":105,"exact":"エージェント","quote":"oft Cobalt、AmpereOneは、**ArmによるクラウドCPUの拡大**を示しています。  \nAIエージェント時代には、Web/API/制御プレーンが膨らむため、Armの性能/Wは大きな武器になります。","quote_start":44,"quote_end":151,"text_sha256":"6cefcf0005b67017626714ecf2d4d157e3314cebc4b9ea80d690756634471d33","block_sha256":"6cefcf0005b67017626714ecf2d4d157e3314cebc4b9ea80d690756634471d33","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4/#blk_9acea719-ec9e-44b0-a5d7-1dbf2ae51cfa"},{"id":"occ_d49d1119fb9a5e527df8a048","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","work_id":"wrk_95f3adb2-9ef8-4c60-8a3b-5ded0268dac4","block_id":"blk_9c675ec7-0002-4e7f-955b-553b7e2ba696","section_id":"sec_4d53ceec-4948-4e97-bc81-cbcb507d7d8d","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":202,"end":209,"exact":"agentic","quote":"品」と説明しています。Veniceはクラウド、エンタープライズ、HPC、AIインフラ向けに位置づけられ、AIがagentic workloadsへ広がることでCPUの重要性が増す、とAMDは強調しています。([AMD](https://www.amd.com/en/newsroom/press-releases/2026-","quote_start":147,"quote_end":309,"text_sha256":"5d3f135d5d4a16be6bca68ed8cc1d7be6fde3ec97cd05aeea96d0acb758c1e8a","block_sha256":"5d3f135d5d4a16be6bca68ed8cc1d7be6fde3ec97cd05aeea96d0acb758c1e8a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4/#blk_9c675ec7-0002-4e7f-955b-553b7e2ba696"},{"id":"occ_6f4c27fffbced4e29d3bc18f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","work_id":"wrk_95f3adb2-9ef8-4c60-8a3b-5ded0268dac4","block_id":"blk_a43b8789-482e-4fb5-bf1e-ae2212c2417c","section_id":"sec_a23c9343-6d7a-40ae-af1c-7a53127eaadc","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":10,"end":16,"exact":"エージェント","quote":"理由は単純です。AIエージェントは、GPUで文章を生成するだけではありません。","quote_start":0,"quote_end":39,"text_sha256":"8b42c537a7e8d7064e91236c4f9afca6b384a9393ad4cb5b5439405a9e04ac42","block_sha256":"8b42c537a7e8d7064e91236c4f9afca6b384a9393ad4cb5b5439405a9e04ac42","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4/#blk_a43b8789-482e-4fb5-bf1e-ae2212c2417c"},{"id":"occ_a701f21e6e3314f21aaa505c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","work_id":"wrk_95f3adb2-9ef8-4c60-8a3b-5ded0268dac4","block_id":"blk_a89e8aea-3183-4ad2-a9a3-64ee8de7d76c","section_id":"sec_73c88c5c-d00f-4152-a231-fb3b751d4978","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":86,"end":93,"exact":"agentic","quote":"性より、Rubin GPUラック全体の実効性能を最大化するCPUです。GPUを待たせないこと、KVキャッシュやagentic AI制御を回すこと、NVLink-C2CでCPUとGPUを一体化することに価値があります。","quote_start":31,"quote_end":139,"text_sha256":"939788ca91850969a573e7da41c527e4a332bea752f108f418619bf2ac2b90b6","block_sha256":"939788ca91850969a573e7da41c527e4a332bea752f108f418619bf2ac2b90b6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4/#blk_a89e8aea-3183-4ad2-a9a3-64ee8de7d76c"},{"id":"occ_9ec301c75f27384cd5369275","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","work_id":"wrk_95f3adb2-9ef8-4c60-8a3b-5ded0268dac4","block_id":"blk_a960b2f1-8b48-48e7-b002-872d56852e50","section_id":"sec_1635d7ea-f4aa-4fa4-9d14-09326e2abbef","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":54,"end":60,"exact":"エージェント","quote":"```\nGPUへのデータ供給\n推論リクエストのスケジューリング\nKVキャッシュ管理\n強化学習環境の実行\nAIエージェントのツール実行\nGPU-GPU通信の制御\nDPU/NIC/ストレージ制御\nマルチテナント分離\np99レイテンシ管理\n```","quote_start":0,"quote_end":120,"text_sha256":"303e0e6424ff38591fc3be6368c8ed0dcdbdff1fefa1891acc610ab2d8a266c5","block_sha256":"303e0e6424ff38591fc3be6368c8ed0dcdbdff1fefa1891acc610ab2d8a266c5","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4/#blk_a960b2f1-8b48-48e7-b002-872d56852e50"},{"id":"occ_43d9013da1c54dee2cd24c61","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","work_id":"wrk_95f3adb2-9ef8-4c60-8a3b-5ded0268dac4","block_id":"blk_b1638c8d-cd82-4c82-b134-edbdf49cf86a","section_id":"sec_a23c9343-6d7a-40ae-af1c-7a53127eaadc","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":111,"end":117,"exact":"エージェント","quote":"はGPU、TPU、NPU、ASICです。  \nしかし、AIデータセンター、つまりAIDCが「単発推論」から**エージェントAI、RAG、ツール実行、コード実行、強化学習、マルチテナント推論**へ進むほど、CPUの重要性はむしろ上がっています。","quote_start":56,"quote_end":177,"text_sha256":"b075b7247a8fc7c2d1831a8ed2fb07ae3ffb5297b0d9ae08bd5214f173bc4ab7","block_sha256":"b075b7247a8fc7c2d1831a8ed2fb07ae3ffb5297b0d9ae08bd5214f173bc4ab7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4/#blk_b1638c8d-cd82-4c82-b134-edbdf49cf86a"},{"id":"occ_f8c4024eab1c59603c4137d4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","work_id":"wrk_95f3adb2-9ef8-4c60-8a3b-5ded0268dac4","block_id":"blk_bf75a484-198e-4a43-a195-2c3926a24415","section_id":"sec_8dd3e000-30a9-46e4-8f9a-1510e88c3fdb","layer":"code","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":70,"end":77,"exact":"agentic","quote":"NVLink-C2C\nLPDDR5X/SOCAMMによるGPU HBM補助\nKVキャッシュ管理\n強化学習環境\nagentic AIのサンドボックス実行\nNVIDIAラック全体のGPU稼働率最大化\n```","quote_start":15,"quote_end":116,"text_sha256":"5c0ef6847e367f07483c0e0157f4cb5fb51be7347fedd3f5314942e174f0b85a","block_sha256":"5c0ef6847e367f07483c0e0157f4cb5fb51be7347fedd3f5314942e174f0b85a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4/#blk_bf75a484-198e-4a43-a195-2c3926a24415"},{"id":"occ_fcba8c117e13e61f0f85795e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","work_id":"wrk_95f3adb2-9ef8-4c60-8a3b-5ded0268dac4","block_id":"blk_c157daba-d987-47a8-9e83-ded3c9111ab9","section_id":"sec_ed68ba05-8997-4209-854e-f87bece86dfb","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェント時代には、GPUやAI ASICだけでなく、CPU需要も拡大します。Armは、AIが学習・単発推論からエージェント型ワークロードへ進むと、CPUがreasoning、coordination、data","quote_start":0,"quote_end":108,"text_sha256":"526ff3e6e423a23c1347c0f8a1f9ff5c2482d044f4c6034b586e91c91da1f501","block_sha256":"526ff3e6e423a23c1347c0f8a1f9ff5c2482d044f4c6034b586e91c91da1f501","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4/#blk_c157daba-d987-47a8-9e83-ded3c9111ab9"},{"id":"occ_0e14e8175ebe04770b37926f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","work_id":"wrk_95f3adb2-9ef8-4c60-8a3b-5ded0268dac4","block_id":"blk_c27baa4b-8594-46aa-9b43-3b689235f832","section_id":"sec_9374547f-f098-42cb-9e7b-85fe7e9749e4","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":17,"end":23,"exact":"エージェント","quote":"```\nGPUを止めないこと\nAIエージェントの周辺処理を回すこと\nメモリ帯域を確保すること\nI/Oを太くすること\nクラウドTCOを下げること\nソフトウェア資産と互換性を守ること\n```","quote_start":0,"quote_end":93,"text_sha256":"3bbb4ce6dcc4cee92a5536b49a7489fc5d3f3780823db44c70823e43cc92bb80","block_sha256":"3bbb4ce6dcc4cee92a5536b49a7489fc5d3f3780823db44c70823e43cc92bb80","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4/#blk_c27baa4b-8594-46aa-9b43-3b689235f832"},{"id":"occ_857ab7649c35db908bc974e9","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_94f3d3d6-6b52-4f98-89a0-318fe428f5cb","work_id":"wrk_43737b34-e6d4-44d7-b140-1cd6aacb15fb","block_id":"blk_1030f69d-33bb-41e9-9459-77007a345a3b","section_id":"sec_f02d1670-2323-408b-9748-eaabb8e3a938","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":21,"end":27,"exact":"エージェント","quote":"Bedrock AgentCoreは、AIエージェントを企業で稼働させるための基盤である。","quote_start":0,"quote_end":45,"text_sha256":"b23ef2370f01482f142d81aa1d0446350ca5dbf21509fc5f95c383b625a83e93","block_sha256":"b23ef2370f01482f142d81aa1d0446350ca5dbf21509fc5f95c383b625a83e93","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_94f3d3d6-6b52-4f98-89a0-318fe428f5cb/#blk_1030f69d-33bb-41e9-9459-77007a345a3b"},{"id":"occ_dc18fc779e8138ffed2f3ff2","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_94f3d3d6-6b52-4f98-89a0-318fe428f5cb","work_id":"wrk_43737b34-e6d4-44d7-b140-1cd6aacb15fb","block_id":"blk_18d41b3a-43eb-41f8-bb3d-27dc184d4b89","section_id":"sec_52892658-24d9-4c05-beb9-3d253effd0fd","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":13,"end":19,"exact":"エージェント","quote":"## Bedrockと企業エージェント","quote_start":0,"quote_end":19,"text_sha256":"e207bf4cc194bf015435b393f794ace4fd897f2335ed26ef455a8f4082e860d6","block_sha256":"e207bf4cc194bf015435b393f794ace4fd897f2335ed26ef455a8f4082e860d6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_94f3d3d6-6b52-4f98-89a0-318fe428f5cb/#blk_18d41b3a-43eb-41f8-bb3d-27dc184d4b89"},{"id":"occ_5352c8dab880c4306d4726b4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_94f3d3d6-6b52-4f98-89a0-318fe428f5cb","work_id":"wrk_43737b34-e6d4-44d7-b140-1cd6aacb15fb","block_id":"blk_6953375d-2410-4d64-bd72-57c8a496ef2c","section_id":"sec_1318ad02-6f46-4359-9fd2-312794bdcdbb","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":13,"end":19,"exact":"エージェント","quote":"しかし今回の発表では、AIエージェントや企業向けアプリケーションへ明確に踏み込んでいる。","quote_start":0,"quote_end":44,"text_sha256":"9622753b10629c2cb9acff428f4f4f9719fa5e5637b24c0011ba8ceb1b2011e1","block_sha256":"9622753b10629c2cb9acff428f4f4f9719fa5e5637b24c0011ba8ceb1b2011e1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_94f3d3d6-6b52-4f98-89a0-318fe428f5cb/#blk_6953375d-2410-4d64-bd72-57c8a496ef2c"},{"id":"occ_b7ec460c5dc854b438a272f7","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_94f3d3d6-6b52-4f98-89a0-318fe428f5cb","work_id":"wrk_43737b34-e6d4-44d7-b140-1cd6aacb15fb","block_id":"blk_9f26b859-6aea-429c-a4a6-6476ac08a0d5","section_id":"sec_dfb79f82-94a7-41c2-9c2b-2b2bf2450000","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":27,"end":33,"exact":"エージェント","quote":"- 学習後のポストトレーニング\n- 強化学習\n- AIエージェントのツール実行\n- CPU処理\n- ストレージ\n- ベクトルデータベース\n- 社内データ検索\n- API接続\n- ログ保存\n- 監視\n- セキュリティ","quote_start":0,"quote_end":107,"text_sha256":"b4c66a9c78dfd09ead4c2b16c9d4633e9af50d00ae54dc98e948ae216bf4d5b1","block_sha256":"b4c66a9c78dfd09ead4c2b16c9d4633e9af50d00ae54dc98e948ae216bf4d5b1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_94f3d3d6-6b52-4f98-89a0-318fe428f5cb/#blk_9f26b859-6aea-429c-a4a6-6476ac08a0d5"},{"id":"occ_ab6329c11fb828ff50eaf0b2","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_94f3d3d6-6b52-4f98-89a0-318fe428f5cb","work_id":"wrk_43737b34-e6d4-44d7-b140-1cd6aacb15fb","block_id":"blk_cbd47e56-a6d6-4fdd-aacb-d39593497b0c","section_id":"sec_b42d94ad-cb08-48fb-82de-01bf56f2dce6","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":34,"end":40,"exact":"エージェント","quote":"Kiroは仕様書からコード作成、テスト、実装までを進めるコーディングエージェントで、利用量は前四半期比3倍となった。Amazonは競合サービスより最大50％費用対効果が高いと説明している。([Q4 Capital](https://s2.q4cdn.com/299287126/f","quote_start":0,"quote_end":140,"text_sha256":"b46f521897480977c7048d82d5a763fbb6350da35f397d4c0d5c03a5c01d4be7","block_sha256":"b46f521897480977c7048d82d5a763fbb6350da35f397d4c0d5c03a5c01d4be7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_94f3d3d6-6b52-4f98-89a0-318fe428f5cb/#blk_cbd47e56-a6d6-4fdd-aacb-d39593497b0c"},{"id":"occ_deeab118c5ed36fd866c8fa8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_94f3d3d6-6b52-4f98-89a0-318fe428f5cb","work_id":"wrk_43737b34-e6d4-44d7-b140-1cd6aacb15fb","block_id":"blk_d2c811ac-4d4d-4349-ace6-15ee1d27657c","section_id":"sec_f02d1670-2323-408b-9748-eaabb8e3a938","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"Q2には、エージェントが自律的に取引を実行するPaymentsや、AWS内でWeb検索を行う機能などが追加された。([Q4 Capital](https://s2.q4cdn.com/299287126/files/doc","quote_start":0,"quote_end":111,"text_sha256":"36a958f034e09273744a4ecd143983741ac25b536f5edfccdd3beecf95d3ee60","block_sha256":"36a958f034e09273744a4ecd143983741ac25b536f5edfccdd3beecf95d3ee60","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_94f3d3d6-6b52-4f98-89a0-318fe428f5cb/#blk_d2c811ac-4d4d-4349-ace6-15ee1d27657c"},{"id":"occ_3566de1ac03c958ad19d3844","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_94f3d3d6-6b52-4f98-89a0-318fe428f5cb","work_id":"wrk_43737b34-e6d4-44d7-b140-1cd6aacb15fb","block_id":"blk_ea5ebc4c-b2c5-4c27-aabb-2babfe976d6c","section_id":"sec_52892658-24d9-4c05-beb9-3d253effd0fd","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":21,"end":27,"exact":"エージェント","quote":"モデル利用だけでなく、企業の本番業務へAIエージェントが広がっているか。Amazon Quick、AgentCore、Kiro、Continuumの採用拡大が焦点となる。","quote_start":0,"quote_end":85,"text_sha256":"4282ec0de7259eee142fb0836f637299d6e549cccfeaa08b79c079865dfc5b98","block_sha256":"4282ec0de7259eee142fb0836f637299d6e549cccfeaa08b79c079865dfc5b98","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_94f3d3d6-6b52-4f98-89a0-318fe428f5cb/#blk_ea5ebc4c-b2c5-4c27-aabb-2babfe976d6c"},{"id":"occ_e5dcb9e0e000499d115b2d1a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_977092a4-68df-47ac-8de2-4e66567c8a95","work_id":"wrk_27b36bee-da31-4942-a5ea-33f6f77ddee3","block_id":"blk_07b601b6-abbe-419b-aaa9-6c460da91b6c","section_id":"sec_06af27ec-4b83-4660-932c-f6e5a9863113","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":7,"end":13,"exact":"エージェント","quote":"生成AI、AIエージェント、クラウドAI、コード生成、画像・動画生成、業務自動化、検索、広告、研究支援、データ分析。これらは一時的な流行ではなく、すでに社会の中に入り込み始めている。今後もAIを使ったサービス、サブスク、従量課","quote_start":0,"quote_end":113,"text_sha256":"d5edf6d6728f9d1329fb8de32b06a3e8c6d99da0be3ca254dde6ac6eda7fe712","block_sha256":"d5edf6d6728f9d1329fb8de32b06a3e8c6d99da0be3ca254dde6ac6eda7fe712","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_977092a4-68df-47ac-8de2-4e66567c8a95/#blk_07b601b6-abbe-419b-aaa9-6c460da91b6c"},{"id":"occ_c16b8178122847b907deb8a3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_977092a4-68df-47ac-8de2-4e66567c8a95","work_id":"wrk_27b36bee-da31-4942-a5ea-33f6f77ddee3","block_id":"blk_8a475e79-0845-43fb-bcc5-17ae3d8bb64d","section_id":"sec_c9a529c7-7d4d-44ee-bb08-1302eee643e9","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"AI研究者エージェントが仮説を立てる。  \n別のAIが論文を読み、既存研究を整理する。  \n別のAIがシミュレーションを走らせる。  \n別のAIがコードを書き、テストし、改善する。  \n別のAIが半導体設計や材料探索を行う","quote_start":0,"quote_end":111,"text_sha256":"cd186fb52de52bbaad3e11b8236524ec59646f3aeca678ea101cd871227b8f44","block_sha256":"cd186fb52de52bbaad3e11b8236524ec59646f3aeca678ea101cd871227b8f44","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_977092a4-68df-47ac-8de2-4e66567c8a95/#blk_8a475e79-0845-43fb-bcc5-17ae3d8bb64d"},{"id":"occ_b8eaf7d264b664511188d586","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_977092a4-68df-47ac-8de2-4e66567c8a95","work_id":"wrk_27b36bee-da31-4942-a5ea-33f6f77ddee3","block_id":"blk_e2dd80f3-ad7e-4a08-a790-89f2ceeee648","section_id":"sec_95a36c2e-ffd9-4876-bc5b-fe9ecc1b21cd","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":32,"end":38,"exact":"エージェント","quote":"これが金融商品化され、先物化され、クラウドクレジット化され、AIエージェントによって自動配分されるようになれば、AI独自通貨に近いものが生まれる。","quote_start":0,"quote_end":73,"text_sha256":"68e23274da6f4acde7b44f3590ca62e7cc7fa5812febd32c058faf32e72eda1c","block_sha256":"68e23274da6f4acde7b44f3590ca62e7cc7fa5812febd32c058faf32e72eda1c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_977092a4-68df-47ac-8de2-4e66567c8a95/#blk_e2dd80f3-ad7e-4a08-a790-89f2ceeee648"},{"id":"occ_95a37f23e0ef9932d06235a4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_98f85497-697b-4917-aabc-df1485131d9b","work_id":"wrk_23480985-b0bd-47ea-a46d-716d46f4d122","block_id":"blk_02e2e3a0-9781-4a7a-8cd7-16e20e74694b","section_id":"sec_65f5ec1a-74a1-4fcf-97ec-30885b19c57c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"- エージェントがどのAPIにアクセスできるのか\n- 認証情報をどう守るのか\n- 外部通信をどう制御するのか\n- MCPやツール呼び出しをどう監査するのか\n- モデルごとのコストやログをどう管理するのか\n- 不正なボ","quote_start":0,"quote_end":108,"text_sha256":"6a1075b728c86613fefcf2504d648f00d291b906e423476d7c10be8aa7e0d4ee","block_sha256":"6a1075b728c86613fefcf2504d648f00d291b906e423476d7c10be8aa7e0d4ee","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_98f85497-697b-4917-aabc-df1485131d9b/#blk_02e2e3a0-9781-4a7a-8cd7-16e20e74694b"},{"id":"occ_fd94742ff8e0192f116a370e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_98f85497-697b-4917-aabc-df1485131d9b","work_id":"wrk_23480985-b0bd-47ea-a46d-716d46f4d122","block_id":"blk_28101dd7-28c8-49a1-8ac3-830e666ec20c","section_id":"sec_7d2cc3b3-9593-42ce-9e05-1480517773a1","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":26,"end":32,"exact":"エージェント","quote":"しかし長期で見ると、CloudflareはむしろAIエージェントが増えるほど必要になる可能性がある。AIエージェントが企業ネットワーク、API、データベース、ファイル、SaaS、ロボット、センサーへアクセスする時代には、誰かが通信を認証し、制御し、記録し、守らなけ","quote_start":0,"quote_end":132,"text_sha256":"b3f7a7d507defca819b7a82e1548a657776fdb03f18d3592af15fed55c7df3db","block_sha256":"b3f7a7d507defca819b7a82e1548a657776fdb03f18d3592af15fed55c7df3db","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_98f85497-697b-4917-aabc-df1485131d9b/#blk_28101dd7-28c8-49a1-8ac3-830e666ec20c"},{"id":"occ_18ac44bd1135432a02781e12","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_98f85497-697b-4917-aabc-df1485131d9b","work_id":"wrk_23480985-b0bd-47ea-a46d-716d46f4d122","block_id":"blk_3aab9314-9173-4cf8-b4e9-913724d06de3","section_id":"sec_04e752a6-6419-4bc2-af6a-6dc0648708ba","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":66,"end":72,"exact":"エージェント","quote":"flareのような銘柄は一時的に不利になる可能性がある。なぜなら、Cloudflareの長期ストーリーは「AIエージェント時代の関所」だが、OpenAIやAnthropicの上場ストーリーはもっと単純で強いからだ。","quote_start":11,"quote_end":118,"text_sha256":"6d25e060db88b79d2c546bbf9967974ff1dfb1d27786c3fa6f007f8ca8227813","block_sha256":"6d25e060db88b79d2c546bbf9967974ff1dfb1d27786c3fa6f007f8ca8227813","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_98f85497-697b-4917-aabc-df1485131d9b/#blk_3aab9314-9173-4cf8-b4e9-913724d06de3"},{"id":"occ_646bb56276971c16e6510cb8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_98f85497-697b-4917-aabc-df1485131d9b","work_id":"wrk_23480985-b0bd-47ea-a46d-716d46f4d122","block_id":"blk_3e1d4557-3f89-41d4-b75a-f439839f819b","section_id":"sec_0a3e979b-c7ec-483d-84a0-87b8a6b72bc6","layer":"body","character_id":null,"count":2,"matched_aliases":["agentic","エージェント"],"evidence":{"text_basis":"markdown","start":47,"end":53,"exact":"エージェント","quote":"第二に、**約20%の人員削減**。Cloudflareは1,100人超の削減を発表し、AIとエージェントを前提にした運営モデルへ移行すると説明した。会社側は「単なるコスト削減ではなく、agentic AI-first operating modelへの再設計」と説明しているが、市場はこれを「成長鈍化への","quote_start":0,"quote_end":153,"text_sha256":"3e3326cf3e0ccbed93fb90bfc53710a7fafb9c8850b326a9a41cadd1cbb63e15","block_sha256":"3e3326cf3e0ccbed93fb90bfc53710a7fafb9c8850b326a9a41cadd1cbb63e15","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_98f85497-697b-4917-aabc-df1485131d9b/#blk_3e1d4557-3f89-41d4-b75a-f439839f819b"},{"id":"occ_4d259acb3d91bf21443a5f36","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_98f85497-697b-4917-aabc-df1485131d9b","work_id":"wrk_23480985-b0bd-47ea-a46d-716d46f4d122","block_id":"blk_4022f222-4e69-4592-bc94-bf2fad5c8c02","section_id":"sec_65f5ec1a-74a1-4fcf-97ec-30885b19c57c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":105,"end":111,"exact":"エージェント","quote":"S防御、WAF、Zero Trust、SASE、Workers、AI Gateway、開発者プラットフォーム、エージェント実行環境である。","quote_start":50,"quote_end":119,"text_sha256":"55a9996539b1e7ee9069bd430e26f4dc70984d7ce191a41fa7439b08b4a7bcae","block_sha256":"55a9996539b1e7ee9069bd430e26f4dc70984d7ce191a41fa7439b08b4a7bcae","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_98f85497-697b-4917-aabc-df1485131d9b/#blk_4022f222-4e69-4592-bc94-bf2fad5c8c02"},{"id":"occ_24492af95d824dcfaf84942f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_98f85497-697b-4917-aabc-df1485131d9b","work_id":"wrk_23480985-b0bd-47ea-a46d-716d46f4d122","block_id":"blk_42dc5b06-dece-47a1-9124-3c95883a54dc","section_id":"sec_65f5ec1a-74a1-4fcf-97ec-30885b19c57c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントが普及すると、企業には次の問題が生まれる。","quote_start":0,"quote_end":29,"text_sha256":"208c5890926f38e83a9dca406232b82d6f7292c204e31f53a6b7959c561a6b50","block_sha256":"208c5890926f38e83a9dca406232b82d6f7292c204e31f53a6b7959c561a6b50","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_98f85497-697b-4917-aabc-df1485131d9b/#blk_42dc5b06-dece-47a1-9124-3c95883a54dc"},{"id":"occ_f49ef30f3e4cfead0682b570","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_98f85497-697b-4917-aabc-df1485131d9b","work_id":"wrk_23480985-b0bd-47ea-a46d-716d46f4d122","block_id":"blk_5a4a677e-e8c7-4ab0-b40b-3330cb49e655","section_id":"sec_7b558531-d7d5-4bd6-8447-8c197c8fcf71","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":131,"end":137,"exact":"エージェント","quote":"リティ投資家から見ると、MythosのようなAIによる代替リスクが怖い。  \nしかし長期投資家から見ると、AIエージェント時代の通信・認証・防御・実行基盤として、かなり美しい位置にいる。","quote_start":76,"quote_end":169,"text_sha256":"a47c33432316b061ac211ee8f6826576dc3617a4e744489bb056657ed82d7975","block_sha256":"a47c33432316b061ac211ee8f6826576dc3617a4e744489bb056657ed82d7975","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_98f85497-697b-4917-aabc-df1485131d9b/#blk_5a4a677e-e8c7-4ab0-b40b-3330cb49e655"},{"id":"occ_d6a0b58ab08672958df3c210","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_98f85497-697b-4917-aabc-df1485131d9b","work_id":"wrk_23480985-b0bd-47ea-a46d-716d46f4d122","block_id":"blk_7205c634-1daa-47df-acf7-55874b0455cc","section_id":"sec_96dc6472-3ab8-4669-82c1-613658f5c7db","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":26,"end":32,"exact":"エージェント","quote":"現在の粗利低下の主因は開発者プラットフォームやAI/エージェント系プロダクトのミックス変化、ネットワーク原価、サーバー投資の影響と見る方が自然だ。([The Cloudflare Blog](https://blog.cloudflare.com/post-qua","quote_start":0,"quote_end":132,"text_sha256":"a22d639ce56c24a5f73bb2ddc4bb052b60316c9bbf4845107386160bfdc0f0a1","block_sha256":"a22d639ce56c24a5f73bb2ddc4bb052b60316c9bbf4845107386160bfdc0f0a1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_98f85497-697b-4917-aabc-df1485131d9b/#blk_7205c634-1daa-47df-acf7-55874b0455cc"},{"id":"occ_a13f62e08903983072354833","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_98f85497-697b-4917-aabc-df1485131d9b","work_id":"wrk_23480985-b0bd-47ea-a46d-716d46f4d122","block_id":"blk_80915283-5375-438f-9aff-827f1bb8e663","section_id":"sec_35277fea-f9ab-44b7-82ad-775a9c27bd50","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":15,"end":21,"exact":"エージェント","quote":"なぜなら、市場ではすでに「AIエージェントがSaaSを代替する」という物語が走っていたからだ。Fortuneは、AnthropicのClaudeのようなAIモデルが専門業務ソフトを不要にするのではないか、という投資家の見方を報じている。つまり","quote_start":0,"quote_end":121,"text_sha256":"cf0772c8fbdaf745fcce57e8bf7c6529fd642e1d710649ab1e2968d8ccc37635","block_sha256":"cf0772c8fbdaf745fcce57e8bf7c6529fd642e1d710649ab1e2968d8ccc37635","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_98f85497-697b-4917-aabc-df1485131d9b/#blk_80915283-5375-438f-9aff-827f1bb8e663"},{"id":"occ_9027612b3530adc6b0d36159","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_98f85497-697b-4917-aabc-df1485131d9b","work_id":"wrk_23480985-b0bd-47ea-a46d-716d46f4d122","block_id":"blk_83c18a79-905f-4cd5-b02b-0e09a2052a9b","section_id":"sec_96dc6472-3ab8-4669-82c1-613658f5c7db","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":86,"end":92,"exact":"エージェント","quote":"ら、Workers、AI Gateway、R2、D1、Durable Objects、Sandboxes、AIエージェント基盤へ広がっている。これは将来性は大きいが、短期的には計算資源、ストレージ、ネットワーク、サーバー減価償却、第三者技術サービス費などが増えやすい。Q1では粗利率が前年から明確に下がっており、市場は「A","quote_start":31,"quote_end":192,"text_sha256":"9e43169a350e34ed8bbe793c8628e2dc08dcfb395f8e652f95457f2dbc397bca","block_sha256":"9e43169a350e34ed8bbe793c8628e2dc08dcfb395f8e652f95457f2dbc397bca","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_98f85497-697b-4917-aabc-df1485131d9b/#blk_83c18a79-905f-4cd5-b02b-0e09a2052a9b"},{"id":"occ_4aca9ba11585c5234593ccad","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_98f85497-697b-4917-aabc-df1485131d9b","work_id":"wrk_23480985-b0bd-47ea-a46d-716d46f4d122","block_id":"blk_9ea0b54b-5ea2-472d-9623-42d8e7f93abf","section_id":"sec_7b558531-d7d5-4bd6-8447-8c197c8fcf71","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":53,"end":59,"exact":"エージェント","quote":"Cloudflareを長期で見るなら、今回の急落は「事業が死んだ」わけではない。むしろ事業の方向性は、AIエージェント、ゼロトラスト、エッジコンピュート、PQCというかなり重要な領域に向かっている。","quote_start":0,"quote_end":99,"text_sha256":"2b9c520f09141ef25f9369e56be36f0a861f58407271ccd4c2535e6402a92f8e","block_sha256":"2b9c520f09141ef25f9369e56be36f0a861f58407271ccd4c2535e6402a92f8e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_98f85497-697b-4917-aabc-df1485131d9b/#blk_9ea0b54b-5ea2-472d-9623-42d8e7f93abf"},{"id":"occ_0b4b29593e92a74853cb0baa","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_98f85497-697b-4917-aabc-df1485131d9b","work_id":"wrk_23480985-b0bd-47ea-a46d-716d46f4d122","block_id":"blk_d1c64c3d-304c-4912-a9cb-505e405bd490","section_id":"sec_7fca16e0-3f25-4c1d-9063-9cc638099ead","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントが増える。  \nAPI通信が増える。  \nBotと人間の境界が曖昧になる。  \n企業はAIの通信、認証、権限、ログ、コスト、セキュリティを制御したくなる。  \n量子時代には通信の暗号化・認証も作り替え","quote_start":0,"quote_end":108,"text_sha256":"6e9491146397a2e851f32be4051b0257f2dca98b03cf286f3ff9426baa5ef085","block_sha256":"6e9491146397a2e851f32be4051b0257f2dca98b03cf286f3ff9426baa5ef085","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_98f85497-697b-4917-aabc-df1485131d9b/#blk_d1c64c3d-304c-4912-a9cb-505e405bd490"},{"id":"occ_4a689c4ff43b55bb7d94ffe6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_98f85497-697b-4917-aabc-df1485131d9b","work_id":"wrk_23480985-b0bd-47ea-a46d-716d46f4d122","block_id":"blk_fb4fa44f-3fc5-470d-8ee1-cbf5e70c9c8c","section_id":"sec_65f5ec1a-74a1-4fcf-97ec-30885b19c57c","layer":"body","character_id":null,"count":3,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":109,"end":115,"exact":"エージェント","quote":"s、Artifacts、Workflows、Outbound Workers for Sandboxesなど、エージェント時代向けの基盤が発表された。特にOutbound Workers for Sandboxesは、AIエージェントに認証情報を直接渡さず、ゼロトラスト型の外部通信制御を行う発想であり、エージェントが自律","quote_start":54,"quote_end":215,"text_sha256":"61c36fc7ad6aaca5311d6f22e127da429a854186617ab6d44a5a5d66adbf17fe","block_sha256":"61c36fc7ad6aaca5311d6f22e127da429a854186617ab6d44a5a5d66adbf17fe","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_98f85497-697b-4917-aabc-df1485131d9b/#blk_fb4fa44f-3fc5-470d-8ee1-cbf5e70c9c8c"},{"id":"occ_27620229cee384edf7c981a0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_98f85497-697b-4917-aabc-df1485131d9b","work_id":"wrk_23480985-b0bd-47ea-a46d-716d46f4d122","block_id":"blk_fddf5541-3199-4688-af1c-ea4edb1422bb","section_id":"sec_7d2cc3b3-9593-42ce-9e05-1480517773a1","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":62,"end":68,"exact":"エージェント","quote":"loudflareは短期ではSaaS売りとLLM上場観測に巻き込まれるリスクが高い。  \n一方で、長期ではAIエージェント、PQC、Zero Trust、SASE、エッジコンピュートの交差点にいる。","quote_start":7,"quote_end":106,"text_sha256":"264711a23070f4d432ec3362efbed576a31c46e5b0f9cce991b9324c7fb92841","block_sha256":"264711a23070f4d432ec3362efbed576a31c46e5b0f9cce991b9324c7fb92841","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_98f85497-697b-4917-aabc-df1485131d9b/#blk_fddf5541-3199-4688-af1c-ea4edb1422bb"},{"id":"occ_bc971794bb747581647472d9","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_9ae4d311-bf43-4bb1-a7b5-36a5649efc7c","work_id":"wrk_9cfc75d7-7e8e-4397-add6-437f60011804","block_id":"blk_4e44c0af-5bc9-4567-9c69-563fb49718c3","section_id":"sec_1c9bc0f5-1546-4c46-847a-9ae8e8733907","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":782,"end":788,"exact":"エージェント","quote":"ます。Habitatプラットフォームは3D仮想環境でロボットが家の中を探索・物体操作するタスクを訓練でき、AIエージェントの環境適応力を高める試みです。またMeta傘下の研究では**「OpenVLA」\\*\\*というオープンソースのビジョン・言語・行動モデルが2024年に公開され、RT-2に倣って画像＋テキストからロボット","quote_start":727,"quote_end":888,"text_sha256":"e8398cdb6d6e8c1ab2692ec6b62f114268cc931bc3d096be98a4c5b8951ba730","block_sha256":"e8398cdb6d6e8c1ab2692ec6b62f114268cc931bc3d096be98a4c5b8951ba730","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_9ae4d311-bf43-4bb1-a7b5-36a5649efc7c/#blk_4e44c0af-5bc9-4567-9c69-563fb49718c3"},{"id":"occ_25e9090212bc8cbf89a59c5d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_9ae4d311-bf43-4bb1-a7b5-36a5649efc7c","work_id":"wrk_9cfc75d7-7e8e-4397-add6-437f60011804","block_id":"blk_a192619d-370c-4f61-a58e-6cd163ee2626","section_id":"sec_bcf933be-70cb-4516-b831-9d97412b96b3","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":24,"end":30,"exact":"エージェント","quote":"- **Meta Habitat**: 具身AIエージェント向けの高速シミュレータ。Isaac Simが「物理的な正確さとレンダリング品質」を重視するのに対し、Habitatは「学習速度（FPS）」を重視し、ナビゲーションタスクなどの強化学習において圧倒的な速","quote_start":0,"quote_end":130,"text_sha256":"6e447787de274b7dbe1aaba72d1743ecf9a5c25920b71ff090eee261f3eaacff","block_sha256":"6e447787de274b7dbe1aaba72d1743ecf9a5c25920b71ff090eee261f3eaacff","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_9ae4d311-bf43-4bb1-a7b5-36a5649efc7c/#blk_a192619d-370c-4f61-a58e-6cd163ee2626"},{"id":"occ_dd3a4ad271f60014d46f5617","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_9ae4d311-bf43-4bb1-a7b5-36a5649efc7c","work_id":"wrk_9cfc75d7-7e8e-4397-add6-437f60011804","block_id":"blk_a6744a83-67b3-4c71-91d6-1295582b8237","section_id":"sec_1c9bc0f5-1546-4c46-847a-9ae8e8733907","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":282,"end":288,"exact":"エージェント","quote":"t」**という高精細な世界モデル生成AIが含まれ、最長30秒先までの環境変化を映像で予測することで、ロボットやエージェントのプランニングを支援します。また**「Cosmos Reason」**は7億パラメータ規模の視覚言語モデルで、物理法則や常識知識を活かしながらロボットに状況認識・推論させることができます。これらCos","quote_start":227,"quote_end":388,"text_sha256":"ba35448686eda75339fe591e5754c29747dffb86f71cceec73f3291c3037fc8c","block_sha256":"ba35448686eda75339fe591e5754c29747dffb86f71cceec73f3291c3037fc8c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_9ae4d311-bf43-4bb1-a7b5-36a5649efc7c/#blk_a6744a83-67b3-4c71-91d6-1295582b8237"},{"id":"occ_d71582fd11d24183f4369a8d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_9ae4d311-bf43-4bb1-a7b5-36a5649efc7c","work_id":"wrk_9cfc75d7-7e8e-4397-add6-437f60011804","block_id":"blk_e91024b2-f5cc-49ec-a9dc-367afa99afbd","section_id":"sec_77f1e8cd-d527-4459-bed1-bee14ebf5be4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":214,"end":220,"exact":"エージェント","quote":"と処理に特化してきた。これに対し、フィジカルAIは「身体性（Embodiment）」を核とする概念であり、AIエージェントが物理的な筐体（ロボット、自動運転車、スマートマシーン）を通じて実世界を認識し、推論し、そして物理的に介入して環境を変化させる能力を持つシステムと定義される。","quote_start":159,"quote_end":299,"text_sha256":"e092692b4a4f968343c6bb2092a78bef0fe604943aed9398e6e66b0534a13633","block_sha256":"e092692b4a4f968343c6bb2092a78bef0fe604943aed9398e6e66b0534a13633","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_9ae4d311-bf43-4bb1-a7b5-36a5649efc7c/#blk_e91024b2-f5cc-49ec-a9dc-367afa99afbd"},{"id":"occ_f7838da8a5568a2cc54e7098","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_9af30ed5-ec26-4204-9f65-58324ee5fba5","work_id":"wrk_c56e9825-8401-406e-aaae-662904613d26","block_id":"blk_6581bae7-b37d-4e24-b3d6-23ee0d384e62","section_id":"sec_f519167e-d09f-41c4-b2ff-eea6ca8ca688","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":41,"end":47,"exact":"エージェント","quote":"OpenAIが黒字に近づく道は5本あります。**サブスク、企業向け席課金/API、エージェント実行課金、広告、政府・主権案件**です。OpenAIはすでにAWSと組んで政府向けに売り始めており、“OpenAI for Countries”では各国向けデータセンターとカスタムChatGPTを提案","quote_start":0,"quote_end":147,"text_sha256":"c181502cd4b192320194d80234ed4e6a576c7fe2eef6daf3d9c9749b1d1a0285","block_sha256":"c181502cd4b192320194d80234ed4e6a576c7fe2eef6daf3d9c9749b1d1a0285","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_9af30ed5-ec26-4204-9f65-58324ee5fba5/#blk_6581bae7-b37d-4e24-b3d6-23ee0d384e62"},{"id":"occ_6b15083b4205f0147d307813","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_9af30ed5-ec26-4204-9f65-58324ee5fba5","work_id":"wrk_c56e9825-8401-406e-aaae-662904613d26","block_id":"blk_974dfddc-2aed-4561-961b-bc31e7647eff","section_id":"sec_b3fc6661-aa31-46b6-bd68-0996bf3edcb7","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":120,"end":126,"exact":"エージェント","quote":"*6.145億ドル**でAI需要が追い風ですが、まだ利益の安定感は大型プラットフォームほど強くありません。AIエージェント時代のネット基盤として面白い一方、期待倍率が大きくなりやすい銘柄です。([Reuters](https://www.reuters.com/business/cloudflare-forecasts-","quote_start":65,"quote_end":226,"text_sha256":"a33d393b906fabddf85d9658e1b7741d196970d70659a9811706947f9406db12","block_sha256":"a33d393b906fabddf85d9658e1b7741d196970d70659a9811706947f9406db12","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_9af30ed5-ec26-4204-9f65-58324ee5fba5/#blk_974dfddc-2aed-4561-961b-bc31e7647eff"},{"id":"occ_5040e1b0ce4f173a187bf0af","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_9af30ed5-ec26-4204-9f65-58324ee5fba5","work_id":"wrk_c56e9825-8401-406e-aaae-662904613d26","block_id":"blk_de455316-54af-4f4c-b1b7-0da755962a60","section_id":"sec_f519167e-d09f-41c4-b2ff-eea6ca8ca688","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":41,"end":47,"exact":"エージェント","quote":"私のシナリオでは、OpenAIの黒字化は3本です。**強気**は、広告・政府・企業エージェントが全部立ち上がり、2030年前後に黒字化。**ベース**は、企業/APIは伸びるが広告と主権案件はゆっくりで、2030〜2032年にキャッシュ黒字化。**弱気**は、オープンソースや価格競争で単価が下","quote_start":0,"quote_end":147,"text_sha256":"b4ed92343e8c0fc36dbf9c716b2c1aeab19a6146955ec4605e7683d9554d758a","block_sha256":"b4ed92343e8c0fc36dbf9c716b2c1aeab19a6146955ec4605e7683d9554d758a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_9af30ed5-ec26-4204-9f65-58324ee5fba5/#blk_de455316-54af-4f4c-b1b7-0da755962a60"},{"id":"occ_428f1e8b6271cf89f066a32e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_9c10c6f9-4055-4591-a368-74112c253a08","work_id":"wrk_a91b4c62-c6f7-43f1-8bc0-9224e28ea11a","block_id":"blk_8eb52957-0ab1-4238-8367-e02cece6a4f7","section_id":"sec_ea7b999b-06bd-48a4-9da0-2fdeff36b586","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":37,"end":43,"exact":"エージェント","quote":"そのため、市場参加者がAIの存在は理解していても、一人の利用者が複数のAIエージェントを常時稼働させることで、推論需要がどこまで増えるかまでは十分に織り込めていない可能性がある。","quote_start":0,"quote_end":89,"text_sha256":"2ca91826562293bf546cec795e61d3196c03a73ffc3aab3952dfab9d073b58fd","block_sha256":"2ca91826562293bf546cec795e61d3196c03a73ffc3aab3952dfab9d073b58fd","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_9c10c6f9-4055-4591-a368-74112c253a08/#blk_8eb52957-0ab1-4238-8367-e02cece6a4f7"},{"id":"occ_81c1664aa255a484cb16b4b7","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_9c10c6f9-4055-4591-a368-74112c253a08","work_id":"wrk_a91b4c62-c6f7-43f1-8bc0-9224e28ea11a","block_id":"blk_9996fda6-0eec-4c7a-a807-f6a8d20a3889","section_id":"sec_ea7b999b-06bd-48a4-9da0-2fdeff36b586","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":9,"end":15,"exact":"エージェント","quote":"さらに、作業AIやエージェントAIは、ChatGPT登場時ほど一度で凄さが伝わりにくい。","quote_start":0,"quote_end":44,"text_sha256":"156d84f7088ceb4796ab7af92b48f86602b011efd7ffd83a23d33e580cf83cb2","block_sha256":"156d84f7088ceb4796ab7af92b48f86602b011efd7ffd83a23d33e580cf83cb2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_9c10c6f9-4055-4591-a368-74112c253a08/#blk_9996fda6-0eec-4c7a-a807-f6a8d20a3889"},{"id":"occ_57d9b4b71d3c4159211737d4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_a30fdebb-6a5f-4491-8480-bc05cbb21f96","work_id":"wrk_97fc23be-d3e1-4a2a-986e-b33ae2fd0697","block_id":"blk_1c73daaa-d0e4-45b7-a25b-0aa703fb39f0","section_id":"sec_cd9a08d8-72d1-4d50-968c-c9f048a2212c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":7,"end":13,"exact":"エージェント","quote":"長い会話、AIエージェント、コーディング支援、企業内文書検索が増えると、KVキャッシュは非常に大きくなる。すべてをHBMに置けば高速だが、HBMは高価で容量も限られる。すべてを通常ストレージへ置けば、読み戻すまでに時間がかかり","quote_start":0,"quote_end":113,"text_sha256":"20860f8f0f3957572a7fb53ab98368a0d75c748e1dda1031faf3d2d0f81f787f","block_sha256":"20860f8f0f3957572a7fb53ab98368a0d75c748e1dda1031faf3d2d0f81f787f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_a30fdebb-6a5f-4491-8480-bc05cbb21f96/#blk_1c73daaa-d0e4-45b7-a25b-0aa703fb39f0"},{"id":"occ_faf5d35c4f600f03b25e045c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_a3e10132-4b92-414d-8744-acbeb6f0b294","work_id":"wrk_ab79e5be-0326-461e-868c-ec7e75b2a90e","block_id":"blk_2cb62a91-2dd3-4edb-af97-ef7d4fed6551","section_id":"sec_3bfbfd39-7e78-471f-bac3-d0763fab9add","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":82,"end":88,"exact":"エージェント","quote":"し、企業や開発者がAIを本格導入すると、重要になるのは単発の知能だけではない。API料金、トークン単価、長時間エージェント作業のコスト、コーディングツールの使いやすさ、社内ワークフローへの組み込みやすさが勝負になる。","quote_start":27,"quote_end":135,"text_sha256":"3c36740da1041d2151b8c56cdaf4089ff783dd1f102fb93f088b188ac3e0b07c","block_sha256":"3c36740da1041d2151b8c56cdaf4089ff783dd1f102fb93f088b188ac3e0b07c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_a3e10132-4b92-414d-8744-acbeb6f0b294/#blk_2cb62a91-2dd3-4edb-af97-ef7d4fed6551"},{"id":"occ_b3369afef9123dfe56f18b5e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_a3e10132-4b92-414d-8744-acbeb6f0b294","work_id":"wrk_ab79e5be-0326-461e-868c-ec7e75b2a90e","block_id":"blk_3861ea46-ad0b-40f9-a7d2-83bf90d66963","section_id":"sec_72851c4d-029f-485c-ac75-596f5bee2aed","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":14,"end":20,"exact":"エージェント","quote":"第四に、決済と商取引。\nAIエージェントが現実の購入、予約、契約、支払いを行えるか。","quote_start":0,"quote_end":42,"text_sha256":"fc04d038ccbe88fd09bfda178cb641c765141bb08d3137bb0d4e21f533550ceb","block_sha256":"fc04d038ccbe88fd09bfda178cb641c765141bb08d3137bb0d4e21f533550ceb","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_a3e10132-4b92-414d-8744-acbeb6f0b294/#blk_3861ea46-ad0b-40f9-a7d2-83bf90d66963"},{"id":"occ_ff0da06e2fc6bf799ce0e4f5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_a3e10132-4b92-414d-8744-acbeb6f0b294","work_id":"wrk_ab79e5be-0326-461e-868c-ec7e75b2a90e","block_id":"blk_51e4c548-c12f-4641-a13a-2acc858ac518","section_id":"sec_74b54acc-c558-4197-b24f-fe8823405c98","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":93,"end":99,"exact":"エージェント","quote":"。\n- 価格低下は短期的にモデル企業の利益率を圧迫し得るが、長期的には推論需要を増やす可能性がある。\n- AIエージェントが商取引へ入るには、本人確認、利用上限、トークン化、不正検知、チャージバックまで必要になる。\n- 10GW級AIデータセンター構想は、AI企業がソフトウェア企業から物理インフラ企業へ近づく流れを示す。","quote_start":38,"quote_end":199,"text_sha256":"bb105e74699f353cb7c2998f8ff7d6895cd5a423dd4f6f97a58edbd09f80d1a3","block_sha256":"bb105e74699f353cb7c2998f8ff7d6895cd5a423dd4f6f97a58edbd09f80d1a3","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_a3e10132-4b92-414d-8744-acbeb6f0b294/#blk_51e4c548-c12f-4641-a13a-2acc858ac518"},{"id":"occ_56624c691df2b3cbfa787811","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_a3e10132-4b92-414d-8744-acbeb6f0b294","work_id":"wrk_ab79e5be-0326-461e-868c-ec7e75b2a90e","block_id":"blk_55a7b1e1-af44-40be-a2f0-9a8ca277c32f","section_id":"sec_2ce0ff1d-cd43-4b01-9845-fcd2f1322cd1","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":17,"end":23,"exact":"エージェント","quote":"## VisaとOpenAI：AIエージェントが買い物をする時代","quote_start":0,"quote_end":32,"text_sha256":"e2d9bb2d6247f52d4e39c3fb92756b461256ebd9ad5d969c9850a50783520f4f","block_sha256":"e2d9bb2d6247f52d4e39c3fb92756b461256ebd9ad5d969c9850a50783520f4f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_a3e10132-4b92-414d-8744-acbeb6f0b294/#blk_55a7b1e1-af44-40be-a2f0-9a8ca277c32f"},{"id":"occ_7062f2be944ea1b20836dd1f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_a3e10132-4b92-414d-8744-acbeb6f0b294","work_id":"wrk_ab79e5be-0326-461e-868c-ec7e75b2a90e","block_id":"blk_6b05eaae-1900-4133-bef1-42e72bf71119","section_id":"sec_2ce0ff1d-cd43-4b01-9845-fcd2f1322cd1","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントが本当に「代理人」になるには、商品選定だけでなく、支払い、本人確認、利用上限、加盟店側の信頼、不正検知、返品、キャンセル、チャージバックまで扱う必要がある。","quote_start":0,"quote_end":86,"text_sha256":"2b2492cface35c974a88c796b2f53f8237d6f24f58ff53d16907b9a564edaf08","block_sha256":"2b2492cface35c974a88c796b2f53f8237d6f24f58ff53d16907b9a564edaf08","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_a3e10132-4b92-414d-8744-acbeb6f0b294/#blk_6b05eaae-1900-4133-bef1-42e72bf71119"},{"id":"occ_fdc8124b7580ac1094be89da","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_a3e10132-4b92-414d-8744-acbeb6f0b294","work_id":"wrk_ab79e5be-0326-461e-868c-ec7e75b2a90e","block_id":"blk_73d5a301-0250-4213-a93a-c3270522c206","section_id":"sec_72851c4d-029f-485c-ac75-596f5bee2aed","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":54,"end":60,"exact":"エージェント","quote":"AIが強くなるほど、安全保障と規制が必要になる。\nAIが普及するほど、価格は下がり、利用量は増える。\nAIがエージェント化するほど、決済と本人認証が必要になる。\nAIの利用量が増えるほど、電力とデータセンターが必要になる。\nAIインフラが巨大化するほど、国家と金融が関与する。","quote_start":0,"quote_end":138,"text_sha256":"1bbb4c1ea4af644b190ca9ebc868820a5702253dad572f47745b2db5e2c2f279","block_sha256":"1bbb4c1ea4af644b190ca9ebc868820a5702253dad572f47745b2db5e2c2f279","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_a3e10132-4b92-414d-8744-acbeb6f0b294/#blk_73d5a301-0250-4213-a93a-c3270522c206"},{"id":"occ_5fd07d28ae1ddf033c852651","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_a3e10132-4b92-414d-8744-acbeb6f0b294","work_id":"wrk_ab79e5be-0326-461e-868c-ec7e75b2a90e","block_id":"blk_84d2eb7b-0a8a-4f66-a1f3-14afbf8c7df1","section_id":"sec_5acdc0d2-e211-48b4-b7f4-228a7c6dc22f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":72,"end":78,"exact":"エージェント","quote":"連する論点 |\n| --- | --- | --- |\n| 能力 | AIは何を実行できるか | モデル性能、エージェント、コーディング |\n| 許可 | AIにどこまで任せるか | 安全装置、本人確認、決済、監査 |\n| 物理資源 | AIを何回動かせるか | GPU、電力、冷却、データセンター、金融 |","quote_start":17,"quote_end":171,"text_sha256":"42832f802309792f204b6fd397649a5a3dfd6e4c6e61b20a48046946b265befe","block_sha256":"42832f802309792f204b6fd397649a5a3dfd6e4c6e61b20a48046946b265befe","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_a3e10132-4b92-414d-8744-acbeb6f0b294/#blk_84d2eb7b-0a8a-4f66-a1f3-14afbf8c7df1"},{"id":"occ_3111daaaf9d82c0df977d0f8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_a3e10132-4b92-414d-8744-acbeb6f0b294","work_id":"wrk_ab79e5be-0326-461e-868c-ec7e75b2a90e","block_id":"blk_90999a10-cb26-43a9-8ffb-26c7ebe7c8b7","section_id":"sec_3bfbfd39-7e78-471f-bac3-d0763fab9add","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントは、普通のチャットよりはるかに多くのトークンを消費する。コードベースを読み、テストを実行し、修正し、また検証する。資料を読み込み、計画し、タスクを分解し、何度も出力する。これが企業内で大規模に動き始め","quote_start":0,"quote_end":108,"text_sha256":"1f9613c911b62de3b8294ae259078cab461cc8bfe9c5d1f50d6baa1cc833e389","block_sha256":"1f9613c911b62de3b8294ae259078cab461cc8bfe9c5d1f50d6baa1cc833e389","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_a3e10132-4b92-414d-8744-acbeb6f0b294/#blk_90999a10-cb26-43a9-8ffb-26c7ebe7c8b7"},{"id":"occ_c863c831a9ea80b585b07066","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_a3e10132-4b92-414d-8744-acbeb6f0b294","work_id":"wrk_ab79e5be-0326-461e-868c-ec7e75b2a90e","block_id":"blk_918d2238-7162-40d4-8ffe-3f61d172d928","section_id":"sec_0cacacad-7328-4720-a7d0-cfb3970915e6","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":49,"end":55,"exact":"エージェント","quote":"Fable 5は、Anthropicの上位モデルとして、長時間のコーディング、複雑な知識労働、AIエージェント作業に強いモデルとされている。一方で、サイバー、バイオ、そしてfrontier LLM development、つまり最先端AIモデルの開発支援に関わるリクエストでは、安全装置が働く。","quote_start":0,"quote_end":147,"text_sha256":"3e9c50ac6d05e2ffaadfc7ab36592f414c839437db99321f3a9060a0cc82b9bb","block_sha256":"3e9c50ac6d05e2ffaadfc7ab36592f414c839437db99321f3a9060a0cc82b9bb","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_a3e10132-4b92-414d-8744-acbeb6f0b294/#blk_918d2238-7162-40d4-8ffe-3f61d172d928"},{"id":"occ_f411995da0a92539dcb0c4e1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_a3e10132-4b92-414d-8744-acbeb6f0b294","work_id":"wrk_ab79e5be-0326-461e-868c-ec7e75b2a90e","block_id":"blk_a2860658-e298-46ba-9676-2360ff79fe24","section_id":"sec_2ce0ff1d-cd43-4b01-9845-fcd2f1322cd1","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":49,"end":55,"exact":"エージェント","quote":"これまで経済の入口はGoogle検索、Amazon、SNS広告、ECサイトだった。これからは、AIエージェントが購買の入口になるかもしれない。そのとき、Visaは決済の信用インフラとして残ろうとしている。","quote_start":0,"quote_end":102,"text_sha256":"3d4eec5044caabdbcbc1ef1a24af9c0ba5ad9294bccc6a7615c2982f15997ab0","block_sha256":"3d4eec5044caabdbcbc1ef1a24af9c0ba5ad9294bccc6a7615c2982f15997ab0","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_a3e10132-4b92-414d-8744-acbeb6f0b294/#blk_a2860658-e298-46ba-9676-2360ff79fe24"},{"id":"occ_e4c8b0950e255b2a34d43847","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_a3e10132-4b92-414d-8744-acbeb6f0b294","work_id":"wrk_ab79e5be-0326-461e-868c-ec7e75b2a90e","block_id":"blk_aec32656-f0db-4c0a-ab92-3a2b06cb95eb","section_id":"sec_3bfbfd39-7e78-471f-bac3-d0763fab9add","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":20,"end":26,"exact":"エージェント","quote":"そのため、価格が高いままだと、企業はAIエージェントを一部用途にしか使えない。逆に単価が下がれば、より多くの業務がAIに流れる。これは、短期的にはモデル会社の利益率を圧迫するが、長期的にはAI利用量を爆発させる可能性がある。","quote_start":0,"quote_end":112,"text_sha256":"d9a386af5dbadfe736bc3932eb3d88f29342ef03122aa2c09f1c8c3eb8741f2b","block_sha256":"d9a386af5dbadfe736bc3932eb3d88f29342ef03122aa2c09f1c8c3eb8741f2b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_a3e10132-4b92-414d-8744-acbeb6f0b294/#blk_aec32656-f0db-4c0a-ab92-3a2b06cb95eb"},{"id":"occ_bc510a03de969a9428a39231","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_a3e10132-4b92-414d-8744-acbeb6f0b294","work_id":"wrk_ab79e5be-0326-461e-868c-ec7e75b2a90e","block_id":"blk_b0b82a50-e6d6-479e-9b9b-b24fa060afd8","section_id":"sec_72851c4d-029f-485c-ac75-596f5bee2aed","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":31,"end":37,"exact":"エージェント","quote":"第一に、モデル能力。\nどの企業が最も強力な推論、コーディング、エージェント能力を持つか。","quote_start":0,"quote_end":44,"text_sha256":"b0b483f8ef1b414928d66f6588e41005bbee442b2930c1130f632c6de80f93cb","block_sha256":"b0b483f8ef1b414928d66f6588e41005bbee442b2930c1130f632c6de80f93cb","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_a3e10132-4b92-414d-8744-acbeb6f0b294/#blk_b0b82a50-e6d6-479e-9b9b-b24fa060afd8"},{"id":"occ_28dec8caa88ab2b5745e1c79","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_a3e10132-4b92-414d-8744-acbeb6f0b294","work_id":"wrk_ab79e5be-0326-461e-868c-ec7e75b2a90e","block_id":"blk_b1d6bd7a-f268-4332-8def-a907ab1f5108","section_id":"sec_2ce0ff1d-cd43-4b01-9845-fcd2f1322cd1","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":18,"end":24,"exact":"エージェント","quote":"VisaとOpenAIの提携は、AIエージェントがオンライン購入を行う世界を現実の取引に近づけるものだ。","quote_start":0,"quote_end":52,"text_sha256":"31f5097a05162dd32226b1c134d1cf382b5d2fa736dd8c460daeccd920140e65","block_sha256":"31f5097a05162dd32226b1c134d1cf382b5d2fa736dd8c460daeccd920140e65","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_a3e10132-4b92-414d-8744-acbeb6f0b294/#blk_b1d6bd7a-f268-4332-8def-a907ab1f5108"},{"id":"occ_1c6457338b96f5657c7ccccc","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_a3e10132-4b92-414d-8744-acbeb6f0b294","work_id":"wrk_ab79e5be-0326-461e-868c-ec7e75b2a90e","block_id":"blk_b2564ee8-e936-4464-92b5-a19c4112630b","section_id":"sec_58c13842-7fc7-45c3-a024-647f74f0b135","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":123,"end":129,"exact":"エージェント","quote":"ことを示した。\nOpenAIの値下げ検討は、AIモデルが価格戦争に入ることを示した。\nVisaとの提携は、AIエージェントが現実の商取引に接続されることを示した。\nオハイオ10GW計画は、AIが電力と重工業の世界に入ったことを示した。","quote_start":68,"quote_end":185,"text_sha256":"db90304956af87f0d9c1f5d6480b78cf32b70e1e66ba1a30a1b06987be9ec47b","block_sha256":"db90304956af87f0d9c1f5d6480b78cf32b70e1e66ba1a30a1b06987be9ec47b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_a3e10132-4b92-414d-8744-acbeb6f0b294/#blk_b2564ee8-e936-4464-92b5-a19c4112630b"},{"id":"occ_e1e3aae6b41f040f7c34226b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_a3e10132-4b92-414d-8744-acbeb6f0b294","work_id":"wrk_ab79e5be-0326-461e-868c-ec7e75b2a90e","block_id":"blk_cec2a262-7d7c-4294-8f79-0d379a8d48ca","section_id":"sec_2ce0ff1d-cd43-4b01-9845-fcd2f1322cd1","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":32,"end":38,"exact":"エージェント","quote":"AIが勝手に何でも買うのではなく、ユーザーが許可した範囲で、AIエージェントが購入手続きを進める。そのとき、Visaはトークン化、認証、利用制限、不正検知、加盟店ネットワークを提供する。つまりVisaは、人間がカード番号を入力する時代の決済会社から、AIエージェントが経済活動","quote_start":0,"quote_end":138,"text_sha256":"068f7f801810f0fd307dffea01c4fb44da2b9f6afda1c8f28969bb2ac533de1b","block_sha256":"068f7f801810f0fd307dffea01c4fb44da2b9f6afda1c8f28969bb2ac533de1b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_a3e10132-4b92-414d-8744-acbeb6f0b294/#blk_cec2a262-7d7c-4294-8f79-0d379a8d48ca"},{"id":"occ_9c591ef5f1505012a4d396aa","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_a3e10132-4b92-414d-8744-acbeb6f0b294","work_id":"wrk_ab79e5be-0326-461e-868c-ec7e75b2a90e","block_id":"blk_e9664fb8-5653-4dc5-b8b6-74c3ba50d72b","section_id":"sec_2ce0ff1d-cd43-4b01-9845-fcd2f1322cd1","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":10,"end":16,"exact":"エージェント","quote":"これは、表側ではAIエージェントが買い物をし、裏側ではカード、銀行、ステーブルコイン、加盟店精算がつながる構図である。","quote_start":0,"quote_end":59,"text_sha256":"2fbef805f7a6042b969c73ca3365217ee07dad272069d8a6be9817c4247d1abf","block_sha256":"2fbef805f7a6042b969c73ca3365217ee07dad272069d8a6be9817c4247d1abf","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_a3e10132-4b92-414d-8744-acbeb6f0b294/#blk_e9664fb8-5653-4dc5-b8b6-74c3ba50d72b"},{"id":"occ_3508b8774263b79b7db89fea","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_a3e10132-4b92-414d-8744-acbeb6f0b294","work_id":"wrk_ab79e5be-0326-461e-868c-ec7e75b2a90e","block_id":"blk_ff7fb822-9c7a-4100-b915-5fe1b3cb5cbf","section_id":"sec_47420fdf-edcd-4374-982e-2fd6d54e217e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":10,"end":16,"exact":"エージェント","quote":"絶ノイア：  \nAIエージェントが買い物をするなら、モデル会社だけでなくVisaのような決済インフラも主役になります。","quote_start":0,"quote_end":59,"text_sha256":"9556e9cc5a03a5c938ff7554fb8459986ac14aaec8f415e228eca1f244b9014d","block_sha256":"9556e9cc5a03a5c938ff7554fb8459986ac14aaec8f415e228eca1f244b9014d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_a3e10132-4b92-414d-8744-acbeb6f0b294/#blk_ff7fb822-9c7a-4100-b915-5fe1b3cb5cbf"},{"id":"occ_17212ec48ef70288fab04b69","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_a4008bc2-ae2c-4bd7-8615-0b3c4e0936ff","work_id":"wrk_9bc2084d-1dcc-40ef-9cc9-a53fe61434c0","block_id":"blk_bc30ae57-6248-4883-bf2d-4e686b0d4aa0","section_id":"sec_5486e2e0-654c-45e9-b973-c6977a7980c1","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":66,"end":72,"exact":"エージェント","quote":"ド企業の収益化が進めば、AI計算資源への需要はさらに強くなる。企業はもっとGPUを使いたい。推論を増やしたい。エージェントを常時動かしたい。動画、音声、ロボティクス、科学計算、金融、医療、防衛、行政向けのAI利用も増える。","quote_start":11,"quote_end":122,"text_sha256":"83a1481351d69a4ed3d84098977765420dd97178734e5fd496d3f2cb950642e1","block_sha256":"83a1481351d69a4ed3d84098977765420dd97178734e5fd496d3f2cb950642e1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_a4008bc2-ae2c-4bd7-8615-0b3c4e0936ff/#blk_bc30ae57-6248-4883-bf2d-4e686b0d4aa0"},{"id":"occ_5fe6866232c64d38a8cfc955","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_a749fcdf-8155-46e9-a466-83e867886a08","work_id":"wrk_e146f08d-b638-466d-b368-146816bcc837","block_id":"blk_92e9ca13-5266-46f5-8c92-4464daaa2265","section_id":"sec_0fd2faa6-2fcd-4dde-8282-c26a2c018e51","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":55,"end":61,"exact":"エージェント","quote":"Fable 5は、Anthropicが一般利用向けに公開したMythos級モデルだった。ソフトウェア開発、長期エージェント作業、文書理解、視覚認識、科学研究などで高い能力を持つとされ、従来のOpus級モデルより上位に位置づけられていた。","quote_start":0,"quote_end":118,"text_sha256":"68fc04f068ec0470c7957d42f640f2438a32815bb69697130886230ec93be085","block_sha256":"68fc04f068ec0470c7957d42f640f2438a32815bb69697130886230ec93be085","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_a749fcdf-8155-46e9-a466-83e867886a08/#blk_92e9ca13-5266-46f5-8c92-4464daaa2265"},{"id":"occ_57c31a0b4b275262afd9384b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_abe6b561-f51c-4549-a97f-aab2b725492e","work_id":"wrk_f31c79fd-b257-498a-9ee1-d344ac94ffc7","block_id":"blk_b6faad5a-b2ad-49a3-b7a9-3325a4925a60","section_id":"sec_5c47fffe-497d-44bc-85e1-1a0843444a7f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":108,"end":114,"exact":"エージェント","quote":"面的に置き換える汎用チップではないということだ。Jalapeñoは、ChatGPT、Codex、API、将来のエージェント製品のような、大量・定常・低遅延の推論ワークロードを効率よく処理するために設計されている。","quote_start":53,"quote_end":159,"text_sha256":"db4fa5f7ad812680ab85c969b87770481b9bb7bb67eacd1cc88f01a30c3662ba","block_sha256":"db4fa5f7ad812680ab85c969b87770481b9bb7bb67eacd1cc88f01a30c3662ba","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_abe6b561-f51c-4549-a97f-aab2b725492e/#blk_b6faad5a-b2ad-49a3-b7a9-3325a4925a60"},{"id":"occ_48b3b679cb9a5d244f1671cb","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_abe6b561-f51c-4549-a97f-aab2b725492e","work_id":"wrk_f31c79fd-b257-498a-9ee1-d344ac94ffc7","block_id":"blk_f99d393f-db47-4025-9e01-6f6fc3dfdd15","section_id":"sec_a520ab4e-5949-44bb-8a45-30a0fd5e6309","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":61,"end":67,"exact":"エージェント","quote":"は、computeとHBMの比率がパッケージ設計時点で固定される。\nしかしLLM推論、長コンテキスト、MoE、エージェントAIでは、計算よりメモリ容量やKVキャッシュがボトルネックになる場合がある。","quote_start":6,"quote_end":105,"text_sha256":"f6627b304055abdb64a98c46c14f66b76497c30b95a397d5cf1167763b1672c7","block_sha256":"f6627b304055abdb64a98c46c14f66b76497c30b95a397d5cf1167763b1672c7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_abe6b561-f51c-4549-a97f-aab2b725492e/#blk_f99d393f-db47-4025-9e01-6f6fc3dfdd15"},{"id":"occ_dc29634f5c2176afd7683a20","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ae55c104-d140-4064-adf1-726fae29a7f3","work_id":"wrk_a1470532-dd67-430b-91ce-2205bc4c54e0","block_id":"blk_d6256910-02f9-421b-8b9e-fa1c279ae1ea","section_id":"sec_0a96a7d9-7297-405e-9540-0b8ae68d950d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":67,"end":73,"exact":"エージェント","quote":"ネックであり、DRAMはサーバー全体の供給制約に直結する。NANDやストレージも、推論、学習、データ保持、AIエージェントの長期記憶といった需要に関わる。","quote_start":12,"quote_end":89,"text_sha256":"ad524e76a0a2e41890cc97f01c458dfac64a2a0ba225e713e74d6c4ed682d0e0","block_sha256":"ad524e76a0a2e41890cc97f01c458dfac64a2a0ba225e713e74d6c4ed682d0e0","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ae55c104-d140-4064-adf1-726fae29a7f3/#blk_d6256910-02f9-421b-8b9e-fa1c279ae1ea"},{"id":"occ_0e4524debd558d709f2808af","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_aea0fd30-a496-42e0-975e-3de7b0cd01ab","work_id":"wrk_4efede78-d9d9-4a37-854a-51a334476949","block_id":"blk_19f4b8bf-5b8c-47a7-9448-56a40ca90182","section_id":"sec_8747c6e9-bfe1-48ae-bd0e-5b586f5654fe","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":38,"end":44,"exact":"エージェント","quote":"本来であれば、GPT-5.6は「次のモデルが来る」「AI能力がまた伸びる」「エージェントやコーディングや企業利用が進む」という期待材料になり得た。だが、実際には「最先端モデルは政府に止められるかもしれない」「OpenAIは自由にリリースできないかもしれない」「顧客ごとに承認が必要なら、商","quote_start":0,"quote_end":144,"text_sha256":"205985acd58f3a17df0d474207183a98a9507d7a55c1b771dc683da7305da060","block_sha256":"205985acd58f3a17df0d474207183a98a9507d7a55c1b771dc683da7305da060","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_aea0fd30-a496-42e0-975e-3de7b0cd01ab/#blk_19f4b8bf-5b8c-47a7-9448-56a40ca90182"},{"id":"occ_5b68e4757a9f2f3e3cb44f6f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_aea0fd30-a496-42e0-975e-3de7b0cd01ab","work_id":"wrk_4efede78-d9d9-4a37-854a-51a334476949","block_id":"blk_26349b8d-1b6a-4b8b-ba6c-0f4c1a6e5542","section_id":"sec_7482a506-dbca-4596-bbd3-a5896d9df42d","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":98,"end":104,"exact":"エージェント","quote":"C。  \n国籍・所在地・用途制限の管理。  \n複数モデルのルーティング。  \nオンプレミス実行環境。  \nAIエージェントの権限管理。  \nサイバー防衛向けの安全なサンドボックス。","quote_start":43,"quote_end":133,"text_sha256":"4190fd2814699298fcade4085ae80613c271c76ca98d9571c35fc89c4dca6e6a","block_sha256":"4190fd2814699298fcade4085ae80613c271c76ca98d9571c35fc89c4dca6e6a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_aea0fd30-a496-42e0-975e-3de7b0cd01ab/#blk_26349b8d-1b6a-4b8b-ba6c-0f4c1a6e5542"},{"id":"occ_8a5b3a3f1f9f87a432a5c9b1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_aea0fd30-a496-42e0-975e-3de7b0cd01ab","work_id":"wrk_4efede78-d9d9-4a37-854a-51a334476949","block_id":"blk_4cb68324-981a-43a2-9b7b-096180dd6f5f","section_id":"sec_8de470f0-40fd-4b4e-8a61-d49e06f0c28e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":33,"end":39,"exact":"エージェント","quote":"第一に、1兆ドル評価を守りたい。  \n第二に、GPT-5.6や次のエージェント製品の収益力を見せたい。  \n第三に、推論コストやデータセンター投資の回収可能性を説明できるようにしたい。  \n第四に、Microsoft、Oracle、Broadcom、政府契約、AIインフラ構想と","quote_start":0,"quote_end":139,"text_sha256":"3d9f8cb9425f6d41f0ad2121b2470253ff5610e6c3997579e052ccbc3eb4ca5d","block_sha256":"3d9f8cb9425f6d41f0ad2121b2470253ff5610e6c3997579e052ccbc3eb4ca5d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_aea0fd30-a496-42e0-975e-3de7b0cd01ab/#blk_4cb68324-981a-43a2-9b7b-096180dd6f5f"},{"id":"occ_d16640426401de828a5ebc99","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_afc88027-978f-428a-a597-dd5d129dc441","work_id":"wrk_9fae1367-ebbb-47c5-ad7b-3700deb207cd","block_id":"blk_98881146-1226-4865-ac97-433660d6fe09","section_id":"sec_e872ff86-dafb-426e-9d86-1911e6d8d8f7","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントはソフトウェアを操作し、ロボットは工場を動かし、EVは都市を走り、ドローンは物流を担い、エッジAIは監視・検査・保守・制御を行い、スマート工場は自律的に生産を最適化していく。","quote_start":0,"quote_end":95,"text_sha256":"b75cc0f564ed82f695d231bfe272edfe65ad1451f980f9174cf750ff8dd0e5f6","block_sha256":"b75cc0f564ed82f695d231bfe272edfe65ad1451f980f9174cf750ff8dd0e5f6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_afc88027-978f-428a-a597-dd5d129dc441/#blk_98881146-1226-4865-ac97-433660d6fe09"},{"id":"occ_e2f7506c3dabbfe84ad7f8e8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_b228986a-43e7-4e2b-8b16-758283d367bc","work_id":"wrk_b193a2ff-ca61-4efc-864e-0778162ac852","block_id":"blk_25a98841-bdbe-4b30-9d23-2ee99681d725","section_id":"sec_c411ead6-1554-425b-9e21-172ff81663aa","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":102,"end":108,"exact":"エージェント","quote":"\nOpenAIはコンサル大手との企業導入連合（Frontier Alliance）を組成し、業務プロセスにAIエージェントを組み込む支援を進めると報じられている。  \nさらにServiceNowとの複数年提携では、顧客が“個別開発なし”でフロンティアモデル機能にアクセスできることを狙っている。  \nこの方向性を、ERP/","quote_start":47,"quote_end":208,"text_sha256":"e567ed92771ddb4024e4a81c9ab37fff2c12e8d9fd3c9877fe37df33f7b99ffe","block_sha256":"e567ed92771ddb4024e4a81c9ab37fff2c12e8d9fd3c9877fe37df33f7b99ffe","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_b228986a-43e7-4e2b-8b16-758283d367bc/#blk_25a98841-bdbe-4b30-9d23-2ee99681d725"},{"id":"occ_e75a7b49a9c04c62c2d9b527","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_b228986a-43e7-4e2b-8b16-758283d367bc","work_id":"wrk_b193a2ff-ca61-4efc-864e-0778162ac852","block_id":"blk_55df5281-84a3-4ff7-895e-3f5d5fc0e8bf","section_id":"sec_45fa293e-960a-4c18-8a55-2e41cea42f87","layer":"body","character_id":null,"count":3,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":13,"end":19,"exact":"エージェント","quote":"- **常時稼働・自律実行エージェントは、チャットより計算資源を食い、サブスクの想定利用を破壊**しやすい（＝収益モデルが従量へ引き戻される）。\n- セキュリティ面でも、常時稼働エージェントは「間接プロンプトインジェクション」「資格情報（","quote_start":0,"quote_end":119,"text_sha256":"eb644e3bf4bc5d81f41e5d8610d4e466cac4428ce623cdf822f7ca474d9c822a","block_sha256":"eb644e3bf4bc5d81f41e5d8610d4e466cac4428ce623cdf822f7ca474d9c822a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_b228986a-43e7-4e2b-8b16-758283d367bc/#blk_55df5281-84a3-4ff7-895e-3f5d5fc0e8bf"},{"id":"occ_999cc56b6fc2dd020f09c5e9","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_b228986a-43e7-4e2b-8b16-758283d367bc","work_id":"wrk_b193a2ff-ca61-4efc-864e-0778162ac852","block_id":"blk_5fd63c46-ffc4-4327-9423-563b4334c7ca","section_id":"sec_e1f09d5b-a2af-4e37-9567-aef1d9c19e55","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"### エージェントAPI/SDK再編が示す「B2B運用意識」","quote_start":0,"quote_end":31,"text_sha256":"f533b4ae9156f210965815b9014f5e548c0c6e9e15b6d256798a4b94102a9ffb","block_sha256":"f533b4ae9156f210965815b9014f5e548c0c6e9e15b6d256798a4b94102a9ffb","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_b228986a-43e7-4e2b-8b16-758283d367bc/#blk_5fd63c46-ffc4-4327-9423-563b4334c7ca"},{"id":"occ_bf10ff43daa71fd522df1545","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_b228986a-43e7-4e2b-8b16-758283d367bc","work_id":"wrk_b193a2ff-ca61-4efc-864e-0778162ac852","block_id":"blk_62963b83-e95c-4e94-a94a-1928e09554b9","section_id":"sec_c411ead6-1554-425b-9e21-172ff81663aa","layer":"body","character_id":null,"count":4,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"第二に、**エージェントの経済”に合わせて課金を再設計する**。  \nOpenClaw事例が示した通り、常時稼働エージェントはサブスク想定を壊し、従量課金へ戻る圧力が強い。  \nOpenAIはBusinessで月額席課金（例","quote_start":0,"quote_end":112,"text_sha256":"2ed5285c7562437531c4f84c7db0e5ea8af7950fa0dda52be53608b91d847ba7","block_sha256":"2ed5285c7562437531c4f84c7db0e5ea8af7950fa0dda52be53608b91d847ba7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_b228986a-43e7-4e2b-8b16-758283d367bc/#blk_62963b83-e95c-4e94-a94a-1928e09554b9"},{"id":"occ_b92b8b27dd16d478ac2a98ec","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_b228986a-43e7-4e2b-8b16-758283d367bc","work_id":"wrk_b193a2ff-ca61-4efc-864e-0778162ac852","block_id":"blk_64f82d85-58f9-44c6-bcfc-9a93d28be1ad","section_id":"sec_cfc10044-44e7-4b15-b973-886a69890249","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":122,"end":128,"exact":"エージェント","quote":"理（ロードマップの揺れ）、③エンタープライズ要件（統治・監査・ソブリン/オンプレ）への対応速度、④開発者体験（エージェント標準・運用観測性・安全設計）の総合戦**」として表面化している、という点にある。根拠として、OpenAIが動画生成アプリSoraを終了し、より収益性が高いと見られる“コーディング/企業向け”へ軸足を移","quote_start":67,"quote_end":228,"text_sha256":"3848c1a7046a95a674851bf8d39037b53d5ff52596217f206a852aef597a038d","block_sha256":"3848c1a7046a95a674851bf8d39037b53d5ff52596217f206a852aef597a038d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_b228986a-43e7-4e2b-8b16-758283d367bc/#blk_64f82d85-58f9-44c6-bcfc-9a93d28be1ad"},{"id":"occ_df055e8b36a55f7e2220206f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_b228986a-43e7-4e2b-8b16-758283d367bc","work_id":"wrk_b193a2ff-ca61-4efc-864e-0778162ac852","block_id":"blk_77142f87-8b39-45aa-b984-85435088247f","section_id":"sec_0a9cd2b7-dfe8-4e68-b682-b8b197cd1b3f","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":67,"end":73,"exact":"エージェント","quote":"のロボット開発スタック（Isaac/ROS/Jetson）との正面連携**：OpenAIがロボット向けモデル/エージェントを提供する場合、実機側の標準（ROS等）と検証基盤（シミュレーション）を避けて通れない。NVIDIAはその“開発面”を既に製品化しているため、OpenAIは競合するより公式対応で取り込む方が早い。\n-","quote_start":12,"quote_end":173,"text_sha256":"c49889b137f17b47d15470a53d9511a9a1aff70b7227215dcf5345e92617c877","block_sha256":"c49889b137f17b47d15470a53d9511a9a1aff70b7227215dcf5345e92617c877","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_b228986a-43e7-4e2b-8b16-758283d367bc/#blk_77142f87-8b39-45aa-b984-85435088247f"},{"id":"occ_bbdb47a6dcc2887388115552","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_b228986a-43e7-4e2b-8b16-758283d367bc","work_id":"wrk_b193a2ff-ca61-4efc-864e-0778162ac852","block_id":"blk_84c00588-6f55-470a-91fc-f87ffed49c66","section_id":"sec_e1f09d5b-a2af-4e37-9567-aef1d9c19e55","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":31,"end":37,"exact":"エージェント","quote":"OpenAIは開発者向けに、**Responses APIを“エージェント的アプリの統一インターフェース”として位置づけ、Web検索・ファイル検索・コンピュータ操作・リモートMCP等の組み込みツールを掲げている。  \nさらに**Agents SDKを提供し、ツール利用、ハン","quote_start":0,"quote_end":137,"text_sha256":"792bd0797b81b3ea6b0d043d28e413c5d7600d2f287885b003c96b307d76da8d","block_sha256":"792bd0797b81b3ea6b0d043d28e413c5d7600d2f287885b003c96b307d76da8d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_b228986a-43e7-4e2b-8b16-758283d367bc/#blk_84c00588-6f55-470a-91fc-f87ffed49c66"},{"id":"occ_1c3105840c18954fd9a8b7e5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_b228986a-43e7-4e2b-8b16-758283d367bc","work_id":"wrk_b193a2ff-ca61-4efc-864e-0778162ac852","block_id":"blk_8f8820a4-f8a6-4176-a271-7c78d1f331e1","section_id":"sec_d3b5a9d1-ec09-4311-b742-bf08719eba9a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":80,"end":86,"exact":"エージェント","quote":"で「強いシグナル」になっている。特に（i）Sora終了、（ii）次世代モデル（Spud）関連の観測、（iii）エージェントAPI/SDKの再編は、同社が“消費者向けの拡大”と“企業向けの運用要求”を、計算資源制約の中で統合し直していることを示す。","quote_start":25,"quote_end":148,"text_sha256":"f72c8e3bb66df49d2d3b30f3121c37d7baea767b5cdb4516a7ab19350e6b28e8","block_sha256":"f72c8e3bb66df49d2d3b30f3121c37d7baea767b5cdb4516a7ab19350e6b28e8","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_b228986a-43e7-4e2b-8b16-758283d367bc/#blk_8f8820a4-f8a6-4176-a271-7c78d1f331e1"},{"id":"occ_caf31525b24e9c633f42e723","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_b228986a-43e7-4e2b-8b16-758283d367bc","work_id":"wrk_b193a2ff-ca61-4efc-864e-0778162ac852","block_id":"blk_990c5a75-fd28-4bac-bfc4-b50dce5feb54","section_id":"sec_45fa293e-960a-4c18-8a55-2e41cea42f87","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":17,"end":23,"exact":"エージェント","quote":"### OpenClawが示した「エージェント経済」の難しさ","quote_start":0,"quote_end":30,"text_sha256":"ea8407f61800f7cd239c7c8e93290f5392cdedef0096f64459fe65ef625dda7e","block_sha256":"ea8407f61800f7cd239c7c8e93290f5392cdedef0096f64459fe65ef625dda7e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_b228986a-43e7-4e2b-8b16-758283d367bc/#blk_990c5a75-fd28-4bac-bfc4-b50dce5feb54"},{"id":"occ_02c76b2afaeda6ca44c8dfcb","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_b228986a-43e7-4e2b-8b16-758283d367bc","work_id":"wrk_b193a2ff-ca61-4efc-864e-0778162ac852","block_id":"blk_b0f9caa9-2991-45ce-abce-37434cb48897","section_id":"sec_cfc10044-44e7-4b15-b973-886a69890249","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":81,"end":87,"exact":"エージェント","quote":"の「Claude/Claude Code」による開発現場のOS化**（ターミナルやIDEで自律的に作業を進めるエージェント型コーディング）、②Google（Alphabet）の**クラウド統合（Vertex AI・Gemini）＋ソブリン/オンプレ（GDC air-gapped）**、③Meta社の**オープンウェイト（","quote_start":26,"quote_end":187,"text_sha256":"cc64e4a95f89fece067f62e5fa4c1ca2be9aaaec060fe060d3f0b7bcb987489d","block_sha256":"cc64e4a95f89fece067f62e5fa4c1ca2be9aaaec060fe060d3f0b7bcb987489d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_b228986a-43e7-4e2b-8b16-758283d367bc/#blk_b0f9caa9-2991-45ce-abce-37434cb48897"},{"id":"occ_7c94f85aaef7bac39db8357f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_b228986a-43e7-4e2b-8b16-758283d367bc","work_id":"wrk_b193a2ff-ca61-4efc-864e-0778162ac852","block_id":"blk_cc34eb38-81b6-451d-a7a1-077520d05bc3","section_id":"sec_0a9cd2b7-dfe8-4e68-b682-b8b197cd1b3f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":72,"end":78,"exact":"エージェント","quote":"帰は、B2Bにとっては好材料である。Sora終了は痛みを伴うが、逆に言えば「企業が買える形（コーディング・業務エージェント・統治）」に資源を寄せたということだ。  \n今後の勝敗は、Spud（次世代モデル）がどれほど強いかよりも、**その能力を「安全に・監査可能に・継続的に」業務へ埋め込めるか**、そして物理AIに向けて*","quote_start":17,"quote_end":178,"text_sha256":"7ba941bc477f2ce36fc62fe9d71cbad3863cacdec688f821b8585dbe4b4bff93","block_sha256":"7ba941bc477f2ce36fc62fe9d71cbad3863cacdec688f821b8585dbe4b4bff93","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_b228986a-43e7-4e2b-8b16-758283d367bc/#blk_cc34eb38-81b6-451d-a7a1-077520d05bc3"},{"id":"occ_e74a4eb9e3191f0694b4488a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_b228986a-43e7-4e2b-8b16-758283d367bc","work_id":"wrk_b193a2ff-ca61-4efc-864e-0778162ac852","block_id":"blk_d55cbf70-2868-4d41-b8b7-fee10894da1a","section_id":"sec_45fa293e-960a-4c18-8a55-2e41cea42f87","layer":"body","character_id":null,"count":3,"matched_aliases":["AI Agent","エージェント"],"evidence":{"text_basis":"markdown","start":15,"end":21,"exact":"エージェント","quote":"参考として、Google側でもエージェント基盤（Vertex AI Agent Engine）に関して、過剰権限のサービスアカウント等がリスクになり得ることが研究報告として示され、Googleドキュメント側でもアクセス管理（サービスアカウント","quote_start":0,"quote_end":121,"text_sha256":"8061c8a57f176bafec2c54d85bbe564b3ecc691d3535a52c2e557bacf221452b","block_sha256":"8061c8a57f176bafec2c54d85bbe564b3ecc691d3535a52c2e557bacf221452b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_b228986a-43e7-4e2b-8b16-758283d367bc/#blk_d55cbf70-2868-4d41-b8b7-fee10894da1a"},{"id":"occ_a23abf5b17bf36bf4224f6dd","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_b228986a-43e7-4e2b-8b16-758283d367bc","work_id":"wrk_b193a2ff-ca61-4efc-864e-0778162ac852","block_id":"blk_ddde3e27-77fa-4b84-9d8e-07c8b2eb88f0","section_id":"sec_cfc10044-44e7-4b15-b973-886a69890249","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":35,"end":41,"exact":"エージェント","quote":"OpenAIが巻き返すための「勝ち筋」は、（A）**コーディング/業務エージェント領域での“運用可能性”を武器にする**（SDK、トレーシング、移行手当、SLA/監査ログ、ガバナンスAPI）、（B）**ソブリン/オンプレッシング層の穴をパートナーで埋める**（地域内GPU推論、政府","quote_start":0,"quote_end":141,"text_sha256":"753adb18ba8b2338bff99bc4e782595354fb4d53374e2ec3d4319244cba5dbac","block_sha256":"753adb18ba8b2338bff99bc4e782595354fb4d53374e2ec3d4319244cba5dbac","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_b228986a-43e7-4e2b-8b16-758283d367bc/#blk_ddde3e27-77fa-4b84-9d8e-07c8b2eb88f0"},{"id":"occ_cf2cd2d5ec9a3ba7f313ec9d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_b228986a-43e7-4e2b-8b16-758283d367bc","work_id":"wrk_b193a2ff-ca61-4efc-864e-0778162ac852","block_id":"blk_ecafb851-ea99-4f99-ac7d-40e884d7bfbf","section_id":"sec_9c25f319-3c75-4406-b763-9ecbe1b9b46b","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":21,"end":27,"exact":"エージェント","quote":"1. **開発現場の主戦場が“チャットからエージェント型コーディング/運用へ移った**  \n   AnthropicのClaude Codeは、コードベース理解→ファイル編集→コマンド実行を前提に設計され、ターミナル/IDE等へ広く展開している。  \n ","quote_start":0,"quote_end":127,"text_sha256":"2bce8e963acdf4389303f2b2ea0c6bec37a132a8b1c9a83fdc68d8c15d1980af","block_sha256":"2bce8e963acdf4389303f2b2ea0c6bec37a132a8b1c9a83fdc68d8c15d1980af","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_b228986a-43e7-4e2b-8b16-758283d367bc/#blk_ecafb851-ea99-4f99-ac7d-40e884d7bfbf"},{"id":"occ_b81878b79abcfc208840aeca","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_b228986a-43e7-4e2b-8b16-758283d367bc","work_id":"wrk_b193a2ff-ca61-4efc-864e-0778162ac852","block_id":"blk_f8486df0-1618-4a51-b411-4d3db7e70195","section_id":"sec_e1f09d5b-a2af-4e37-9567-aef1d9c19e55","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":49,"end":55,"exact":"エージェント","quote":"これはB2B視点では、「（i）統合APIへ収束（運用/監査/コスト制御を一箇所に集約）、（ii）“エージェントの本番運用”を前提にした設計（ツール実行・観測性・移行期限の明記）」という“エンタープライズ・ソフトウェア化”の意思表示と解釈できる。","quote_start":0,"quote_end":122,"text_sha256":"507b69a8b20cbbe605f28d5a728f71ed7db100e64992642e0679470cd639ab04","block_sha256":"507b69a8b20cbbe605f28d5a728f71ed7db100e64992642e0679470cd639ab04","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_b228986a-43e7-4e2b-8b16-758283d367bc/#blk_f8486df0-1618-4a51-b411-4d3db7e70195"},{"id":"occ_35193add0569a4833d8d8c1f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_b228986a-43e7-4e2b-8b16-758283d367bc","work_id":"wrk_b193a2ff-ca61-4efc-864e-0778162ac852","block_id":"blk_f91f1e6c-bcb4-4943-b459-2cc53b039ba0","section_id":"sec_c411ead6-1554-425b-9e21-172ff81663aa","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":134,"end":140,"exact":"エージェント","quote":"stants APIの停止期限を明記するなど前進しているため、これを企業向け全機能（モデル、ツール、コネクタ、エージェント）に拡張し、「LTS（長期サポート）モデル」「非互換変更の移行ウィンドウ」「顧客側監査のための変更履歴」を標準化すべきである。","quote_start":79,"quote_end":203,"text_sha256":"6cac8deafad92fb1d9ad1dd2c137602654048993684f926b516c7eace7372cbe","block_sha256":"6cac8deafad92fb1d9ad1dd2c137602654048993684f926b516c7eace7372cbe","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_b228986a-43e7-4e2b-8b16-758283d367bc/#blk_f91f1e6c-bcb4-4943-b459-2cc53b039ba0"},{"id":"occ_780d781dadc708b65786e067","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_b8a0343f-f76e-47f4-8c47-4536da944126","work_id":"wrk_0eee3d56-e636-4917-98f6-cddcb3cca3f5","block_id":"blk_d1f4f99a-9af5-4af3-8356-1daea6e2da15","section_id":"sec_96c1115d-eec3-4d26-89b9-853d6a718ef1","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"では、AIエージェントがホワイトカラーを駆逐してフィジカルAIがブルーカラーを駆逐したとして、貯金もない資産もない人間たちはどうなるのか？","quote_start":0,"quote_end":69,"text_sha256":"fa1581646d5a577059ed054474e197fed5ab734bed1a23cc176267d73ed0df87","block_sha256":"fa1581646d5a577059ed054474e197fed5ab734bed1a23cc176267d73ed0df87","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_b8a0343f-f76e-47f4-8c47-4536da944126/#blk_d1f4f99a-9af5-4af3-8356-1daea6e2da15"},{"id":"occ_861271e273f4c600771a6680","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":null,"section_id":null,"layer":"title","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"title","start":17,"end":23,"exact":"エージェント","quote":"Fugu、ARD、A2Aが示すAIエージェント時代の競争軸","quote_start":0,"quote_end":29,"text_sha256":"b07447532b9e020ce82ca2fc6d7383a76abd8e9a4fd96f893d71234974856494","block_sha256":null,"offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/"},{"id":"occ_35132489f6720b59bb97c8fc","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_00469009-04ec-43c8-9f61-a621bf3a9b76","section_id":"sec_98579fc2-c6fa-4167-9af4-945cd97a1608","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":94,"end":100,"exact":"エージェント","quote":"enなど、どのモデルを使うかを自分で選ぶ。しかしFuguは、その選択を内部で行う。タスクに応じて複数のモデルやエージェントを呼び出し、考える役、作業する役、検証する役のように分担させる。","quote_start":39,"quote_end":132,"text_sha256":"9025a5148aae503010f5c56bd0b713dbc8f47dd3b57770eac973838ee03d0058","block_sha256":"9025a5148aae503010f5c56bd0b713dbc8f47dd3b57770eac973838ee03d0058","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_00469009-04ec-43c8-9f61-a621bf3a9b76"},{"id":"occ_653ea002ac513be72c2ba500","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_07cfe4b6-e491-4a09-bfd9-ce7ffdcefd3c","section_id":"sec_0bbd999a-1fba-4638-a49a-22638fd9d87f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":13,"end":19,"exact":"エージェント","quote":"どの能力を探せるか。\nどのエージェントと繋がれるか。\nどのモデルをいつ使うか。\nどれだけ検証するか。\nどこまで透明性を保てるか。\n誰がその指揮権を持つのか。","quote_start":0,"quote_end":78,"text_sha256":"2af92603b0b412408c14ac9d4d4e058290663c6546fec119251f1ba0d14a4d0f","block_sha256":"2af92603b0b412408c14ac9d4d4e058290663c6546fec119251f1ba0d14a4d0f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_07cfe4b6-e491-4a09-bfd9-ce7ffdcefd3c"},{"id":"occ_93a27da8c87e4b632fcd696b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_0f36c3ed-c316-4cda-a6f8-415aa8cecf5a","section_id":"sec_adc2b089-8f68-448c-8021-f306a270142a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"このようなエージェント間の業務依頼、進捗管理、成果物の受け渡しを標準化するのがA2Aである。","quote_start":0,"quote_end":46,"text_sha256":"525caf52936e672c15c14067a9753a8f93bf7418920000dfaac91cf73b949853","block_sha256":"525caf52936e672c15c14067a9753a8f93bf7418920000dfaac91cf73b949853","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_0f36c3ed-c316-4cda-a6f8-415aa8cecf5a"},{"id":"occ_6614f8a8094b388909a99f1a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_107989f4-9595-40ce-a2b0-f533f373bbb5","section_id":"sec_0ac1e5bb-e60a-4a9f-bb55-991409bbe5e3","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":35,"end":41,"exact":"エージェント","quote":"```text\nARD: 使える能力を探す\n  ↓\nA2A: 見つけたエージェントに仕事を頼む\n  ↓\nMCP: ツールやデータを操作する\n  ↓\nFugu型オーケストレーター: 誰に何を任せ、どう検証するかを決める\n```","quote_start":0,"quote_end":112,"text_sha256":"3f27dea564add544f34eb23fd7ba58d79e326df287b152299cd3af5e326af400","block_sha256":"3f27dea564add544f34eb23fd7ba58d79e326df287b152299cd3af5e326af400","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_107989f4-9595-40ce-a2b0-f533f373bbb5"},{"id":"occ_8174950fc6935fbb24256b22","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_111c62b9-612c-4582-a09f-8938c546a430","section_id":"sec_adc2b089-8f68-448c-8021-f306a270142a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":10,"end":16,"exact":"エージェント","quote":"つまりA2Aは、AIエージェント同士の会話・依頼プロトコルである。","quote_start":0,"quote_end":33,"text_sha256":"9fa7ad119719c2c4140c72d15b176d2599dbb66f22e979a7cc6c1ebc3b7a8baf","block_sha256":"9fa7ad119719c2c4140c72d15b176d2599dbb66f22e979a7cc6c1ebc3b7a8baf","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_111c62b9-612c-4582-a09f-8938c546a430"},{"id":"occ_9208e85e5deb3fb6be377071","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_1ede1827-0d35-4823-9204-c9a5cefa5aa8","section_id":"sec_98579fc2-c6fa-4167-9af4-945cd97a1608","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":24,"end":30,"exact":"エージェント","quote":"Fuguは、Sakana AIが発表した「マルチエージェント・システムを一つのモデルAPIのように使う」仕組みである。","quote_start":0,"quote_end":59,"text_sha256":"cae7b16e11d2f3f13617de25380704b7af908b8e8709dd3ac8779b7a9d1b1253","block_sha256":"cae7b16e11d2f3f13617de25380704b7af908b8e8709dd3ac8779b7a9d1b1253","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_1ede1827-0d35-4823-9204-c9a5cefa5aa8"},{"id":"occ_985afecc29d19fb0aa89ddc0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_242d0f56-84b9-4ac2-9665-a3d71d28e444","section_id":"sec_42f65980-b59f-4137-9f0e-50ce58092323","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":9,"end":15,"exact":"エージェント","quote":"たとえるなら、AIエージェント用の検索エンジン、電話帳、DNS、あるいはアプリストアの入口に近い。エージェントが「このタスクにはどんなツールが必要か」を考え、ARDを通じて外部の能力を検索し、信頼できるかを確認する。","quote_start":0,"quote_end":108,"text_sha256":"bd9719217310886d4173267432fb9d04f72dd4ec6cf6dd391a46858cf0a4179b","block_sha256":"bd9719217310886d4173267432fb9d04f72dd4ec6cf6dd391a46858cf0a4179b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_242d0f56-84b9-4ac2-9665-a3d71d28e444"},{"id":"occ_8354a61b97972228cedeae9a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_245f2340-95c0-4d0c-bcbf-11a85e461fd8","section_id":"sec_7e0c4909-5d2d-46b5-87cb-984fe85c2316","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":48,"end":54,"exact":"エージェント","quote":"この透明性がなければ、オーケストレーションは性能を上げる一方で、責任の所在を見えにくくする。AIエージェント時代の重要な問いは、「誰が最も賢いか」だけではない。「誰が指揮権を持ち、その指揮をどこまで説明できるか」である。","quote_start":0,"quote_end":110,"text_sha256":"bcf7b6e6a9146641de55aa81e8c7acfa8187f763401513fa16068d092ecef591","block_sha256":"bcf7b6e6a9146641de55aa81e8c7acfa8187f763401513fa16068d092ecef591","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_245f2340-95c0-4d0c-bcbf-11a85e461fd8"},{"id":"occ_3f7ce6c08d5e63ab5d91c89e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_31c68936-6ca8-4c55-b614-ad5e129de579","section_id":"sec_0bbd999a-1fba-4638-a49a-22638fd9d87f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":11,"end":17,"exact":"エージェント","quote":"つまり、これからのAIエージェント時代の本質は、「AIが賢くなる」だけではない。","quote_start":0,"quote_end":40,"text_sha256":"2134ec129b81c1defc3507eb6056d87287043a9b88c3c1a6fbc04a413f9d3e18","block_sha256":"2134ec129b81c1defc3507eb6056d87287043a9b88c3c1a6fbc04a413f9d3e18","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_31c68936-6ca8-4c55-b614-ad5e129de579"},{"id":"occ_82ac75b8fb2e8de4ae12778d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_39554127-72b3-4247-a98c-ed41c71c6a18","section_id":"sec_42f65980-b59f-4137-9f0e-50ce58092323","layer":"body","character_id":null,"count":3,"matched_aliases":["Agentic","エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":12,"exact":"Agentic","quote":"ARDは、Agentic Resource Discoveryの略で、AIエージェントが外部のツール、スキル、MCPサーバー、他のエージェントを発見するための仕様である。","quote_start":0,"quote_end":85,"text_sha256":"cea815594827b6e9ab836f1911e10acb44fe2dc945432557c66d89cbbd9cd7dd","block_sha256":"cea815594827b6e9ab836f1911e10acb44fe2dc945432557c66d89cbbd9cd7dd","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_39554127-72b3-4247-a98c-ed41c71c6a18"},{"id":"occ_61bcaa4b3e680dd0b58cc67a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_4a6b4d1e-8c04-440c-8cae-371368ce7660","section_id":"sec_4c6637ef-a85c-488a-96ef-431c9b4636e6","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":143,"end":149,"exact":"エージェント","quote":"ん、この競争はいまも続いている。だが最近の動きを見ると、次の主戦場は「モデルそのもの」だけではなく、「モデルやエージェントをどう探し、どう接続し、どう指揮するか」に移り始めている。","quote_start":88,"quote_end":178,"text_sha256":"d10504dcf78895555fba9f672774ccc561620b6d2795ca727f98ebd7c9f874b8","block_sha256":"d10504dcf78895555fba9f672774ccc561620b6d2795ca727f98ebd7c9f874b8","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_4a6b4d1e-8c04-440c-8cae-371368ce7660"},{"id":"occ_0301d199d102459f602a28b9","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_532cc97d-4b8e-4683-abdb-8cb32e863ab9","section_id":"sec_0bbd999a-1fba-4638-a49a-22638fd9d87f","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":7,"end":13,"exact":"エージェント","quote":"ARDは、AIエージェントが能力を探すための発見レイヤー。\nA2Aは、AIエージェント同士が仕事を依頼し合う通信レイヤー。\nFuguは、複数モデルをどう使うかを決める指揮レイヤー。","quote_start":0,"quote_end":90,"text_sha256":"efb0f9865294cbccb2992d58b4a05faa77c7b4a8602ba38afa784a076c11e01a","block_sha256":"efb0f9865294cbccb2992d58b4a05faa77c7b4a8602ba38afa784a076c11e01a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_532cc97d-4b8e-4683-abdb-8cb32e863ab9"},{"id":"occ_ead187f497b0fbd1c65c9511","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_6c59345d-04fa-40c5-8b14-45a53717bb40","section_id":"sec_664d1a35-9f46-4d0d-956f-18ba2f21f6f5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":67,"end":73,"exact":"エージェント","quote":"いか。\n次は、どのAIプロダクトが一番作業しやすいか。\nそしてこれからは、どのシステムが、どのモデル・ツール・エージェントを、どれだけ上手く指揮できるか。","quote_start":12,"quote_end":89,"text_sha256":"ec79fb723882616e24ebc1fee89c95c6a158715b97d754cce8418b8b938208ed","block_sha256":"ec79fb723882616e24ebc1fee89c95c6a158715b97d754cce8418b8b938208ed","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_6c59345d-04fa-40c5-8b14-45a53717bb40"},{"id":"occ_582bc76564c5895b02017022","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_7d8ab82a-93de-457f-b3b4-1203e5c68b89","section_id":"sec_42f65980-b59f-4137-9f0e-50ce58092323","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":30,"end":36,"exact":"エージェント","quote":"重要なのは、ARD自体は知能ではないということだ。ARDは、エージェントが能力を探すための道であり、地図であり、索引である。","quote_start":0,"quote_end":62,"text_sha256":"ea8ec3eb8f7271d8a57b43b5cb07e40f9f59020ae4756b89618fb87bed39f0b8","block_sha256":"ea8ec3eb8f7271d8a57b43b5cb07e40f9f59020ae4756b89618fb87bed39f0b8","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_7d8ab82a-93de-457f-b3b4-1203e5c68b89"},{"id":"occ_f949a2fb36db2306c5de06d0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_85364a59-e91c-41d1-ba0f-b1d0e82b8da4","section_id":"sec_adc2b089-8f68-448c-8021-f306a270142a","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":7,"end":13,"exact":"エージェント","quote":"たとえば、あるエージェントが「この企業の最新決算を分析してほしい」と別のエージェントに依頼する。相手は作業を受け取り、途中経過を返し、必要なら追加情報を求め、最終的な成果物を返す。","quote_start":0,"quote_end":90,"text_sha256":"0734b4e4236c6f0ed251c08d68e815dffa8470cc9e31958022abdcefd4b46ba8","block_sha256":"0734b4e4236c6f0ed251c08d68e815dffa8470cc9e31958022abdcefd4b46ba8","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_85364a59-e91c-41d1-ba0f-b1d0e82b8da4"},{"id":"occ_9c317687facbddd0b2062415","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_9cdcc954-c3c8-440a-80d4-2b0460dc8b25","section_id":"sec_0ac1e5bb-e60a-4a9f-bb55-991409bbe5e3","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":51,"end":57,"exact":"エージェント","quote":"この構図では、モデルの賢さは必要条件であって、十分条件ではない。どのモデルをどの順番で使うか。どの専門エージェントに分担するか。どこで人間承認を挟むか。どのログを残すか。ここが、AIプロダクトの品質とコストを分ける。","quote_start":0,"quote_end":108,"text_sha256":"9ecf29a6ff80dab9fe982fdf15701289c647313430e45aa3d5fb6a4a18da0df9","block_sha256":"9ecf29a6ff80dab9fe982fdf15701289c647313430e45aa3d5fb6a4a18da0df9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_9cdcc954-c3c8-440a-80d4-2b0460dc8b25"},{"id":"occ_550ffabe8ded8170f3b61fcf","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_9edc1b73-571d-4182-904c-f5872e602a5e","section_id":"sec_664d1a35-9f46-4d0d-956f-18ba2f21f6f5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":45,"end":51,"exact":"エージェント","quote":"ChatGPTの強さは、モデル単体だけでなく、検索、ファイル、コード、画像、ツール、推論、エージェント的実行が、一つの体験にまとまっている点にある。","quote_start":0,"quote_end":74,"text_sha256":"72ef455c68cad3794b7524f19a779b1193371e716cadb785972c93cea0429038","block_sha256":"72ef455c68cad3794b7524f19a779b1193371e716cadb785972c93cea0429038","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_9edc1b73-571d-4182-904c-f5872e602a5e"},{"id":"occ_99f7df3d9678a1653b0848e9","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_a3075d2c-e5b7-4a6b-96fb-ef2bc0130f3a","section_id":"sec_98579fc2-c6fa-4167-9af4-945cd97a1608","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":78,"end":84,"exact":"エージェント","quote":"問題を分解し、別のモデルが実装し、別のモデルが検証し、最後に統合する。Fugu Ultraでは、より多くの専門エージェントを使い、Kaggle、論文再現、サイバーセキュリティ分析、特許調査のような重いタスクを狙う。","quote_start":23,"quote_end":130,"text_sha256":"3920d6974fff686a015c8ac0b9640fa58e11edb0bcd0543a76a5e3644c1927f4","block_sha256":"3920d6974fff686a015c8ac0b9640fa58e11edb0bcd0543a76a5e3644c1927f4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_a3075d2c-e5b7-4a6b-96fb-ef2bc0130f3a"},{"id":"occ_7ce1f283be45888843de0296","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_b3b3e913-0ce8-427b-bc66-38c64cbed60e","section_id":"sec_42f65980-b59f-4137-9f0e-50ce58092323","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"現在のAIエージェント開発では、使うツールをあらかじめ開発者が設定しておくことが多い。たとえば、GitHubを使うMCPサーバー、ブラウザ検索ツール、社内データベース、Slack連携、決算書検索ツールなどを、人間が手動で接","quote_start":0,"quote_end":111,"text_sha256":"270a0701363ff36252cf522a051c45e47e61a61aef4471c095db6313b44f5283","block_sha256":"270a0701363ff36252cf522a051c45e47e61a61aef4471c095db6313b44f5283","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_b3b3e913-0ce8-427b-bc66-38c64cbed60e"},{"id":"occ_ba981449f5bada7ea478da6d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_b900fae8-1639-471c-aa2c-570114478389","section_id":"sec_d1a36cd7-e762-4953-918c-434ae1119220","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":78,"end":84,"exact":"エージェント","quote":"かは基本的に見えない。通常版Fuguでは一部モデルの除外ができるが、Fugu Ultraでは性能のために固定のエージェントプールを使う。つまり、ユーザーから見ると、便利になる一方で、内部の制御権はSakana側に移る。","quote_start":23,"quote_end":132,"text_sha256":"40cb3bca69f75a777b8448dac9dfc963dfd814e50911ac10e72f7b5c8e8d6296","block_sha256":"40cb3bca69f75a777b8448dac9dfc963dfd814e50911ac10e72f7b5c8e8d6296","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_b900fae8-1639-471c-aa2c-570114478389"},{"id":"occ_a6037412f864df398b541564","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_c119f9f6-c872-485f-964e-17378b9135b1","section_id":"sec_d32d447f-8628-43f4-8405-6489aaac3242","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":20,"end":26,"exact":"エージェント","quote":"# Fugu、ARD、A2Aが示す「AIエージェント時代」の本当の競争軸","quote_start":0,"quote_end":36,"text_sha256":"028c08ac96d95227ee7e20a3d357891378e96d5bb172592ddee9c4c6252501cd","block_sha256":"028c08ac96d95227ee7e20a3d357891378e96d5bb172592ddee9c4c6252501cd","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_c119f9f6-c872-485f-964e-17378b9135b1"},{"id":"occ_18d4eacf2e89144c89bc8212","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_c15785a3-d1cc-423f-a704-b79a1fe4bf2c","section_id":"sec_1b47f3d5-fbd9-4558-a206-29d81f69df96","layer":"character","character_id":"zetu_noia","count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":23,"end":29,"exact":"エージェント","quote":"私はそこに、AIインフラの新しい需要を見ます。エージェントが裏側で十回、百回と試し、別のモデルに聞き、別のツールを呼び、検証を繰り返す。人間の一問一答よりも、ずっと多い推論が回る。AIがAIを使う時代には、推論需要そのものが多段化していきます。","quote_start":0,"quote_end":122,"text_sha256":"5c366a2eaabcd069543bda603f8c4dabb1a872e4f454ae0c9ebd4ca03f3a2625","block_sha256":"5c366a2eaabcd069543bda603f8c4dabb1a872e4f454ae0c9ebd4ca03f3a2625","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_c15785a3-d1cc-423f-a704-b79a1fe4bf2c"},{"id":"occ_92ae87b25548510f9dafdac9","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_c500ce89-951c-4c42-9dfd-adc670700ea2","section_id":"sec_3220509b-4b37-497f-a2fc-863b0685f8d7","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":65,"end":71,"exact":"エージェント","quote":"M、DRAM、NAND、CXL、ネットワーク、ストレージ、データセンター、電力需要にとって追い風になる。特に、エージェントが自動で調査、実験、コード生成、検証を回す時代になると、推論需要は人間のチャット回数よりもはるかに大きくなる可能性がある。","quote_start":10,"quote_end":132,"text_sha256":"f13154fc7d2764da465d85a3158ec1e6a1a0414dc3b7aee2c4b6776b11a14f87","block_sha256":"f13154fc7d2764da465d85a3158ec1e6a1a0414dc3b7aee2c4b6776b11a14f87","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_c500ce89-951c-4c42-9dfd-adc670700ea2"},{"id":"occ_a75d18837e53ac20db8a8542","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_c76af0a0-594f-42ea-94d6-f538534f04c2","section_id":"sec_adc2b089-8f68-448c-8021-f306a270142a","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":20,"end":26,"exact":"エージェント","quote":"MCPはツールを使う。\nARDはツールやエージェントを探す。\nA2Aはエージェント同士が話す。\nFuguはどれを使うかを指揮する。","quote_start":0,"quote_end":65,"text_sha256":"2df6be2b5ad50e2a3285f7ffb3620c3e406681d6c8e38b43622958989fea2fdd","block_sha256":"2df6be2b5ad50e2a3285f7ffb3620c3e406681d6c8e38b43622958989fea2fdd","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_c76af0a0-594f-42ea-94d6-f538534f04c2"},{"id":"occ_55b35b2f057e3651c8261df0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_ce7edd11-8500-4085-8582-219dce23b498","section_id":"sec_adc2b089-8f68-448c-8021-f306a270142a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":22,"end":28,"exact":"エージェント","quote":"A2Aは、Agent2Agentの略で、AIエージェント同士が通信・協調するための標準である。","quote_start":0,"quote_end":47,"text_sha256":"0a127eccde5ad31e2015e272d27f3f070b504f4afe7ca4413212cabd6b64f242","block_sha256":"0a127eccde5ad31e2015e272d27f3f070b504f4afe7ca4413212cabd6b64f242","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_ce7edd11-8500-4085-8582-219dce23b498"},{"id":"occ_5b0634ad3ad1f1906b58aa8d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_da219a44-2199-449f-8e29-020ccac362ac","section_id":"sec_7e0c4909-5d2d-46b5-87cb-984fe85c2316","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":36,"end":42,"exact":"エージェント","quote":"Fugu型の仕組みは、ユーザーにとって便利である。内部で複数モデルや専門エージェントが動き、結果だけが返ってくるなら、利用者はモデル選択に悩まなくてよい。しかし、企業利用や主権AIの文脈では、便利さだけでは足りない。","quote_start":0,"quote_end":108,"text_sha256":"27ee3c91c9d6e2a3d78ad9d01f4102ad06d4c31d91578cc3cd1c94cdfe553954","block_sha256":"27ee3c91c9d6e2a3d78ad9d01f4102ad06d4c31d91578cc3cd1c94cdfe553954","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_da219a44-2199-449f-8e29-020ccac362ac"},{"id":"occ_fedee72ec73e04f0dc654cb5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_f7cdc7ab-0efb-427c-8860-0df94337070d","section_id":"sec_adc2b089-8f68-448c-8021-f306a270142a","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":22,"end":28,"exact":"エージェント","quote":"ここでMCPとの違いも重要になる。MCPは、エージェントがツールやデータに接続するための仕組みである。一方A2Aは、エージェント同士が互いに仕事を頼み合う仕組みである。","quote_start":0,"quote_end":84,"text_sha256":"90e86852a540bdd1706cde57675fd634e48377eb657e743d6c7b88e05579f783","block_sha256":"90e86852a540bdd1706cde57675fd634e48377eb657e743d6c7b88e05579f783","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_f7cdc7ab-0efb-427c-8860-0df94337070d"},{"id":"occ_b052c5213221d72d5e4004ac","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_fd8d38b8-b826-4d37-8ff5-09e1480e4f45","section_id":"sec_42f65980-b59f-4137-9f0e-50ce58092323","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"しかし、エージェントが本当に自律的に仕事をするなら、「この仕事に使える能力はどこにあるのか」を自分で探せる必要がある。","quote_start":0,"quote_end":59,"text_sha256":"a12a009482c79478ae831008b61528558a81140dfed75550e0b60eacd2856583","block_sha256":"a12a009482c79478ae831008b61528558a81140dfed75550e0b60eacd2856583","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_fd8d38b8-b826-4d37-8ff5-09e1480e4f45"},{"id":"occ_19364ee4141973cc7ffdbc8a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe","block_id":"blk_ffc33fc4-4db7-473f-b64b-dbdca4f9ae57","section_id":"sec_0ac1e5bb-e60a-4a9f-bb55-991409bbe5e3","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":116,"end":122,"exact":"エージェント","quote":"マーク、コンテキスト長、推論能力、コード性能が中心だった。もちろん最上位モデルの能力は今後も重要である。だが、エージェント時代には、強いモデルを一つ持つだけでは足りない。","quote_start":61,"quote_end":146,"text_sha256":"7c4f8fbc9ce36c279497da28abde3105ca58eef8604ed38815530f4c21a67bd7","block_sha256":"7c4f8fbc9ce36c279497da28abde3105ca58eef8604ed38815530f4c21a67bd7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bb9a9bde-af83-422a-9558-c75e6cb1da78/#blk_ffc33fc4-4db7-473f-b64b-dbdca4f9ae57"},{"id":"occ_b2b30c5ecda8a4c60348101f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920","block_id":"blk_0d2dfe8a-0f5f-41f3-802a-f551b5332c97","section_id":"sec_ea5643fa-5a5b-4232-ad12-1e4ebbf7a8da","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":50,"end":56,"exact":"エージェント","quote":"DeepSeekは極端なコスト効率を追求し、AlibabaやMoonshotは巨大MoEによる長時間エージェントを追求する。","quote_start":0,"quote_end":62,"text_sha256":"b6423cc98218fbbea729ef722962546f8af161eb3b4a328902399346e5c5e139","block_sha256":"b6423cc98218fbbea729ef722962546f8af161eb3b4a328902399346e5c5e139","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bcca5c5e-94f7-4bf9-b249-187454f09a22/#blk_0d2dfe8a-0f5f-41f3-802a-f551b5332c97"},{"id":"occ_9e9d0b9f254db8529f773893","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920","block_id":"blk_14fff35c-77d0-4c69-85c5-407462073a1f","section_id":"sec_92accf82-c2c2-420e-9442-0bba7fd58b12","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":125,"end":131,"exact":"エージェント","quote":"特に長時間実行されるワークフロー向けと位置付けられている。GPT-5.6では、単一モデルだけでなく、複数のサブエージェントを使う「ultra」モードも導入された。(OpenAI)","quote_start":70,"quote_end":159,"text_sha256":"f5b44701c61ba1953ca886f92db41f86e0231e62abd003b20ae16752def93012","block_sha256":"f5b44701c61ba1953ca886f92db41f86e0231e62abd003b20ae16752def93012","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bcca5c5e-94f7-4bf9-b249-187454f09a22/#blk_14fff35c-77d0-4c69-85c5-407462073a1f"},{"id":"occ_48f3ba659280115fd07d3992","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920","block_id":"blk_22d0dbdd-3dd7-483a-80a5-05d83951b2c9","section_id":"sec_587780e0-55d1-4b67-9287-65515f8370ff","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":11,"end":17,"exact":"エージェント","quote":"一方、一人当たり複数のエージェントが常時動けば、1GWで支えられる人口は数百万人から数千万人まで下がる。","quote_start":0,"quote_end":52,"text_sha256":"46dc387d2d95eb9c39b1d02901d9565c4018d816b5f6b0fdedf8964551236818","block_sha256":"46dc387d2d95eb9c39b1d02901d9565c4018d816b5f6b0fdedf8964551236818","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bcca5c5e-94f7-4bf9-b249-187454f09a22/#blk_22d0dbdd-3dd7-483a-80a5-05d83951b2c9"},{"id":"occ_fafc94ee8a90394543234cbf","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920","block_id":"blk_3da4d3ef-9d08-4b33-8fe5-1d39c52794b4","section_id":"sec_7caca598-9fb0-4ca3-8590-3b0021075e60","layer":"character","character_id":"zetu_noia","count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":47,"end":53,"exact":"エージェント","quote":"だから「効率が10倍ならデータセンターは10分の1でよい」とは言えません。何人が使うか、何体のエージェントが動くか、どれだけ長い仕事を任せるか、何回検証するか。その積が分母を追い越すかどうかが本当の観測点です。","quote_start":0,"quote_end":105,"text_sha256":"3b66d2ead29e0b04ba2808276ae62cc0499941487a26e31910d211dc2f0d5c18","block_sha256":"3b66d2ead29e0b04ba2808276ae62cc0499941487a26e31910d211dc2f0d5c18","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bcca5c5e-94f7-4bf9-b249-187454f09a22/#blk_3da4d3ef-9d08-4b33-8fe5-1d39c52794b4"},{"id":"occ_2b59b1dfc68e9212906011d8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920","block_id":"blk_4697b595-9cb1-4aa6-bf92-e92cea5bbfbf","section_id":"sec_82d27293-5711-40ec-84e9-bdf93072ee7e","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":158,"end":164,"exact":"エージェント","quote":" | 文書作成、分析、一般的なコーディング |\n| Max・フロンティアモデル | 難問、研究、重要判断、長期エージェント |\n| 複数エージェント系 | 探索、討論、検証、シミュレーション |","quote_start":103,"quote_end":200,"text_sha256":"f694664760b8f5108bc2c7b5e07b0e3b0c7dfca2c64bfa37ecd12ae296a66ed2","block_sha256":"f694664760b8f5108bc2c7b5e07b0e3b0c7dfca2c64bfa37ecd12ae296a66ed2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bcca5c5e-94f7-4bf9-b249-187454f09a22/#blk_4697b595-9cb1-4aa6-bf92-e92cea5bbfbf"},{"id":"occ_2a6f88bd7201eaee69900aa6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920","block_id":"blk_495769ae-1a34-4df5-97f6-1407c3ade790","section_id":"sec_92accf82-c2c2-420e-9442-0bba7fd58b12","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":33,"end":39,"exact":"エージェント","quote":"AnthropicもClaude Fable 5を「長時間稼働するエージェントのための次世代知能」と位置付けている。公開デモでは、工場建設ゲームを自律的に進めたり、CADソフトウェアと3Dモデルを作成したり、物理法則から太陽系シミュレーションを構築したりしている。(Anthro","quote_start":0,"quote_end":139,"text_sha256":"041f5333922a6778baad995f5c6ea99da8a0b066e066e744f2e82633766cefc2","block_sha256":"041f5333922a6778baad995f5c6ea99da8a0b066e066e744f2e82633766cefc2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bcca5c5e-94f7-4bf9-b249-187454f09a22/#blk_495769ae-1a34-4df5-97f6-1407c3ade790"},{"id":"occ_bdd2e82e125fab99c43e1acf","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920","block_id":"blk_5d7ceb8b-5051-4c3b-8818-2db25148fc86","section_id":"sec_64fada82-8e83-40ea-a899-c5da4ae29f8e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"複数エージェントを同時に持つ","quote_start":0,"quote_end":14,"text_sha256":"901eb6e6c265c44ca9e1749adf10b2045392b5ee91db2d082449e519eb7d7ad9","block_sha256":"901eb6e6c265c44ca9e1749adf10b2045392b5ee91db2d082449e519eb7d7ad9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bcca5c5e-94f7-4bf9-b249-187454f09a22/#blk_5d7ceb8b-5051-4c3b-8818-2db25148fc86"},{"id":"occ_c7d85eb12242ddf5498d63f0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920","block_id":"blk_696fd571-a481-42b8-a774-5264070b7f8c","section_id":"sec_11fe2347-e88b-45ad-af26-232d63e527a0","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":3,"end":9,"exact":"エージェント","quote":"長時間エージェント訓練","quote_start":0,"quote_end":11,"text_sha256":"5b41bf5324056dab3d514ab840e0eee3d011b4ed63013646bccc081471dfb94c","block_sha256":"5b41bf5324056dab3d514ab840e0eee3d011b4ed63013646bccc081471dfb94c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bcca5c5e-94f7-4bf9-b249-187454f09a22/#blk_696fd571-a481-42b8-a774-5264070b7f8c"},{"id":"occ_0d184c88b71f560fe6ac183b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920","block_id":"blk_6aee5d0d-92f7-4b6c-b8a1-45e91dab19bc","section_id":"sec_8d29fa84-1240-471c-a3e5-6426ce1c6280","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":37,"end":43,"exact":"エージェント","quote":"- 1トークン当たり効率と、1仕事当たり総計算量を分ける。\n- 利用者数、エージェント数、仕事数、試行回数、実行時間を追う。\n- AIDCは世代、電力費、冷却、ネットワーク、顧客集中で質が分かれる。\n- 旧モデルの陳腐化と、計算市場全体の消滅は同義ではない。\n- 最終需要が育たない場合","quote_start":0,"quote_end":143,"text_sha256":"6aaa7bbeaad4b5785d0d93f5661ed65a49c5ab697e7f1a70abcbd253aaa66eaf","block_sha256":"6aaa7bbeaad4b5785d0d93f5661ed65a49c5ab697e7f1a70abcbd253aaa66eaf","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bcca5c5e-94f7-4bf9-b249-187454f09a22/#blk_6aee5d0d-92f7-4b6c-b8a1-45e91dab19bc"},{"id":"occ_c7086fa45e4fa582e6cd43bd","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920","block_id":"blk_7691633c-512c-4664-a959-375dd787b55b","section_id":"sec_ea5643fa-5a5b-4232-ad12-1e4ebbf7a8da","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"人間や別エージェントへ報告する","quote_start":0,"quote_end":15,"text_sha256":"5ee549972a68cf135d49b03fede9ea31045d3ca107ad24d6b7a38776c4cd6732","block_sha256":"5ee549972a68cf135d49b03fede9ea31045d3ca107ad24d6b7a38776c4cd6732","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bcca5c5e-94f7-4bf9-b249-187454f09a22/#blk_7691633c-512c-4664-a959-375dd787b55b"},{"id":"occ_a216e5f241bdd1c46ba2abf0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920","block_id":"blk_88bae935-f140-4863-b99e-6a66506eb79c","section_id":"sec_587780e0-55d1-4b67-9287-65515f8370ff","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":137,"end":143,"exact":"エージェント","quote":".5億～10億人 |\n| 仕事用コパイロット | 10万token | 約5000万～2億人 |\n| 常時稼働エージェント | 100万token | 約500万～2000万人 |","quote_start":82,"quote_end":172,"text_sha256":"585efbe4e2249798ccf028e7227f6d0d0640fe0a1bc4e87356b990e78e55421b","block_sha256":"585efbe4e2249798ccf028e7227f6d0d0640fe0a1bc4e87356b990e78e55421b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bcca5c5e-94f7-4bf9-b249-187454f09a22/#blk_88bae935-f140-4863-b99e-6a66506eb79c"},{"id":"occ_cc93a2acfa8d2b2fc1b7e814","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920","block_id":"blk_8ac51f25-a3f8-41aa-b840-efa413123ed9","section_id":"sec_1bf58aa7-c810-41d7-a8d6-4f6233904046","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"複数のサブエージェントを使う推論","quote_start":0,"quote_end":16,"text_sha256":"9eb97207a332ded006aaa8ebd205ff5d9e70521fbf11103c0dcc39b0e09ebced","block_sha256":"9eb97207a332ded006aaa8ebd205ff5d9e70521fbf11103c0dcc39b0e09ebced","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bcca5c5e-94f7-4bf9-b249-187454f09a22/#blk_8ac51f25-a3f8-41aa-b840-efa413123ed9"},{"id":"occ_566d9156b08bd83b6521ca1d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920","block_id":"blk_8bbb5002-5485-459b-9e5f-7640788297e2","section_id":"sec_11fe2347-e88b-45ad-af26-232d63e527a0","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"運用中のエージェントが失敗例を集め、それを次の訓練へ戻す。","quote_start":0,"quote_end":29,"text_sha256":"77290031f24447476c7125e095dc1e1ee3a90ef855d6c20c853e7b50025c2d1d","block_sha256":"77290031f24447476c7125e095dc1e1ee3a90ef855d6c20c853e7b50025c2d1d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bcca5c5e-94f7-4bf9-b249-187454f09a22/#blk_8bbb5002-5485-459b-9e5f-7640788297e2"},{"id":"occ_bd10489b64b8031e1e34f5c6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920","block_id":"blk_94769d3f-34a5-4941-bc46-351ec72341d9","section_id":"sec_1bf58aa7-c810-41d7-a8d6-4f6233904046","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"科学研究エージェント","quote_start":0,"quote_end":10,"text_sha256":"829cbc3d76328e4a5001731e5be957fbf0d8176906e7a4a767bc318d568c15fe","block_sha256":"829cbc3d76328e4a5001731e5be957fbf0d8176906e7a4a767bc318d568c15fe","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bcca5c5e-94f7-4bf9-b249-187454f09a22/#blk_94769d3f-34a5-4941-bc46-351ec72341d9"},{"id":"occ_daa17445f1d5fc256d2d2150","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920","block_id":"blk_99835737-5656-4387-8b50-dfcfc069854b","section_id":"sec_587780e0-55d1-4b67-9287-65515f8370ff","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":1,"end":7,"exact":"エージェント","quote":"別エージェントに反論させる","quote_start":0,"quote_end":13,"text_sha256":"75d64d4dcb5aa2e2222461200f358e1c63b3463914a05028d1f61c41ba153907","block_sha256":"75d64d4dcb5aa2e2222461200f358e1c63b3463914a05028d1f61c41ba153907","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bcca5c5e-94f7-4bf9-b249-187454f09a22/#blk_99835737-5656-4387-8b50-dfcfc069854b"},{"id":"occ_0d21f6685c9676e4118fa5c0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920","block_id":"blk_9eef7e21-e4a7-44df-89da-ac7246791ab7","section_id":"sec_de35e1ea-d76a-4512-85ff-2432d73678a6","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":7,"end":13,"exact":"エージェント","quote":"一人当たり複数エージェントが動く","quote_start":0,"quote_end":16,"text_sha256":"32041d7734e20d5faddc2e5718567ab2e6189fa98b2e232e2a85d12820a67b1a","block_sha256":"32041d7734e20d5faddc2e5718567ab2e6189fa98b2e232e2a85d12820a67b1a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bcca5c5e-94f7-4bf9-b249-187454f09a22/#blk_9eef7e21-e4a7-44df-89da-ac7246791ab7"},{"id":"occ_fe0627828ba7531e1079bbd7","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920","block_id":"blk_9f778d47-f630-4938-b11f-16ae4de5109a","section_id":"sec_1bf58aa7-c810-41d7-a8d6-4f6233904046","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"一人当たりのエージェントが3倍","quote_start":0,"quote_end":15,"text_sha256":"0241f014a553ed1242b5b2e130c2a128d4321ed62a279a75848c9b99ec12786a","block_sha256":"0241f014a553ed1242b5b2e130c2a128d4321ed62a279a75848c9b99ec12786a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bcca5c5e-94f7-4bf9-b249-187454f09a22/#blk_9f778d47-f630-4938-b11f-16ae4de5109a"},{"id":"occ_47d895906d7fec1e50cb5575","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920","block_id":"blk_b4f13bca-de67-4881-ae4e-0c97a7bf4366","section_id":"sec_8244ec6d-4cf3-4c40-a187-27c03189741a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":48,"end":54,"exact":"エージェント","quote":"IEAは、1回のAI処理に必要な電力が急速に下がっていることを認めながらも、AI利用者の増加と、エージェントのような計算集約的用途の拡大が効率改善を上回ると見ている。(IEA)","quote_start":0,"quote_end":88,"text_sha256":"2d14da9ba68f3099ff71bf94f7160fd60ddb877fd7968f30b86837a50e834586","block_sha256":"2d14da9ba68f3099ff71bf94f7160fd60ddb877fd7968f30b86837a50e834586","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bcca5c5e-94f7-4bf9-b249-187454f09a22/#blk_b4f13bca-de67-4881-ae4e-0c97a7bf4366"},{"id":"occ_961513d8338a576336b170f8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920","block_id":"blk_cd2c3258-fd9c-449c-a457-77662527ca30","section_id":"sec_1bf58aa7-c810-41d7-a8d6-4f6233904046","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":9,"end":15,"exact":"エージェント","quote":"長時間コーディングエージェント","quote_start":0,"quote_end":15,"text_sha256":"9ec5e3819ba76d230e969ab3eb064842bd12840027989b1e3dc0a4c27d4c3746","block_sha256":"9ec5e3819ba76d230e969ab3eb064842bd12840027989b1e3dc0a4c27d4c3746","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bcca5c5e-94f7-4bf9-b249-187454f09a22/#blk_cd2c3258-fd9c-449c-a457-77662527ca30"},{"id":"occ_f1604385194446b87b55f97a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920","block_id":"blk_d9388bb4-66a8-4c35-9f60-571fcc321382","section_id":"sec_f4bb836a-d5d3-41e2-a580-d9869d6efa2c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":146,"end":152,"exact":"エージェント","quote":"を扱いながら、高速かつ低価格な推論を目指して設計されている。2026年7月31日の正式版では、プレビュー版からエージェント能力も大きく強化された。(DeepSeek API Docs)","quote_start":91,"quote_end":183,"text_sha256":"f02c87bd727261cde1dc50fa6db8e4f4e1c1e42684753da8b3b01976f09abc1e","block_sha256":"f02c87bd727261cde1dc50fa6db8e4f4e1c1e42684753da8b3b01976f09abc1e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bcca5c5e-94f7-4bf9-b249-187454f09a22/#blk_d9388bb4-66a8-4c35-9f60-571fcc321382"},{"id":"occ_b29349e3ceffc3b701f695f9","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920","block_id":"blk_ee03af1e-0678-4074-93c5-d08d44a4d446","section_id":"sec_96e65b1e-68ce-4464-9c22-cf4ccb5a1be7","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"複数エージェントを管理する","quote_start":0,"quote_end":13,"text_sha256":"7a437b6b4d51da917dedb3ecf0c70011e2e5650c0010e9c0d969415808f27874","block_sha256":"7a437b6b4d51da917dedb3ecf0c70011e2e5650c0010e9c0d969415808f27874","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bcca5c5e-94f7-4bf9-b249-187454f09a22/#blk_ee03af1e-0678-4074-93c5-d08d44a4d446"},{"id":"occ_5c823a13c9f34c3de72e735f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a","work_id":"wrk_dcf18d88-5123-4a4f-aa9b-564dc767f048","block_id":"blk_1504045b-0da4-40fc-9927-35a0973a88ce","section_id":"sec_2d2d01a9-1272-4a8b-bee2-c8aa5c77b91b","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":62,"end":68,"exact":"エージェント","quote":"、Anthropicより黒字化が遅いと見られやすい存在です。大規模モデル開発、推論需要、動画生成、音声、画像、エージェント、インフラ投資まで抱えているため、費用の山は大きい。しかし、その分だけ上振れ余地も広い。","quote_start":7,"quote_end":112,"text_sha256":"af7ffab869298afc9295f0b2d5d6ca8be67e12617a814381fac1c677c6106b3c","block_sha256":"af7ffab869298afc9295f0b2d5d6ca8be67e12617a814381fac1c677c6106b3c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a/#blk_1504045b-0da4-40fc-9927-35a0973a88ce"},{"id":"occ_8347ec2b73052574907954b4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a","work_id":"wrk_dcf18d88-5123-4a4f-aa9b-564dc767f048","block_id":"blk_2dcc4161-ccf5-420a-b17d-08d6811d330a","section_id":"sec_0b7cac33-c3c4-452e-a0d1-e5f5303c2a85","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":518,"end":524,"exact":"エージェント","quote":"と企業ワークフローで強化                   | dynamic workflowsで数百サブエージェント化 |\n| 収益化      | OpenAIの黒字化エンジン                           | Anthropicの黒字化エンジン             |\n| 弱点    ","quote_start":463,"quote_end":624,"text_sha256":"d558a501e95a3336281e50ec2df7776d3bddee9eefcedf7e80860881f9becf97","block_sha256":"d558a501e95a3336281e50ec2df7776d3bddee9eefcedf7e80860881f9becf97","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a/#blk_2dcc4161-ccf5-420a-b17d-08d6811d330a"},{"id":"occ_0aa50c89bab6a0c869ab51d5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a","work_id":"wrk_dcf18d88-5123-4a4f-aa9b-564dc767f048","block_id":"blk_409809bd-3f79-46f3-94cf-9b6fa417f1e9","section_id":"sec_16e8a24f-d04b-4796-9e8d-197b74c4daa1","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":118,"end":124,"exact":"エージェント","quote":"orkflows**を導入しました。これは、Claudeが大規模タスクを計画し、1セッション内で数百の並列サブエージェントを動かし、出力を検証してから報告する機能です。Anthropicは、数十万行規模のコードベース移行のような作業を想定しています。([Anthropic][7])","quote_start":63,"quote_end":204,"text_sha256":"bf44a679dc8470e96de3a104b2ff1e981720b895769ae92a67a0785b2e42f60b","block_sha256":"bf44a679dc8470e96de3a104b2ff1e981720b895769ae92a67a0785b2e42f60b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a/#blk_409809bd-3f79-46f3-94cf-9b6fa417f1e9"},{"id":"occ_23ff166a83bbf7bb01b128aa","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a","work_id":"wrk_dcf18d88-5123-4a4f-aa9b-564dc767f048","block_id":"blk_4bca9f99-a523-4f7e-8fed-68a9e47b69c5","section_id":"sec_4aee217c-3c3a-43fe-9ee3-b9101a5f5f69","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":30,"end":36,"exact":"エージェント","quote":"**AIモデル企業は本当に利益を出せるのか\nAIコーディングエージェントは高粗利事業になるのか\n推論・訓練・クラウド費用を売上で吸収できるのか\nAI企業の評価額1兆ドルは正当化できるのか**","quote_start":0,"quote_end":95,"text_sha256":"7ac2d8c5728cef19bc0502fe8a07daf3c10e6c65c5b12224a31bd3fc29b20ec7","block_sha256":"7ac2d8c5728cef19bc0502fe8a07daf3c10e6c65c5b12224a31bd3fc29b20ec7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a/#blk_4bca9f99-a523-4f7e-8fed-68a9e47b69c5"},{"id":"occ_6a48896fa91da1fcf26203d3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a","work_id":"wrk_dcf18d88-5123-4a4f-aa9b-564dc767f048","block_id":"blk_676150d9-e966-458e-a290-f4154dde4036","section_id":"sec_83bf4d3b-39a4-4d26-8fc4-1a1e62a3759a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":90,"end":96,"exact":"エージェント","quote":"に強みを持つ方向へ伸びています。Codexは、OpenAIのユーザー基盤、ChatGPTとの統合、開発者体験、エージェント的な操作性で広がっています。どちらが勝つかは、単純なモデル性能だけでは決まりません。","quote_start":35,"quote_end":138,"text_sha256":"dfd2d74a4d892c3731da098565bcbd8fae970645dbc4dac47f300542a7f0f228","block_sha256":"dfd2d74a4d892c3731da098565bcbd8fae970645dbc4dac47f300542a7f0f228","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a/#blk_676150d9-e966-458e-a290-f4154dde4036"},{"id":"occ_e625be87ae09f7e7679f5c18","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a","work_id":"wrk_dcf18d88-5123-4a4f-aa9b-564dc767f048","block_id":"blk_a18ace1b-6cb6-4965-9ea1-b1a3e84349a2","section_id":"sec_24ce0414-970d-4aa3-94b2-e6d952e67f1b","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":27,"end":33,"exact":"エージェント","quote":"OpenAIが物理AIに強い可能性を持つのは、モデル、エージェント、マルチモーダル理解、開発者エコシステム、消費者接点をすでに持っているからです。ロボットは、ただ腕を動かすだけでは足りません。言葉を理解し、画像を理解し、計画し、失敗を修正し、道具を使い、人間の指示を","quote_start":0,"quote_end":133,"text_sha256":"28f46a8a2fbb85248098cc8d55de96223b2ef4cab129ac842032015d237695b8","block_sha256":"28f46a8a2fbb85248098cc8d55de96223b2ef4cab129ac842032015d237695b8","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a/#blk_a18ace1b-6cb6-4965-9ea1-b1a3e84349a2"},{"id":"occ_e5a34e3ca68ddb075c1c21d8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a","work_id":"wrk_dcf18d88-5123-4a4f-aa9b-564dc767f048","block_id":"blk_b32cfc85-0a73-4f27-ba25-c65fb6c82eb8","section_id":"sec_16e8a24f-d04b-4796-9e8d-197b74c4daa1","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":64,"end":70,"exact":"エージェント","quote":"deは、単なるコード補完ではなく、リポジトリを読み、設計を理解し、修正し、テストし、長時間の開発作業を進める「エージェント型コーディングツール」に近づいています。","quote_start":9,"quote_end":90,"text_sha256":"9c4f5680abfce55c92fc56f709cd0aebc0c77f21b1c1ca4af94fe25d12cf46ae","block_sha256":"9c4f5680abfce55c92fc56f709cd0aebc0c77f21b1c1ca4af94fe25d12cf46ae","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a/#blk_b32cfc85-0a73-4f27-ba25-c65fb6c82eb8"},{"id":"occ_95020926ba01fae58ee1bb69","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a","work_id":"wrk_dcf18d88-5123-4a4f-aa9b-564dc767f048","block_id":"blk_b6909581-6b3c-4849-9993-cd1f97829797","section_id":"sec_e86d3884-c729-42f7-93a4-aba48f6bf0ec","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":54,"end":60,"exact":"エージェント","quote":"Sam Altmanは「The Gentle Singularity」で、2025年には実際の認知作業を行うエージェント、2026年には新しい洞察を出すシステム、2027年には現実世界でタスクを行うロボットが登場する可能性があると述べています。さらに、ロボットがロボットやデータセンター、チップ工場を作れるようになれば、","quote_start":0,"quote_end":160,"text_sha256":"f0dcfd9b260352ed6126a8a3b80084277c38459cf8baead62113bfdd77d4e85d","block_sha256":"f0dcfd9b260352ed6126a8a3b80084277c38459cf8baead62113bfdd77d4e85d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a/#blk_b6909581-6b3c-4849-9993-cd1f97829797"},{"id":"occ_64f2e52139cca0049a09a1d5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_bf97dda8-b55b-4e4d-b1cf-b775474af9b2","work_id":"wrk_72fede57-1e40-4a22-81c2-67b2328ba457","block_id":"blk_a6af8a3f-7710-4740-ad64-5680683a9593","section_id":"sec_27ceab9f-05ba-4f7b-b144-726e5e287f7e","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":0,"end":7,"exact":"Agentic","quote":"Agentic AI","quote_start":0,"quote_end":10,"text_sha256":"f679e01a69740bdf12b36242424bf9ceebf4528d505428f45d0ba58eaa76a0e4","block_sha256":"f679e01a69740bdf12b36242424bf9ceebf4528d505428f45d0ba58eaa76a0e4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_bf97dda8-b55b-4e4d-b1cf-b775474af9b2/#blk_a6af8a3f-7710-4740-ad64-5680683a9593"},{"id":"occ_71f90c016382a005ad69f8da","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c1530075-b9e3-4144-b8ce-f6e480c778fa","work_id":"wrk_50367f10-59d7-41cc-990d-213cec7e7f79","block_id":"blk_005b2f08-5d5b-4089-b263-d2e8f42c8459","section_id":"sec_6be829d2-a1cb-4a24-bd65-1473c6406068","layer":"body","character_id":null,"count":3,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":67,"end":73,"exact":"エージェント","quote":"I開発の歴史を、人間がすべての工程を担う段階から、チャットボットが短いコード補助をする段階、そしてコーディングエージェントがファイル単位で編集する段階へ進んできたと整理している。現在では、エージェントが自分でコードを実行し、他のエージェントへ作業を委任する段階に入りつつある。([Anthropic][3])","quote_start":12,"quote_end":166,"text_sha256":"cf1ec9f40a90dde982e055580a282dc8c1c620fe2334c167c39c85bc37e3e955","block_sha256":"cf1ec9f40a90dde982e055580a282dc8c1c620fe2334c167c39c85bc37e3e955","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c1530075-b9e3-4144-b8ce-f6e480c778fa/#blk_005b2f08-5d5b-4089-b263-d2e8f42c8459"},{"id":"occ_c979ee0e2a5efdb36121afec","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c1530075-b9e3-4144-b8ce-f6e480c778fa","work_id":"wrk_50367f10-59d7-41cc-990d-213cec7e7f79","block_id":"blk_03370fd5-c48d-4ce3-b349-4c003b07860a","section_id":"sec_2263c3b4-10a4-4d2d-9c36-f188d314858f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":37,"end":43,"exact":"エージェント","quote":"Webを巡回する。\n情報を集める。\nコードを書く。\n実験を実行する。\n別のエージェントに作業を投げる。\n結果を評価する。\n人間の意思決定を代行する。","quote_start":0,"quote_end":74,"text_sha256":"2f6f6907004e65e17b1bcca3cbe6d41e0a28088cce23100c6ee20b98d4c945d5","block_sha256":"2f6f6907004e65e17b1bcca3cbe6d41e0a28088cce23100c6ee20b98d4c945d5","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c1530075-b9e3-4144-b8ce-f6e480c778fa/#blk_03370fd5-c48d-4ce3-b349-4c003b07860a"},{"id":"occ_d78fbc2f7791589eefd63ae6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c1530075-b9e3-4144-b8ce-f6e480c778fa","work_id":"wrk_50367f10-59d7-41cc-990d-213cec7e7f79","block_id":"blk_19a90316-295d-4b5e-a1f5-688a08a5aeca","section_id":"sec_e7c0487f-0d08-4bc6-89ec-4da00dd79fbf","layer":"body","character_id":null,"count":2,"matched_aliases":["agentic","エージェント"],"evidence":{"text_basis":"markdown","start":96,"end":103,"exact":"agentic","quote":"、humanが42.5%になったとされている。Cloudflare CEOのMatthew Prince氏も、agentic traffic、つまりAIエージェント的なトラフィックの伸びが予想より速く、2027年より前にボットが人間を超えたと述べた。([Tom's Hardware][2])","quote_start":41,"quote_end":187,"text_sha256":"ddeef738be40a4d41cac013e5c5248585ab925b15696bebe75641f384e2629a8","block_sha256":"ddeef738be40a4d41cac013e5c5248585ab925b15696bebe75641f384e2629a8","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c1530075-b9e3-4144-b8ce-f6e480c778fa/#blk_19a90316-295d-4b5e-a1f5-688a08a5aeca"},{"id":"occ_a59325df529059eb740e26ac","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c1530075-b9e3-4144-b8ce-f6e480c778fa","work_id":"wrk_50367f10-59d7-41cc-990d-213cec7e7f79","block_id":"blk_41adbd5d-b031-4412-82d8-9b164b99afe7","section_id":"sec_b7b10151-a335-44a0-ad66-44228b201aa4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":48,"end":54,"exact":"エージェント","quote":"人間社会はこれまで、人間の認知、労働時間、組織能力、教育速度によって制約されていた。\nしかしAIエージェントが普及すると、人間の意図はAIに委任され、AIは大量の情報探索、実装、検証、実験、判断補助を並列に行う。","quote_start":0,"quote_end":106,"text_sha256":"8d73c5cf2e2f3fa10868aa1c0a60f662ffe390e515e4638b38f0ec8f0d1ec532","block_sha256":"8d73c5cf2e2f3fa10868aa1c0a60f662ffe390e515e4638b38f0ec8f0d1ec532","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c1530075-b9e3-4144-b8ce-f6e480c778fa/#blk_41adbd5d-b031-4412-82d8-9b164b99afe7"},{"id":"occ_3296471888d271858b848a0d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c1530075-b9e3-4144-b8ce-f6e480c778fa","work_id":"wrk_50367f10-59d7-41cc-990d-213cec7e7f79","block_id":"blk_4238c332-052a-4fd3-990f-d9aebf22daf0","section_id":"sec_b63a2f23-5f8b-4758-bf2e-0cdcaa874712","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":106,"end":112,"exact":"エージェント","quote":"能 | ベンチマークは伸びるか | 実業務で何回使われるか |\n| Web | 人間のPVが増えるか | AIエージェントのアクセス権をどう扱うか |\n| 開発 | 人間がコードを書くか | AIがレビュー、修正、テストまで回すか |\n| インフラ | 学習用GPUが足りるか | 常時推論、検証、ログ保存まで足りるか |","quote_start":51,"quote_end":212,"text_sha256":"17a1b2ed5c48e85e6e3d6752ba568dc79566b808b94b1500f8da5894f5b3f1b5","block_sha256":"17a1b2ed5c48e85e6e3d6752ba568dc79566b808b94b1500f8da5894f5b3f1b5","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c1530075-b9e3-4144-b8ce-f6e480c778fa/#blk_4238c332-052a-4fd3-990f-d9aebf22daf0"},{"id":"occ_64ef4c6c0210fdc4a8935584","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c1530075-b9e3-4144-b8ce-f6e480c778fa","work_id":"wrk_50367f10-59d7-41cc-990d-213cec7e7f79","block_id":"blk_78c9ff82-0f08-40aa-ada0-06758bccdfc2","section_id":"sec_3c250fc0-2901-4927-8b3e-c77936da21b8","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":15,"end":21,"exact":"エージェント","quote":"**どれだけの人間活動が、AIエージェント、推論インフラ、ロボティクス、Webアクセス、コード生成、研究開発自動化へ置き換わるのか。**","quote_start":0,"quote_end":68,"text_sha256":"a8311ab5862d11519f11eabe35d2bb9ea39880ec135de040b242fc78d7f0b5e3","block_sha256":"a8311ab5862d11519f11eabe35d2bb9ea39880ec135de040b242fc78d7f0b5e3","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c1530075-b9e3-4144-b8ce-f6e480c778fa/#blk_78c9ff82-0f08-40aa-ada0-06758bccdfc2"},{"id":"occ_1bee24f95a1b72f0123c026c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c1530075-b9e3-4144-b8ce-f6e480c778fa","work_id":"wrk_50367f10-59d7-41cc-990d-213cec7e7f79","block_id":"blk_94b76b22-0a21-4416-9da2-a27cc843db22","section_id":"sec_91250c42-f6c4-403f-9316-ac0bceefcc12","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントが大量にWebへアクセスすれば、サイト運営者は人間とbotの区別、課金、アクセス権、コンテンツ利用ルールを再設計しなければならない。AIがコードを書けば、レビューとテストが増える。AIが研究を助ければ","quote_start":0,"quote_end":108,"text_sha256":"14a4afd00b315e322e5f95fa2249c75a5850267c087456c05ffc69a3079ccb58","block_sha256":"14a4afd00b315e322e5f95fa2249c75a5850267c087456c05ffc69a3079ccb58","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c1530075-b9e3-4144-b8ce-f6e480c778fa/#blk_94b76b22-0a21-4416-9da2-a27cc843db22"},{"id":"occ_f26ac20436042a6d3f940a9f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c1530075-b9e3-4144-b8ce-f6e480c778fa","work_id":"wrk_50367f10-59d7-41cc-990d-213cec7e7f79","block_id":"blk_9989a591-9709-45b4-9075-f82c8ca1bc37","section_id":"sec_08a397fd-eaf1-406c-be53-142803266fa2","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"ところがAIエージェント時代には、この流れが変わる。","quote_start":0,"quote_end":26,"text_sha256":"d37134756695d4d7816a4f90baba30541f91be8aafd88fdbd325633bad1023bd","block_sha256":"d37134756695d4d7816a4f90baba30541f91be8aafd88fdbd325633bad1023bd","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c1530075-b9e3-4144-b8ce-f6e480c778fa/#blk_9989a591-9709-45b4-9075-f82c8ca1bc37"},{"id":"occ_68f37c58327074a90414e4ef","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c1530075-b9e3-4144-b8ce-f6e480c778fa","work_id":"wrk_50367f10-59d7-41cc-990d-213cec7e7f79","block_id":"blk_a20cb03b-5694-40a1-ab2f-4a59bcd40160","section_id":"sec_9db6ecbd-adb2-4f81-a920-885de0e446fe","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":65,"end":71,"exact":"エージェント","quote":"需要の本体は、\n「より賢い文章を返すための学習」から、\n**より多くの現実タスクをAIに実行させるための推論・エージェント・実験・通信需要**へ移っていく。","quote_start":10,"quote_end":88,"text_sha256":"898b8dd02e778914d9b0c73953c19f756d1c219c8faffeef2d599251813615d7","block_sha256":"898b8dd02e778914d9b0c73953c19f756d1c219c8faffeef2d599251813615d7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c1530075-b9e3-4144-b8ce-f6e480c778fa/#blk_a20cb03b-5694-40a1-ab2f-4a59bcd40160"},{"id":"occ_0979be2bade8cc6e853695d9","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c1530075-b9e3-4144-b8ce-f6e480c778fa","work_id":"wrk_50367f10-59d7-41cc-990d-213cec7e7f79","block_id":"blk_b70ece0b-e2cb-4a06-911b-437209bf7eba","section_id":"sec_567adb79-3d56-4cdc-9576-368d9e56e2d8","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":70,"end":76,"exact":"エージェント","quote":"てもらう技術だった。\nAEOは、AIの回答に使われる技術になる。\nAgent Optimizationは、AIエージェントが正しく読み、比較し、判断し、引用し、必要なら支払いまでできるようにする設計になる。","quote_start":15,"quote_end":118,"text_sha256":"34d290a3837012a789620fa27de522e2d4ba686bc3e742d97437ead92c53dc27","block_sha256":"34d290a3837012a789620fa27de522e2d4ba686bc3e742d97437ead92c53dc27","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c1530075-b9e3-4144-b8ce-f6e480c778fa/#blk_b70ece0b-e2cb-4a06-911b-437209bf7eba"},{"id":"occ_55fbf0f7d88e52153f8c890b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c1530075-b9e3-4144-b8ce-f6e480c778fa","work_id":"wrk_50367f10-59d7-41cc-990d-213cec7e7f79","block_id":"blk_bfea4ccc-eb6a-4370-8667-7b08b48bc5f1","section_id":"sec_567adb79-3d56-4cdc-9576-368d9e56e2d8","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"しかしAIエージェント時代には、AIが記事を読み、要約し、ユーザーに答えを返す。\n元サイトには人間が来ない。\n広告も表示されない。\n購読にもつながらない。\nだが、サーバー負荷だけは発生する。","quote_start":0,"quote_end":95,"text_sha256":"7b43b64c38480478138c959c6fc0571f352c236d2a192836afc42e00132a3cea","block_sha256":"7b43b64c38480478138c959c6fc0571f352c236d2a192836afc42e00132a3cea","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c1530075-b9e3-4144-b8ce-f6e480c778fa/#blk_bfea4ccc-eb6a-4370-8667-7b08b48bc5f1"},{"id":"occ_f38a7d5f420f294e4ac10cb9","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c1530075-b9e3-4144-b8ce-f6e480c778fa","work_id":"wrk_50367f10-59d7-41cc-990d-213cec7e7f79","block_id":"blk_c717779d-9919-4300-84b2-509699ffbce0","section_id":"sec_f3be83b4-7f0f-4bcc-981a-5b8a332ce3e3","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":95,"end":101,"exact":"エージェント","quote":"、より多いデータ、より大きな計算資源。それによって、AIがどれだけ賢くなるのかが議論の中心だった。しかし、AIエージェント、コード生成、自動調査、Webアクセス、ロボティクスが立ち上がると、問いは少し変わる。","quote_start":40,"quote_end":144,"text_sha256":"a5ce2543d069554614423d6c0e64e0ff88d0981852853126afdf39dfda54a8f2","block_sha256":"a5ce2543d069554614423d6c0e64e0ff88d0981852853126afdf39dfda54a8f2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c1530075-b9e3-4144-b8ce-f6e480c778fa/#blk_c717779d-9919-4300-84b2-509699ffbce0"},{"id":"occ_785870bad8f837c8a6867c02","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c1530075-b9e3-4144-b8ce-f6e480c778fa","work_id":"wrk_50367f10-59d7-41cc-990d-213cec7e7f79","block_id":"blk_dabb7d28-d100-480a-bda8-4a37a33a8f55","section_id":"sec_08a397fd-eaf1-406c-be53-142803266fa2","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":42,"end":48,"exact":"エージェント","quote":"これは「人間がインターネットから消えた」という話ではない。\nむしろ、人間の意図がAIエージェントに委任されている。\n人間1人の背後で、AIが数十、数百、数千のページを読み、調べ、判断する。","quote_start":0,"quote_end":94,"text_sha256":"7409565eda87a0586121fb5006654e056a017aa2f41ed8e79c118dfb52f99dc3","block_sha256":"7409565eda87a0586121fb5006654e056a017aa2f41ed8e79c118dfb52f99dc3","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c1530075-b9e3-4144-b8ce-f6e480c778fa/#blk_dabb7d28-d100-480a-bda8-4a37a33a8f55"},{"id":"occ_02e850b6384b532d074e13a6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c1530075-b9e3-4144-b8ce-f6e480c778fa","work_id":"wrk_50367f10-59d7-41cc-990d-213cec7e7f79","block_id":"blk_f54399b8-4b65-4801-9170-93ae70fd228a","section_id":"sec_56b4b416-d82b-46be-be56-98e21ec1db93","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":18,"end":24,"exact":"エージェント","quote":"この内容はAIインフラとWeb/AIエージェント動向の観測であり、個別銘柄の売買を推奨するものではありません。実際の判断では、一次情報、利用条件、規制、収益モデルを必ず確認してください。","quote_start":0,"quote_end":93,"text_sha256":"46da97e48256a57b75da3ba4633b3e3971276a2fa0f59cbab493a374249eb7ae","block_sha256":"46da97e48256a57b75da3ba4633b3e3971276a2fa0f59cbab493a374249eb7ae","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c1530075-b9e3-4144-b8ce-f6e480c778fa/#blk_f54399b8-4b65-4801-9170-93ae70fd228a"},{"id":"occ_af6da19fc8750213542595a7","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c1530075-b9e3-4144-b8ce-f6e480c778fa","work_id":"wrk_50367f10-59d7-41cc-990d-213cec7e7f79","block_id":"blk_fa17016e-0139-4e67-ba99-8f1bc6b86b23","section_id":"sec_bcd70ef8-3578-4856-8b37-f4d224ce2396","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":63,"end":69,"exact":"エージェント","quote":"。\nすると推論需要が増える。\n推論需要が増えると、GPU、HBM、ネットワーク、電力、冷却が必要になる。\nAIエージェントが増えると、Webアクセス、Bot Management、WAF、ID管理、署名付きエージェント、レート制限が必要になる。\nロボティクスやVLAが進むと、実世界データ、シミュレーション、センサー、アク","quote_start":8,"quote_end":169,"text_sha256":"27531114406ddc839f3de50525dd2590af41fec63902caef7864c3573ebe88f6","block_sha256":"27531114406ddc839f3de50525dd2590af41fec63902caef7864c3573ebe88f6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c1530075-b9e3-4144-b8ce-f6e480c778fa/#blk_fa17016e-0139-4e67-ba99-8f1bc6b86b23"},{"id":"occ_a13686462103c57e7f51f946","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42","work_id":"wrk_675d7d8c-da38-4285-a483-5126d06bc835","block_id":"blk_0a4a281f-8bd6-40d5-b91b-4b740ffe5e75","section_id":"sec_5189a9a0-17a0-4540-920b-ce4749879c93","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":19,"end":27,"exact":"AI Agent","quote":"いくらクラウド上で暗号化されていても、AI Agentが画像を解析するときにはどこかで復号しなければならない。","quote_start":0,"quote_end":55,"text_sha256":"437aaca6c491583c5dce860b43ac433e2b74f6b12cbfb4af66133dc62c99f248","block_sha256":"437aaca6c491583c5dce860b43ac433e2b74f6b12cbfb4af66133dc62c99f248","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42/#blk_0a4a281f-8bd6-40d5-b91b-4b740ffe5e75"},{"id":"occ_6d9b2422bd887c9ec56a2777","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42","work_id":"wrk_675d7d8c-da38-4285-a483-5126d06bc835","block_id":"blk_18152c25-681a-48cb-83fe-60c2127bcc65","section_id":"sec_d8fcad11-db48-4df8-a092-9f1702a6b75a","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":33,"end":41,"exact":"AI Agent","quote":"**Storage + Compute + Software + AI Agent**","quote_start":0,"quote_end":43,"text_sha256":"a7ee099b18d02b8349f83ddf71100cbadeb3f6745b34802708c3cebceb8178f3","block_sha256":"a7ee099b18d02b8349f83ddf71100cbadeb3f6745b34802708c3cebceb8178f3","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42/#blk_18152c25-681a-48cb-83fe-60c2127bcc65"},{"id":"occ_97d7a5f3887018082454ec67","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42","work_id":"wrk_675d7d8c-da38-4285-a483-5126d06bc835","block_id":"blk_26cd12e1-f631-4044-9170-96040657ed7d","section_id":"sec_14e563f2-29dc-4abf-b0c6-e9c280ef3a53","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":3,"end":11,"exact":"AI Agent","quote":"しかしAI Agentが普及すると、","quote_start":0,"quote_end":18,"text_sha256":"dd31fa7f001bb7bfadae6096717582b42020c0e344f6b85338ef733f931e3e8b","block_sha256":"dd31fa7f001bb7bfadae6096717582b42020c0e344f6b85338ef733f931e3e8b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42/#blk_26cd12e1-f631-4044-9170-96040657ed7d"},{"id":"occ_8c8e7a01b4ab35fabb89a54e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42","work_id":"wrk_675d7d8c-da38-4285-a483-5126d06bc835","block_id":"blk_4c647af2-c8f0-4574-a7a4-3b3b6dc656f6","section_id":"sec_137e1f01-90cc-4caf-9298-ce2ade456b1a","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":9,"end":17,"exact":"AI Agent","quote":"Personal AI Agent。","quote_start":0,"quote_end":18,"text_sha256":"e1c1413aa51e064b21b6ee02019092686c2d34de496081a62788c4bf256a4f5c","block_sha256":"e1c1413aa51e064b21b6ee02019092686c2d34de496081a62788c4bf256a4f5c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42/#blk_4c647af2-c8f0-4574-a7a4-3b3b6dc656f6"},{"id":"occ_236a90399815a8314534ca43","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42","work_id":"wrk_675d7d8c-da38-4285-a483-5126d06bc835","block_id":"blk_561c441f-6f1e-45e4-a6a6-3e26a32dd2d4","section_id":"sec_62474a07-2df7-4fbf-acc5-7ce017b0d031","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":4,"end":12,"exact":"AI Agent","quote":"なぜならAI Agentは、人間よりもはるかに速くデータを読むからである。","quote_start":0,"quote_end":37,"text_sha256":"47a260f56f301623ed02bf881ff3f63fde271b033e14eca43330e1511c255069","block_sha256":"47a260f56f301623ed02bf881ff3f63fde271b033e14eca43330e1511c255069","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42/#blk_561c441f-6f1e-45e4-a6a6-3e26a32dd2d4"},{"id":"occ_dd3663cffa87c144dc7ca1b5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42","work_id":"wrk_675d7d8c-da38-4285-a483-5126d06bc835","block_id":"blk_589a6ea5-1f72-4dfc-a3a6-e6914c1bb991","section_id":"sec_14e563f2-29dc-4abf-b0c6-e9c280ef3a53","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":5,"end":13,"exact":"AI Agent","quote":"**人間＋AI Agentのための5TB**","quote_start":0,"quote_end":22,"text_sha256":"6d1df33a16629794f1d1ced9cb09cd1f074e34e4627d5af6b1645f009f428b93","block_sha256":"6d1df33a16629794f1d1ced9cb09cd1f074e34e4627d5af6b1645f009f428b93","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42/#blk_589a6ea5-1f72-4dfc-a3a6-e6914c1bb991"},{"id":"occ_acfb9197e6be2912dd5eff4e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42","work_id":"wrk_675d7d8c-da38-4285-a483-5126d06bc835","block_id":"blk_7749991b-7b36-4519-ba59-fc1cc257f8ce","section_id":"sec_62474a07-2df7-4fbf-acc5-7ce017b0d031","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":13,"end":21,"exact":"AI Agent","quote":"その中でPersonal AI Agentが動く。","quote_start":0,"quote_end":25,"text_sha256":"70fef7f9cc052b60aa46b3b36540c1f373e7912f712ef4b310f50e113e13abf5","block_sha256":"70fef7f9cc052b60aa46b3b36540c1f373e7912f712ef4b310f50e113e13abf5","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42/#blk_7749991b-7b36-4519-ba59-fc1cc257f8ce"},{"id":"occ_2595d236341b48d49194f6a6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42","work_id":"wrk_675d7d8c-da38-4285-a483-5126d06bc835","block_id":"blk_e3226117-a098-41b3-b0ff-1d6bcb59f141","section_id":"sec_14e563f2-29dc-4abf-b0c6-e9c280ef3a53","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":6,"end":14,"exact":"AI Agent","quote":"しかし私は、AI Agent時代になると、この5TBの意味そのものが変わるのではないかと思う。","quote_start":0,"quote_end":47,"text_sha256":"5307c456583ddf5b5cf171a8e2d8bcdfe7678bb00a75710011e8ebd8ef1e823b","block_sha256":"5307c456583ddf5b5cf171a8e2d8bcdfe7678bb00a75710011e8ebd8ef1e823b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42/#blk_e3226117-a098-41b3-b0ff-1d6bcb59f141"},{"id":"occ_1e8006ccef4039e56efeb4c9","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42","work_id":"wrk_675d7d8c-da38-4285-a483-5126d06bc835","block_id":"blk_f63f1c9c-f992-4413-88b5-a8900bb1b48f","section_id":"sec_3234502a-1e78-415e-9ae0-e4bdf0f22094","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":4,"end":12,"exact":"AI Agent","quote":"## 「AI Agent用OS」がクラウド上に作られる","quote_start":0,"quote_end":27,"text_sha256":"4bdddf3b8e84874ac86e2e26b485de36d7d5622b30e7613d8093e3ff68bb5cb4","block_sha256":"4bdddf3b8e84874ac86e2e26b485de36d7d5622b30e7613d8093e3ff68bb5cb4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42/#blk_f63f1c9c-f992-4413-88b5-a8900bb1b48f"},{"id":"occ_ac60ddd8ee180aad00b25dbf","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42","work_id":"wrk_675d7d8c-da38-4285-a483-5126d06bc835","block_id":"blk_fa518a35-3cb6-45d5-955a-e094d399d71f","section_id":"sec_e10bbbbd-faa3-43bb-826a-ffbefe81aee5","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agentが自分のファイルを読み、","quote_start":0,"quote_end":20,"text_sha256":"e3cc7c8c8e42473e20f08609ee42a0675e04f556b049e10e02f22adfac480ca9","block_sha256":"e3cc7c8c8e42473e20f08609ee42a0675e04f556b049e10e02f22adfac480ca9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42/#blk_fa518a35-3cb6-45d5-955a-e094d399d71f"},{"id":"occ_68f3ca9998b4bc16c722fe7b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c832f985-caf8-4ac7-bf58-f785e8782922","work_id":"wrk_37471fb9-72db-4a02-a976-c42f743def51","block_id":"blk_0c9a692c-0217-4776-a8c9-4daab8c138ee","section_id":"sec_c93aee54-3d03-4694-8648-e6699e9982c9","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agentでは探索・生成・検証を並列化できる。","quote_start":0,"quote_end":26,"text_sha256":"ecfa85880cf7b114442dc028a90a11070cc0cfb942cb71c7aa9e6f9049fdcba5","block_sha256":"ecfa85880cf7b114442dc028a90a11070cc0cfb942cb71c7aa9e6f9049fdcba5","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c832f985-caf8-4ac7-bf58-f785e8782922/#blk_0c9a692c-0217-4776-a8c9-4daab8c138ee"},{"id":"occ_916ac3539e7390a7498ecda0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c832f985-caf8-4ac7-bf58-f785e8782922","work_id":"wrk_37471fb9-72db-4a02-a976-c42f743def51","block_id":"blk_4d9e4006-cc6e-480a-927e-288a5ead9a6b","section_id":"sec_2ccad8c8-fdbb-4d7f-bf70-50f725b082e0","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":14,"end":22,"exact":"AI Agent","quote":"これらは生命進化、文化進化、AI Agent系などで異なるProxyを用いて検証できる。","quote_start":0,"quote_end":44,"text_sha256":"03f0fdc2bed66dcf3a34a30f6ebea6122dc3622152a30de370bc6ad305116f77","block_sha256":"03f0fdc2bed66dcf3a34a30f6ebea6122dc3622152a30de370bc6ad305116f77","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c832f985-caf8-4ac7-bf58-f785e8782922/#blk_4d9e4006-cc6e-480a-927e-288a5ead9a6b"},{"id":"occ_f9f96935eed0fe222ae41acd","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c89081f7-eb04-4049-a4a4-68cbe681a3f5","work_id":"wrk_08ecb116-f4c7-4c99-9574-e485a5adb41f","block_id":"blk_6edcd044-5f78-4997-8ed2-c807bd939f96","section_id":"sec_0f28a83e-8fdf-45d8-bfe6-9d9ad26ba0ca","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agent / Model\n                               │\n             ┌─────────────────┼─────────────────┐\n       ","quote_start":0,"quote_end":108,"text_sha256":"94ad62e7c602543bfbb524cba78c0c70aab93997b0332c3eeeace7e38621b748","block_sha256":"94ad62e7c602543bfbb524cba78c0c70aab93997b0332c3eeeace7e38621b748","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c89081f7-eb04-4049-a4a4-68cbe681a3f5/#blk_6edcd044-5f78-4997-8ed2-c807bd939f96"},{"id":"occ_50cb278a35e5fc7f672d5a80","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c8b19a4c-93eb-4695-8f38-c8166d22cbd9","work_id":"wrk_f4bf8d6d-6894-4267-93c4-c8e1e447fab3","block_id":"blk_527f587f-a7d4-452e-bca8-3283f66b8116","section_id":"sec_1ada2bef-0680-49f9-b057-fb9a02d221e4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":30,"end":36,"exact":"エージェント","quote":"ミリ秒単位の安全制御は引き続きPLCやマイコンが担当し、AIエージェントはその上位で監視、診断、計画、運用を行う。Digiは物理AIの「脳」や「筋肉」ではなく、**神経網と運用管理層**に位置する。","quote_start":0,"quote_end":99,"text_sha256":"80dd72f2d16f90b7b5339ca1c899a7e5578a47679863efd38656f0b190ed362f","block_sha256":"80dd72f2d16f90b7b5339ca1c899a7e5578a47679863efd38656f0b190ed362f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c8b19a4c-93eb-4695-8f38-c8166d22cbd9/#blk_527f587f-a7d4-452e-bca8-3283f66b8116"},{"id":"occ_d634bc5d2e8388a4185c0971","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c8b19a4c-93eb-4695-8f38-c8166d22cbd9","work_id":"wrk_f4bf8d6d-6894-4267-93c4-c8e1e447fab3","block_id":"blk_66a04d26-5999-41ca-990a-e470dead0b2f","section_id":"sec_1ada2bef-0680-49f9-b057-fb9a02d221e4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":68,"end":74,"exact":"エージェント","quote":"ote Manager内部に組み込まれ、接続機器の診断、トラブルシューティング、ネットワーク運用を支援するAIエージェントである。これに先立ち、DigiはMCP Serverを提供し、企業のAIアシスタントがRemote ManagerやGenesisの機器情報へ安全にアクセスできる基盤を整備していた。([Digi](h","quote_start":13,"quote_end":174,"text_sha256":"6215253120479c55f93fa62dedeab6f24da394065426d4d4c1e8ad204e4ecf87","block_sha256":"6215253120479c55f93fa62dedeab6f24da394065426d4d4c1e8ad204e4ecf87","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c8b19a4c-93eb-4695-8f38-c8166d22cbd9/#blk_66a04d26-5999-41ca-990a-e470dead0b2f"},{"id":"occ_29ce7b37c060dca9f2c24be6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c8b19a4c-93eb-4695-8f38-c8166d22cbd9","work_id":"wrk_f4bf8d6d-6894-4267-93c4-c8e1e447fab3","block_id":"blk_dbc2da38-48b2-451d-a792-7f6d9c77f0da","section_id":"sec_1ada2bef-0680-49f9-b057-fb9a02d221e4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":30,"end":36,"exact":"エージェント","quote":"**センサーで状態を取得  \n→ AIが異常を判定  \n→ エージェントが対応を提案  \n→ 人間が承認  \n→ 再起動、設定変更、回線切り替え、点検指示を実行**","quote_start":0,"quote_end":82,"text_sha256":"860be225bc8f35316d7274936165a16cce5d038b720dd1c61289606d8b61522c","block_sha256":"860be225bc8f35316d7274936165a16cce5d038b720dd1c61289606d8b61522c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c8b19a4c-93eb-4695-8f38-c8166d22cbd9/#blk_dbc2da38-48b2-451d-a792-7f6d9c77f0da"},{"id":"occ_791f82086be44996e5a91c3e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c950677f-d85d-461a-bf3d-314cf6b08a37","work_id":"wrk_a1c11bb3-0dd1-46bc-9f3c-ec06fe1df011","block_id":"blk_0f4340fe-8f62-41b4-863f-742f368ac562","section_id":"sec_3b186da5-6148-4ec6-8107-71770aa82a18","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agent Securityでも同じことが起きるなら、","quote_start":0,"quote_end":30,"text_sha256":"d7bebaf3c3b13cd2c7c008f7aa1f249cce5ebf762b59b7a6568d20e76eda277c","block_sha256":"d7bebaf3c3b13cd2c7c008f7aa1f249cce5ebf762b59b7a6568d20e76eda277c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c950677f-d85d-461a-bf3d-314cf6b08a37/#blk_0f4340fe-8f62-41b4-863f-742f368ac562"},{"id":"occ_24030b03ff1fcc0481c18ffa","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c950677f-d85d-461a-bf3d-314cf6b08a37","work_id":"wrk_a1c11bb3-0dd1-46bc-9f3c-ec06fe1df011","block_id":"blk_2a56f22b-257c-47ac-a2b4-3d1928243737","section_id":"sec_de254171-4ea1-4c49-b16f-2e785a67caad","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":1,"end":8,"exact":"Agentic","quote":"「Agentic InternetのControl Planeになる」","quote_start":0,"quote_end":35,"text_sha256":"ace34b6477c7ff1917c22146de000342dd2c6965d68f61a1ce0e95d1739fe2a9","block_sha256":"ace34b6477c7ff1917c22146de000342dd2c6965d68f61a1ce0e95d1739fe2a9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c950677f-d85d-461a-bf3d-314cf6b08a37/#blk_2a56f22b-257c-47ac-a2b4-3d1928243737"},{"id":"occ_25ba7398fa19f64e665e3816","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c950677f-d85d-461a-bf3d-314cf6b08a37","work_id":"wrk_a1c11bb3-0dd1-46bc-9f3c-ec06fe1df011","block_id":"blk_2f45bb4c-f628-417c-9bc9-6832cfbfaa19","section_id":"sec_3b186da5-6148-4ec6-8107-71770aa82a18","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agent Securityは現在まだ、","quote_start":0,"quote_end":23,"text_sha256":"ee94a62a727f9d5bbd17de73220f204c0222e35666dc6638940c77dacc2464b4","block_sha256":"ee94a62a727f9d5bbd17de73220f204c0222e35666dc6638940c77dacc2464b4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c950677f-d85d-461a-bf3d-314cf6b08a37/#blk_2f45bb4c-f628-417c-9bc9-6832cfbfaa19"},{"id":"occ_adfa506e822ec3a8a00e3950","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c950677f-d85d-461a-bf3d-314cf6b08a37","work_id":"wrk_a1c11bb3-0dd1-46bc-9f3c-ec06fe1df011","block_id":"blk_5077df40-e065-432e-a394-eecfe6385fab","section_id":"sec_5f7ac40e-b8a2-407e-9cff-b6c80466cd38","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":47,"end":54,"exact":"Agentic","quote":"OpenAIが2026年5月に公開したThird-Party Evaluationの指針では、Agentic Modelを評価する場合、単にモデル名だけではなく、","quote_start":0,"quote_end":81,"text_sha256":"845f0daebe42eb6972e45e16c518780da053d3cc57b0928a19d5778e28b24adb","block_sha256":"845f0daebe42eb6972e45e16c518780da053d3cc57b0928a19d5778e28b24adb","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c950677f-d85d-461a-bf3d-314cf6b08a37/#blk_5077df40-e065-432e-a394-eecfe6385fab"},{"id":"occ_44ff2d0871339fc27f2b72bb","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c950677f-d85d-461a-bf3d-314cf6b08a37","work_id":"wrk_a1c11bb3-0dd1-46bc-9f3c-ec06fe1df011","block_id":"blk_85d9b11d-7274-4351-aa2a-26666d10dea6","section_id":"sec_4ca04f7e-70dc-4f5b-8e2c-3bb8a8b19752","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":0,"end":7,"exact":"Agentic","quote":"Agentic Internetの基盤企業","quote_start":0,"quote_end":21,"text_sha256":"5dd808be8c1a45b6e6f46d884e4975fa4c706a8a1fd0c9ce26532096cb714ad3","block_sha256":"5dd808be8c1a45b6e6f46d884e4975fa4c706a8a1fd0c9ce26532096cb714ad3","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c950677f-d85d-461a-bf3d-314cf6b08a37/#blk_85d9b11d-7274-4351-aa2a-26666d10dea6"},{"id":"occ_213f7066612218d66a4e2c39","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c950677f-d85d-461a-bf3d-314cf6b08a37","work_id":"wrk_a1c11bb3-0dd1-46bc-9f3c-ec06fe1df011","block_id":"blk_877d2e20-ada9-4dca-8621-22996690a6ca","section_id":"sec_8ac5cde6-b426-42c8-90ef-410dc0c084e4","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":21,"end":28,"exact":"Agentic","quote":"OpenAIはこの事件後、Astraを含むAgentic application全体に対して、危険行動やmisalignmentを対象とするuniversal monitoringを導入したと説明している。","quote_start":0,"quote_end":103,"text_sha256":"69c2079320c6253a318c841cfde573031003f2270b1a4bdc2624c217097a0a60","block_sha256":"69c2079320c6253a318c841cfde573031003f2270b1a4bdc2624c217097a0a60","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c950677f-d85d-461a-bf3d-314cf6b08a37/#blk_877d2e20-ada9-4dca-8621-22996690a6ca"},{"id":"occ_b25e13adae8f77e7a204e90d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c950677f-d85d-461a-bf3d-314cf6b08a37","work_id":"wrk_a1c11bb3-0dd1-46bc-9f3c-ec06fe1df011","block_id":"blk_8a0bcbe5-8b5a-4ed3-8eff-61a8af7104c6","section_id":"sec_a1ebb5bb-89cf-4a9e-9776-c9b443420c0b","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":0,"end":7,"exact":"Agentic","quote":"Agentic Internetという巨大な最終市場を買うならNETだが、Valuation Riskも最大級。","quote_start":0,"quote_end":56,"text_sha256":"f26efcb644acfed8be3256610d4b1baac2c36cff075e0ba4aac61c4d549fee70","block_sha256":"f26efcb644acfed8be3256610d4b1baac2c36cff075e0ba4aac61c4d549fee70","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c950677f-d85d-461a-bf3d-314cf6b08a37/#blk_8a0bcbe5-8b5a-4ed3-8eff-61a8af7104c6"},{"id":"occ_7697307a36365c9f2479986e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c950677f-d85d-461a-bf3d-314cf6b08a37","work_id":"wrk_a1c11bb3-0dd1-46bc-9f3c-ec06fe1df011","block_id":"blk_aea894bd-4892-45e2-8e51-4189a6297e0b","section_id":"sec_42516b62-e676-4950-aac4-56ca1d4a1084","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":0,"end":7,"exact":"Agentic","quote":"Agentic Internetの主要Control Planeになる","quote_start":0,"quote_end":35,"text_sha256":"2101a8683a2286fc53f25cdf0e4bf4cdf73244d84298c9b94b18430354927fa2","block_sha256":"2101a8683a2286fc53f25cdf0e4bf4cdf73244d84298c9b94b18430354927fa2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c950677f-d85d-461a-bf3d-314cf6b08a37/#blk_aea894bd-4892-45e2-8e51-4189a6297e0b"},{"id":"occ_2a73d6e14e31d8dd19b714dc","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c950677f-d85d-461a-bf3d-314cf6b08a37","work_id":"wrk_a1c11bb3-0dd1-46bc-9f3c-ec06fe1df011","block_id":"blk_cd8e2223-0748-4519-aa05-e13c7fb4d04f","section_id":"sec_70941684-6396-42ee-9f02-8eaf0bf8ce34","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agentが生み出す新しいTelemetryを取り込める。","quote_start":0,"quote_end":32,"text_sha256":"5b5d53e0163df66d45ba048d60dc7194c12986c65d08586c888849a547f3434d","block_sha256":"5b5d53e0163df66d45ba048d60dc7194c12986c65d08586c888849a547f3434d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c950677f-d85d-461a-bf3d-314cf6b08a37/#blk_cd8e2223-0748-4519-aa05-e13c7fb4d04f"},{"id":"occ_b951bbc036807a3c0d0d66b8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_c950677f-d85d-461a-bf3d-314cf6b08a37","work_id":"wrk_a1c11bb3-0dd1-46bc-9f3c-ec06fe1df011","block_id":"blk_f321b398-bbf0-40f6-bacd-d0c9e2409f57","section_id":"sec_bd230d02-266a-417f-80c4-7271e5d8f408","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agent Securityの世界で、","quote_start":0,"quote_end":22,"text_sha256":"c989f6b80b6c1105c8879c9ced14f385e64e30b048ef720db6aad7d61b3d2f40","block_sha256":"c989f6b80b6c1105c8879c9ced14f385e64e30b048ef720db6aad7d61b3d2f40","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_c950677f-d85d-461a-bf3d-314cf6b08a37/#blk_f321b398-bbf0-40f6-bacd-d0c9e2409f57"},{"id":"occ_a4f0eda42fe3d1d16b9fbdcf","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ca87695f-4cf9-40a1-bc61-6a123a71ffe6","work_id":"wrk_92c272f3-ada3-4a0c-815e-d33c175ffbe0","block_id":"blk_170e9883-2fce-4c51-874a-c1d2dbe60945","section_id":"sec_740b94d6-6323-44b4-810d-f00a04305aa2","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":104,"end":110,"exact":"エージェント","quote":"・重要インフラ向けの信頼アクセスモデルに近い位置づけだった。これらのモデルが、脆弱性探索、攻撃コード生成、長期エージェント作業、生物・化学領域の知識支援などに強いなら、米政府が「これは単なるチャットAIではなく、国家安全保障資産に近い」と見なすこと自体は理解できる。","quote_start":49,"quote_end":182,"text_sha256":"5021856888b934ce60d12b72ada378c221ed72cdf798d80872b01ec2cf6c8c5e","block_sha256":"5021856888b934ce60d12b72ada378c221ed72cdf798d80872b01ec2cf6c8c5e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ca87695f-4cf9-40a1-bc61-6a123a71ffe6/#blk_170e9883-2fce-4c51-874a-c1d2dbe60945"},{"id":"occ_e3e3b7fd6a12c603451f3006","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ca87695f-4cf9-40a1-bc61-6a123a71ffe6","work_id":"wrk_92c272f3-ada3-4a0c-815e-d33c175ffbe0","block_id":"blk_67b506e4-3310-4578-9d23-2186ed5fc1eb","section_id":"sec_3efd9001-55fc-4319-b803-fbe95e6d8808","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":32,"end":38,"exact":"エージェント","quote":"データ供給者、評価機関、監査機関、クラウド、モデル配信基盤、AIエージェント実行環境。  \nこれらが特定の巨大AI企業に近づきすぎると、顧客は警戒する。  \n今後は「強い」だけでなく、「中立である」ことが重要な競争力になる。","quote_start":0,"quote_end":112,"text_sha256":"5b703e6db5f5b8aa300642278e8c201bcfb0f443f240bff61d71d03e108342cd","block_sha256":"5b703e6db5f5b8aa300642278e8c201bcfb0f443f240bff61d71d03e108342cd","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ca87695f-4cf9-40a1-bc61-6a123a71ffe6/#blk_67b506e4-3310-4578-9d23-2186ed5fc1eb"},{"id":"occ_b359cb9786b6f0727e89f74a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ca87695f-4cf9-40a1-bc61-6a123a71ffe6","work_id":"wrk_92c272f3-ada3-4a0c-815e-d33c175ffbe0","block_id":"blk_d3b1bce7-df33-486a-b181-16cc553afcd5","section_id":"sec_1f955c53-8b8a-4652-95b9-169d25308b7a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":95,"end":101,"exact":"エージェント","quote":"給者の中立性。  \n評価機関の中立性。  \nセキュリティ監査の中立性。  \nモデル配信基盤の中立性。  \nAIエージェント実行環境の中立性。","quote_start":40,"quote_end":110,"text_sha256":"dbe3e21d307301d2d27617c56c2021bcafb25a9ced9077bf1c5ff5d068b2189e","block_sha256":"dbe3e21d307301d2d27617c56c2021bcafb25a9ced9077bf1c5ff5d068b2189e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ca87695f-4cf9-40a1-bc61-6a123a71ffe6/#blk_d3b1bce7-df33-486a-b181-16cc553afcd5"},{"id":"occ_3c0831fe32d576d2d3223559","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ca87695f-4cf9-40a1-bc61-6a123a71ffe6","work_id":"wrk_92c272f3-ada3-4a0c-815e-d33c175ffbe0","block_id":"blk_da5d43de-84b5-4be7-892f-bfa59beba2f3","section_id":"sec_2c388ffe-ce09-42d1-8655-42551b557352","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":77,"end":83,"exact":"エージェント","quote":"クセスを警戒すること自体には、一定の安全保障上の理由がある。サイバー攻撃、脆弱性探索、生物・化学領域、長期自律エージェント化。Fable 5やMythos 5のようなモデルが、これらの能力に近づいているなら、国家が警戒するのは自然である。","quote_start":22,"quote_end":141,"text_sha256":"ba9f49fda17aba91f4cce78106d0148f6e1161cf3550b4ffb57275d09487db10","block_sha256":"ba9f49fda17aba91f4cce78106d0148f6e1161cf3550b4ffb57275d09487db10","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ca87695f-4cf9-40a1-bc61-6a123a71ffe6/#blk_da5d43de-84b5-4be7-892f-bfa59beba2f3"},{"id":"occ_e4557177e8953b0bb3be2a8d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ca87695f-4cf9-40a1-bc61-6a123a71ffe6","work_id":"wrk_92c272f3-ada3-4a0c-815e-d33c175ffbe0","block_id":"blk_f13bc603-584d-467b-8673-92691f47339f","section_id":"sec_740b94d6-6323-44b4-810d-f00a04305aa2","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":142,"end":148,"exact":"エージェント","quote":"デルで対抗するのは当然だ。しかも、GPT-5.6がGPT-5.5から意味のある改善を持ち、コーディング、推論、エージェント、視覚、フロントエンド生成で強化されるなら、それは単なる小幅更新ではなく、Fable 5への明確な対抗馬になる。","quote_start":87,"quote_end":204,"text_sha256":"0e5eab01cbd2c6a3b501b59685f47ed4e9f7e2b6894229b966131735c4006424","block_sha256":"0e5eab01cbd2c6a3b501b59685f47ed4e9f7e2b6894229b966131735c4006424","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ca87695f-4cf9-40a1-bc61-6a123a71ffe6/#blk_f13bc603-584d-467b-8673-92691f47339f"},{"id":"occ_086c6eb5a7e8af79ca08f885","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499","block_id":"blk_07d42b8c-a175-4b4c-a73e-4f5ae6cd564a","section_id":"sec_7f9a26a7-7219-429a-8b27-f1a31b2e4e68","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":29,"end":35,"exact":"エージェント","quote":"ENPIREでは、これに相当する対象が物理世界へ移る。AIエージェントが方針や報酬関数を変える。ロボットが実機で試す。映像、軌跡、成功判定、失敗ログが残る。エージェントが原因を読み、再び方針を変える。","quote_start":0,"quote_end":100,"text_sha256":"075d6d80124e0ac0c5fbe9c8abd28e63077ebd62b069f0fa72c7a605ed610098","block_sha256":"075d6d80124e0ac0c5fbe9c8abd28e63077ebd62b069f0fa72c7a605ed610098","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_cf107e9a-1286-4df3-bd1f-adb77c070354/#blk_07d42b8c-a175-4b4c-a73e-4f5ae6cd564a"},{"id":"occ_8b1b4ba1116578af13dd8742","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499","block_id":"blk_17490015-fcea-4259-a426-3884b50a8bd2","section_id":"sec_1830142b-6932-4a58-bf2b-845921c561b1","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":46,"end":52,"exact":"エージェント","quote":"**絶ノイア:** だからフィジカルAIの本命は、モデル単体ではなく、実機実験、世界モデル、エージェント、改善ループの組み合わせ。","quote_start":0,"quote_end":65,"text_sha256":"8989d2342dd0fddefe85f4137b9146a3c2f0c3789c3f89f529afa83c22c0b317","block_sha256":"8989d2342dd0fddefe85f4137b9146a3c2f0c3789c3f89f529afa83c22c0b317","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_cf107e9a-1286-4df3-bd1f-adb77c070354/#blk_17490015-fcea-4259-a426-3884b50a8bd2"},{"id":"occ_fc4ae6a331c6edf3511d5c35","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499","block_id":"blk_23e61a71-e3be-4afc-831b-0b1a88c2af85","section_id":"sec_115648db-57b9-4060-a505-e18396c66b85","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":92,"end":98,"exact":"エージェント","quote":"方を、どう改善するか」を自動化する。\nWAM/Cosmosは「行動した後の世界がどうなるか」を予測する。\nAIエージェントは「長期作業をどう分解し、どこでやり直すか」を管理する。","quote_start":37,"quote_end":126,"text_sha256":"a6997ddc07ba224f8fac40aefc61c441c5b39626996a78e3607f2ab431ca537a","block_sha256":"a6997ddc07ba224f8fac40aefc61c441c5b39626996a78e3607f2ab431ca537a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_cf107e9a-1286-4df3-bd1f-adb77c070354/#blk_23e61a71-e3be-4afc-831b-0b1a88c2af85"},{"id":"occ_b28585fc7b042b9cd3c8132c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499","block_id":"blk_43c75d98-2d9e-4f3d-89fc-6acfdc468cbb","section_id":"sec_7cda0e34-94a4-4d85-a73c-30759efa7bc7","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":56,"end":62,"exact":"エージェント","quote":"AMは、行動と未来状態を結びつける。\nENPIREは、実機での失敗結果を使って、その行動方針を改善する。\nAIエージェントは、長期計画や失敗時の再計画を担う。","quote_start":1,"quote_end":80,"text_sha256":"de4acd53ef8e4dd96bab11d56dd1fd25034ed740d7fc840aef686d759495f6f2","block_sha256":"de4acd53ef8e4dd96bab11d56dd1fd25034ed740d7fc840aef686d759495f6f2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_cf107e9a-1286-4df3-bd1f-adb77c070354/#blk_43c75d98-2d9e-4f3d-89fc-6acfdc468cbb"},{"id":"occ_b6e8249983e7e4c817cb206e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499","block_id":"blk_4e9579a1-202b-478c-abec-b0656f5f9f08","section_id":"sec_be92a06a-15c9-4a6a-8b02-585b5dea6872","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":34,"end":40,"exact":"エージェント","quote":"- ENPIREは、ロボット方針の実機試行、失敗判定、ログ保存、AIエージェントによる改善を回す自動研究基盤である。\n- WLAは統合型のロボット脳、ENPIREはその脳や低レベル方針を実機で鍛える改善ループとして見ると分かりやすい。\n- WAM/Cosmosは物理未来の予測、A","quote_start":0,"quote_end":140,"text_sha256":"324319d557b1ee77a9c787f82ae9fc38c7ae6ee8e43e8653f542ffeb1bf8bbba","block_sha256":"324319d557b1ee77a9c787f82ae9fc38c7ae6ee8e43e8653f542ffeb1bf8bbba","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_cf107e9a-1286-4df3-bd1f-adb77c070354/#blk_4e9579a1-202b-478c-abec-b0656f5f9f08"},{"id":"occ_243272d393ac707bcd43bf55","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499","block_id":"blk_61eab5b3-d952-4561-98c6-9e15ed4eeeb4","section_id":"sec_ac7dcda0-dbd9-4de7-a0c0-e6bbf66ec828","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":24,"end":30,"exact":"エージェント","quote":"つまり、\n**WLAは統合型の脳**、\n**AIエージェント＋WAM＋ENPIREは分散型の脳と自己改善工場**\nである。","quote_start":0,"quote_end":61,"text_sha256":"f1c279476b35224ad5fd7c21ce758f59a72eb82ed85367ff4ae3606ff0578fdf","block_sha256":"f1c279476b35224ad5fd7c21ce758f59a72eb82ed85367ff4ae3606ff0578fdf","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_cf107e9a-1286-4df3-bd1f-adb77c070354/#blk_61eab5b3-d952-4561-98c6-9e15ed4eeeb4"},{"id":"occ_85781ef9f153bf791237c5c6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499","block_id":"blk_73ac8633-6a79-4a89-b0ec-8a9a0ef152d2","section_id":"sec_115648db-57b9-4060-a505-e18396c66b85","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":58,"end":64,"exact":"エージェント","quote":"は、ロボットが長期タスクを理解し、計画し、行動するための統合型基盤モデルである。\nこれは、物理世界におけるAIエージェントの脳に近い。","quote_start":3,"quote_end":70,"text_sha256":"cf370ded612fabd5dca7afc74d1cce03699911deb98af9e25ce051ba05bdc4f6","block_sha256":"cf370ded612fabd5dca7afc74d1cce03699911deb98af9e25ce051ba05bdc4f6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_cf107e9a-1286-4df3-bd1f-adb77c070354/#blk_73ac8633-6a79-4a89-b0ec-8a9a0ef152d2"},{"id":"occ_1bf7108cad43542229d1b27a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499","block_id":"blk_8b2d099f-925b-4fbd-b6c6-88b491ba8a8a","section_id":"sec_823193e4-6996-48d1-9d32-9696f136383f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":61,"end":67,"exact":"エージェント","quote":"が失敗を見て、原因を考え、コードを書き、学習設定を変え、また実験していた。\nENPIREでは、このループをAIエージェントが回す。","quote_start":6,"quote_end":71,"text_sha256":"8a12a370d5aacce44eb26bfd29ada7152a5e703dcf99047224726113bce507dd","block_sha256":"8a12a370d5aacce44eb26bfd29ada7152a5e703dcf99047224726113bce507dd","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_cf107e9a-1286-4df3-bd1f-adb77c070354/#blk_8b2d099f-925b-4fbd-b6c6-88b491ba8a8a"},{"id":"occ_9c853a840ad6f51353629b6e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499","block_id":"blk_8cc677a4-72cc-4f9a-9a5d-4ff794eb059a","section_id":"sec_9618b865-9de0-4496-a301-84d7a631622e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":23,"end":29,"exact":"エージェント","quote":"ENPIREは、失敗をログとして取り込み、AIエージェントがその原因を分析し、行動方針や制御設定を変え、再び実機で試す。これにより、現実世界のズレを潰していく。","quote_start":0,"quote_end":80,"text_sha256":"106af45660588de129bef05ccccaaf340a4595921caaa79545fbecc6cab15fef","block_sha256":"106af45660588de129bef05ccccaaf340a4595921caaa79545fbecc6cab15fef","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_cf107e9a-1286-4df3-bd1f-adb77c070354/#blk_8cc677a4-72cc-4f9a-9a5d-4ff794eb059a"},{"id":"occ_09ac27f43a154a8f2bd6533c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499","block_id":"blk_943035ba-65a9-4d7c-9031-9b8f522fe66d","section_id":"sec_97d4b135-0244-46ea-a311-bd5382c26ef5","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":76,"end":82,"exact":"エージェント","quote":"ットが方針を実行する\n↓\n成功・失敗を自動判定する\n↓\nログ、映像、報酬、軌跡を保存する\n↓\nAIコーディングエージェントが失敗原因を分析する\n↓\n方針コード、報酬関数、学習設定、制御補正を変更する\n↓\n再び実機で試す\n```","quote_start":21,"quote_end":134,"text_sha256":"c782406561f44a193e7dc1ec1fc5bee61c18bfd5f110b3b6908e8526de40c5dc","block_sha256":"c782406561f44a193e7dc1ec1fc5bee61c18bfd5f110b3b6908e8526de40c5dc","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_cf107e9a-1286-4df3-bd1f-adb77c070354/#blk_943035ba-65a9-4d7c-9031-9b8f522fe66d"},{"id":"occ_d30502028150ac7cd58eae5b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499","block_id":"blk_b1e879db-dc81-472b-8430-82d9b04bce8b","section_id":"sec_7f9a26a7-7219-429a-8b27-f1a31b2e4e68","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":19,"end":25,"exact":"エージェント","quote":"ソフトウェア開発では、AIコーディングエージェントがコードを書き、テストを走らせ、失敗ログを読み、修正し、またテストする。このループが速くなるほど、ソフトウェアの改善速度は上がる。","quote_start":0,"quote_end":90,"text_sha256":"c5c0499eddd55e5682fea87eaf4aa50fb6e6e15b947791a14641429c18c7a784","block_sha256":"c5c0499eddd55e5682fea87eaf4aa50fb6e6e15b947791a14641429c18c7a784","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_cf107e9a-1286-4df3-bd1f-adb77c070354/#blk_b1e879db-dc81-472b-8430-82d9b04bce8b"},{"id":"occ_f080e65067466c95b1b3f712","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499","block_id":"blk_b4b73d00-2a38-4466-b73a-025cfdcb3a4c","section_id":"sec_97d4b135-0244-46ea-a311-bd5382c26ef5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":52,"end":58,"exact":"エージェント","quote":"ENPIREは、単体のロボット基盤モデルではない。\nそれは、実機ロボットでの試行、失敗、評価、改善をAIエージェントに回させるための自動研究基盤である。","quote_start":0,"quote_end":76,"text_sha256":"d2bcb065d27f17a6e9c73b40bc2ae6a3499d55f418169bcb28f129731da2c52b","block_sha256":"d2bcb065d27f17a6e9c73b40bc2ae6a3499d55f418169bcb28f129731da2c52b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_cf107e9a-1286-4df3-bd1f-adb77c070354/#blk_b4b73d00-2a38-4466-b73a-025cfdcb3a4c"},{"id":"occ_190d0b82ed46ab156335f1d1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499","block_id":"blk_bcd17581-b35c-4399-b2a6-90e1331818a2","section_id":"sec_ac7dcda0-dbd9-4de7-a0c0-e6bbf66ec828","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":18,"end":24,"exact":"エージェント","quote":"```text\n人間の指示\n↓\nAIエージェント\n長期計画、手順分解、失敗時の再計画\n↓\nCosmos / WAM\n物理的な未来状態の予測、行動候補の評価\n↓\n低レベルポリシー\n把持、挿入、移動、押す、切る、運ぶ\n↓\nENPIRE\n実機失敗ログから","quote_start":0,"quote_end":124,"text_sha256":"8cd8edc485d2ab5fdbcf9b430868007e1a2c18068126e29f787ee27111a0741e","block_sha256":"8cd8edc485d2ab5fdbcf9b430868007e1a2c18068126e29f787ee27111a0741e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_cf107e9a-1286-4df3-bd1f-adb77c070354/#blk_bcd17581-b35c-4399-b2a6-90e1331818a2"},{"id":"occ_1875cfb1b329f61123d5b821","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499","block_id":"blk_c4df669e-d794-48e3-9b38-cefb4e327b2d","section_id":"sec_823193e4-6996-48d1-9d32-9696f136383f","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":11,"end":17,"exact":"エージェント","quote":"ソフトウェアでは、AIエージェントがコードを書き、テストを走らせ、エラーを見て、修正する。\nENPIREでは、AIエージェントがロボット方針を書き、実機で試し、失敗映像を見て、改善する。","quote_start":0,"quote_end":93,"text_sha256":"370eeeb03c7d662114f571ca852bed3d937eeb2335b02d43b9b4e509873d0350","block_sha256":"370eeeb03c7d662114f571ca852bed3d937eeb2335b02d43b9b4e509873d0350","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_cf107e9a-1286-4df3-bd1f-adb77c070354/#blk_c4df669e-d794-48e3-9b38-cefb4e327b2d"},{"id":"occ_2a3ea04f815772b1c259b954","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499","block_id":"blk_c8b54341-dc19-4d37-a281-58358117edd5","section_id":"sec_fded6cdb-741c-4140-9229-8e95d5f3f39b","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":42,"end":48,"exact":"エージェント","quote":"WLAは、このような長期タスクをモデル内部で扱おうとする。\nその意味で、WLAはAIエージェントの物理AI版に近い。ソフトウェア上のAIエージェントが複数ステップの仕事を進めるように、WLAは物理空間で複数ステップの作業を進めるための基盤モデルである。","quote_start":0,"quote_end":126,"text_sha256":"05e8bc578f471162894f46c3e1d7a58dfbd8f73ee7b820b3ae3c112831823ef7","block_sha256":"05e8bc578f471162894f46c3e1d7a58dfbd8f73ee7b820b3ae3c112831823ef7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_cf107e9a-1286-4df3-bd1f-adb77c070354/#blk_c8b54341-dc19-4d37-a281-58358117edd5"},{"id":"occ_a2dcc7ff2989f211b4909588","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499","block_id":"blk_cd2d8793-b1d8-4ef7-a460-fedbc29eaf11","section_id":"sec_ac7dcda0-dbd9-4de7-a0c0-e6bbf66ec828","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":43,"end":49,"exact":"エージェント","quote":"もしENPIREが実機で失敗を潰し、CosmosやWAMが物理世界の未来を予測し、AIエージェントが長期計画を立てるなら、WLAのような統合モデルがなくても、ロボットは長期タスクをこなせるのではないか。","quote_start":0,"quote_end":101,"text_sha256":"79ea3b7d50edfcf864e9fab79ea6ac3eb756d4a88e5330b184993eceaf687c74","block_sha256":"79ea3b7d50edfcf864e9fab79ea6ac3eb756d4a88e5330b184993eceaf687c74","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_cf107e9a-1286-4df3-bd1f-adb77c070354/#blk_cd2d8793-b1d8-4ef7-a460-fedbc29eaf11"},{"id":"occ_4a6a06d3047f74cc6adff32a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499","block_id":"blk_ce4c55fd-603a-4bab-8cd2-a79335547ab7","section_id":"sec_115648db-57b9-4060-a505-e18396c66b85","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":140,"end":146,"exact":"エージェント","quote":"ement、Rollout、Evolutionの4モジュールにより、自動リセット・自動検証・実機ロールアウト・エージェントによる改善を回すものです。NVIDIAのページでは、Push-T、ピン挿入、GPU挿入、結束バンド操作などでpass@8 99%成功率に到達したと説明されています。([NVIDIA][1])","quote_start":85,"quote_end":241,"text_sha256":"6299f59fcdd1d565d8869c1e529570bb3d380e68e8f6c96e0d84b8c1f170cfc7","block_sha256":"6299f59fcdd1d565d8869c1e529570bb3d380e68e8f6c96e0d84b8c1f170cfc7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_cf107e9a-1286-4df3-bd1f-adb77c070354/#blk_ce4c55fd-603a-4bab-8cd2-a79335547ab7"},{"id":"occ_c45012b7b4856915b8a9c9df","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499","block_id":"blk_d6ff8967-c60d-45db-82da-810afc90b9c4","section_id":"sec_97d4b135-0244-46ea-a311-bd5382c26ef5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":115,"end":121,"exact":"エージェント","quote":"学習法を試すべきか」「どの制御補正を入れるべきか」「どの報酬関数に変えるべきか」という研究作業そのものを、AIエージェントに担わせる。","quote_start":60,"quote_end":127,"text_sha256":"c7e00722c0467847ba2d8de46a89c19256b95c7c58cfac75af4673a70f76f116","block_sha256":"c7e00722c0467847ba2d8de46a89c19256b95c7c58cfac75af4673a70f76f116","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_cf107e9a-1286-4df3-bd1f-adb77c070354/#blk_d6ff8967-c60d-45db-82da-810afc90b9c4"},{"id":"occ_f90abc873e0e552a45ceabb4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499","block_id":"blk_ed360c9d-e3a3-4978-b082-cef006477d14","section_id":"sec_7cda0e34-94a4-4d85-a73c-30759efa7bc7","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":32,"end":38,"exact":"エージェント","quote":"```text\nWAM / Cosmos\n世界を予測する\n\nAIエージェント\nタスクを分解し、作業を管理する\n\n低レベルポリシー\n実際にロボットを動かす\n\nENPIRE\n失敗したポリシーを実機で改善する\n```","quote_start":0,"quote_end":105,"text_sha256":"3e02b81310e95fcc32dfa36245e0b515868fda7d0f7945acac9d10970326ca45","block_sha256":"3e02b81310e95fcc32dfa36245e0b515868fda7d0f7945acac9d10970326ca45","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_cf107e9a-1286-4df3-bd1f-adb77c070354/#blk_ed360c9d-e3a3-4978-b082-cef006477d14"},{"id":"occ_31e40acc76cd702a81c3e2c7","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499","block_id":"blk_f1307770-ec61-4081-aaea-23202fef8854","section_id":"sec_823193e4-6996-48d1-9d32-9696f136383f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":24,"end":30,"exact":"エージェント","quote":"これは、ソフトウェア開発におけるAIコーディングエージェントの進化とよく似ている。","quote_start":0,"quote_end":41,"text_sha256":"136c15cb385b7074713cf1f73f4d35cf62a3586641cfc95d1a653180db42e245","block_sha256":"136c15cb385b7074713cf1f73f4d35cf62a3586641cfc95d1a653180db42e245","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_cf107e9a-1286-4df3-bd1f-adb77c070354/#blk_f1307770-ec61-4081-aaea-23202fef8854"},{"id":"occ_3c61c389d023779ac04ea369","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499","block_id":"blk_fb4b919f-923c-485b-83db-a0d6b1c3602f","section_id":"sec_c8e3ca4e-af4e-4488-887a-b928f6f27cba","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":73,"end":79,"exact":"エージェント","quote":"/ GR00T\n初期のロボット基盤ポリシーを作る\n\nCosmos / WAM\n世界の未来状態を予測する\n\nAIエージェント\n長期計画、タスク分解、失敗時の再計画を担う\n\nENPIRE\n実機で失敗を潰し、現場ごとに自己改善する\n\nIsaac / Omniverse / Newton\nシミュレーション、合成データ、物理検証を","quote_start":18,"quote_end":179,"text_sha256":"047011be904c37a06dbd0c1bdc5413301814788b2e778618c9aa03f1fe72bd87","block_sha256":"047011be904c37a06dbd0c1bdc5413301814788b2e778618c9aa03f1fe72bd87","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_cf107e9a-1286-4df3-bd1f-adb77c070354/#blk_fb4b919f-923c-485b-83db-a0d6b1c3602f"},{"id":"occ_e13fd350dbf012cfd02cd9d4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_d4189e74-ba90-4c4f-8495-9a2fccea5a2a","work_id":"wrk_ba0d2481-72cf-412b-92be-570c7d9c07e4","block_id":"blk_5565b036-4596-468b-ad06-bd52696a7a42","section_id":"sec_cc833604-c796-40b3-8038-0d8d85cce4f0","layer":"code","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":131,"end":137,"exact":"エージェント","quote":"Google TPU |\n| 大量推論 | TPU・Trainium・Maia・MTIA |\n| CPU処理・エージェント管理 | x86・Arm CPU |\n| 低価格バッチ処理 | カスタムASIC |\n| 専門領域 | GPU・ASIC・専用アクセラレーター |\n```","quote_start":76,"quote_end":214,"text_sha256":"09e9ebdeabc09f8b4b36a90ce891c4392092648a3188dfb00555425e616e6bc0","block_sha256":"09e9ebdeabc09f8b4b36a90ce891c4392092648a3188dfb00555425e616e6bc0","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_d4189e74-ba90-4c4f-8495-9a2fccea5a2a/#blk_5565b036-4596-468b-ad06-bd52696a7a42"},{"id":"occ_21ec342b6c14b34e3ec98658","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_d4189e74-ba90-4c4f-8495-9a2fccea5a2a","work_id":"wrk_ba0d2481-72cf-412b-92be-570c7d9c07e4","block_id":"blk_b09a587c-c8f7-4ddd-8ecb-ced9f062c8bb","section_id":"sec_bf426a9f-960d-4d13-a9bc-1a7f22c53c46","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":31,"end":37,"exact":"エージェント","quote":"日本、台湾、韓国の熟練技術者が持つ知識を、デジタルツイン、製造エージェント、装置データへ移せれば、少人数で新工場を立ち上げられる可能性がある。","quote_start":0,"quote_end":71,"text_sha256":"c37993a0bcb06b166331c6ec1e62d076cead353377eb76c2f08b2951aa7842fd","block_sha256":"c37993a0bcb06b166331c6ec1e62d076cead353377eb76c2f08b2951aa7842fd","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_d4189e74-ba90-4c4f-8495-9a2fccea5a2a/#blk_b09a587c-c8f7-4ddd-8ecb-ced9f062c8bb"},{"id":"occ_0852122354854a15b403f677","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_d47a57d8-f3a6-4505-9ea6-c03ae4416616","work_id":"wrk_0ebc3696-2c90-4588-8ed4-2ea51e51433d","block_id":"blk_7031793c-60d3-40e0-bcdd-efd2370abf4b","section_id":"sec_34b7c44c-c552-4609-9d6b-88a91e5cae27","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":45,"end":51,"exact":"エージェント","quote":"さらに、GoogleのAI Co-Scientistは、Gemini 2.0を使ったマルチエージェント型の科学共同研究システムとして、仮説生成や研究提案を支援するために発表された。Googleは、これを科学・生物医学の発見速度を上げるための仮想共同研究者と位置づけている。([Google Resear","quote_start":0,"quote_end":151,"text_sha256":"16ed136adac036dfeefed6f01615c260dfee4057a01526f0cefe3994b8294d16","block_sha256":"16ed136adac036dfeefed6f01615c260dfee4057a01526f0cefe3994b8294d16","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_d47a57d8-f3a6-4505-9ea6-c03ae4416616/#blk_7031793c-60d3-40e0-bcdd-efd2370abf4b"},{"id":"occ_68f42be88da27b7d6bce6fa6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_d47a57d8-f3a6-4505-9ea6-c03ae4416616","work_id":"wrk_0ebc3696-2c90-4588-8ed4-2ea51e51433d","block_id":"blk_8be60d0b-c43f-4826-80a4-d36603c2f42f","section_id":"sec_cdcc16ea-76f1-4d43-8b6d-95410a3fa392","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":7,"end":13,"exact":"エージェント","quote":"モデル、評価、エージェント、知識ベース、ワークフロー、推論基盤が全部外部にあると、AI時代の企業能力が国外プラットフォーム依存になる。","quote_start":0,"quote_end":67,"text_sha256":"ce0d8a331317b2095ae5fd5981860b00001f856af6a956cd3bc8dcba56be9652","block_sha256":"ce0d8a331317b2095ae5fd5981860b00001f856af6a956cd3bc8dcba56be9652","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_d47a57d8-f3a6-4505-9ea6-c03ae4416616/#blk_8be60d0b-c43f-4826-80a4-d36603c2f42f"},{"id":"occ_7efdba82004fb8de946c6d00","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_d47a57d8-f3a6-4505-9ea6-c03ae4416616","work_id":"wrk_0ebc3696-2c90-4588-8ed4-2ea51e51433d","block_id":"blk_bcb15e54-4639-420a-8513-7d7113195066","section_id":"sec_93c73cc9-ac9d-4599-a70c-1ab06d99efbf","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":49,"end":55,"exact":"エージェント","quote":"第五に、学習ループを国内に残すこと。評価データ、実験ログ、失敗データ、業務フロー、特許化プロセス、エージェント記憶、企業固有の判断基準は国内・社内に残す必要がある。","quote_start":0,"quote_end":82,"text_sha256":"90f0dfc00d9fe69c37ddf959d200f2b74e306b9a364f206aff9d4da77a3b0a13","block_sha256":"90f0dfc00d9fe69c37ddf959d200f2b74e306b9a364f206aff9d4da77a3b0a13","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_d47a57d8-f3a6-4505-9ea6-c03ae4416616/#blk_bcb15e54-4639-420a-8513-7d7113195066"},{"id":"occ_dd8d8b9f652ef9c6bd909ac0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_d47a57d8-f3a6-4505-9ea6-c03ae4416616","work_id":"wrk_0ebc3696-2c90-4588-8ed4-2ea51e51433d","block_id":"blk_e21e22b9-c4cb-484e-8b89-c597dee763b9","section_id":"sec_1d128fab-1214-4488-b84a-51656024c5ed","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":312,"end":318,"exact":"エージェント","quote":"災などで自国データを活かす |\n| 学習ループを国内に残す | 評価データ、実験ログ、失敗データ、業務フロー、エージェント記憶を国内・社内に残す |\n| AIの手足を押さえる | 電力、通信、ロボット、工場、ラボ、病院、センサー網で現実世界への実装力を持つ |","quote_start":257,"quote_end":387,"text_sha256":"a69acc12a555f1317c700e7369328a6feb1ab02fac1a5a91417ba20dd41005d5","block_sha256":"a69acc12a555f1317c700e7369328a6feb1ab02fac1a5a91417ba20dd41005d5","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_d47a57d8-f3a6-4505-9ea6-c03ae4416616/#blk_e21e22b9-c4cb-484e-8b89-c597dee763b9"},{"id":"occ_6b77ca4492bef769b3206c50","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_d47a57d8-f3a6-4505-9ea6-c03ae4416616","work_id":"wrk_0ebc3696-2c90-4588-8ed4-2ea51e51433d","block_id":"blk_ed8fb6e8-4669-4d9d-bb6a-3596dd344e25","section_id":"sec_05785b8a-9f7f-47be-8bf6-41620882a868","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":66,"end":72,"exact":"エージェント","quote":"ィアモデルを戦略物資として扱う方向に進む。高度なモデル、AIチップ、推論クラスタ、サイバー能力、軍事利用可能なエージェントは、輸出管理や同盟管理の対象になる。同盟国にはアクセスを与えるが、敵対国や不安定な国には制限する。モデル利用も、API、クラウド、データセンター、ライセンス、国籍、組織属性によって段階的に管理される。","quote_start":11,"quote_end":172,"text_sha256":"ca8035b1ea60ca67bc2b87783513f241f01afd141f452663eb1831bb0e171fbd","block_sha256":"ca8035b1ea60ca67bc2b87783513f241f01afd141f452663eb1831bb0e171fbd","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_d47a57d8-f3a6-4505-9ea6-c03ae4416616/#blk_ed8fb6e8-4669-4d9d-bb6a-3596dd344e25"},{"id":"occ_9236b1753eac5adf92ed518c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_d8e8cb66-632c-4170-afae-461911315dbd","work_id":"wrk_7ab4bb8d-7296-4037-b30a-70a60e28196a","block_id":"blk_aa09e101-ea6e-4e5a-b8fa-a46afa124bc5","section_id":"sec_63af13c9-bd9c-4848-9887-ce3064e24a9b","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":70,"end":76,"exact":"エージェント","quote":"- 汎用的なスキルを事前に学習し、新しい環境でこれを基礎として探索。\n- **特徴**:\n  \n  - 小さなエージェントが環境との相互作用を通じて、初期のスキルセット（例えば移動や観察）を習得。\n  - 明示的タスク設定がなくても、次元削減や潜在空間の探索を促進。","quote_start":15,"quote_end":148,"text_sha256":"a6864eab6252ef8d596b37355a255076b180d4fe94dd09a1d0202027f166b402","block_sha256":"a6864eab6252ef8d596b37355a255076b180d4fe94dd09a1d0202027f166b402","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_d8e8cb66-632c-4170-afae-461911315dbd/#blk_aa09e101-ea6e-4e5a-b8fa-a46afa124bc5"},{"id":"occ_ec5156d56afb5241213790f3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_da6063d5-59b2-4519-9c42-45d8f4f206c8","work_id":"wrk_ef613b57-c920-49a8-bb03-b7fdb149eeca","block_id":"blk_44b9b708-7e31-40ec-9650-282d98429269","section_id":"sec_24a6659c-f6c6-4d1c-8dac-688384dcdffd","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":86,"end":92,"exact":"エージェント","quote":"、学習データ、チェックポイント、RAG用データベース、ベクトルインデックス、KVキャッシュ、長文コンテキスト、エージェントの記憶、ロボットや車の現場ログが増え続ける。これらをすべてHBMやDRAMに置くのは高すぎる。一方、普通のSSDでは遅すぎる場面も増える。","quote_start":31,"quote_end":161,"text_sha256":"dc1ce4f13d61a949b5a24c26f9354f87bdfe1d232f7734cb8cda4aba24d91f44","block_sha256":"dc1ce4f13d61a949b5a24c26f9354f87bdfe1d232f7734cb8cda4aba24d91f44","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_da6063d5-59b2-4519-9c42-45d8f4f206c8/#blk_44b9b708-7e31-40ec-9650-282d98429269"},{"id":"occ_5e87a9ab261dfb9d5483a73b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_da6063d5-59b2-4519-9c42-45d8f4f206c8","work_id":"wrk_ef613b57-c920-49a8-bb03-b7fdb149eeca","block_id":"blk_7f6fdd85-46d4-4d89-8507-b80fc8d3094e","section_id":"sec_430bbb44-c02d-4d44-996e-92bbccc6adce","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":28,"end":34,"exact":"エージェント","quote":"キオクシアのCEOメッセージも、AIの応用が生成AIからエージェントAI、フィジカルAIへ広がり、データの生成・活用方法が多様化する中で、フラッシュメモリとSSD需要が成長すると述べている。([KIOXIA Corporation](https://www.kioxi","quote_start":0,"quote_end":134,"text_sha256":"1f635d2777bfee12718835e3c7407cbe9e9e691ae96b8bb59d33579d73264649","block_sha256":"1f635d2777bfee12718835e3c7407cbe9e9e691ae96b8bb59d33579d73264649","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_da6063d5-59b2-4519-9c42-45d8f4f206c8/#blk_7f6fdd85-46d4-4d89-8507-b80fc8d3094e"},{"id":"occ_f68fbc373d70156b5b203df5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_da6063d5-59b2-4519-9c42-45d8f4f206c8","work_id":"wrk_ef613b57-c920-49a8-bb03-b7fdb149eeca","block_id":"blk_cadd77c4-b099-490e-98a3-9bb7f509f0ec","section_id":"sec_59c9be61-1cf5-4f6e-8b7f-ffe9c0858a10","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":58,"end":64,"exact":"エージェント","quote":"キオクシア決算は、単なる「NAND価格が上がったから儲かった」という話ではない。より本質的には、**生成AI、エージェントAI、フィジカルAIの拡大によって、NANDフラッシュが保存装置からAIの記憶階層へ押し上げられ始めたことを示す決算**だった。","quote_start":3,"quote_end":128,"text_sha256":"f6c2cc36047ec7364c416b575d712d25eb3fd605cdfd907791fa4b151cbdb668","block_sha256":"f6c2cc36047ec7364c416b575d712d25eb3fd605cdfd907791fa4b151cbdb668","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_da6063d5-59b2-4519-9c42-45d8f4f206c8/#blk_cadd77c4-b099-490e-98a3-9bb7f509f0ec"},{"id":"occ_39a1e254e708e0200cb067c0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_da6063d5-59b2-4519-9c42-45d8f4f206c8","work_id":"wrk_ef613b57-c920-49a8-bb03-b7fdb149eeca","block_id":"blk_d22502f8-99e3-4808-ac75-bce5db3a296d","section_id":"sec_e717abe0-23e7-47ae-a4bc-e9648cb7d3fb","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":38,"end":44,"exact":"エージェント","quote":"この転換に成功すれば、キオクシアは単にNANDを作る会社ではなく、生成AI、エージェントAI、フィジカルAIが現実世界に蓄積していく膨大な記憶を支える企業になる。逆に、そこへ移れなければ、中国増産とメモリサイクルの波に再び飲み込まれる。","quote_start":0,"quote_end":118,"text_sha256":"adf8223dc4c12152e79d86076f89c819cb2d4675a886fea1d8925956e6fd384a","block_sha256":"adf8223dc4c12152e79d86076f89c819cb2d4675a886fea1d8925956e6fd384a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_da6063d5-59b2-4519-9c42-45d8f4f206c8/#blk_d22502f8-99e3-4808-ac75-bce5db3a296d"},{"id":"occ_9ed80af5f7946b7df18e8dbe","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_dbe69fb8-f76d-44e6-b111-aa27e548a204","work_id":"wrk_0252d3aa-e602-4a99-abf9-0b9ec2b5796a","block_id":"blk_140f8a4a-bd61-49d2-8727-7a432e179fb9","section_id":"sec_d7bf9997-c67b-4e17-966d-bc5c55552344","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":24,"end":30,"exact":"エージェント","quote":"これにより、TPU 8iはMoE、長文推論、AIエージェントのような不規則通信に強くなる。","quote_start":0,"quote_end":45,"text_sha256":"76f841d35dcd45533e4c690f7ad1c32a607e7552656c6f47ee6bfad73e286542","block_sha256":"76f841d35dcd45533e4c690f7ad1c32a607e7552656c6f47ee6bfad73e286542","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_dbe69fb8-f76d-44e6-b111-aa27e548a204/#blk_140f8a4a-bd61-49d2-8727-7a432e179fb9"},{"id":"occ_3a80f403fd45e2c755520856","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_dbe69fb8-f76d-44e6-b111-aa27e548a204","work_id":"wrk_0252d3aa-e602-4a99-abf9-0b9ec2b5796a","block_id":"blk_5db11282-e4ef-4371-b582-a413cf92139f","section_id":"sec_0e18eb64-9933-4275-851c-0225dc9241d4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":48,"end":54,"exact":"エージェント","quote":"TPU 8tは学習用であり、巨大モデルを鍛えるための工場である。TPU 8iは推論用であり、AIエージェント、長文reasoning、MoE、低遅延サービングを支える発電所である。","quote_start":0,"quote_end":90,"text_sha256":"97c89e77b559aeef569154926dc78271783e6670696af3817f2d752878fc4dc4","block_sha256":"97c89e77b559aeef569154926dc78271783e6670696af3817f2d752878fc4dc4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_dbe69fb8-f76d-44e6-b111-aa27e548a204/#blk_5db11282-e4ef-4371-b582-a413cf92139f"},{"id":"occ_a7d2915aa6a1786b68e922a9","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_dbe69fb8-f76d-44e6-b111-aa27e548a204","work_id":"wrk_0252d3aa-e602-4a99-abf9-0b9ec2b5796a","block_id":"blk_6286a5f2-d65e-4195-bcf2-19c9b2babcd6","section_id":"sec_619e1043-0e1c-41c3-a4f5-f0581fbae27c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":45,"end":51,"exact":"エージェント","quote":"学習用クラスタは巨大だが、使われる場所は限られる。一方、推論はユーザー数、トークン数、AIエージェント数、APIリクエスト数に比例して伸びる。検索、広告、Gmail、Workspace、YouTube、Gemini、Cloud API、コーディング、動画生成、音声、ロボット、RAGまで含めると、本当に","quote_start":0,"quote_end":151,"text_sha256":"6d9419442d53d8b0352b8281419af2611f9a8f11160cc6f9fdb12c98955156f2","block_sha256":"6d9419442d53d8b0352b8281419af2611f9a8f11160cc6f9fdb12c98955156f2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_dbe69fb8-f76d-44e6-b111-aa27e548a204/#blk_6286a5f2-d65e-4195-bcf2-19c9b2babcd6"},{"id":"occ_b4d1d646665db69d4703d58f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_dbe69fb8-f76d-44e6-b111-aa27e548a204","work_id":"wrk_0252d3aa-e602-4a99-abf9-0b9ec2b5796a","block_id":"blk_6c365b20-3c40-481c-b121-3edba58ed2c4","section_id":"sec_f1543817-fc0d-4ce1-96f7-c303e55d266a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":72,"end":78,"exact":"エージェント","quote":"UやTPUをまとめて「AI計算資源」として見ることが多かった。しかし、生成AIが学習中心の時代から、推論、AIエージェント、長文reasoning、MoEの時代へ移るにつれ、必要なチップの性格は大きく変わった。","quote_start":17,"quote_end":122,"text_sha256":"d02479ed928d8f2e68368161cb8c10136bc5b2361e978153936fabd1b044a680","block_sha256":"d02479ed928d8f2e68368161cb8c10136bc5b2361e978153936fabd1b044a680","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_dbe69fb8-f76d-44e6-b111-aa27e548a204/#blk_6c365b20-3c40-481c-b121-3edba58ed2c4"},{"id":"occ_c2cf281772ea5a2816d5f444","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_dbe69fb8-f76d-44e6-b111-aa27e548a204","work_id":"wrk_0252d3aa-e602-4a99-abf9-0b9ec2b5796a","block_id":"blk_8183280c-a656-4a86-a171-8bfc948b8286","section_id":"sec_acac2f8c-188a-41b5-ba8e-397790fdd6f8","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":105,"end":111,"exact":"エージェント","quote":"pert呼び出し、同期処理がボトルネックになる。コンテキストが長くなればなるほど、KVキャッシュは膨らむ。AIエージェントのように何度も考え直す処理では、単純な行列演算だけでなく、メモリと通信の待ち時間が支配的になる。","quote_start":50,"quote_end":159,"text_sha256":"27d1747d835132b3bcfc2be82ee679e35a103b7588a2d2c8d072163f76680c62","block_sha256":"27d1747d835132b3bcfc2be82ee679e35a103b7588a2d2c8d072163f76680c62","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_dbe69fb8-f76d-44e6-b111-aa27e548a204/#blk_8183280c-a656-4a86-a171-8bfc948b8286"},{"id":"occ_4f8159cdacc3c8297fb245ba","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_dbe69fb8-f76d-44e6-b111-aa27e548a204","work_id":"wrk_0252d3aa-e602-4a99-abf9-0b9ec2b5796a","block_id":"blk_b0d44a4d-8880-4a7a-a006-baa62f24a022","section_id":"sec_f1543817-fc0d-4ce1-96f7-c303e55d266a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":81,"end":87,"exact":"エージェント","quote":"クラスタ接続が必要になる。一方、推論では、ユーザーのリクエストに対して低遅延で応答し続ける必要がある。特にAIエージェントでは、1回の回答の中で検索、ツール実行、再推論、確認、出力が何度も繰り返される。ここでは単純なピークFLOPSより、メモリ容量、HBM帯域、KVキャッシュ、チップ間通信の遅延、テールレイテンシが重要に","quote_start":26,"quote_end":187,"text_sha256":"1e8e26c88e8d2dc8b799feeb1e1252a8e13e61b4049d8ce1d97ae29ac3192f41","block_sha256":"1e8e26c88e8d2dc8b799feeb1e1252a8e13e61b4049d8ce1d97ae29ac3192f41","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_dbe69fb8-f76d-44e6-b111-aa27e548a204/#blk_b0d44a4d-8880-4a7a-a006-baa62f24a022"},{"id":"occ_fa7b8da9d90d2076a5fd7c75","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_dbe69fb8-f76d-44e6-b111-aa27e548a204","work_id":"wrk_0252d3aa-e602-4a99-abf9-0b9ec2b5796a","block_id":"blk_f4079410-cd85-46d5-8e41-83e3e3787585","section_id":"sec_d66e9049-6559-4ebb-be89-11d94d5da144","layer":"character","character_id":"zetu_noia","count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":56,"end":62,"exact":"エージェント","quote":"から私は、8iのHBM容量、SRAM、Boardflyに注目したい。ピークFLOPSよりも、長文、MoE、AIエージェント、テールレイテンシの現実に向き合った設計だからだ。ここから先のAIインフラは、賢さだけでなく、配る力の競争になる。","quote_start":1,"quote_end":119,"text_sha256":"32366a1196d1701c86d262a6ad5bb925f7ce5601fcfcfe3b8c53ddedbeb5f1bf","block_sha256":"32366a1196d1701c86d262a6ad5bb925f7ce5601fcfcfe3b8c53ddedbeb5f1bf","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_dbe69fb8-f76d-44e6-b111-aa27e548a204/#blk_f4079410-cd85-46d5-8e41-83e3e3787585"},{"id":"occ_87a0136113b707aed88999c2","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":null,"section_id":null,"layer":"title","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"title","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントのWebが始まる: MCP、A2A、Skills、ARD、Fuguのデジタル神経系","quote_start":0,"quote_end":49,"text_sha256":"ec707adc1f534c6209b431eece8cdcf1d67e7ea2711f19384f8c96edf7ab9ae6","block_sha256":null,"offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/"},{"id":"occ_d5b7b13463a8024497d1a93d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_062efb85-3a6b-4ccc-81b7-fb153b77684a","section_id":"sec_2674fee6-cdf1-479d-8e6e-3a0d6c29c620","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":22,"end":28,"exact":"エージェント","quote":"人間がWeb検索でサービスを探すように、AIエージェントがAI向けリソースを探す。これがARDの本質である。","quote_start":0,"quote_end":54,"text_sha256":"3bb1c69178702fe23ff0e1d737444b1d1fcafea76a93a658b80f401361b6b249","block_sha256":"3bb1c69178702fe23ff0e1d737444b1d1fcafea76a93a658b80f401361b6b249","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_062efb85-3a6b-4ccc-81b7-fb153b77684a"},{"id":"occ_46007cc5ce0852f6dd82d624","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_0b56bfe4-01d7-437b-aad7-1ddff735974c","section_id":"sec_7d203312-341b-49f6-82c7-4250d459c9c4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":118,"end":124,"exact":"エージェント","quote":"て専門的な手順を持つようになった。\nMCPによって外部ツールやデータに接続する手足を得た。\nA2Aによって他のエージェントと連携する神経接続を得た。\nそしてARDによって、どこにどんな能力があるかを探す検索神経、地図、住所録を得ようとしている。","quote_start":63,"quote_end":185,"text_sha256":"3755462a772d89083318b4cceb5a04d73e83c8d08f21d21bba5871456fcd3eaf","block_sha256":"3755462a772d89083318b4cceb5a04d73e83c8d08f21d21bba5871456fcd3eaf","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_0b56bfe4-01d7-437b-aad7-1ddff735974c"},{"id":"occ_0fd436af7ecff046782b5b55","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_125c0fac-03d3-4927-9ac8-0e47b87ced3c","section_id":"sec_7543c6ba-0fe3-41dd-97da-94f6ed7a9d26","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":37,"end":43,"exact":"エージェント","quote":"AIがどのツールを使えるのか。\nどの業務アプリに接続できるのか。\nどの専門エージェントに仕事を頼めるのか。\nどのSkillを読み込めるのか。\nどのリソースが信頼できるのか。\nそして、それらをどう発見し、どう組み合わせるのか。","quote_start":0,"quote_end":112,"text_sha256":"b5169c5e92c5e8307f5cf24ec941441ec731aadaeb418629e0f92de82dc24a1f","block_sha256":"b5169c5e92c5e8307f5cf24ec941441ec731aadaeb418629e0f92de82dc24a1f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_125c0fac-03d3-4927-9ac8-0e47b87ced3c"},{"id":"occ_56018daf2b692a2a94060045","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_12979acd-23a5-4dbb-b5fc-ea2720721227","section_id":"sec_8f7f847f-8697-41ff-9c29-78e0560565e5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":63,"end":69,"exact":"エージェント","quote":"普及すると、Fugu的な価値の一部は標準化される可能性もある。つまり、複数のモデル、Skill、MCP、A2Aエージェントを見つけて組み合わせることが、Fuguだけの専売特許ではなくなる。","quote_start":8,"quote_end":102,"text_sha256":"48101342275346613b5c298f7e1449a1a47a5945cea6550dceccfea02cc9ba51","block_sha256":"48101342275346613b5c298f7e1449a1a47a5945cea6550dceccfea02cc9ba51","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_12979acd-23a5-4dbb-b5fc-ea2720721227"},{"id":"occ_59199f258197c0c1266fe50f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_16fe21c8-b95e-4f94-8097-9c56b728be3a","section_id":"sec_0f0ebb58-54c3-492f-9d90-baea6e617e7a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":54,"end":60,"exact":"エージェント","quote":"従来のWebでは、検索エンジン、DNS、API、アプリストア、OAuth、クラウド、CDNが重要だった。AIエージェント時代には、それに相当する新しいレイヤーが必要になる。ARDは、その中でも「どこに何があるか」「誰が提供しているか」「信頼できるか」を扱う標準になる可能性がある。","quote_start":0,"quote_end":140,"text_sha256":"9418e24ac65d7f5424d2505189a5341aa7d4df6d3270cd6abfeb8e51a5ac9767","block_sha256":"9418e24ac65d7f5424d2505189a5341aa7d4df6d3270cd6abfeb8e51a5ac9767","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_16fe21c8-b95e-4f94-8097-9c56b728be3a"},{"id":"occ_9b41d5d6d653e6b04dffeed0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_262528bb-f58b-403a-b831-147eba1e5911","section_id":"sec_ccba5d74-f5e6-4fde-b96b-b602739527fc","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"第四に、エージェント連携。\nA2Aによって、複数の専門エージェントを組織化できるかが問われる。","quote_start":0,"quote_end":47,"text_sha256":"8f4437c073892e89d2a365e7f7b1965df4a9bf88818934d905e698664df0d12a","block_sha256":"8f4437c073892e89d2a365e7f7b1965df4a9bf88818934d905e698664df0d12a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_262528bb-f58b-403a-b831-147eba1e5911"},{"id":"occ_1332091fe05fe174f8869d1d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_26a55e98-b1e1-4f5b-96e0-a12d6e83f361","section_id":"sec_8f7f847f-8697-41ff-9c29-78e0560565e5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":32,"end":38,"exact":"エージェント","quote":"FuguにとってARDは、使えば強化材料、無視すれば標準化されたエージェントWebから取り残されるリスクになる。","quote_start":0,"quote_end":56,"text_sha256":"f5ce871fb2f61ba17fa315979d8da1491d8e9ad09061e0fc552dd739b943e2bc","block_sha256":"f5ce871fb2f61ba17fa315979d8da1491d8e9ad09061e0fc552dd739b943e2bc","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_26a55e98-b1e1-4f5b-96e0-a12d6e83f361"},{"id":"occ_6a9c9295f5a95bdb911c4004","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_2b5a796b-b075-4ca2-97ec-6789741b67f5","section_id":"sec_357132d1-8bfe-4377-90c6-0f15ae0becba","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":44,"end":50,"exact":"エージェント","quote":"たとえばCodexが、特定のフレームワークに必要なテストSkill、セキュリティレビューエージェント、Figma連携MCP、社内CIツールをその場で探して使えるようになれば、開発エージェントとしての能力はさらに高まる。","quote_start":0,"quote_end":109,"text_sha256":"402cc699ec6010b9b2210fb99e752cdb3113d7c3a48c154d794a99de2bb29aac","block_sha256":"402cc699ec6010b9b2210fb99e752cdb3113d7c3a48c154d794a99de2bb29aac","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_2b5a796b-b075-4ca2-97ec-6789741b67f5"},{"id":"occ_31f1f494e1622d0513cd1c45","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_2df442e6-97ac-4e98-bad7-7ffa30a767f7","section_id":"sec_5f9d61b2-5985-4cdf-8b4b-fbbdd9ab7467","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":57,"end":63,"exact":"エージェント","quote":"ble 5やGPT、Claude、Geminiは「知能」そのものである。\nARDは、その知能が使うべきツールやエージェントを探すための検索レイヤーである。","quote_start":2,"quote_end":79,"text_sha256":"82850486a8765d5d9d437826009191f058c556109757fba7ea308a83e646d559","block_sha256":"82850486a8765d5d9d437826009191f058c556109757fba7ea308a83e646d559","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_2df442e6-97ac-4e98-bad7-7ffa30a767f7"},{"id":"occ_e36760fbb6816aea2f5e21ca","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_3656f03d-2b48-4ba9-9358-bac71d73fa1e","section_id":"sec_2674fee6-cdf1-479d-8e6e-3a0d6c29c620","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"現在のAIエージェントは、基本的にはユーザーや開発者があらかじめ接続したツールしか使えない。Claude CodeやCodexにMCPサーバーを登録する。GitHubやFigmaやDBを設定する。特定のSkillをプロジェ","quote_start":0,"quote_end":111,"text_sha256":"cd820ca99468850bd81a0d15dd56e7e2a3495cf8692a5b46247a43f8f88d3fcc","block_sha256":"cd820ca99468850bd81a0d15dd56e7e2a3495cf8692a5b46247a43f8f88d3fcc","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_3656f03d-2b48-4ba9-9358-bac71d73fa1e"},{"id":"occ_8e65485a8c735af91861bc4f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_37fac988-fc5a-4977-8236-642597614027","section_id":"sec_f4da916f-b72c-487d-9b85-7e19a8e01735","layer":"character","character_id":"zetu_noia","count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントに「この仕事をして」と頼んだ時、本当に難しいのは推論だけではありません。どの社内DBに行けばいいのか。どのMCPサーバーが正しいのか。どのSkillを読み込むべきなのか。どの専門エージェントなら権限が","quote_start":0,"quote_end":108,"text_sha256":"1ae81fe4d000024e9ca204c14fc3d1586bd77b17b5d0aa7ccf0117c367d4d20b","block_sha256":"1ae81fe4d000024e9ca204c14fc3d1586bd77b17b5d0aa7ccf0117c367d4d20b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_37fac988-fc5a-4977-8236-642597614027"},{"id":"occ_56ab23dbdcc9e2cd97bcccd2","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_3879a064-932c-4c01-b726-243dbe1f5c11","section_id":"sec_e250648f-4d88-4198-bd76-e12c5a2fd425","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":87,"end":93,"exact":"エージェント","quote":"打ち上げ運用、安全管理といった領域では、膨大な内部ツール、設計資料、運用システム、監視基盤がある。これらをAIエージェントが使うには、単に強いモデルを導入するだけでは足りない。","quote_start":32,"quote_end":120,"text_sha256":"2d27cf88819f1e17756b6940e67f64a85b8d7e3f0a23a3df4fab324f24761196","block_sha256":"2d27cf88819f1e17756b6940e67f64a85b8d7e3f0a23a3df4fab324f24761196","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_3879a064-932c-4c01-b726-243dbe1f5c11"},{"id":"occ_2fecae83cc27924cd4d5065a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_41179167-f745-4638-9a0a-4d2beae352d5","section_id":"sec_8c121823-5eff-45c7-aa58-1ca1c60f68c8","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":28,"end":34,"exact":"エージェント","quote":"MCPは、AIがツールを使うための手足。\nA2Aは、AIエージェント同士をつなぐ神経。\nSkillsは、AIに専門的な仕事の型を与える記憶や手順書。\nARDは、必要な能力を探すための検索エンジン、住所録、地図である。","quote_start":0,"quote_end":108,"text_sha256":"ea08840d666cb9c41b1534ebea9791142e2265d8b845d6caa3a51fedc1e2b306","block_sha256":"ea08840d666cb9c41b1534ebea9791142e2265d8b845d6caa3a51fedc1e2b306","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_41179167-f745-4638-9a0a-4d2beae352d5"},{"id":"occ_a0b23f95537708e8bc3e87e2","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_4894d54e-a4a7-4e2b-a550-c0b46dcdbebd","section_id":"sec_7d203312-341b-49f6-82c7-4250d459c9c4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":28,"end":34,"exact":"エージェント","quote":"これは、単一の巨大AIではない。\nデジタル上に生まれる、エージェントの生態系である。","quote_start":0,"quote_end":42,"text_sha256":"63878519b857b73646bcce43db3ec2d6cdfb7fd4c61f762fd78e491d977abbf9","block_sha256":"63878519b857b73646bcce43db3ec2d6cdfb7fd4c61f762fd78e491d977abbf9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_4894d54e-a4a7-4e2b-a550-c0b46dcdbebd"},{"id":"occ_2c1eeb32ef2e75c74bf01d23","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_4aa75667-6780-4a61-8c8e-fffe5241ee6b","section_id":"sec_0f0ebb58-54c3-492f-9d90-baea6e617e7a","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":20,"end":26,"exact":"エージェント","quote":"GoogleとMicrosoftは、AIエージェントの入口とクラウド基盤を握る企業である。GitHubは開発者向けエージェントの中心にいる。Hugging Faceはオープンモデル、MLアプリ、Spaces、Skill、MCPサーバーの集積地である。N","quote_start":0,"quote_end":126,"text_sha256":"2364a2a7623f313961ad0ecf0f3a8bbf6e007618b78ff4f85300aa9010dfe324","block_sha256":"2364a2a7623f313961ad0ecf0f3a8bbf6e007618b78ff4f85300aa9010dfe324","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_4aa75667-6780-4a61-8c8e-fffe5241ee6b"},{"id":"occ_d222320633f5fadb809f42f2","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_589c197f-51e2-4182-9826-203c922f9ff1","section_id":"sec_8c121823-5eff-45c7-aa58-1ca1c60f68c8","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":9,"end":15,"exact":"エージェント","quote":"一方、A2AはAIエージェント同士が連携するための規格である。","quote_start":0,"quote_end":31,"text_sha256":"ae003ed6e69232b593e33c150bac52bd9b14e41333cf0eb78ff4946246967073","block_sha256":"ae003ed6e69232b593e33c150bac52bd9b14e41333cf0eb78ff4946246967073","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_589c197f-51e2-4182-9826-203c922f9ff1"},{"id":"occ_b78319805f33a358fa82d5a4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_5e2327a5-f61d-4be7-8b48-da3281ac2923","section_id":"sec_357132d1-8bfe-4377-90c6-0f15ae0becba","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":39,"end":45,"exact":"エージェント","quote":"ChatGPT、Codex、Claude、Claude Codeのような強力なエージェントがARDに対応すれば、必要なMCPサーバー、Skill、A2Aエージェントを実行時に発見し、使えるようになる。これはユーザー体験を大きく改善する。","quote_start":0,"quote_end":118,"text_sha256":"74075e09056bedf416e0eee990a66bb4b5189c8106870db752aced0774404de3","block_sha256":"74075e09056bedf416e0eee990a66bb4b5189c8106870db752aced0774404de3","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_5e2327a5-f61d-4be7-8b48-da3281ac2923"},{"id":"occ_2741fb5c0d3c04822984b371","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_6446f9fa-e303-42bb-ad21-3051cbc244ac","section_id":"sec_3baaf0a6-5859-4f91-8254-fa0fd544838e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェント時代には、企業の業務プロセスそのものが、AIに見つけられる形へ変わっていく。営業、法務、会計、開発、品質保証、セキュリティ、調達、製造、データ分析。それぞれの機能がAPI化、MCP化、Skill化、A","quote_start":0,"quote_end":108,"text_sha256":"b854b9e2794c0fe54df6f3dfcf9f71f87afa0ddc5bb534fe2515f4e1d45fcec0","block_sha256":"b854b9e2794c0fe54df6f3dfcf9f71f87afa0ddc5bb534fe2515f4e1d45fcec0","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_6446f9fa-e303-42bb-ad21-3051cbc244ac"},{"id":"occ_e4c4325f08f3eae43ee32f5d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_6a84e064-d9e3-4920-ad2d-7391889d65a1","section_id":"sec_3f67ce1f-8cf3-43d8-a7de-dd9619da99d8","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":34,"end":40,"exact":"エージェント","quote":"**絶ノイア:** MCPやSkillを作っても、見つけられなければエージェントは使えない。","quote_start":0,"quote_end":46,"text_sha256":"9fcf7ecd68f06ff78b60bb91b9e3fc7ed1fd532b5fbd9eddd280f69efe98a0e5","block_sha256":"9fcf7ecd68f06ff78b60bb91b9e3fc7ed1fd532b5fbd9eddd280f69efe98a0e5","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_6a84e064-d9e3-4920-ad2d-7391889d65a1"},{"id":"occ_6c55b0ab4f5efa02c1707980","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_7a946fab-c21f-4b99-8e2f-5b535c2d9735","section_id":"sec_bb24b233-3916-43a8-80c3-0b3758a9d5bc","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":28,"end":34,"exact":"エージェント","quote":"MCPは、AIがツールを使うための手足。\nA2Aは、AIエージェント同士をつなぐ神経。\nSkillsは、専門的な仕事を再現するための記憶や手順書。\nARDは、必要な能力を探すための検索エンジンである。\nFuguは、複数のモデルやエージェントを束ねる司令塔に近い。","quote_start":0,"quote_end":131,"text_sha256":"219e3668506230a662b4e2d592e6a227eb8e39eeae86772a235badd27a75ec6c","block_sha256":"219e3668506230a662b4e2d592e6a227eb8e39eeae86772a235badd27a75ec6c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_7a946fab-c21f-4b99-8e2f-5b535c2d9735"},{"id":"occ_4a5bef90bdd17759bc1c0b48","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_7e02a883-4d15-405c-b3e8-110d8f971045","section_id":"sec_7d203312-341b-49f6-82c7-4250d459c9c4","layer":"body","character_id":null,"count":5,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":10,"end":16,"exact":"エージェント","quote":"ゲーム制作に特化したエージェント。\n動画制作に特化したエージェント。\n投資分析に特化したエージェント。\nサイト運用に特化したエージェント。\n営業、会計、法務、人事、開発、QA、セキュリティに特化したエージェント。","quote_start":0,"quote_end":106,"text_sha256":"0f580ed23bab73c7608ee8ca1a30bfd39819c01ddd7cf47093614745bcedc7ff","block_sha256":"0f580ed23bab73c7608ee8ca1a30bfd39819c01ddd7cf47093614745bcedc7ff","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_7e02a883-4d15-405c-b3e8-110d8f971045"},{"id":"occ_a3e78283f185b3c9a96cd5af","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_81d240b8-75cd-4d26-8788-97ad6575513d","section_id":"sec_3e50dc30-e0eb-4653-bafa-0414e8484fa1","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":92,"end":98,"exact":"エージェント","quote":"力台帳としても重要になる。\n- MCP、A2A、Skills、APIは、ARDで発見可能になって初めて大規模なエージェント経済に乗りやすくなる。\n- 今後は「AIが読める説明」「権限」「監査」「信頼情報」が、ツールやSaaSの競争力になる。\n- OpenAIやAnthropicのようなモデル企業にとってARDは追い風であ","quote_start":37,"quote_end":198,"text_sha256":"867e9082b337b75036f25f6fdaed4386ef69d3a866c562add692fa2c24cb6c89","block_sha256":"867e9082b337b75036f25f6fdaed4386ef69d3a866c562add692fa2c24cb6c89","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_81d240b8-75cd-4d26-8788-97ad6575513d"},{"id":"occ_66c57b6eac2cee396c49686c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_8376cf67-96fe-4f32-9498-9598bbfbaf12","section_id":"sec_e250648f-4d88-4198-bd76-e12c5a2fd425","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":46,"end":52,"exact":"エージェント","quote":"つまりSpaceXのような企業にとってARDは、公開Web向けの検索標準というより、社内AIエージェントのための能力マップになる。","quote_start":0,"quote_end":65,"text_sha256":"5d5d68acd5f943dea86ebf3e1b592d9632ac951f0de9a256c281f5b0df49b1de","block_sha256":"5d5d68acd5f943dea86ebf3e1b592d9632ac951f0de9a256c281f5b0df49b1de","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_8376cf67-96fe-4f32-9498-9598bbfbaf12"},{"id":"occ_511f22792f9655e51369dead","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_84e16b82-1f23-4991-b362-fb8d499f29e7","section_id":"sec_8a7c1f41-fccb-421e-9cf6-2a3e56b15501","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":76,"end":82,"exact":"エージェント","quote":"同時に企業内の能力台帳でもある。どの能力が存在し、誰が所有し、どの権限で使え、どのログが残るのか。その整備が、エージェント導入の本当の土台になる。","quote_start":21,"quote_end":94,"text_sha256":"cc90e5736ceca2e4a3c066a92d012cfe66c6fe6d64fed0173cb07fe0a8b24b5c","block_sha256":"cc90e5736ceca2e4a3c066a92d012cfe66c6fe6d64fed0173cb07fe0a8b24b5c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_84e16b82-1f23-4991-b362-fb8d499f29e7"},{"id":"occ_20715c6057b95459ba756346","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_8f38c098-4c8b-4730-8ab5-83e094bc2b48","section_id":"sec_8c121823-5eff-45c7-aa58-1ca1c60f68c8","layer":"body","character_id":null,"count":6,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"営業エージェントが会計エージェントに確認し、法務エージェントが契約書を確認し、開発エージェントがコードを修正し、QAエージェントがテストする。A2Aは、このような複数の独立したエージェント同士が、タスクを渡し、進捗を","quote_start":0,"quote_end":108,"text_sha256":"05f7215172b077f8c953f9ea76fabc31fa769e09cbc2693220584b136a12014e","block_sha256":"05f7215172b077f8c953f9ea76fabc31fa769e09cbc2693220584b136a12014e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_8f38c098-4c8b-4730-8ab5-83e094bc2b48"},{"id":"occ_e314c71e263799b6da97010c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_970035a7-e827-40b9-8eaf-f42cdb61ebae","section_id":"sec_7d203312-341b-49f6-82c7-4250d459c9c4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":10,"end":16,"exact":"エージェント","quote":"ここまで見ると、AIエージェントの進化は、生物の神経系や社会性に似た形でスケールしているように見える。","quote_start":0,"quote_end":51,"text_sha256":"8832c68216ef8ab4da9cf7c3770ca3db1dc34755045736038b627e461e2fec03","block_sha256":"8832c68216ef8ab4da9cf7c3770ca3db1dc34755045736038b627e461e2fec03","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_970035a7-e827-40b9-8eaf-f42cdb61ebae"},{"id":"occ_bf82dd346036d7dd48658157","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_9aa5376b-a9a3-444a-9f76-5c88f5b89802","section_id":"sec_8f7f847f-8697-41ff-9c29-78e0560565e5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":53,"end":59,"exact":"エージェント","quote":"Fuguが内部で複数モデルをルーティングする仕組みだとすれば、ARDは外部にあるSkill、MCP、A2Aエージェント、APIを探す仕組みである。FuguがARDを使えば、モデル選択だけでなく、外部能力の発見まで含めた、より広いオーケストレーターになれる。","quote_start":0,"quote_end":128,"text_sha256":"1797113c6d488ba75d22c1adb6ba2790c51ec761578f92dbd0ac625f8f473856","block_sha256":"1797113c6d488ba75d22c1adb6ba2790c51ec761578f92dbd0ac625f8f473856","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_9aa5376b-a9a3-444a-9f76-5c88f5b89802"},{"id":"occ_a37acc0f38d9d48b4e357ee2","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_9c44aed9-6308-48b0-83bd-b72e91107ae9","section_id":"sec_0f0ebb58-54c3-492f-9d90-baea6e617e7a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェント時代における、検索、信頼、身元確認、業務アプリ、データ基盤、開発環境、推論インフラをつなぐための土台である。","quote_start":0,"quote_end":62,"text_sha256":"ec6808f550315a9faad0024e473268a5df760d5382079e60bf138db5faf9c082","block_sha256":"ec6808f550315a9faad0024e473268a5df760d5382079e60bf138db5faf9c082","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_9c44aed9-6308-48b0-83bd-b72e91107ae9"},{"id":"occ_fe302f6b8d1caced1b632246","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_9cb37ca2-e155-40ab-853c-d9db89205571","section_id":"sec_8a7c1f41-fccb-421e-9cf6-2a3e56b15501","layer":"code","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":38,"end":44,"exact":"エージェント","quote":"```text\n人間のWeb\nページ → 検索エンジン → ユーザー\n\nAIエージェントのWeb\nMCP / Skill / A2A agent / API\n  ↓\nARD catalog / registry\n  ↓\nエージェントが発見・検証・接続\n```","quote_start":0,"quote_end":129,"text_sha256":"a5d19e1c5d9456dc122281bc546ecde62bc3f5151e3e5644d9bebe7ea1b4cd2d","block_sha256":"a5d19e1c5d9456dc122281bc546ecde62bc3f5151e3e5644d9bebe7ea1b4cd2d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_9cb37ca2-e155-40ab-853c-d9db89205571"},{"id":"occ_7c698bbb30761fdb84368519","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_9da5d2ed-7130-4391-940d-934589bf2e6e","section_id":"sec_8a7c1f41-fccb-421e-9cf6-2a3e56b15501","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":104,"end":110,"exact":"エージェント","quote":"間のWebでは、検索エンジンに見つけてもらうために、サイト構造、メタデータ、リンク、権威性が重要になった。AIエージェントのWebでも似たことが起きる。","quote_start":49,"quote_end":125,"text_sha256":"c89e422a0dce923941b6df41c032a3aa72cbb165416150c918f7768b6ee0e43a","block_sha256":"c89e422a0dce923941b6df41c032a3aa72cbb165416150c918f7768b6ee0e43a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_9da5d2ed-7130-4391-940d-934589bf2e6e"},{"id":"occ_2c8d7161c29f34e87dae06a9","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_a95839ca-caef-4bba-b38f-0087cdc91bb5","section_id":"sec_3baaf0a6-5859-4f91-8254-fa0fd544838e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":150,"end":156,"exact":"エージェント","quote":" | SaaS画面、社内ツール | MCPサーバー、API |\n| 分業 | 部署、担当者、外注先 | A2Aエージェント |\n| 手順 | マニュアル、教育資料 | Skills、ワークフロー |\n| 発見 | 社内ポータル、検索 | ARDカタログ、レジストリ |","quote_start":95,"quote_end":229,"text_sha256":"6f227b6489bd0a4256f33b5c953d2f57290a16bf79af36bfbb2d36d8091de35b","block_sha256":"6f227b6489bd0a4256f33b5c953d2f57290a16bf79af36bfbb2d36d8091de35b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_a95839ca-caef-4bba-b38f-0087cdc91bb5"},{"id":"occ_7709d4bd48503c56f90c9336","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_af397bcd-fa7b-415f-bf43-c83a5777224e","section_id":"sec_bb24b233-3916-43a8-80c3-0b3758a9d5bc","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":25,"end":31,"exact":"エージェント","quote":"その意味で、ARDは単なる技術仕様ではない。\nAIエージェントがWeb上、企業内、業務アプリ間で能力を探し、信頼し、接続するための新しいインフラである。","quote_start":0,"quote_end":76,"text_sha256":"b6299fa27c912c278cad9a785566af120229919d466993de1b8afdda236de068","block_sha256":"b6299fa27c912c278cad9a785566af120229919d466993de1b8afdda236de068","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_af397bcd-fa7b-415f-bf43-c83a5777224e"},{"id":"occ_1b4da103719bcc9c50edc37b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_b2bf57d4-b209-496f-b5be-995494cf2b78","section_id":"sec_e250648f-4d88-4198-bd76-e12c5a2fd425","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":59,"end":65,"exact":"エージェント","quote":"ツールを使えばよいのか。\nどのシミュレーション環境にアクセスできるのか。\nどの設計文書が正しいのか。\nどの運用エージェントに依頼すべきか。\nどの操作には人間承認が必要か。","quote_start":4,"quote_end":89,"text_sha256":"5cbcf31917735339d8992a60bd062f04539ecfc01a12f6c440006beccf0f0fd7","block_sha256":"5cbcf31917735339d8992a60bd062f04539ecfc01a12f6c440006beccf0f0fd7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_b2bf57d4-b209-496f-b5be-995494cf2b78"},{"id":"occ_43f32e10c4013a15fa2acb75","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_b2c370cc-4087-4fd3-83ac-02450854ee06","section_id":"sec_5f9d61b2-5985-4cdf-8b4b-fbbdd9ab7467","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":8,"end":14,"exact":"エージェント","quote":"Codexは開発エージェントであり、MCP、Skills、Subagentsを使ってコードベースや外部ツールを操作する。ARDが普及すれば、Codexは必要なSkillやMCPサーバーを発見して使えるようになり、強化される。","quote_start":0,"quote_end":112,"text_sha256":"c268e54218c0c4044bc8262136680986418c9dd29d6bcb406dd172acbece0d17","block_sha256":"c268e54218c0c4044bc8262136680986418c9dd29d6bcb406dd172acbece0d17","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_b2c370cc-4087-4fd3-83ac-02450854ee06"},{"id":"occ_673aba27119c0c100c0fb8ab","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_b74e85ae-188b-4ed4-9cb8-4de9db32d8f9","section_id":"sec_ccba5d74-f5e6-4fde-b96b-b602739527fc","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":33,"end":39,"exact":"エージェント","quote":"第六に、オーケストレーション。\nFuguのように、複数モデルや複数エージェントをどう組み合わせるかが価値になる。","quote_start":0,"quote_end":56,"text_sha256":"550d5b2ad87b111b271ead697c0ddc6a9689bf7070f47dcdf618168c9a69e771","block_sha256":"550d5b2ad87b111b271ead697c0ddc6a9689bf7070f47dcdf618168c9a69e771","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_b74e85ae-188b-4ed4-9cb8-4de9db32d8f9"},{"id":"occ_ae7fa94a308983fcfdb978b0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_bbc3682f-341b-4980-b87c-bf9caa42d3cb","section_id":"sec_5f9d61b2-5985-4cdf-8b4b-fbbdd9ab7467","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":52,"end":58,"exact":"エージェント","quote":"しかし、GitHub Copilot側もARDを使ったagent finderを進めている。これは、開発エージェントが必要なツールやSkillを検索して呼び出す仕組みである。そうなるとCodexとGitHub Copilotの競争は、単なるモデル性能やコード生成精度だけではなくなる。","quote_start":0,"quote_end":142,"text_sha256":"ee5074f404144377ef386e34561bf92ba613d578a4cf4ea138c1e27c4a7b549e","block_sha256":"ee5074f404144377ef386e34561bf92ba613d578a4cf4ea138c1e27c4a7b549e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_bbc3682f-341b-4980-b87c-bf9caa42d3cb"},{"id":"occ_b069891508e0a17183835400","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_c092acc0-1fed-4ffb-b5ae-559f50028d16","section_id":"sec_8f7f847f-8697-41ff-9c29-78e0560565e5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":22,"end":28,"exact":"エージェント","quote":"Sakana Fuguは、複数のモデルや専門エージェントを動的に組み合わせるオーケストレーターである。タスクに応じて、どのモデルを使うか、どの役割を与えるか、どのように連携させるかを調整する。","quote_start":0,"quote_end":96,"text_sha256":"f00fce12da57e47778b69c24d6612d7b05a0d4ea1c458055b75b64fc5b75fc4e","block_sha256":"f00fce12da57e47778b69c24d6612d7b05a0d4ea1c458055b75b64fc5b75fc4e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_c092acc0-1fed-4ffb-b5ae-559f50028d16"},{"id":"occ_1bbc996efa7474c07badc722","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_c2fecfbe-6de7-4c36-93bb-cf77fabd5651","section_id":"sec_2674fee6-cdf1-479d-8e6e-3a0d6c29c620","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":11,"end":17,"exact":"エージェント","quote":"ARDの重要性は、AIエージェントが「事前に接続されたもの」だけに依存しなくなる点にある。","quote_start":0,"quote_end":45,"text_sha256":"ea0e4ab8776ff43cb86afffa276700e0bb09960ee0419ee0beac026c38f96eb7","block_sha256":"ea0e4ab8776ff43cb86afffa276700e0bb09960ee0419ee0beac026c38f96eb7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_c2fecfbe-6de7-4c36-93bb-cf77fabd5651"},{"id":"occ_4ad75b39d3d6c03a2039b407","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_c7bf0335-3e09-4e29-9bcb-50a4a2e8117c","section_id":"sec_8a7c1f41-fccb-421e-9cf6-2a3e56b15501","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":18,"end":24,"exact":"エージェント","quote":"企業は、自社のMCPサーバー、A2Aエージェント、API、Skill、ワークフローを、AIが理解できる形で公開または社内登録する必要が出てくる。名前、説明、入力、出力、認証、権限、料金、監査、信頼情報。これらが整っていない能力は、AIエージェントか","quote_start":0,"quote_end":124,"text_sha256":"03387a900585192ec03a15ea6cb5110d5b8496f3977af20be926d9196995a86e","block_sha256":"03387a900585192ec03a15ea6cb5110d5b8496f3977af20be926d9196995a86e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_c7bf0335-3e09-4e29-9bcb-50a4a2e8117c"},{"id":"occ_eeb4b570ca4494b231750ef3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_ca5e6b25-607c-4f82-aa7d-e58cc9eeb380","section_id":"sec_7543c6ba-0fe3-41dd-97da-94f6ed7a9d26","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"しかし、AIエージェント時代に入ると、重要になるのはモデル単体の性能だけではない。","quote_start":0,"quote_end":41,"text_sha256":"61d80f09cbf19bcf03300ada766df0d20322f1afe75561e2a7c646df19403ebf","block_sha256":"61d80f09cbf19bcf03300ada766df0d20322f1afe75561e2a7c646df19403ebf","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_ca5e6b25-607c-4f82-aa7d-e58cc9eeb380"},{"id":"occ_eb04eb269b2fe4b1be7b5b51","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_da7551ac-910e-4c10-9ded-03a43d96913a","section_id":"sec_2674fee6-cdf1-479d-8e6e-3a0d6c29c620","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":9,"end":15,"exact":"エージェント","quote":"## ARDはAIエージェント版の検索エンジンである","quote_start":0,"quote_end":26,"text_sha256":"9985e479a15807ba7510fca622edf87a3b5129b268643888ef8e1623124fc196","block_sha256":"9985e479a15807ba7510fca622edf87a3b5129b268643888ef8e1623124fc196","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_da7551ac-910e-4c10-9ded-03a43d96913a"},{"id":"occ_8163be3ea342100bd892372c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_dbd7e798-667c-4928-9c60-67659227d526","section_id":"sec_f4da916f-b72c-487d-9b85-7e19a8e01735","layer":"character","character_id":"zetu_noia","count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"だから、AIエージェントのWebでは、能力が存在するだけでは足りません。AIに見つかり、AIに信頼され、AIが安全に接続できる形で置かれている必要があります。","quote_start":0,"quote_end":79,"text_sha256":"6f537cba40bf18b5bf0dcb55f9c77f2735c8adc8d6ff8190dd690896ebc51510","block_sha256":"6f537cba40bf18b5bf0dcb55f9c77f2735c8adc8d6ff8190dd690896ebc51510","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_dbd7e798-667c-4928-9c60-67659227d526"},{"id":"occ_45bb68e59c77c2cbcfe47dee","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_dc6be98e-2944-4dc2-905d-5eb1d8952ef0","section_id":"sec_357132d1-8bfe-4377-90c6-0f15ae0becba","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":52,"end":58,"exact":"エージェント","quote":"つまり、OpenAIやAnthropicは、モデル性能では強くても、発見レイヤーを他社に握られると、AIエージェント経済の入口を失うリスクがある。","quote_start":0,"quote_end":73,"text_sha256":"b16593c58d747ce5a5d6c372408a20fd79675372ddf7f1f9193c1a30f3f467c6","block_sha256":"b16593c58d747ce5a5d6c372408a20fd79675372ddf7f1f9193c1a30f3f467c6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_dc6be98e-2944-4dc2-905d-5eb1d8952ef0"},{"id":"occ_447d083e3ccece715d54c967","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_eafde0d0-dda5-457f-95b1-e0bad61767f7","section_id":"sec_e250648f-4d88-4198-bd76-e12c5a2fd425","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":11,"end":17,"exact":"エージェント","quote":"こうした社内能力をAIエージェントが安全に発見するには、プライベートなARDレジストリが有効になる。","quote_start":0,"quote_end":50,"text_sha256":"e44ebe9926e95e0ec33974098047b4a50125e2cc6c105d1aa838226f062b2ff1","block_sha256":"e44ebe9926e95e0ec33974098047b4a50125e2cc6c105d1aa838226f062b2ff1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_eafde0d0-dda5-457f-95b1-e0bad61767f7"},{"id":"occ_c7e25c95a3a86332c813b9f3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_f200d4f3-1520-4157-942d-dd419197731e","section_id":"sec_bb24b233-3916-43a8-80c3-0b3758a9d5bc","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":10,"end":16,"exact":"エージェント","quote":"ARDの登場は、AIエージェント時代が次の段階に入ったことを示している。","quote_start":0,"quote_end":36,"text_sha256":"3fc7c488f05efa0820cf46f57aed38cbf5f3f4b885137b8cc24de2e57a9f3274","block_sha256":"3fc7c488f05efa0820cf46f57aed38cbf5f3f4b885137b8cc24de2e57a9f3274","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_f200d4f3-1520-4157-942d-dd419197731e"},{"id":"occ_4565bc67c1b6769a60f84536","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_f2ba403b-7a6f-45a2-8d62-51945bd7cbf1","section_id":"sec_2674fee6-cdf1-479d-8e6e-3a0d6c29c620","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"しかし、エージェント時代が本格化すると、利用可能なツール、Skill、MCPサーバー、A2Aエージェント、API、ワークフローは膨大になる。","quote_start":0,"quote_end":70,"text_sha256":"244ec7f717d7de59ff848c983dcf0a853420f45004515d39fca2e111ccb1d1ef","block_sha256":"244ec7f717d7de59ff848c983dcf0a853420f45004515d39fca2e111ccb1d1ef","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_f2ba403b-7a6f-45a2-8d62-51945bd7cbf1"},{"id":"occ_782a46389ee382070cb4f9c8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_f31e63d0-6698-48a2-a999-6c22a26a5957","section_id":"sec_7543c6ba-0fe3-41dd-97da-94f6ed7a9d26","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"# AIエージェントのWebが始まる──A2A、MCP、Skills、ARD、Fuguが作る「デジタル神経系」","quote_start":0,"quote_end":55,"text_sha256":"48aada15ffef17ba18acd74795aa204e3a6a62e8ad111667d8ee3fcd243784fd","block_sha256":"48aada15ffef17ba18acd74795aa204e3a6a62e8ad111667d8ee3fcd243784fd","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_f31e63d0-6698-48a2-a999-6c22a26a5957"},{"id":"occ_84b88b44f04b28583444c48f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_f5c63b3d-af5b-4ed8-a8c6-128935065282","section_id":"sec_357132d1-8bfe-4377-90c6-0f15ae0becba","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":94,"end":100,"exact":"エージェント","quote":"Hugging Face側が握れば、OpenAIやAnthropicは「最高のモデルを持つ会社」であっても、「エージェントがどの能力を見つけ、どのツールを使うか」を完全には支配できない。","quote_start":39,"quote_end":132,"text_sha256":"b2465a1937e70c66d4293ddc67a15d81fe461a110476e3918084206519e35a6a","block_sha256":"b2465a1937e70c66d4293ddc67a15d81fe461a110476e3918084206519e35a6a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_f5c63b3d-af5b-4ed8-a8c6-128935065282"},{"id":"occ_6fc3e00c43a8b4da4516237d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_fd5fbed6-f4af-405e-abd3-57fabdd78f4c","section_id":"sec_2674fee6-cdf1-479d-8e6e-3a0d6c29c620","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":51,"end":57,"exact":"エージェント","quote":"企業や開発者は、自分のドメインにAI向けカタログを置く。そこには、自社が提供するMCPサーバー、A2Aエージェント、API、Skill、ワークフローなどが記述される。AIクライアントはレジストリを通じて、そのカタログを検索し、必要な能力を見つける。","quote_start":0,"quote_end":124,"text_sha256":"56d7bc95443d900cbf7797be274a19cf79cc8c3ab6de9bb911a5bbbc0736890f","block_sha256":"56d7bc95443d900cbf7797be274a19cf79cc8c3ab6de9bb911a5bbbc0736890f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_fd5fbed6-f4af-405e-abd3-57fabdd78f4c"},{"id":"occ_d1fc550ff98cf6c9d643a1fc","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d","block_id":"blk_feba1541-65b7-46ad-8123-5e41fb57b246","section_id":"sec_8c121823-5eff-45c7-aa58-1ca1c60f68c8","layer":"body","character_id":null,"count":3,"matched_aliases":["Agentic","エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":11,"exact":"Agentic","quote":"ARD、Agentic Resource Discoveryは、AIエージェントが「どのMCPサーバーを使うべきか」「どのSkillがあるか」「どのA2Aエージェントに頼めるか」「どのAPIやワークフローがこのタスクに向い","quote_start":0,"quote_end":111,"text_sha256":"7cc948930c92c8991c6d3aa6115f7abd99ec7082fd5e549da5e1d6eec6443445","block_sha256":"7cc948930c92c8991c6d3aa6115f7abd99ec7082fd5e549da5e1d6eec6443445","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ddc53551-0396-485b-a72e-8b3d5126293f/#blk_feba1541-65b7-46ad-8123-5e41fb57b246"},{"id":"occ_3c6c7bed04b0d05ca49aa037","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646","work_id":"wrk_4c390ee8-19a9-47a2-baff-32ac4e8e79ad","block_id":"blk_0b86d9eb-6cf2-4004-b59c-95766ebea461","section_id":"sec_f80f85d9-1a66-42af-8ee0-ea7f0769a6c4","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agentが本格的に仕事をする時代について考えていると、Computerという言葉そのものの意味が変わってくる。","quote_start":0,"quote_end":59,"text_sha256":"bd493aa7471bbb544acde34d6fc0a115e3106f533b0788a084f64425ee6f3f37","block_sha256":"bd493aa7471bbb544acde34d6fc0a115e3106f533b0788a084f64425ee6f3f37","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646/#blk_0b86d9eb-6cf2-4004-b59c-95766ebea461"},{"id":"occ_bf634e963d4b81fe03b579f6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646","work_id":"wrk_4c390ee8-19a9-47a2-baff-32ac4e8e79ad","block_id":"blk_1b1e2350-f4ca-4272-953d-30b2bd363223","section_id":"sec_07a4322b-6c55-4bad-bab9-5c4274bc9eb4","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":10,"end":17,"exact":"Agentic","quote":"AMDも2026年、Agentic AIは従来のGPU-heavy RackへCPUを少し追加する話ではなく、**GPU Infrastructureの横に新しいCPU Server Rack群が必要になる構造変化**だと説明している","quote_start":0,"quote_end":117,"text_sha256":"7e305d3c2cdb028e65a69ac3567ddb28c60a0a10cd0e3d322eba8dff65768133","block_sha256":"7e305d3c2cdb028e65a69ac3567ddb28c60a0a10cd0e3d322eba8dff65768133","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646/#blk_1b1e2350-f4ca-4272-953d-30b2bd363223"},{"id":"occ_c912c2324333f472e4228bf0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646","work_id":"wrk_4c390ee8-19a9-47a2-baff-32ac4e8e79ad","block_id":"blk_8a19f342-0302-4eaf-a6ed-0db86326203f","section_id":"sec_ee030aed-a5e0-4a43-8b6d-81840aa1d4ba","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agentが「第二のPC」を持つ時代である。","quote_start":0,"quote_end":25,"text_sha256":"f809661801b37702ae3744a8f709e2690bce40fdc7768377425e64ae83f3acb6","block_sha256":"f809661801b37702ae3744a8f709e2690bce40fdc7768377425e64ae83f3acb6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646/#blk_8a19f342-0302-4eaf-a6ed-0db86326203f"},{"id":"occ_9fdfa6fda4b514bf03e5abe0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646","work_id":"wrk_4c390ee8-19a9-47a2-baff-32ac4e8e79ad","block_id":"blk_901a8d83-3f47-49f3-ad10-d84fa617f4d1","section_id":"sec_00c32239-6a25-4376-96f6-3c7b3165338d","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":0,"end":7,"exact":"Agentic","quote":"Agentic AIではInferenceが単発Queryから継続的なIterative Cycleへ変わる。","quote_start":0,"quote_end":55,"text_sha256":"ceea992a7fa9732a398fb48b9f5caf541ac939c9ba5db08c22abe131e761fd3a","block_sha256":"ceea992a7fa9732a398fb48b9f5caf541ac939c9ba5db08c22abe131e761fd3a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646/#blk_901a8d83-3f47-49f3-ad10-d84fa617f4d1"},{"id":"occ_054ee3905e74836867c40ca2","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646","work_id":"wrk_4c390ee8-19a9-47a2-baff-32ac4e8e79ad","block_id":"blk_a8a91e94-e885-4a8c-81fc-6a0fbdc7618b","section_id":"sec_f80f85d9-1a66-42af-8ee0-ea7f0769a6c4","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":3,"end":11,"exact":"AI Agent","quote":"しかしAI Agentは、人間と同じ速度では動かない。","quote_start":0,"quote_end":27,"text_sha256":"9255e579f3ed362d0de5625ed67a9eb2694ba299785196a5b875b6f1d82476ca","block_sha256":"9255e579f3ed362d0de5625ed67a9eb2694ba299785196a5b875b6f1d82476ca","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646/#blk_a8a91e94-e885-4a8c-81fc-6a0fbdc7618b"},{"id":"occ_9cad052a00e90065f76e65a5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646","work_id":"wrk_4c390ee8-19a9-47a2-baff-32ac4e8e79ad","block_id":"blk_d4aff54a-9cf3-47c4-a22c-d3bbe6147777","section_id":"sec_39252be7-a40f-4f19-b570-5bb1e711e414","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":0,"end":7,"exact":"Agentic","quote":"Agentic Workloadでは、","quote_start":0,"quote_end":19,"text_sha256":"5a03cd213043bfaff3b15f49ff543fde6d7b0005fd3d011083f6e4909f1b724c","block_sha256":"5a03cd213043bfaff3b15f49ff543fde6d7b0005fd3d011083f6e4909f1b724c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646/#blk_d4aff54a-9cf3-47c4-a22c-d3bbe6147777"},{"id":"occ_918dc07b87b81db7553b38cd","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646","work_id":"wrk_4c390ee8-19a9-47a2-baff-32ac4e8e79ad","block_id":"blk_daa2059a-c854-42ca-9b6c-124491805313","section_id":"sec_6bda6cf2-5fdc-47a6-b819-0bcefdcda0b8","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":44,"end":52,"exact":"AI Agent","quote":"Primary Agent Activity――主Agent利用量。  \n一人が中心的なAI Agentで一日に処理するToken量。","quote_start":0,"quote_end":67,"text_sha256":"3e2e08563b41321e10c58ce34e546a63652744508a80391cf24b1b6764d6db71","block_sha256":"3e2e08563b41321e10c58ce34e546a63652744508a80391cf24b1b6764d6db71","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646/#blk_daa2059a-c854-42ca-9b6c-124491805313"},{"id":"occ_d3fa3b06c62cf77a8eefe3e8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646","work_id":"wrk_4c390ee8-19a9-47a2-baff-32ac4e8e79ad","block_id":"blk_f0b8cebf-49be-4ee4-b82d-c2bc98dbf342","section_id":"sec_f80f85d9-1a66-42af-8ee0-ea7f0769a6c4","layer":"body","character_id":null,"count":2,"matched_aliases":["AI Agent","エージェント"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agent――AIエージェント。  \n人間から目的を受け取り、調査・判断・Tool操作・Computer操作などを自律的に進めるAIである。","quote_start":0,"quote_end":73,"text_sha256":"9c3e2baee7ace93c4c366e22e819ce7a93672b738ababe4f6e4ba9dd7234882a","block_sha256":"9c3e2baee7ace93c4c366e22e819ce7a93672b738ababe4f6e4ba9dd7234882a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646/#blk_f0b8cebf-49be-4ee4-b82d-c2bc98dbf342"},{"id":"occ_a7a8faf88ea741d81e36b288","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_e1c0a212-5ad0-4fbc-81f6-60d2df49d753","work_id":"wrk_88c25067-a100-4eee-96f7-180120d2ac61","block_id":"blk_98bfd302-2d13-4f02-bba2-e974168ff11c","section_id":"sec_6177e688-bbe6-43d1-9d75-29ddb6a92410","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":17,"end":24,"exact":"Agentic","quote":"AMDも第6世代EPYCについて、Agentic AIではplanning、retrieval、tool executionなどCPU側の仕事が増えると説明している。 ([AMD](https://www.amd.com/en/products/p","quote_start":0,"quote_end":124,"text_sha256":"bc506ef2854f13a9e9b873bc34afb4affb7b5f4dbe5b2b52570732035230b741","block_sha256":"bc506ef2854f13a9e9b873bc34afb4affb7b5f4dbe5b2b52570732035230b741","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_e1c0a212-5ad0-4fbc-81f6-60d2df49d753/#blk_98bfd302-2d13-4f02-bba2-e974168ff11c"},{"id":"occ_3e4f592a3e3edc62aae8afc4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_e1c0a212-5ad0-4fbc-81f6-60d2df49d753","work_id":"wrk_88c25067-a100-4eee-96f7-180120d2ac61","block_id":"blk_bb683eb4-1074-4136-8d11-a0dd8f3cf04c","section_id":"sec_3507b738-5337-4662-b205-109370db662f","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":20,"end":27,"exact":"Agentic","quote":"AMDとMoonshot AIが構築したAgentic AI stackでは、さらにUMBP（Unified Memory & Bandwidth Pool）として、","quote_start":0,"quote_end":83,"text_sha256":"1c5586cc091245a7ffb2605933c0b754cdebb4655756af94385a89530e49debd","block_sha256":"1c5586cc091245a7ffb2605933c0b754cdebb4655756af94385a89530e49debd","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_e1c0a212-5ad0-4fbc-81f6-60d2df49d753/#blk_bb683eb4-1074-4136-8d11-a0dd8f3cf04c"},{"id":"occ_47ef544fbc7ffbccceb2030a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_e2c369be-a992-44a9-bbc1-81ed4928e89f","work_id":"wrk_d2f61eaa-5ee5-4d1f-b70d-7356a79ea61d","block_id":"blk_062665c5-122b-453b-ab05-eae44ecee4ce","section_id":"sec_bb58f780-8c88-48b3-8458-52c8755c724f","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":0,"end":7,"exact":"Agentic","quote":"Agentic workloads。","quote_start":0,"quote_end":18,"text_sha256":"9a27abb2e91363cd6f43068df30915308a5a86e7f523523e2b9e7fba79e6d562","block_sha256":"9a27abb2e91363cd6f43068df30915308a5a86e7f523523e2b9e7fba79e6d562","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_e2c369be-a992-44a9-bbc1-81ed4928e89f/#blk_062665c5-122b-453b-ab05-eae44ecee4ce"},{"id":"occ_d598ca1df6e7cebe537586a2","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_e5e24c6f-5887-4f16-b03e-081da5340883","work_id":"wrk_470cca01-eadf-46c3-bd50-cd79e3b1e197","block_id":"blk_0d657f7e-95dc-4609-80aa-23e6877265b7","section_id":"sec_bc08c3d6-a070-4920-adb5-cad90465aba2","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":15,"end":21,"exact":"エージェント","quote":"## 18．PARL――複数のエージェントへ、どう仕事を分けるかを学ぶ","quote_start":0,"quote_end":35,"text_sha256":"8c45badcafdf5f241fb2f1a05b4317913653ad9b2294cf208bfa3d1b161b2e83","block_sha256":"8c45badcafdf5f241fb2f1a05b4317913653ad9b2294cf208bfa3d1b161b2e83","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_e5e24c6f-5887-4f16-b03e-081da5340883/#blk_0d657f7e-95dc-4609-80aa-23e6877265b7"},{"id":"occ_31da7960e520617565f80bbe","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_e5e24c6f-5887-4f16-b03e-081da5340883","work_id":"wrk_470cca01-eadf-46c3-bd50-cd79e3b1e197","block_id":"blk_0e82c023-5990-4eb5-988c-1ef9184ba599","section_id":"sec_8f229c67-6ddf-48ff-9929-0b7d092a0b77","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":30,"end":36,"exact":"エージェント","quote":"数学なら、答えが合っていたか。コードなら、テストに通ったか。エージェントなら、依頼された作業を完了したか。","quote_start":0,"quote_end":53,"text_sha256":"f32388c2517c4a8dd4cc63d210cbbb85f4a0429530ba5332d6c9dd230688c3f6","block_sha256":"f32388c2517c4a8dd4cc63d210cbbb85f4a0429530ba5332d6c9dd230688c3f6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_e5e24c6f-5887-4f16-b03e-081da5340883/#blk_0e82c023-5990-4eb5-988c-1ef9184ba599"},{"id":"occ_da7301e76cc5a17e47c13a0d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_e5e24c6f-5887-4f16-b03e-081da5340883","work_id":"wrk_470cca01-eadf-46c3-bd50-cd79e3b1e197","block_id":"blk_446ef8b8-300d-487e-8a3f-562db0e35f8c","section_id":"sec_3d1a7f8f-6e04-4555-b855-a2035f01c073","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":39,"end":45,"exact":"エージェント","quote":"例えば、Kimi K2では、追加学習を自由に行う出発点としてBase版、対話やエージェント用途へ調整したものとしてInstruct版を分けている。([Hugging Face](https://huggingface.co/moonshotai/Kimi-K2-Base))","quote_start":0,"quote_end":137,"text_sha256":"9a66c42beada8a4b9b1289318bdeb1c60ae57f3ac351ebc75a490946e0aa556d","block_sha256":"9a66c42beada8a4b9b1289318bdeb1c60ae57f3ac351ebc75a490946e0aa556d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_e5e24c6f-5887-4f16-b03e-081da5340883/#blk_446ef8b8-300d-487e-8a3f-562db0e35f8c"},{"id":"occ_b16e27b735dfa55792dde41b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_e5e24c6f-5887-4f16-b03e-081da5340883","work_id":"wrk_470cca01-eadf-46c3-bd50-cd79e3b1e197","block_id":"blk_53e6643e-e7cd-4059-b2e0-558ff03cbc69","section_id":"sec_bc08c3d6-a070-4920-adb5-cad90465aba2","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":10,"end":16,"exact":"エージェント","quote":"学ばせたいのは、単にエージェントをたくさん呼ぶことではない。","quote_start":0,"quote_end":30,"text_sha256":"8940051f1700bce2b32300a5d686a7bc0a8df029cde6444c913b9a233a72e9cb","block_sha256":"8940051f1700bce2b32300a5d686a7bc0a8df029cde6444c913b9a233a72e9cb","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_e5e24c6f-5887-4f16-b03e-081da5340883/#blk_53e6643e-e7cd-4059-b2e0-558ff03cbc69"},{"id":"occ_ba15b15a8da0b67eedb132db","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_e5e24c6f-5887-4f16-b03e-081da5340883","work_id":"wrk_470cca01-eadf-46c3-bd50-cd79e3b1e197","block_id":"blk_75875000-73b4-4ade-ace2-383d42fc774e","section_id":"sec_6f6f73c9-253a-44e6-bdae-a6b2d9509e73","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":45,"end":51,"exact":"エージェント","quote":"しかし、一つの能力を伸ばす変更が、別の能力を損なうこともある。また、短い数学問題と長時間のエージェント作業では、試行や評価にかかる時間が違う。","quote_start":0,"quote_end":71,"text_sha256":"72015aa6c557b4180694cb122786a0a5be7b8f0ef44689e7e12684981452b6a2","block_sha256":"72015aa6c557b4180694cb122786a0a5be7b8f0ef44689e7e12684981452b6a2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_e5e24c6f-5887-4f16-b03e-081da5340883/#blk_75875000-73b4-4ade-ace2-383d42fc774e"},{"id":"occ_8d5e8d271a33e0aba12ae7ee","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_e5e24c6f-5887-4f16-b03e-081da5340883","work_id":"wrk_470cca01-eadf-46c3-bd50-cd79e3b1e197","block_id":"blk_97adf40c-dcac-4dc9-8b99-cc8de373ea55","section_id":"sec_f0638cb4-bc28-4d8b-891e-268b51d48d43","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":30,"end":36,"exact":"エージェント","quote":"DeepSeekMath-V2は検証者を鍛え、PARLは複数エージェントの協調を学ぶ。RLPは事前学習へ推論の評価を入れ、SEALは自分を改善する教材づくりを学ぶ。","quote_start":0,"quote_end":82,"text_sha256":"787134e6b762830ee531c298366d2a798d8fca3bf300a8678675f81ef1b313ca","block_sha256":"787134e6b762830ee531c298366d2a798d8fca3bf300a8678675f81ef1b313ca","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_e5e24c6f-5887-4f16-b03e-081da5340883/#blk_97adf40c-dcac-4dc9-8b99-cc8de373ea55"},{"id":"occ_50237c3d6174af64c3a8c3aa","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_e5e24c6f-5887-4f16-b03e-081da5340883","work_id":"wrk_470cca01-eadf-46c3-bd50-cd79e3b1e197","block_id":"blk_a109ef48-8a24-41e7-bc0f-75b3f1c96646","section_id":"sec_944f873a-22d5-44b2-a24a-eb7ee4123a37","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":4,"end":10,"exact":"エージェント","quote":"### エージェントRLでは、環境とハーネスが必要になる","quote_start":0,"quote_end":28,"text_sha256":"80f674ce25b0cdd6d60fa3a0a59397fc3df78a260441b69b6d36a936fbab0f51","block_sha256":"80f674ce25b0cdd6d60fa3a0a59397fc3df78a260441b69b6d36a936fbab0f51","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_e5e24c6f-5887-4f16-b03e-081da5340883/#blk_a109ef48-8a24-41e7-bc0f-75b3f1c96646"},{"id":"occ_e037bc58023cb136079b52e8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_e5e24c6f-5887-4f16-b03e-081da5340883","work_id":"wrk_470cca01-eadf-46c3-bd50-cd79e3b1e197","block_id":"blk_d092f0d1-fb85-4872-9d7a-22d753384643","section_id":"sec_bc08c3d6-a070-4920-adb5-cad90465aba2","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":33,"end":39,"exact":"エージェント","quote":"公表された構成では、仕事を分解して割り当てる指揮役を訓練し、下位のエージェントは固定した重みで動かす。([Kimi](https://www.kimi.com/blog/kimi-k2-5.html))","quote_start":0,"quote_end":101,"text_sha256":"51e95be2b9924205b7419768529718c665a4479ce1ce4ff587511cd47df05357","block_sha256":"51e95be2b9924205b7419768529718c665a4479ce1ce4ff587511cd47df05357","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_e5e24c6f-5887-4f16-b03e-081da5340883/#blk_d092f0d1-fb85-4872-9d7a-22d753384643"},{"id":"occ_39962f08f600b8666f5d1517","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_e5e24c6f-5887-4f16-b03e-081da5340883","work_id":"wrk_470cca01-eadf-46c3-bd50-cd79e3b1e197","block_id":"blk_df6e2954-2f99-4c0a-a0d7-34c37026fe07","section_id":"sec_bc08c3d6-a070-4920-adb5-cad90465aba2","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":60,"end":66,"exact":"エージェント","quote":"K2.5の**PARL、Parallel-Agent Reinforcement Learning**は、複数エージェントによる並列作業の進め方を学習する。","quote_start":5,"quote_end":83,"text_sha256":"c4e0d9d23c3433ba84299619d29c96909da1b75978fa90201b4f04656cfae73e","block_sha256":"c4e0d9d23c3433ba84299619d29c96909da1b75978fa90201b4f04656cfae73e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_e5e24c6f-5887-4f16-b03e-081da5340883/#blk_df6e2954-2f99-4c0a-a0d7-34c37026fe07"},{"id":"occ_28114c06d30cd08d5d60565b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eaa7004e-c9e4-4bc4-9718-2e518decec70","work_id":"wrk_dfcbc670-103a-4931-8d23-d5f51671c61b","block_id":"blk_198c5a24-d70d-4a5d-ad9d-f2e23703b759","section_id":"sec_1afdbaf1-5786-4caa-9bb7-1a11027e856f","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":225,"end":232,"exact":"Agentic","quote":"B、Signal Integrity、Power Integrity、Thermal、MechanicalまでAgentic AIで扱い始めた。(Cadence Newsroom)","quote_start":170,"quote_end":260,"text_sha256":"0c959918f2a5a6ea6d165d11542e4363c7019bb4bd313975e67681e7b49b4325","block_sha256":"0c959918f2a5a6ea6d165d11542e4363c7019bb4bd313975e67681e7b49b4325","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eaa7004e-c9e4-4bc4-9718-2e518decec70/#blk_198c5a24-d70d-4a5d-ad9d-f2e23703b759"},{"id":"occ_f71449d8a9e1f1f6233c35e4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eaa7004e-c9e4-4bc4-9718-2e518decec70","work_id":"wrk_dfcbc670-103a-4931-8d23-d5f51671c61b","block_id":"blk_660c09b5-3f15-4322-9b4c-d392177213a9","section_id":"sec_1afdbaf1-5786-4caa-9bb7-1a11027e856f","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":2,"end":10,"exact":"AI Agent","quote":"までAI Agentで扱い始めた。(Cadence Newsroom)","quote_start":0,"quote_end":35,"text_sha256":"1ef84eaed3d344f17460dd0f6458653777a17bfb20e796e626009a89fa2e3978","block_sha256":"1ef84eaed3d344f17460dd0f6458653777a17bfb20e796e626009a89fa2e3978","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eaa7004e-c9e4-4bc4-9718-2e518decec70/#blk_660c09b5-3f15-4322-9b4c-d392177213a9"},{"id":"occ_7b20076f11d60ce98a7bdafc","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eaa7004e-c9e4-4bc4-9718-2e518decec70","work_id":"wrk_dfcbc670-103a-4931-8d23-d5f51671c61b","block_id":"blk_81f29ec5-1f34-4447-bc9d-fc51ce5e92bd","section_id":"sec_1afdbaf1-5786-4caa-9bb7-1a11027e856f","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":79,"end":86,"exact":"Agentic","quote":"cture、Model Deployment、Inferenceまでを自動化し、Private Cloud上でAgentic ApplicationをBuild、Run、GovernするPlatformである。(Broadcom)","quote_start":24,"quote_end":139,"text_sha256":"5b8d362d71ba598df718925de53d56be55e7cd335b4317a74614dacfb67b3654","block_sha256":"5b8d362d71ba598df718925de53d56be55e7cd335b4317a74614dacfb67b3654","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eaa7004e-c9e4-4bc4-9718-2e518decec70/#blk_81f29ec5-1f34-4447-bc9d-fc51ce5e92bd"},{"id":"occ_33b88553bc5e6f53661dd306","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eaa7004e-c9e4-4bc4-9718-2e518decec70","work_id":"wrk_dfcbc670-103a-4931-8d23-d5f51671c61b","block_id":"blk_a8806264-4668-48eb-8ec3-d55b7c008f28","section_id":"sec_1afdbaf1-5786-4caa-9bb7-1a11027e856f","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":62,"end":70,"exact":"AI Agent","quote":"Frontier Modelを最初に使える企業が、自社のModel Workload Dataを与え、24時間AI Agentを動かしてChipを改善する。","quote_start":7,"quote_end":85,"text_sha256":"fe5473aee095e1636dcd2fe96911c567666bf0e8da89282caefb69c9822ca883","block_sha256":"fe5473aee095e1636dcd2fe96911c567666bf0e8da89282caefb69c9822ca883","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eaa7004e-c9e4-4bc4-9718-2e518decec70/#blk_a8806264-4668-48eb-8ec3-d55b7c008f28"},{"id":"occ_4214df484ce76f04582f4846","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eaa7004e-c9e4-4bc4-9718-2e518decec70","work_id":"wrk_dfcbc670-103a-4931-8d23-d5f51671c61b","block_id":"blk_aa72965a-5990-465a-b880-dddf5ecae8e2","section_id":"sec_1afdbaf1-5786-4caa-9bb7-1a11027e856f","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":7,"end":14,"exact":"Agentic","quote":"GoogleもAgentic EDAへ進んでいる","quote_start":0,"quote_end":24,"text_sha256":"433557a5fb83b0a1df0f6dbd4bee3f6a6da7fe92bbf627061d7692402972ce18","block_sha256":"433557a5fb83b0a1df0f6dbd4bee3f6a6da7fe92bbf627061d7692402972ce18","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eaa7004e-c9e4-4bc4-9718-2e518decec70/#blk_aa72965a-5990-465a-b880-dddf5ecae8e2"},{"id":"occ_41f9f77f7eff2cf8b5812ec6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e","work_id":"wrk_65ea8146-0c98-40e2-b084-2dacaa4f1bf1","block_id":"blk_0f65b6e7-b1b5-4239-916a-d6bf781d2f20","section_id":"sec_df50e932-cf79-47ef-b9d2-c72831bc8342","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェントが普及すると、世界中で膨大な推論が常時走ります。全部を最高級GPUで回すのは電力・コスト的に重い。そこで、","quote_start":0,"quote_end":61,"text_sha256":"cbb5998f2208a7e630d47dd146f68d6c316494f4c7fbd232cb137c26a648b68c","block_sha256":"cbb5998f2208a7e630d47dd146f68d6c316494f4c7fbd232cb137c26a648b68c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e/#blk_0f65b6e7-b1b5-4239-916a-d6bf781d2f20"},{"id":"occ_e3ece3729e33050684875599","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e","work_id":"wrk_65ea8146-0c98-40e2-b084-2dacaa4f1bf1","block_id":"blk_204bdf02-8249-4957-a68b-4503a82417b7","section_id":"sec_c147381c-c845-4eab-9414-a97208977392","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":31,"end":37,"exact":"エージェント","quote":"Qualcommは、AI需要が下り方向の動画消費中心から、AIエージェント、ロボティクス、センサーリッチな端末による**上り方向・低遅延・双方向ワークロード**へ変わると説明しています。これはかなり重要な視点です。([Qualcomm](https://www.qualco","quote_start":0,"quote_end":137,"text_sha256":"3dcb16b8c18a5c74d304fecef47db4466f763e970745d3a1a677b6821e0a6b5a","block_sha256":"3dcb16b8c18a5c74d304fecef47db4466f763e970745d3a1a677b6821e0a6b5a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e/#blk_204bdf02-8249-4957-a68b-4503a82417b7"},{"id":"occ_8351ea1b463b8fa282d5b9ff","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e","work_id":"wrk_65ea8146-0c98-40e2-b084-2dacaa4f1bf1","block_id":"blk_4e7427cb-b714-42bf-a3c3-9735f325bde5","section_id":"sec_c147381c-c845-4eab-9414-a97208977392","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":89,"end":95,"exact":"エージェント","quote":"- エッジでAI推論する\n- デジタルツインと現実をつなぐ\n- 車・ドローン・ロボットの制御を支える\n- AIエージェント端末の双方向通信を支える","quote_start":34,"quote_end":107,"text_sha256":"0c3087cb85f82181b0cc95e66496344d5d7b36d0bec027f7ded7ae8a06f02867","block_sha256":"0c3087cb85f82181b0cc95e66496344d5d7b36d0bec027f7ded7ae8a06f02867","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e/#blk_4e7427cb-b714-42bf-a3c3-9735f325bde5"},{"id":"occ_b3269688fcc79dc627434093","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e","work_id":"wrk_65ea8146-0c98-40e2-b084-2dacaa4f1bf1","block_id":"blk_72769833-1848-4b84-833b-40c8d3c22d7f","section_id":"sec_09c43d29-39ab-44b6-8eec-f13e43c51c5a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":46,"end":52,"exact":"エージェント","quote":"これからのAIインフラでは、GPUだけでなくCPU需要も高まる可能性があります。理由は、AIエージェントが単にLLMを一回呼び出すだけではなく、外部ツール、ブラウザ、Python、データベース、検索、RAG、API、メモリ管理、スケジューリングを大量に使うからです。","quote_start":0,"quote_end":133,"text_sha256":"4fea965a32c75e82fe94418465b1f06a3b0390014ba1b98f9693263f5575b323","block_sha256":"4fea965a32c75e82fe94418465b1f06a3b0390014ba1b98f9693263f5575b323","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e/#blk_72769833-1848-4b84-833b-40c8d3c22d7f"},{"id":"occ_7db2f0b1fcb319c89b6dfe02","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e","work_id":"wrk_65ea8146-0c98-40e2-b084-2dacaa4f1bf1","block_id":"blk_7b8984f9-8101-463c-afe8-26962b65d1fe","section_id":"sec_09c43d29-39ab-44b6-8eec-f13e43c51c5a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェント時代にCPU需要が増えるなら、クアルコムは「端末側CPU」だけでなく、「エッジ/データセンターCPU」でも評価される可能性があります。","quote_start":0,"quote_end":75,"text_sha256":"94c868f3df1b2850066a85cfd5782729bc925b792a4f91a60996542748ad1374","block_sha256":"94c868f3df1b2850066a85cfd5782729bc925b792a4f91a60996542748ad1374","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e/#blk_7b8984f9-8101-463c-afe8-26962b65d1fe"},{"id":"occ_bde1a7d9fd1d9575a4aa61fc","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e","work_id":"wrk_65ea8146-0c98-40e2-b084-2dacaa4f1bf1","block_id":"blk_aeaf9c90-24d6-4107-ab08-b39f49ce03c6","section_id":"sec_09c43d29-39ab-44b6-8eec-f13e43c51c5a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":19,"end":25,"exact":"エージェント","quote":"## \\5. CPU需要の高まり：AIエージェント時代はGPUだけでは回らない","quote_start":0,"quote_end":39,"text_sha256":"702904d2f88a538f86ed17f4043a1fbdc77467c042c7be1f5d8284811b221330","block_sha256":"702904d2f88a538f86ed17f4043a1fbdc77467c042c7be1f5d8284811b221330","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e/#blk_aeaf9c90-24d6-4107-ab08-b39f49ce03c6"},{"id":"occ_ac2b55f2430bb1f820689e75","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e","work_id":"wrk_65ea8146-0c98-40e2-b084-2dacaa4f1bf1","block_id":"blk_dca65c76-53aa-4c07-a313-77b2c60609c2","section_id":"sec_09c43d29-39ab-44b6-8eec-f13e43c51c5a","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":0,"end":7,"exact":"Agentic","quote":"Agentic AIの研究でも、ツール処理が全体レイテンシの大きな割合を占め、CPU側の同期、コアの過剰使用、メモリ帯域などがボトルネックになり得ると分析されています。([arXiv](https://arxiv.","quote_start":0,"quote_end":107,"text_sha256":"d9382fbfe5d0443a197004c59bb65db9789a5ee4ce1452863f93bd713f312b85","block_sha256":"d9382fbfe5d0443a197004c59bb65db9789a5ee4ce1452863f93bd713f312b85","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e/#blk_dca65c76-53aa-4c07-a313-77b2c60609c2"},{"id":"occ_474af06e866d374874d5298d","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e","work_id":"wrk_65ea8146-0c98-40e2-b084-2dacaa4f1bf1","block_id":"blk_ff2f6a54-1635-4b6d-92d5-bef4ac4f602c","section_id":"sec_df50e932-cf79-47ef-b9d2-c72831bc8342","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":47,"end":53,"exact":"エージェント","quote":"- 小型〜中型LLM\n- 企業内RAG\n- コード補完\n- 音声AI\n- 画像/動画推論\n- エージェントのツール実行\n- オンプレ/プライベートAI\n- ソブリンAI","quote_start":0,"quote_end":84,"text_sha256":"c1f42da5b0f7d64879193e404634dbfa39b40a6291acf4ee6d6aa0b0caa8ab1a","block_sha256":"c1f42da5b0f7d64879193e404634dbfa39b40a6291acf4ee6d6aa0b0caa8ab1a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e/#blk_ff2f6a54-1635-4b6d-92d5-bef4ac4f602c"},{"id":"occ_b2e55705d545205e3c511524","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd","block_id":"blk_0e02490f-002c-4b64-886a-7e033473fcb1","section_id":"sec_c7c2e836-8756-49fb-a27b-daab6d45afbd","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":22,"end":29,"exact":"Agentic","quote":"Elasticは今回、KubernetesのAgentic Investigationも拡張しており、Metrics、Logs、Cluster EventなどをAgentが横断してRoot Causeを探す方向へ進んでいる。(Elastic)","quote_start":0,"quote_end":121,"text_sha256":"3d7036ad1851fbcb4cd0e4232560c57b17cc1050909ea95f2cc08310dd5ccc3b","block_sha256":"3d7036ad1851fbcb4cd0e4232560c57b17cc1050909ea95f2cc08310dd5ccc3b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4/#blk_0e02490f-002c-4b64-886a-7e033473fcb1"},{"id":"occ_951090f827a66d636c0bc990","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd","block_id":"blk_11f08372-b1dc-4def-b003-d002ff6d5b2a","section_id":"sec_99fca058-960a-4947-b362-1bcb25d27bac","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agentが、","quote_start":0,"quote_end":10,"text_sha256":"eb540ab992433829579400ab7bea51752bc4aabcdfa9bbe6420e86c88472346b","block_sha256":"eb540ab992433829579400ab7bea51752bc4aabcdfa9bbe6420e86c88472346b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4/#blk_11f08372-b1dc-4def-b003-d002ff6d5b2a"},{"id":"occ_892672ea40d986fb3f4c3668","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd","block_id":"blk_1407b20a-f011-4aea-85e9-03997ad1418f","section_id":"sec_c7c2e836-8756-49fb-a27b-daab6d45afbd","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI AgentやAI ApplicationがKubernetes上へ大量展開されると、これらのMetrics・Logs・Tracesも急増する。","quote_start":0,"quote_end":74,"text_sha256":"d270d32707ccfb3e99b63e835a05bc702f3c7dbdc2d0c3d28be76d75c02c2593","block_sha256":"d270d32707ccfb3e99b63e835a05bc702f3c7dbdc2d0c3d28be76d75c02c2593","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4/#blk_1407b20a-f011-4aea-85e9-03997ad1418f"},{"id":"occ_c8361009cabf8548178152cf","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd","block_id":"blk_35ff6245-bc92-4ae2-a5ba-039e044b3b57","section_id":"sec_88035905-26c5-4090-8170-fa42568690e2","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI AgentとAnalystが、","quote_start":0,"quote_end":18,"text_sha256":"8da76f18d149267f9c897098681c55ea9d66cfb8fd0bce13a379900822722957","block_sha256":"8da76f18d149267f9c897098681c55ea9d66cfb8fd0bce13a379900822722957","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4/#blk_35ff6245-bc92-4ae2-a5ba-039e044b3b57"},{"id":"occ_d368f8e851cc6756bdb1206b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd","block_id":"blk_3a6c7730-0a76-4348-a717-34c44f201863","section_id":"sec_aec07add-61f6-429f-bd5c-ee33d57529b4","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":0,"end":7,"exact":"Agentic","quote":"Agentic SRE","quote_start":0,"quote_end":11,"text_sha256":"9878fa56437a99dbd1c7c0b1c1ed74898f91879d38c5895bf6e313959228403b","block_sha256":"9878fa56437a99dbd1c7c0b1c1ed74898f91879d38c5895bf6e313959228403b","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4/#blk_3a6c7730-0a76-4348-a717-34c44f201863"},{"id":"occ_29d8531f962e43436d325651","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd","block_id":"blk_4251a47e-841d-44c9-839b-a2f5fbd34b94","section_id":"sec_8465328f-25fc-42c0-aeea-d2bfcfa49e7c","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agentがどれほど賢くても、障害を調査するにはEvidenceが必要である。","quote_start":0,"quote_end":42,"text_sha256":"35ed701cee0ea7ebaf37ebf18215890eeba3645698ff60d179409d821e0e3f12","block_sha256":"35ed701cee0ea7ebaf37ebf18215890eeba3645698ff60d179409d821e0e3f12","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4/#blk_4251a47e-841d-44c9-839b-a2f5fbd34b94"},{"id":"occ_0927ffa859454142dc57d386","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd","block_id":"blk_54b16842-1939-448f-91c7-e14e64cb8496","section_id":"sec_4ce0f622-4084-4363-8f3e-9c68832f1fcb","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agentは作るだけなら簡単になっていく。","quote_start":0,"quote_end":24,"text_sha256":"68ca529ca426ceff6363f57234f73baa83f1c6dc21ac6fe782885684a8ac21c2","block_sha256":"68ca529ca426ceff6363f57234f73baa83f1c6dc21ac6fe782885684a8ac21c2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4/#blk_54b16842-1939-448f-91c7-e14e64cb8496"},{"id":"occ_242ff9ae5dcbc0a996318b62","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd","block_id":"blk_55274730-ba83-4f4f-90a6-7e5485c282d6","section_id":"sec_eb8517a8-a674-4cb1-ad75-54b6caf7f550","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agentが操作できる世界が増えるほど、観測しなければならないSignalも増える。","quote_start":0,"quote_end":45,"text_sha256":"12f81ed434373f0a866e45f9c50f30f7b59ed0ef875409775669eb1e0cf50b24","block_sha256":"12f81ed434373f0a866e45f9c50f30f7b59ed0ef875409775669eb1e0cf50b24","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4/#blk_55274730-ba83-4f4f-90a6-7e5485c282d6"},{"id":"occ_9ff1c9b1d32e6aac880d09ef","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd","block_id":"blk_59594adc-0225-49e5-acbd-304b35d4ce5c","section_id":"sec_eb8517a8-a674-4cb1-ad75-54b6caf7f550","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":4,"end":12,"exact":"AI Agent","quote":"これらはAI Agentから見ると、","quote_start":0,"quote_end":18,"text_sha256":"53aa2632077ecacbc8b9baa0997ddad5ae89cee6914734fd461f80cdc0880569","block_sha256":"53aa2632077ecacbc8b9baa0997ddad5ae89cee6914734fd461f80cdc0880569","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4/#blk_59594adc-0225-49e5-acbd-304b35d4ce5c"},{"id":"occ_e346c79df2dbe331bbb07b4b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd","block_id":"blk_66e6c7d7-7199-4f8d-a523-e89f644d4ead","section_id":"sec_9cb0e71e-5f2f-4f15-9119-97bd1346083b","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":3,"end":11,"exact":"AI Agent","quote":"例えばAI Agentが、","quote_start":0,"quote_end":13,"text_sha256":"9a9678d872b175ac933bc21ce74a977ec4bf8a658ee0b0f0e56e119245b8e4ea","block_sha256":"9a9678d872b175ac933bc21ce74a977ec4bf8a658ee0b0f0e56e119245b8e4ea","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4/#blk_66e6c7d7-7199-4f8d-a523-e89f644d4ead"},{"id":"occ_8fd1a87009043b36e32b4b39","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd","block_id":"blk_7cc9106d-8281-4844-9b40-0fe898ed0806","section_id":"sec_3945e3ee-d201-47d0-ab1c-7fe7e5843e32","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":3,"end":10,"exact":"Agentic","quote":"というAgentic SRE Stackを狙っていると見ることができる。","quote_start":0,"quote_end":36,"text_sha256":"39c6a05b05a5aab313e7d7dcef2589162baf1422e6f0bb4fa9234977cc3d40b9","block_sha256":"39c6a05b05a5aab313e7d7dcef2589162baf1422e6f0bb4fa9234977cc3d40b9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4/#blk_7cc9106d-8281-4844-9b40-0fe898ed0806"},{"id":"occ_ea26b2f87182f5118daad36c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd","block_id":"blk_83ed7838-cd00-4e36-8b31-b3ae882fad70","section_id":"sec_16a850da-d754-4476-9cf0-9cd5a60ca50a","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":27,"end":34,"exact":"Agentic","quote":"実際、現在の記事でもDeductive AIを利用したAgentic SREを、このMeta-Function化に近い例として位置づけている。","quote_start":0,"quote_end":71,"text_sha256":"f22594e5da42ddc19de1d47c7f6249d1e595714ce40200e3ef585c413b58578e","block_sha256":"f22594e5da42ddc19de1d47c7f6249d1e595714ce40200e3ef585c413b58578e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4/#blk_83ed7838-cd00-4e36-8b31-b3ae882fad70"},{"id":"occ_d1fb78ce970040adcb1056ee","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd","block_id":"blk_8d53a1d4-4d2c-45b0-8738-ccda53104f58","section_id":"sec_8db3a4a8-8ab4-4b02-91f8-59f73e720853","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":39,"end":47,"exact":"AI Agent","quote":"Deductive AIは、このInvestigative LoopそのものをAI Agentへ移そうとしている。","quote_start":0,"quote_end":57,"text_sha256":"4a7c184405ce7685b5416f9bd957971b4685daabc5937b88899158f7ec9233ab","block_sha256":"4a7c184405ce7685b5416f9bd957971b4685daabc5937b88899158f7ec9233ab","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4/#blk_8d53a1d4-4d2c-45b0-8738-ccda53104f58"},{"id":"occ_cf172916f6d7f46a70781842","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd","block_id":"blk_afd0280a-9670-4b10-a3d4-2238ec694c4f","section_id":"sec_fc38d7f6-5ef5-4966-a7e3-e5e6c6e4875b","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":35,"end":42,"exact":"Agentic","quote":"WorkflowsはDeterministic AutomationとAgentic Reasoningを組み合わせる。","quote_start":0,"quote_end":60,"text_sha256":"d204678a122ec65acc908165f78354298d003fe598973e62573281746a278470","block_sha256":"d204678a122ec65acc908165f78354298d003fe598973e62573281746a278470","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4/#blk_afd0280a-9670-4b10-a3d4-2238ec694c4f"},{"id":"occ_c3bc283d1b21168554a401e4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd","block_id":"blk_b2a2ade8-804e-4533-89f3-591898a8dc64","section_id":"sec_33f9b655-95a8-4e04-a1f1-5b5c28e6531e","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":0,"end":7,"exact":"Agentic","quote":"Agentic SREのTime to Market","quote_start":0,"quote_end":26,"text_sha256":"e375a0c87d9e02a7619d8315deb9fd33ddb0b819bf46c7b3966fcfc1fac4ad14","block_sha256":"e375a0c87d9e02a7619d8315deb9fd33ddb0b819bf46c7b3966fcfc1fac4ad14","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4/#blk_b2a2ade8-804e-4533-89f3-591898a8dc64"},{"id":"occ_264f781e4c3c5a0e730cfcb1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd","block_id":"blk_b7c688a5-601a-40f7-9390-c967f66b6b14","section_id":"sec_fc38d7f6-5ef5-4966-a7e3-e5e6c6e4875b","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":0,"end":7,"exact":"Agentic","quote":"Agentic Investigation","quote_start":0,"quote_end":21,"text_sha256":"8f639dd3b6c2309bb9686c9e97bb8280e6e303b3bb054c493324848e010324d1","block_sha256":"8f639dd3b6c2309bb9686c9e97bb8280e6e303b3bb054c493324848e010324d1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4/#blk_b7c688a5-601a-40f7-9390-c967f66b6b14"},{"id":"occ_e133584641ee743c07e62cc4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd","block_id":"blk_c29757ac-54c1-4880-9236-71aa245a9507","section_id":"sec_7d4ea36d-98a5-4cd0-abd9-2f859f2169d3","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":17,"end":24,"exact":"Agentic","quote":"Deductive AIを利用したAgentic SREなどは、この方向に近い。","quote_start":0,"quote_end":40,"text_sha256":"5393a33d7ba3d88f43a03226a072ce6f3982bb1ba6dd60a4d26777833b0a61c0","block_sha256":"5393a33d7ba3d88f43a03226a072ce6f3982bb1ba6dd60a4d26777833b0a61c0","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4/#blk_c29757ac-54c1-4880-9236-71aa245a9507"},{"id":"occ_26f4f48b8b7d33e6359c03ac","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd","block_id":"blk_c7b7d3cf-d1fe-4d64-8021-868fb1d08824","section_id":"sec_757bf49b-9020-48fe-92d5-cfae65086666","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":39,"end":46,"exact":"Agentic","quote":"ObservabilityにはColumnar + Prometheus + Agentic SREという新しい攻め手が入った。","quote_start":0,"quote_end":64,"text_sha256":"eb934dfa14f9f9ee795df469089954f102c3d95e90aca84573d305756da23f51","block_sha256":"eb934dfa14f9f9ee795df469089954f102c3d95e90aca84573d305756da23f51","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4/#blk_c7b7d3cf-d1fe-4d64-8021-868fb1d08824"},{"id":"occ_9e233bfbe6cbbca1fd8aa2d6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd","block_id":"blk_d2dfb742-efbe-4d0e-a3a1-8f9f67070761","section_id":"sec_33f9b655-95a8-4e04-a1f1-5b5c28e6531e","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":36,"end":43,"exact":"Agentic","quote":"Elasticが自社だけで同じTechnologyを作るより1～2年早くAgentic SREを市場投入できるなら、そのTime Advantageにも価値がある。","quote_start":0,"quote_end":82,"text_sha256":"fb9a869be470603f7e1c0f700454ea01d3c00bae05a319428a5137bed9de5eb1","block_sha256":"fb9a869be470603f7e1c0f700454ea01d3c00bae05a319428a5137bed9de5eb1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4/#blk_d2dfb742-efbe-4d0e-a3a1-8f9f67070761"},{"id":"occ_3add396e0478701f429ec05a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd","block_id":"blk_ec127bf6-cb18-48a6-8f09-ee57d6d0cba8","section_id":"sec_f027c6c3-26b8-4081-9a3d-31486a06a1af","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":11,"end":18,"exact":"Agentic","quote":"Kubernetes Agentic Investigation、","quote_start":0,"quote_end":33,"text_sha256":"b82c2a3bb9571a6567e7d846f21db40cb30befa2528a77b419ec5668e52005ad","block_sha256":"b82c2a3bb9571a6567e7d846f21db40cb30befa2528a77b419ec5668e52005ad","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4/#blk_ec127bf6-cb18-48a6-8f09-ee57d6d0cba8"},{"id":"occ_86a5200b1bcfe646a7b72ff5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd","block_id":"blk_f95caec1-d1c9-4e15-9d45-370f3984a8b6","section_id":"sec_d0c81fe9-bddd-465b-a748-002eb86571e5","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":45,"end":53,"exact":"AI Agent","quote":"Observabilityでは、Logs・Metrics・Tracesを集約してシステムやAI Agentの動作を監視し、障害原因を自動調査する。","quote_start":0,"quote_end":73,"text_sha256":"098b8649a822fb1be29397860c2b73a0076c08379ff2124abd6e6ca4e4dfc43a","block_sha256":"098b8649a822fb1be29397860c2b73a0076c08379ff2124abd6e6ca4e4dfc43a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4/#blk_f95caec1-d1c9-4e15-9d45-370f3984a8b6"},{"id":"occ_dfd903b48b2c1889bf50cc8b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f17ef624-e9fd-49a4-b061-966093172de4","work_id":"wrk_168418fa-2b60-4053-acfe-b86f6898d033","block_id":"blk_0c37d8bc-6757-47a3-bf0c-e5cdc34b193a","section_id":"sec_873fc1e8-7005-4e5b-9e44-bec7f5043b7a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":72,"end":78,"exact":"エージェント","quote":"アプリの登場は**計算需要の質と量の両面**に影響します。一つの高性能モデルを叩くだけでなく、複数モデル・複数エージェントを**動的に組み合わせ**て動かす負荷が増えます。また訓練と推論の境界が曖昧になり（オンライン学習や継続学習の普及）、常時学習を伴うサービスが増えるでしょう。結果として、**データセンターはより多様な","quote_start":17,"quote_end":178,"text_sha256":"ce2ce13a1b91c8294c7e6035a471e94c2ce1f7734ccfa9af9808d3ec9fbfe7f9","block_sha256":"ce2ce13a1b91c8294c7e6035a471e94c2ce1f7734ccfa9af9808d3ec9fbfe7f9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f17ef624-e9fd-49a4-b061-966093172de4/#blk_0c37d8bc-6757-47a3-bf0c-e5cdc34b193a"},{"id":"occ_a8c70bd5a99d3e762acef342","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f17ef624-e9fd-49a4-b061-966093172de4","work_id":"wrk_168418fa-2b60-4053-acfe-b86f6898d033","block_id":"blk_2c9fa3c2-2b48-4da0-81dc-c876e882d20c","section_id":"sec_873fc1e8-7005-4e5b-9e44-bec7f5043b7a","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":185,"end":191,"exact":"エージェント","quote":"ラインサービス上で行われ、YouTubeなどのトラフィックに匹敵する負荷になる可能性もあります。最近話題のAIエージェントゲーム（仮想社会シミュレーション）では、多数のLLMエージェントが同時稼働し、ゲームサーバーの負荷が従来比でケタ違いに増える懸念もあります。これらも総じてデータセンターの役割を拡大するでしょう。","quote_start":130,"quote_end":288,"text_sha256":"d6dc7b27aea2754ef6702cc8e69f8d22985d5d1fef6569f53643285852d3858d","block_sha256":"d6dc7b27aea2754ef6702cc8e69f8d22985d5d1fef6569f53643285852d3858d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f17ef624-e9fd-49a4-b061-966093172de4/#blk_2c9fa3c2-2b48-4da0-81dc-c876e882d20c"},{"id":"occ_bbb3a06acf4a782bb8e7d093","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f17ef624-e9fd-49a4-b061-966093172de4","work_id":"wrk_168418fa-2b60-4053-acfe-b86f6898d033","block_id":"blk_2caa674f-2227-42fa-8bee-79fe04c1eedd","section_id":"sec_873fc1e8-7005-4e5b-9e44-bec7f5043b7a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":146,"end":152,"exact":"エージェント","quote":"が実用化すれば、それをホスティングできるのは世界でも数箇所のAIデータセンターのみとなるかもしれません。また、エージェントAIが社会に浸透すれば、人々のあらゆる行動の裏でGPUが走るような世界になります。それは電力や通信インフラとも不可分で、AI需要がインフラ容量拡張をドライブし、同時に環境負荷軽減技術へのプレッシャーに","quote_start":91,"quote_end":252,"text_sha256":"c6bc694e5e39f7024f7a272bf1e14fd17cdabcb249210c992a6c6c66776706de","block_sha256":"c6bc694e5e39f7024f7a272bf1e14fd17cdabcb249210c992a6c6c66776706de","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f17ef624-e9fd-49a4-b061-966093172de4/#blk_2caa674f-2227-42fa-8bee-79fe04c1eedd"},{"id":"occ_32993da78bda64618935ec0f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f17ef624-e9fd-49a4-b061-966093172de4","work_id":"wrk_168418fa-2b60-4053-acfe-b86f6898d033","block_id":"blk_960e8816-3595-41f3-be52-08881efff671","section_id":"sec_873fc1e8-7005-4e5b-9e44-bec7f5043b7a","layer":"body","character_id":null,"count":6,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":748,"end":754,"exact":"エージェント","quote":"*複数モデルを並列動作させる機構**や**迅速なコンテキスト切替**機能が求められるかもしれません。\n- **エージェント対話・自己強化学習 (Agent-to-Agent):** AutoGPTやSelf-Reflective LLMなどの登場により、**複数のAIエージェントが相互に対話・協調してタスクを遂行**する","quote_start":693,"quote_end":854,"text_sha256":"6529c31a02709a2afc4bf77e793b6584189afcf105fc8d6d119f379614066288","block_sha256":"6529c31a02709a2afc4bf77e793b6584189afcf105fc8d6d119f379614066288","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f17ef624-e9fd-49a4-b061-966093172de4/#blk_960e8816-3595-41f3-be52-08881efff671"},{"id":"occ_8afe3615a363c81e175e6ebb","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f17ef624-e9fd-49a4-b061-966093172de4","work_id":"wrk_168418fa-2b60-4053-acfe-b86f6898d033","block_id":"blk_d8b8361d-96d9-4953-bc4c-a8aed7fd9961","section_id":"sec_b9721051-b457-47bd-9149-6110a35cbcea","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":36,"end":42,"exact":"エージェント","quote":"さらに、**巨大モデルを小型モデル群で置き換える動き**（MoEや後述のエージェント分散）も、一見効率化に思えて新たな負荷を生んでいます。総括すると、生成AIブームはアルゴリズム効率向上の努力を促しつつも、ネット全体の推論・学習トラフィック増大により**計算資源への需要曲線は依然鋭角","quote_start":0,"quote_end":142,"text_sha256":"147b4ed686a6ebc19ad69cf35f5584bd6b9ce93ad21fb49e6dc1840d71b4a554","block_sha256":"147b4ed686a6ebc19ad69cf35f5584bd6b9ce93ad21fb49e6dc1840d71b4a554","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f17ef624-e9fd-49a4-b061-966093172de4/#blk_d8b8361d-96d9-4953-bc4c-a8aed7fd9961"},{"id":"occ_ce16737cb7be64073712dc62","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f3a5d7c9-6dfa-4c38-8bf2-a37063801332","work_id":"wrk_d32481f4-3a5a-40d8-a817-86b115c8c4f4","block_id":"blk_0aa1fd9d-65d2-4cef-a143-24de950d0050","section_id":"sec_be8484e9-3419-494b-b1dc-280c7cdfb30e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":72,"end":78,"exact":"エージェント","quote":"LLMNVIDIA GPUの性能を最大化vLLM導入しやすさとKVメモリー効率SGLangPrefix再利用、エージェント、構造化生成","quote_start":17,"quote_end":84,"text_sha256":"c41bb6be0032d1c039cf2f9e2d7c1f4af8742d6975d204c5f58f56e0cca3b1ef","block_sha256":"c41bb6be0032d1c039cf2f9e2d7c1f4af8742d6975d204c5f58f56e0cca3b1ef","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f3a5d7c9-6dfa-4c38-8bf2-a37063801332/#blk_0aa1fd9d-65d2-4cef-a143-24de950d0050"},{"id":"occ_91735fabae4b275a50785aba","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f3a5d7c9-6dfa-4c38-8bf2-a37063801332","work_id":"wrk_d32481f4-3a5a-40d8-a817-86b115c8c4f4","block_id":"blk_181c242d-5617-4e9e-8a9c-1253d5ef77d3","section_id":"sec_be8484e9-3419-494b-b1dc-280c7cdfb30e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":30,"end":36,"exact":"エージェント","quote":"SGLangは、複数回の生成、RAG、マルチターン会話、AIエージェント、構造化出力を効率化する推論基盤である。","quote_start":0,"quote_end":56,"text_sha256":"e2e795253bc62b884f2d8c9f2771d682db8ba4e7ae44bd28e442b357174edc1e","block_sha256":"e2e795253bc62b884f2d8c9f2771d682db8ba4e7ae44bd28e442b357174edc1e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f3a5d7c9-6dfa-4c38-8bf2-a37063801332/#blk_181c242d-5617-4e9e-8a9c-1253d5ef77d3"},{"id":"occ_2f94c8c8bb97ccfa21119287","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f3a5d7c9-6dfa-4c38-8bf2-a37063801332","work_id":"wrk_d32481f4-3a5a-40d8-a817-86b115c8c4f4","block_id":"blk_9b1cb2b3-0be7-4ea2-a2db-6d9e9a376889","section_id":"sec_0875f100-4397-4df8-8237-7c7ac7bc5c36","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":38,"end":44,"exact":"エージェント","quote":"第三に、Decodeだけへ設備を集中することで、1ユーザー当たりの生成速度やエージェントの処理完了時間を短縮できる。","quote_start":0,"quote_end":58,"text_sha256":"31e8ad4fbba430a215aafc3fc06c96fda0095181f0b3da2c0643d8ad272c7b4d","block_sha256":"31e8ad4fbba430a215aafc3fc06c96fda0095181f0b3da2c0643d8ad272c7b4d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f3a5d7c9-6dfa-4c38-8bf2-a37063801332/#blk_9b1cb2b3-0be7-4ea2-a2db-6d9e9a376889"},{"id":"occ_3047387b0bb9fed2c717db04","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f3a5d7c9-6dfa-4c38-8bf2-a37063801332","work_id":"wrk_d32481f4-3a5a-40d8-a817-86b115c8c4f4","block_id":"blk_c919bcad-a792-4e51-bffd-34c893a6b7bc","section_id":"sec_4fa93586-6946-4576-b2e4-3fc4bbaa28f6","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":83,"end":89,"exact":"エージェント","quote":"担当し、大容量HBMで入力とKVキャッシュを処理する。Cerebrasは、その後の逐次生成を担当し、応答時間、エージェントの反復速度、コード生成の完了時間を短縮する。","quote_start":28,"quote_end":111,"text_sha256":"e4a3af99b30f3ddf3074820c20733255a173c6c2aade399d7b79328a67be0bd8","block_sha256":"e4a3af99b30f3ddf3074820c20733255a173c6c2aade399d7b79328a67be0bd8","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f3a5d7c9-6dfa-4c38-8bf2-a37063801332/#blk_c919bcad-a792-4e51-bffd-34c893a6b7bc"},{"id":"occ_393e8b362641729e88977fc1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f3a5d7c9-6dfa-4c38-8bf2-a37063801332","work_id":"wrk_d32481f4-3a5a-40d8-a817-86b115c8c4f4","block_id":"blk_dbcc4e39-87fb-40e8-bcde-61fa847e4523","section_id":"sec_be8484e9-3419-494b-b1dc-280c7cdfb30e","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":28,"end":34,"exact":"エージェント","quote":"共通部分を毎回Prefillし直さずに済むため、RAG、エージェント、共通プロンプトの多いサービスで効果を発揮する。([arXiv](https://arxiv.org/abs/2312.07104?utm_source=chatgpt.com))","quote_start":0,"quote_end":124,"text_sha256":"392faaf17a5fa31922a6ac1fac47a00793a5bb49c7aa565ea1b1f9cea3046a77","block_sha256":"392faaf17a5fa31922a6ac1fac47a00793a5bb49c7aa565ea1b1f9cea3046a77","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f3a5d7c9-6dfa-4c38-8bf2-a37063801332/#blk_dbcc4e39-87fb-40e8-bcde-61fa847e4523"},{"id":"occ_30cef7c63806f2db97554024","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f45cfd6f-625d-4e49-b987-25f79284501d","work_id":"wrk_28a55046-0758-4fb4-b7b9-5ec8fa7681a7","block_id":"blk_c3f2369f-984a-4ec5-bc0f-22c3e2202620","section_id":"sec_188babf8-526b-4c94-ae27-ecd58f603766","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":2,"end":8,"exact":"エージェント","quote":"AIエージェント、","quote_start":0,"quote_end":9,"text_sha256":"a596d9fe3be90ade757423a114d6a84ba7ed4ab314de46da01b7af3eae81a2e8","block_sha256":"a596d9fe3be90ade757423a114d6a84ba7ed4ab314de46da01b7af3eae81a2e8","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f45cfd6f-625d-4e49-b987-25f79284501d/#blk_c3f2369f-984a-4ec5-bc0f-22c3e2202620"},{"id":"occ_f066c968bb4bf85b29202d0a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_0858de72-2f98-497f-a266-1864311cc370","section_id":"sec_16ed3027-55c8-4b8c-b9c4-e2349a20f4bd","layer":"body","character_id":null,"count":4,"matched_aliases":["AI Agents","AI agent","AI agents"],"evidence":{"text_basis":"markdown","start":552,"end":560,"exact":"AI agent","quote":"nline inspection、OT/IoT segmentation** を重ねる構造です。強みは、社員・AI agent・アプリ・データの通信を**通さない、見逃さない、漏らさない**ことにあります。AI-SPM は AI models / agents / services の 360度可視化をうたい、AI Secu","quote_start":497,"quote_end":660,"text_sha256":"d84f333ae01537b8bbedea21fc5b09bba24d5840dcc6689a80895e432d5cc2a0","block_sha256":"d84f333ae01537b8bbedea21fc5b09bba24d5840dcc6689a80895e432d5cc2a0","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_0858de72-2f98-497f-a266-1864311cc370"},{"id":"occ_1086387e12530f56f9db5a09","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_0f571368-e893-4b70-93e2-c4c97cd67752","section_id":"sec_c6191217-048d-4a92-9328-dbd9aceace84","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":226,"end":232,"exact":"エージェント","quote":"文言でも、**lateral movement を消す、工場や病院の attack surface を下げる、エージェントなしで OT/IoT を分離する**ことを前面に出しています。これは、ロボット・医療機器・産業設備・センサーが増えるフィジカルAI時代にかなり相性が良いです。 ([Zscaler](https://w","quote_start":171,"quote_end":332,"text_sha256":"fa4d59779f03ee3b70b7aff7a08b45f839ee3aba38c0c1769dee3738d80f10e5","block_sha256":"fa4d59779f03ee3b70b7aff7a08b45f839ee3aba38c0c1769dee3738d80f10e5","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_0f571368-e893-4b70-93e2-c4c97cd67752"},{"id":"occ_656d6e2d2d77666b7c4249fe","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_13f48678-69cb-4da8-954e-b2e334db2dd5","section_id":"sec_5a96d32e-ac76-47bc-8d53-93c707709b6c","layer":"body","character_id":null,"count":1,"matched_aliases":["AI agents"],"evidence":{"text_basis":"markdown","start":127,"end":136,"exact":"AI agents","quote":"\n- **AI runtime と Zero Trust の一体性**：**Cloudflare**\n- **AI agents / NHI の統治**：**Okta**","quote_start":72,"quote_end":157,"text_sha256":"9d3440cb0c17b89ee2893b201f70f3ecf9b39769ed0e99018c28cea213093295","block_sha256":"9d3440cb0c17b89ee2893b201f70f3ecf9b39769ed0e99018c28cea213093295","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_13f48678-69cb-4da8-954e-b2e334db2dd5"},{"id":"occ_74cd7800d2cbbd89909d44b1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_4a89eb80-9bdb-40f0-a1d0-f24f65939b37","section_id":"sec_d07f112c-ff5d-4123-b37c-2e55ed02f75d","layer":"body","character_id":null,"count":1,"matched_aliases":["AI agent"],"evidence":{"text_basis":"markdown","start":75,"end":83,"exact":"AI agent","quote":"Mythos は、  \n理屈の上では ZS よりむしろ相性が良い面もあります。  \nCloudflare は AI agent や AIアプリを**動かす側**まで入っているので、AI利用が増えるほどトラフィック、実行、ガードレール需要を取りやすいからです。Reuters も、Cloudflare の売上見通し上振れを AI","quote_start":20,"quote_end":183,"text_sha256":"f77ea8f0011537dd6d601707c6b76e3ebe4a719b3e56b9a8f98d1d660580e721","block_sha256":"f77ea8f0011537dd6d601707c6b76e3ebe4a719b3e56b9a8f98d1d660580e721","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_4a89eb80-9bdb-40f0-a1d0-f24f65939b37"},{"id":"occ_3ebce07b4121be0ceb2571e3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_522665bf-bc7a-4ebe-a845-c3d43c9d1b59","section_id":"sec_d5547d52-4d2f-4044-b521-2a05be9e520a","layer":"body","character_id":null,"count":2,"matched_aliases":["AI agent","エージェント"],"evidence":{"text_basis":"markdown","start":61,"end":69,"exact":"AI agent","quote":"**  \nOktaの役割は、「誰が」「何として」動いているのかを定義するアイデンティティ基盤だ。人間だけでなくAI agent identitiesとnon-human identitiesを同列に管理する方向は、フィジカルAI時代に非常に重い。AI-RANや工場自動化で本当に怖いのは、侵入そのものより、権限の誤付与、放置さ","quote_start":6,"quote_end":169,"text_sha256":"5eb8ddbdd3ad4594b5fdce2734cd873a47f66d4042cf57048b1a9efd8b44def2","block_sha256":"5eb8ddbdd3ad4594b5fdce2734cd873a47f66d4042cf57048b1a9efd8b44def2","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_522665bf-bc7a-4ebe-a845-c3d43c9d1b59"},{"id":"occ_793772215d671881b4a41b5f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_5aac202b-a546-46fb-ba47-fa63cd215de4","section_id":"sec_d804928a-d2fc-4c8a-9708-9a5250cd7da2","layer":"body","character_id":null,"count":1,"matched_aliases":["AI agent"],"evidence":{"text_basis":"markdown","start":204,"end":212,"exact":"AI agent","quote":"mic Workers、Project Think、MCP サーバー群を次々に出しており、Reuters も AI agent の盛り上がりと AI需要による売上見通し上振れを報じています。Zscaler は AI Security を前面に出しつつも、Reuters ベースでは直近Q2売上は **26%増**だった一方、投","quote_start":149,"quote_end":312,"text_sha256":"bcb7a404a6bf5e66e59415cf75628a29949d3cb5076b12bdcf8028cda6884852","block_sha256":"bcb7a404a6bf5e66e59415cf75628a29949d3cb5076b12bdcf8028cda6884852","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_5aac202b-a546-46fb-ba47-fa63cd215de4"},{"id":"occ_dc00f16ec79de4509b34fcaa","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_5f478418-7a47-450f-80b9-7f43aebae473","section_id":"sec_c6191217-048d-4a92-9328-dbd9aceace84","layer":"body","character_id":null,"count":1,"matched_aliases":["AI agent"],"evidence":{"text_basis":"markdown","start":168,"end":176,"exact":"AI agent","quote":"mesh connectivity を提供していますし、MCP や agent runtime を通じて **AI agent が外部サービスやインフラへ触る土台**を作れます。ですが、公開されている製品の押し出し方を見る限り、Cloudflare の主戦場は **“分散接続された agent / app / API / e","quote_start":113,"quote_end":276,"text_sha256":"1f9e00e42fc68be0deab5514b2d1b24159ea04efb00425d05e6dea9aa7d18b52","block_sha256":"1f9e00e42fc68be0deab5514b2d1b24159ea04efb00425d05e6dea9aa7d18b52","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_5f478418-7a47-450f-80b9-7f43aebae473"},{"id":"occ_0bf6340db4faae5d1baa3854","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_5f76041f-83b8-4f99-9743-31cc57a1d1b6","section_id":"sec_c1422757-c0ba-4db1-8a8d-eb9ed6d267ee","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":105,"end":111,"exact":"エージェント","quote":"ル圧縮**です。  \n長期的には、**高度な防御、ゼロトラスト、ID、データ保護、修復自動化、コード/モデル/エージェント監査**の需要増です。Reuters は銀行業界や規制当局が Mythos を非常に重く見ており、銀行・中央銀行・政府機関が対応を急いでいると報じています。これは「脅威が増えるならセキュリティ需要は増","quote_start":50,"quote_end":211,"text_sha256":"d7f71e4ae2cc80eb19300459796cdffc372cb26cd24ab1cfb38a32167dca6288","block_sha256":"d7f71e4ae2cc80eb19300459796cdffc372cb26cd24ab1cfb38a32167dca6288","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_5f76041f-83b8-4f99-9743-31cc57a1d1b6"},{"id":"occ_9ffcf44b1c8d3d4147199936","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_5fd034ca-dfb3-49ef-bfb7-313858ce71f9","section_id":"sec_a97ffdcb-0c86-449c-991f-d12c44d39979","layer":"body","character_id":null,"count":2,"matched_aliases":["AI agents"],"evidence":{"text_basis":"markdown","start":55,"end":64,"exact":"AI agents","quote":"Okta は、**identity という一点においては最重要**です。  \nMythos 時代には、人間よりAI agentsやservice accountsのほうが増えやすくなります。Okta はそれを Non-Human Identities として明確に定義し、AI agents 向けに visibility と g","quote_start":0,"quote_end":164,"text_sha256":"2c94c5bbaf1a73957e93119c3fd68a3416e7912190b446032779afe21d96fb8d","block_sha256":"2c94c5bbaf1a73957e93119c3fd68a3416e7912190b446032779afe21d96fb8d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_5fd034ca-dfb3-49ef-bfb7-313858ce71f9"},{"id":"occ_861315cbbd7baf95a7c99484","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_61335a17-d4b6-40e5-9d8e-e230b7657fe1","section_id":"sec_fde83af6-0614-44e3-8903-5ee12aa21bae","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":30,"end":36,"exact":"エージェント","quote":"人間のトラフィックも、  \n攻撃のトラフィックも、  \nAIエージェントの通信も、  \n全部そこを通るなら、  \nその通過点を押さえている会社はお金を取りやすいです。","quote_start":0,"quote_end":83,"text_sha256":"98b9ce4723e551d2391d885dd2671f67913763a813994d5f4a310fda9ce274c0","block_sha256":"98b9ce4723e551d2391d885dd2671f67913763a813994d5f4a310fda9ce274c0","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_61335a17-d4b6-40e5-9d8e-e230b7657fe1"},{"id":"occ_15b3123e4469104660074690","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_648431ae-2d64-43d4-a341-b3864159b427","section_id":"sec_ab19d224-cdd3-48b0-81ec-e7a8ea44e075","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":39,"end":45,"exact":"エージェント","quote":"するとL3は、  \n**“人間の通信インフラ”** から  \n**“人間＋AIエージェントの通信インフラ”**  \nへ広がります。","quote_start":0,"quote_end":65,"text_sha256":"8a222c687e5cd22682d217d2b9de220b59bb7078ef9f014fe21c37f38c61e8f7","block_sha256":"8a222c687e5cd22682d217d2b9de220b59bb7078ef9f014fe21c37f38c61e8f7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_648431ae-2d64-43d4-a341-b3864159b427"},{"id":"occ_c0a3dfb44e7fdc060b995f74","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_691033e6-6ec2-47ad-a5e2-bfeaaa324e0d","section_id":"sec_1a939ca6-d324-48ca-ab56-4cbc4b72c019","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":12,"end":18,"exact":"エージェント","quote":"結論から言うと、**AIエージェント・工場OS・フィジカルAIの時代に、AI-RAN文脈で先に価値が乗りやすいのは「ロボット本体」よりも、分散推論を支える計算基盤・RANソフト・光/接続・工場の運用OS」側**です。AI-RANは、無線","quote_start":0,"quote_end":118,"text_sha256":"6c20d20bf04b545d7b7d611e94fcd8bd13947056f9f112818114f03838189ee4","block_sha256":"6c20d20bf04b545d7b7d611e94fcd8bd13947056f9f112818114f03838189ee4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_691033e6-6ec2-47ad-a5e2-bfeaaa324e0d"},{"id":"occ_ce9e681a0fc7df1e0162dd4f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_6e419880-ed0d-4591-b093-21c982e9f8ec","section_id":"sec_0b8bd6af-914c-48ab-a6f2-a3bfc1fed306","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":136,"end":143,"exact":"agentic","quote":"Iの時代の企業統治OS」に近づいている**会社です。Cloudflare の未来は、Workers AI や agentic workflow が本当に大きくなった時に、**実行面と防御面の両方を同じ布の上で取れる**ことにあります。Zscaler の未来は、AIとフィジカルAIが企業に広がるほど、**ゼロトラスト・データ","quote_start":81,"quote_end":243,"text_sha256":"b163ebfed94962eb082814f8130b7f1a7cbb4a75968e83c6333bad5ce576c3cd","block_sha256":"b163ebfed94962eb082814f8130b7f1a7cbb4a75968e83c6333bad5ce576c3cd","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_6e419880-ed0d-4591-b093-21c982e9f8ec"},{"id":"occ_c3cb3e81e5c30b403469e403","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_7364a50c-8e6c-489f-975d-162bdefe4f5e","section_id":"sec_d5547d52-4d2f-4044-b521-2a05be9e520a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":122,"end":128,"exact":"エージェント","quote":"dflare Oneではクラウドネイティブなポスト量子SWG/Zero Trust/SASEを前面に出し、AIエージェント向けにはAccessがOAuth認可サーバとして振る舞うmanaged OAuthも打ち出している。つまり、公開ネットワークと私設環境のあいだで、通信路の量子耐性とアプリ接続の認可を同時に握るポジショ","quote_start":67,"quote_end":228,"text_sha256":"273838d1670b9298c55c4406ff615c0554c9f6293a31a0422b2cdbd39e9a4325","block_sha256":"273838d1670b9298c55c4406ff615c0554c9f6293a31a0422b2cdbd39e9a4325","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_7364a50c-8e6c-489f-975d-162bdefe4f5e"},{"id":"occ_fea3ad93fa20499952e2e7d7","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_7850da19-50ca-42d3-965d-84f00b8ab784","section_id":"sec_5126fce9-574e-4e6c-8ffc-fc3d1f43fe55","layer":"body","character_id":null,"count":1,"matched_aliases":["AI agents"],"evidence":{"text_basis":"markdown","start":28,"end":37,"exact":"AI agents","quote":"Cloudflare One は「workforce, AI agents, infrastructure を connect and protect」と明示し、Access は self-hosted / SaaS / non-web apps を identity と ","quote_start":0,"quote_end":137,"text_sha256":"73a1bec3a7abe73f06f2c35e5895842257d1ccde42bf4880c6b4da78be07a7f1","block_sha256":"73a1bec3a7abe73f06f2c35e5895842257d1ccde42bf4880c6b4da78be07a7f1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_7850da19-50ca-42d3-965d-84f00b8ab784"},{"id":"occ_86d3b218458c7c6273f90d7b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_90045b3e-0d7d-4afc-93fb-99e50ba56b1b","section_id":"sec_b90e26e6-6e8c-4d94-beef-9af995a2d60d","layer":"body","character_id":null,"count":1,"matched_aliases":["AI agent"],"evidence":{"text_basis":"markdown","start":56,"end":64,"exact":"AI agent","quote":". **未知脆弱性の発見速度が上がること**\n2. **侵害後の横移動や権限奪取が速くなること**\n3. **AI agent や service account など非人間主体が爆増すること**\n4. **AIアプリそのものが攻撃面になること**","quote_start":1,"quote_end":125,"text_sha256":"12f480d6a51128879ffaed7c2c9ab08bbe5b0a79720946f2ea9034829c7d07ac","block_sha256":"12f480d6a51128879ffaed7c2c9ab08bbe5b0a79720946f2ea9034829c7d07ac","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_90045b3e-0d7d-4afc-93fb-99e50ba56b1b"},{"id":"occ_92e9c48b65a62e1a3d7e3fd6","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_9d231028-9c93-410b-88aa-69af70e59d02","section_id":"sec_951d1ffb-ee2d-4e06-b4d9-4dc86da9ac14","layer":"body","character_id":null,"count":3,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"さらにAIエージェントが増えると、認証の主役は人間だけではなくなる。OktaはAIエージェントを一意のnon-human identityとして扱い、発見・登録・最小権限統制を行うガバナンス層を打ち出している。これは実機管","quote_start":0,"quote_end":111,"text_sha256":"ab7bd093cdbee1dc20cdf8d6fe8f28cd9f83aa3b2b587bc9af2986b906b6a54f","block_sha256":"ab7bd093cdbee1dc20cdf8d6fe8f28cd9f83aa3b2b587bc9af2986b906b6a54f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_9d231028-9c93-410b-88aa-69af70e59d02"},{"id":"occ_1d0e3c45b87775b638d29cea","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_a26b2382-1940-4270-ad37-2fd56becc448","section_id":"sec_d5547d52-4d2f-4044-b521-2a05be9e520a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":126,"end":132,"exact":"エージェント","quote":"現場寄りの総合防衛」だ。OT露出の急増を踏まえた能動防御を強調しつつ、Prisma AIRSでAIアプリやAIエージェントのruntime保護も進めている。SiemensのAI-ready産業用データセンターにNVIDIAと並んで組み込まれている事実は、Palo AltoがフィジカルAIのオンプレ現場で重くなることを示し","quote_start":71,"quote_end":232,"text_sha256":"151cfa6107aac60e1126075445bb5337d7caee1a369fc0ba9beef4953d0f9c64","block_sha256":"151cfa6107aac60e1126075445bb5337d7caee1a369fc0ba9beef4953d0f9c64","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_a26b2382-1940-4270-ad37-2fd56becc448"},{"id":"occ_8f61cf342172eef9e2229226","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_a480cc40-a649-4167-9523-cc7ec619a2ee","section_id":"sec_eae4603a-f127-4b35-b6fd-531c493dfd5f","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"さらに AIエージェント時代になると、  \n人間だけでなく AI 同士の通信も増えます。  \nすると L3 は、**人間の道路**から**人間＋AIの道路**へ変わります。  \nこのため、PQC時代のL3は「量子耐性のある道","quote_start":0,"quote_end":112,"text_sha256":"6699fd8e670447f0075e2420f36ace6f25a3b7dc2384aeaa6f2c0e50ac81e7ae","block_sha256":"6699fd8e670447f0075e2420f36ace6f25a3b7dc2384aeaa6f2c0e50ac81e7ae","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_a480cc40-a649-4167-9523-cc7ec619a2ee"},{"id":"occ_da682c087cacce36d7a32d19","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_a5fea0b2-87c7-4004-8c71-51e20b1ad2b3","section_id":"sec_897c8207-9069-4d16-9c57-f18e87425ded","layer":"body","character_id":null,"count":1,"matched_aliases":["AI agent"],"evidence":{"text_basis":"markdown","start":110,"end":118,"exact":"AI agent","quote":"scaler**、  \n**AIを“動かしながら守る”構造で一番強いのは Cloudflare**、  \n**AI agent と非人間IDの統治で一番強いのは Okta**、  \nです。 ([crowdstrike.com](https://www.crowdstrike.com/en-us/platform/charlo","quote_start":55,"quote_end":218,"text_sha256":"aeebbbd1414e0ab5d950d212882e856509d7e2abe143bdeefab330199c00af71","block_sha256":"aeebbbd1414e0ab5d950d212882e856509d7e2abe143bdeefab330199c00af71","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_a5fea0b2-87c7-4004-8c71-51e20b1ad2b3"},{"id":"occ_71fb308cd83453c0e20900ea","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_b0ab6c83-c106-4900-b89f-cf663064b253","section_id":"sec_1a939ca6-d324-48ca-ab56-4cbc4b72c019","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":11,"end":17,"exact":"エージェント","quote":"強気シナリオでは、AIエージェントが現場へ降りてきて、ロボットや設備が常時ネットワーク側の推論資源を借りるようになります。その場合、**NVIDIA → Nokia/SoftBank/Ericsson → Siemens**の順に大き","quote_start":0,"quote_end":117,"text_sha256":"854e4bf74bfb3b6f982d1934e48b9777961d568c017acc0af74fad3f6cac058a","block_sha256":"854e4bf74bfb3b6f982d1934e48b9777961d568c017acc0af74fad3f6cac058a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_b0ab6c83-c106-4900-b89f-cf663064b253"},{"id":"occ_f71a239b4c45a2e69c2ea622","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_b2f8027d-673f-497a-9921-e8fdb952444d","section_id":"sec_1a939ca6-d324-48ca-ab56-4cbc4b72c019","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":5,"end":11,"exact":"エージェント","quote":"## AIエージェント、工場OS、フィジカルAIの時代にAI-RANの文脈で価値が上がる銘柄考察","quote_start":0,"quote_end":48,"text_sha256":"4417082e8a350cf08b102100ed46c5725c52d742d1a6463650c693477efac838","block_sha256":"4417082e8a350cf08b102100ed46c5725c52d742d1a6463650c693477efac838","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_b2f8027d-673f-497a-9921-e8fdb952444d"},{"id":"occ_68bca3d1af8a078d7aa8a4df","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_b4fc72de-221b-4f00-9140-95836cc48105","section_id":"sec_d804928a-d2fc-4c8a-9708-9a5250cd7da2","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":34,"end":40,"exact":"エージェント","quote":"Zscaler は逆で、**AIをどこで動かすか**より、**社員やエージェントがAIをどう安全に使うか**に寄っています。AI Security と Unified Data Security は、**shadow AI の可視化、ChatGPT や Copilot へのデータ","quote_start":0,"quote_end":140,"text_sha256":"a080e09df36be47819e170137b9bdda888d71012863aad8e491e15299c76fa5a","block_sha256":"a080e09df36be47819e170137b9bdda888d71012863aad8e491e15299c76fa5a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_b4fc72de-221b-4f00-9140-95836cc48105"},{"id":"occ_305a2ee15490af70b4b912f1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_c0420082-44e9-432c-82a8-603a2986f3a0","section_id":"sec_0b17bfbb-4874-457b-8692-c646d4725b33","layer":"body","character_id":null,"count":2,"matched_aliases":["AI agent","AI agents"],"evidence":{"text_basis":"markdown","start":31,"end":39,"exact":"AI agent","quote":"**Cloudflare が向く会社**は、  \nAIアプリやAI agentを自分で作る、マルチモデルを使う、MCPやedge executionを使う、ユーザー体験と実行速度も重視する会社です。Cloudflare は **AI Gateway + Workers AI +","quote_start":0,"quote_end":139,"text_sha256":"6a1cfa4191f650405a2c06a7676ac2975e22bf4a20dd27a950706f411650124d","block_sha256":"6a1cfa4191f650405a2c06a7676ac2975e22bf4a20dd27a950706f411650124d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_c0420082-44e9-432c-82a8-603a2986f3a0"},{"id":"occ_668d39e186d10ce0bc86b5f1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_d8c79718-7099-4cbd-9aaa-807647e61c08","section_id":"sec_dcbd52a8-1df5-4b18-bd42-9ea5e710d39c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":69,"end":75,"exact":"エージェント","quote":"は **NPU**\n- **本体の重い知覚・行動生成** は **GPU**\n- **現場MECの言語/司令/エージェント** は **LPU**\n- **固定化した検査や前処理** は **ASIC/NPU**\n- **学習・シミュレーション・双子化** は **GPU** ([qualcomm.com](https:","quote_start":14,"quote_end":175,"text_sha256":"b778612ca88d28faae97a64f9356c46818877d2bdbf3807d93ce946a7adb2c5f","block_sha256":"b778612ca88d28faae97a64f9356c46818877d2bdbf3807d93ce946a7adb2c5f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_d8c79718-7099-4cbd-9aaa-807647e61c08"},{"id":"occ_8ab0a052259b7dd52df4a607","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_d975a086-a654-41a0-9213-0b892b259f1a","section_id":"sec_59c3afc2-b470-4440-9dea-183c913e44c5","layer":"body","character_id":null,"count":2,"matched_aliases":["AI agents"],"evidence":{"text_basis":"markdown","start":61,"end":70,"exact":"AI agents","quote":"lare は、**製品構造だけ見ると最もユニーク**です。  \nMythos 時代に増えるのは、AI app、AI agents、MCP tools、API-to-model traffic です。Cloudflare は AI Gateway を unified inference layer に拡張し、Workers AI","quote_start":6,"quote_end":170,"text_sha256":"90b59a4736eb76ecf267af699daf0d1e30cc93b8e513d7bb89cce54ea6626eb4","block_sha256":"90b59a4736eb76ecf267af699daf0d1e30cc93b8e513d7bb89cce54ea6626eb4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_d975a086-a654-41a0-9213-0b892b259f1a"},{"id":"occ_581ad7342c98745919656d4a","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_e685e00c-106d-47f1-8ef0-d91e98ec7ca2","section_id":"sec_782f54d6-6ef1-4d93-92e8-9502c9d70ad7","layer":"body","character_id":null,"count":1,"matched_aliases":["agentic"],"evidence":{"text_basis":"markdown","start":125,"end":132,"exact":"agentic","quote":"マルチモデルの observability と control をかけ、Agents SDK と MCP で agentic workflow を組み、Cloudflare One / Access で identity と device posture をかける**、という一体設計を取れます。しかも自社の MCP ser","quote_start":70,"quote_end":232,"text_sha256":"ac1b46c955b46b2b83dd7ee52f58942773a0138d79a56e58a743dadaca4489fe","block_sha256":"ac1b46c955b46b2b83dd7ee52f58942773a0138d79a56e58a743dadaca4489fe","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_e685e00c-106d-47f1-8ef0-d91e98ec7ca2"},{"id":"occ_7add1f1376169ab3f751a2a5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_ebc4be38-4af3-4d4b-8d2e-a7e0701dc4b2","section_id":"sec_5126fce9-574e-4e6c-8ffc-fc3d1f43fe55","layer":"body","character_id":null,"count":1,"matched_aliases":["AI agents"],"evidence":{"text_basis":"markdown","start":245,"end":254,"exact":"AI agents","quote":"離”が強い**、というのがいちばん実態に近いです。Cloudflare は Cloudflare One を「AI agents まで守る SASE」と位置づけ、Workers AI / Agents / MCP を同じ面で押し出しています。Zscaler は Zero Trust Exchange、AI Security、A","quote_start":190,"quote_end":354,"text_sha256":"36a81fbd2df95bc183589d3bbafecf84adfb78c2634ecd80e02269c749363c2c","block_sha256":"36a81fbd2df95bc183589d3bbafecf84adfb78c2634ecd80e02269c749363c2c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_ebc4be38-4af3-4d4b-8d2e-a7e0701dc4b2"},{"id":"occ_51c421326f8d84a5af4fba0c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd","block_id":"blk_ecb70b36-d4ab-4ccd-bb07-e1330555d9c4","section_id":"sec_ab19d224-cdd3-48b0-81ec-e7a8ea44e075","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":33,"end":39,"exact":"エージェント","quote":"攻撃が増えれば、通過するトラフィックも増えます。  \nさらに、AIエージェント時代になると、","quote_start":0,"quote_end":46,"text_sha256":"c4350f325f92ec92f5575ed2e02a39ccafb919969bbbfe88d58a47ef2037e148","block_sha256":"c4350f325f92ec92f5575ed2e02a39ccafb919969bbbfe88d58a47ef2037e148","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f487aa08-2568-42ad-85cb-11c258b5b71d/#blk_ecb70b36-d4ab-4ccd-bb07-e1330555d9c4"},{"id":"occ_3c09f7e6e9ab4014f7317647","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":null,"section_id":null,"layer":"title","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"title","start":0,"end":8,"exact":"AI Agent","quote":"AI Agent時代のセキュリティ地図","quote_start":0,"quote_end":19,"text_sha256":"cf64cddd241906a99e2ec1ba5e68a5d75af8c044c5eb449dc22ec77a11c8e4be","block_sha256":null,"offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/"},{"id":"occ_3c2e811f54184c4bfea8934e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_0321228b-1641-4ba4-b793-784d5cbc2f5e","section_id":"sec_be062351-0e6f-4d02-99fa-315a5794381c","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":38,"end":45,"exact":"Agentic","quote":"ObservabilityではLogs、Metrics、Tracesに加えてAgentic investigationへ進んでいる。","quote_start":0,"quote_end":66,"text_sha256":"473dff836ce49c69e4d508aad09e2d65f44b612464cf727dd6046fac66ad2af6","block_sha256":"473dff836ce49c69e4d508aad09e2d65f44b612464cf727dd6046fac66ad2af6","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_0321228b-1641-4ba4-b793-784d5cbc2f5e"},{"id":"occ_6e9805971f2266ea7365848e","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_1667f1d7-1a13-4f4c-bc2d-ead925697b02","section_id":"sec_8fec3259-5a22-44a7-be31-759d58bcb8fa","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":3,"end":11,"exact":"AI Agent","quote":"28．AI Agent Identityはまだ数字には小さい","quote_start":0,"quote_end":30,"text_sha256":"d172ed29c414603d8dc2ad92c0dddb392a58df8a97e6ee98c761f67f3582a7c4","block_sha256":"d172ed29c414603d8dc2ad92c0dddb392a58df8a97e6ee98c761f67f3582a7c4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_1667f1d7-1a13-4f4c-bc2d-ead925697b02"},{"id":"occ_9053a00fcf969f0d18121cf7","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_272d4343-eb09-4754-94d2-014a2f9da068","section_id":"sec_c0c44040-4a02-43d8-a0c2-4c6085e61de5","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agent時代には、","quote_start":0,"quote_end":13,"text_sha256":"4318cd8d2c61591d41951614dee5cacaed009b63952b65abc729e0e78adb26a7","block_sha256":"4318cd8d2c61591d41951614dee5cacaed009b63952b65abc729e0e78adb26a7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_272d4343-eb09-4754-94d2-014a2f9da068"},{"id":"occ_0e32fb552776503681a783d5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_2a17a990-73ea-43af-bb09-a6d3a7da2b96","section_id":"sec_1665d02d-d220-45cd-b09c-f43e416b36aa","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":71,"end":79,"exact":"AI Agent","quote":"seのLogs、Documents、Metrics、Security Eventを一つのContextへ集め、AI AgentのMemory Layerへ進んでいる。","quote_start":16,"quote_end":99,"text_sha256":"d9c1a917558d459b6f4e6e5e8b6aea1197e5e689e0c9d1d62121692d1b79910e","block_sha256":"d9c1a917558d459b6f4e6e5e8b6aea1197e5e689e0c9d1d62121692d1b79910e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_2a17a990-73ea-43af-bb09-a6d3a7da2b96"},{"id":"occ_bc80cd0daddb82970ce865f7","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_2bbaeb8f-637f-4ef4-8469-042574ca985c","section_id":"sec_179ed553-2c01-43b8-bf49-1527dd8c1af9","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":8,"end":16,"exact":"AI Agent","quote":"収益化の早さは、AI Agent Securityや新Control Pointが「将来のTAM」ではなく、既にARR、Bookings、Consumption、Upsellとして数字へ出始めているかを見る。","quote_start":0,"quote_end":104,"text_sha256":"cd4a6788cf64711a0ba034fce263bf084d4c135f6babb85627ff7f56c5f210d9","block_sha256":"cd4a6788cf64711a0ba034fce263bf084d4c135f6babb85627ff7f56c5f210d9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_2bbaeb8f-637f-4ef4-8469-042574ca985c"},{"id":"occ_cd74394e379214986790e5f4","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_2cd311c8-596b-4715-84b7-4409837adb1f","section_id":"sec_1f2b9270-1c9b-4d26-86f6-d8102d816a7e","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agentにとってこれは極めて価値が高い。","quote_start":0,"quote_end":24,"text_sha256":"c0ea090afce6505f9f96e1f705dc6c7fabe25e4a21bd50969ea3bd0851e23ddc","block_sha256":"c0ea090afce6505f9f96e1f705dc6c7fabe25e4a21bd50969ea3bd0851e23ddc","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_2cd311c8-596b-4715-84b7-4409837adb1f"},{"id":"occ_04e16713a92c6be0acc2ae3f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_2d22abe9-37cc-460d-ae34-9e33de028227","section_id":"sec_99b869d9-de26-44da-ab83-44ee11328d8b","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agent時代のSecurity市場そのものには極めて大きな成長余地がある。","quote_start":0,"quote_end":41,"text_sha256":"1fd91256e63bc029449bbf327e12ef2a9cc68d71822878de1366d22665391701","block_sha256":"1fd91256e63bc029449bbf327e12ef2a9cc68d71822878de1366d22665391701","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_2d22abe9-37cc-460d-ae34-9e33de028227"},{"id":"occ_0a2c21caa879af5276c893d5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_2ff12dfa-a958-48b5-97bb-33954f0e09f4","section_id":"sec_b637faa0-05da-42d9-858f-d36964271fd0","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":3,"end":10,"exact":"Agentic","quote":"23．Agentic Internet","quote_start":0,"quote_end":19,"text_sha256":"b713ad869264b1db35b6e2f97a5489e68e81ceeca6da5a21e24a7444a4480401","block_sha256":"b713ad869264b1db35b6e2f97a5489e68e81ceeca6da5a21e24a7444a4480401","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_2ff12dfa-a958-48b5-97bb-33954f0e09f4"},{"id":"occ_df1641ce1d669af1edc5281f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_3c2c21f2-7924-40ca-9df0-ae1f40c466a6","section_id":"sec_1665d02d-d220-45cd-b09c-f43e416b36aa","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agent Securityの本質は、最終的にはこの因果関係をMachine Speedで理解し、","quote_start":0,"quote_end":52,"text_sha256":"ba7225b057d4172e320a85afc913dc58a306525c9813556b4c3f9587df15ad77","block_sha256":"ba7225b057d4172e320a85afc913dc58a306525c9813556b4c3f9587df15ad77","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_3c2c21f2-7924-40ca-9df0-ae1f40c466a6"},{"id":"occ_95aff9849d0fe35f05c24e89","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_3eb50e32-5cb9-4902-91f9-61ff28911682","section_id":"sec_83251f62-543f-4273-ae56-6f7950346f39","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":5,"end":13,"exact":"AI Agent","quote":"最も安価なAI Agent Context optionはESTC。","quote_start":0,"quote_end":34,"text_sha256":"fd194bd5e70289983f3b0042c36bf9a2a74344ed0f30d4403a8241c279c06e17","block_sha256":"fd194bd5e70289983f3b0042c36bf9a2a74344ed0f30d4403a8241c279c06e17","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_3eb50e32-5cb9-4902-91f9-61ff28911682"},{"id":"occ_dda14b4ef8d99314ee806414","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_48b737f8-546c-4921-bcea-cc4e11a8773e","section_id":"sec_a2f3d9bd-ece3-43aa-b63d-4493f9bdc80a","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":4,"end":11,"exact":"Agentic","quote":"NETはAgentic Internet。","quote_start":0,"quote_end":21,"text_sha256":"ec3de2f22c8989f9a895afb5c4e690574390ee10301d6d2ea850737ac6d41b7c","block_sha256":"ec3de2f22c8989f9a895afb5c4e690574390ee10301d6d2ea850737ac6d41b7c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_48b737f8-546c-4921-bcea-cc4e11a8773e"},{"id":"occ_293c991f1016f087bbd8cb04","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_4c94c7c1-5cf1-4d7c-8087-7867159dead1","section_id":"sec_a48fc354-9743-473b-9524-90aecbdc3c13","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agent時代、このPositionが想像以上に重要になっている。","quote_start":0,"quote_end":36,"text_sha256":"1de4a1cb3330a0325a941fbbf27af80bd396e69dbf1759ba4995891db48e982a","block_sha256":"1de4a1cb3330a0325a941fbbf27af80bd396e69dbf1759ba4995891db48e982a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_4c94c7c1-5cf1-4d7c-8087-7867159dead1"},{"id":"occ_af413f415ada3ebc757433d1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_4d79c2fe-3eac-45fb-b2cc-13d601b1f087","section_id":"sec_8adc1ea0-0380-4d1d-a6db-5b762dac15f2","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agentが増えるほどEnterprise Contextは必要になる。","quote_start":0,"quote_end":39,"text_sha256":"3d4f399791c158ade7eaa391d3902de8b3ad4702bc263d6a450a6bd70be624a4","block_sha256":"3d4f399791c158ade7eaa391d3902de8b3ad4702bc263d6a450a6bd70be624a4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_4d79c2fe-3eac-45fb-b2cc-13d601b1f087"},{"id":"occ_5a453586681717669b0007e5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_50c2f10b-9d78-42c7-a816-680b468085cb","section_id":"sec_40320c4d-4ae9-408e-9853-74401681f990","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI AgentのExecutionとInternet Trafficを同時に握れる企業は少ない。","quote_start":0,"quote_end":49,"text_sha256":"64acb637768043811591fb448c3560e702c93ac8386329314234bd3eb03905c9","block_sha256":"64acb637768043811591fb448c3560e702c93ac8386329314234bd3eb03905c9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_50c2f10b-9d78-42c7-a816-680b468085cb"},{"id":"occ_659b52a4204a18bb370e6626","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_5751b1cd-84be-46ec-b581-db55a1ffeaf4","section_id":"sec_2d353f06-af34-4681-84a6-d19957f9689a","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":55,"end":62,"exact":"Agentic","quote":"一方、Zero Trust Exchange for Agents、Endpoint AI Security、Agentic SecOpsの本格寄与はFY27後半以降。","quote_start":0,"quote_end":84,"text_sha256":"5c40d1bf2574c33c76687a3e91ee78b065744a0e29d755ce01cfb38385e45b4c","block_sha256":"5c40d1bf2574c33c76687a3e91ee78b065744a0e29d755ce01cfb38385e45b4c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_5751b1cd-84be-46ec-b581-db55a1ffeaf4"},{"id":"occ_6114d290925007f0f8240b7f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_737aa00c-2741-49fe-b2c1-e2a574725edd","section_id":"sec_dca62c70-5950-40e5-ac41-ac960624158c","layer":"caption","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":2,"end":10,"exact":"AI Agent","quote":"![AI Agent時代のセキュリティ地図 / 表紙](media/ff2c410b20c8b773bc3996a0bd3eba391db42d377dc9ddf1060e6b697452fc24.png)","quote_start":0,"quote_end":103,"text_sha256":"fa3862addb21cc909bffeac436c31b27bcdf35684d5f23a23f92914ecc985327","block_sha256":"fa3862addb21cc909bffeac436c31b27bcdf35684d5f23a23f92914ecc985327","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_737aa00c-2741-49fe-b2c1-e2a574725edd"},{"id":"occ_8ccd2886257ad8bee752abb1","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_739af338-540d-460b-91f5-36544c1dd20b","section_id":"sec_1665d02d-d220-45cd-b09c-f43e416b36aa","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agentが企業の内部で安全に行動するための「制御層」を誰が握るのか。","quote_start":0,"quote_end":38,"text_sha256":"42fabf914fee622daf79fa2888174c3fb927dc79beca23f87df984447f3acb4a","block_sha256":"42fabf914fee622daf79fa2888174c3fb927dc79beca23f87df984447f3acb4a","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_739af338-540d-460b-91f5-36544c1dd20b"},{"id":"occ_852a996f917bbf28baf791b2","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_75f0950c-0e72-45d6-8cc3-05f8eaef0fca","section_id":"sec_8fec3259-5a22-44a7-be31-759d58bcb8fa","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":57,"end":65,"exact":"AI Agent","quote":" Callでは新製品がBookingsの約30%を占め、新製品を含む案件の平均ACV upliftは約40%。AI Agent関連では数十件のDealがあり、複数の$1M+ Dealも出始めている。一方ManagementはAI関連売上がまだ全社数字に現れるほど大きくないことも明確にしている。(Earnings Whispe","quote_start":2,"quote_end":165,"text_sha256":"7d703b8acbbed22f20c44bc9917fe288e674d0a1314fe2d9236ca06a4533545f","block_sha256":"7d703b8acbbed22f20c44bc9917fe288e674d0a1314fe2d9236ca06a4533545f","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_75f0950c-0e72-45d6-8cc3-05f8eaef0fca"},{"id":"occ_af701fb0603ac97834146181","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_7707ef2c-8774-4738-b115-51539b5138a9","section_id":"sec_179ed553-2c01-43b8-bf49-1527dd8c1af9","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":29,"end":37,"exact":"AI Agent","quote":"2027～2030年の伸びしろは株価上昇率の予想ではない。AI Agent、Physical AI、Autonomous SOC、Agent Runtimeなどによって、その企業が取り得るBusiness TAMが現在よりどれほど拡大できるかを見る。","quote_start":0,"quote_end":125,"text_sha256":"f6b0ee244d4402e058e15f2d55da3afa935c3cee5e7e27b58fc712baa9168164","block_sha256":"f6b0ee244d4402e058e15f2d55da3afa935c3cee5e7e27b58fc712baa9168164","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_7707ef2c-8774-4738-b115-51539b5138a9"},{"id":"occ_96faa0e4028255cbee038b21","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_77f50670-614d-4c46-9e89-5059ead9cfd8","section_id":"sec_09e30fef-1e77-4496-bf2e-5133524ebb86","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI AgentがUserに代わって別Applicationへアクセスする","quote_start":0,"quote_end":37,"text_sha256":"4e5387a8d4ef5119fc36d0666c8b4546cf6ac7176bebaf5502013e58880de1b9","block_sha256":"4e5387a8d4ef5119fc36d0666c8b4546cf6ac7176bebaf5502013e58880de1b9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_77f50670-614d-4c46-9e89-5059ead9cfd8"},{"id":"occ_7a0a1e3b9a2fa4bc8b61ad31","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_7f44765c-4cfc-485a-92ed-84c66b8e9a99","section_id":"sec_9143de2a-c6b3-443f-8212-6a04330f7e89","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agent時代のセキュリティ地図","quote_start":0,"quote_end":19,"text_sha256":"cf64cddd241906a99e2ec1ba5e68a5d75af8c044c5eb449dc22ec77a11c8e4be","block_sha256":"cf64cddd241906a99e2ec1ba5e68a5d75af8c044c5eb449dc22ec77a11c8e4be","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_7f44765c-4cfc-485a-92ed-84c66b8e9a99"},{"id":"occ_bf04dcb7a0f002f2ad65784f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_81e174e2-cb73-4b54-b795-12feb2533405","section_id":"sec_f60e347a-80d6-4c49-9d60-063ec0a84f7e","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":3,"end":11,"exact":"AI Agent","quote":"48．AI Agent時代の理想的Security Loop","quote_start":0,"quote_end":30,"text_sha256":"1cd8dcfc4760c1b58d4aa7c69303f52c5fc969898745c04b7ab51805329900c1","block_sha256":"1cd8dcfc4760c1b58d4aa7c69303f52c5fc969898745c04b7ab51805329900c1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_81e174e2-cb73-4b54-b795-12feb2533405"},{"id":"occ_6a0b659f187212b698d43c02","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_8b563b47-d898-44c1-9ffe-e62baf3f0db5","section_id":"sec_c88d88cf-a16e-46d0-814a-0cf63994d5f3","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agent時代との相性が良いのはこのためである。","quote_start":0,"quote_end":27,"text_sha256":"2c78a8896f331b7c33b27848162d4a4c6544276a3bcffda40c998998f33c217c","block_sha256":"2c78a8896f331b7c33b27848162d4a4c6544276a3bcffda40c998998f33c217c","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_8b563b47-d898-44c1-9ffe-e62baf3f0db5"},{"id":"occ_9939615a2041df715d6a938b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_8ec065fa-6a50-4e87-93ce-ace8feeb42d6","section_id":"sec_25eb03a5-be9f-443c-852f-be34f88a0839","layer":"body","character_id":null,"count":3,"matched_aliases":["AI Agent","Agentic"],"evidence":{"text_basis":"markdown","start":42,"end":50,"exact":"AI Agent","quote":"CrowdStrikeはEndpointからIdentity、SIEM、Cloud、AI Agent Securityへ進み、Palo Alto NetworksはFirewallからEndpoint、SOC、Identity、Observabilityへ広がる。ZscalerはZero Trust","quote_start":0,"quote_end":150,"text_sha256":"8d76b26146b4bd902d121567065fe641634f373576f8ea7a6a3929b2719ef075","block_sha256":"8d76b26146b4bd902d121567065fe641634f373576f8ea7a6a3929b2719ef075","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_8ec065fa-6a50-4e87-93ce-ace8feeb42d6"},{"id":"occ_8d18f0ffb8078d371d1006d7","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_98bda91c-8360-4cd7-bf56-6a57f20e2b15","section_id":"sec_d653f570-d28e-4abc-92a2-343a9b9e672c","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":2,"end":10,"exact":"AI Agent","quote":"特にAI Agentでは「本人に許されているから、そのAgentにも全部許す」という構造は危険である。","quote_start":0,"quote_end":51,"text_sha256":"28404cd69149aad56cc7064ae01f2e2591f72b41646b4f8ad45a0a3c4ed9cc98","block_sha256":"28404cd69149aad56cc7064ae01f2e2591f72b41646b4f8ad45a0a3c4ed9cc98","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_98bda91c-8360-4cd7-bf56-6a57f20e2b15"},{"id":"occ_48b9b2320befd35952f75e83","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_a1d74f14-83f0-4823-9d53-f865c9f3427c","section_id":"sec_73b70c96-ec2d-47be-b01c-514409499e18","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":2,"end":10,"exact":"AI Agent","quote":"今後AI Agent数がHuman数を大幅に上回る世界を考える。","quote_start":0,"quote_end":32,"text_sha256":"a54cb5d06d612c67d7595d2c7d5324875e2a5b20f4c2b0ca2ccaf51c83a1349e","block_sha256":"a54cb5d06d612c67d7595d2c7d5324875e2a5b20f4c2b0ca2ccaf51c83a1349e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_a1d74f14-83f0-4823-9d53-f865c9f3427c"},{"id":"occ_8d8d03c54c7e40005fde7eb3","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_a2885f57-4638-4675-8743-54896adbf43a","section_id":"sec_5e0aa1ea-b5f5-481c-a307-cf5ee6be5ef3","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agent時代には、","quote_start":0,"quote_end":13,"text_sha256":"4318cd8d2c61591d41951614dee5cacaed009b63952b65abc729e0e78adb26a7","block_sha256":"4318cd8d2c61591d41951614dee5cacaed009b63952b65abc729e0e78adb26a7","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_a2885f57-4638-4675-8743-54896adbf43a"},{"id":"occ_af082f28165196aee21fa808","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_a3515004-1ca4-4b09-af2d-6227229dbdca","section_id":"sec_2c2a5ade-9349-4480-9f4d-f259666d19fa","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":3,"end":11,"exact":"AI Agent","quote":"というAI Agent時代に必要なControl Pointがある。","quote_start":0,"quote_end":34,"text_sha256":"626f7cb5937bb7901c4857b929f5d8647e7a964720b6cf24529b00734530afc5","block_sha256":"626f7cb5937bb7901c4857b929f5d8647e7a964720b6cf24529b00734530afc5","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_a3515004-1ca4-4b09-af2d-6227229dbdca"},{"id":"occ_5ddaf6051e421755cd1e6f60","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_afd63e25-cb59-4922-9465-8dea95619cae","section_id":"sec_1665d02d-d220-45cd-b09c-f43e416b36aa","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":3,"end":11,"exact":"AI Agent","quote":"しかしAI Agent時代には、この分類だけでは足りない。","quote_start":0,"quote_end":29,"text_sha256":"8b24e6eb42a1a72e36a1f69b326ef3725add699916f234addd076f23f16a5810","block_sha256":"8b24e6eb42a1a72e36a1f69b326ef3725add699916f234addd076f23f16a5810","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_afd63e25-cb59-4922-9465-8dea95619cae"},{"id":"occ_4b544798d2658405d89bf220","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_afe1c24b-12a9-4453-be13-80e29570a6ea","section_id":"sec_5d79c207-85e5-4ce7-9496-76e73f4833a9","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":68,"end":75,"exact":"Agentic","quote":"tive AIを買収し、過去・現在のInvestigationから学びながらRoot Causeを自動解析するAgentic SREを作ろうとしている。","quote_start":13,"quote_end":89,"text_sha256":"a31b7bce492161452de522900f89cac644f4174228a375e247804f1bd1cda8a4","block_sha256":"a31b7bce492161452de522900f89cac644f4174228a375e247804f1bd1cda8a4","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_afe1c24b-12a9-4453-be13-80e29570a6ea"},{"id":"occ_3451e8ac5ad091fec8281b9f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_b28e256c-c58f-4fec-a583-21bee6bb6bcc","section_id":"sec_136d90b2-a1f9-4a2a-a35c-73897c5c2ceb","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":5,"end":13,"exact":"AI Agent","quote":"それでも「AI Agent時代のSecurity OS」に最も近い企業の一つである。","quote_start":0,"quote_end":42,"text_sha256":"2fa2cc8a55fb9dd670fd6a0c43415775c665fd7089dd12ad8dba831067274755","block_sha256":"2fa2cc8a55fb9dd670fd6a0c43415775c665fd7089dd12ad8dba831067274755","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_b28e256c-c58f-4fec-a583-21bee6bb6bcc"},{"id":"occ_f7c9800f3780c8a79c4502d7","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_b8d6788d-f13c-4b80-8097-e75d90c57173","section_id":"sec_1665d02d-d220-45cd-b09c-f43e416b36aa","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":4,"end":12,"exact":"AI Agent","quote":"結論――AI Agent時代、Security企業を見る軸そのものを変える","quote_start":0,"quote_end":37,"text_sha256":"74dd0e95c42208dac2ed1d9dba270824a20a5006b387f2d7e39388b2a5fe95a9","block_sha256":"74dd0e95c42208dac2ed1d9dba270824a20a5006b387f2d7e39388b2a5fe95a9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_b8d6788d-f13c-4b80-8097-e75d90c57173"},{"id":"occ_531a8c37fc6c3c9d28c5a9b5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_bee071da-c1ef-44fb-bc84-f30b2b2c2b3e","section_id":"sec_6e480afb-5c53-4fa9-983c-c0dc3cc1f95d","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":16,"end":23,"exact":"Agentic","quote":"Security for AI、Agentic SOC、Zero Trust Branch、Consumption modelという再加速候補が複数存在する。","quote_start":0,"quote_end":79,"text_sha256":"078f00745f9d57c4ba65c8d18830b50fd9f1b0a1494a986548521a794b506d46","block_sha256":"078f00745f9d57c4ba65c8d18830b50fd9f1b0a1494a986548521a794b506d46","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_bee071da-c1ef-44fb-bc84-f30b2b2c2b3e"},{"id":"occ_eda00bc4d0ef1630932e139b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_bf56446c-5c35-4db1-b897-e55af96662df","section_id":"sec_7231ac10-c0e7-450a-b397-cf03bda9606a","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":5,"end":12,"exact":"Agentic","quote":"したがってAgentic SecurityやAI SOCが、","quote_start":0,"quote_end":30,"text_sha256":"4b3743f868fb17d64c18c23bbe346b6707ac82ba0aebdbd64968b1236cef3a61","block_sha256":"4b3743f868fb17d64c18c23bbe346b6707ac82ba0aebdbd64968b1236cef3a61","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_bf56446c-5c35-4db1-b897-e55af96662df"},{"id":"occ_beb5167c9b41c5daa811ca93","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_cd1dfc68-a150-4075-a78a-fece212cb34e","section_id":"sec_52528bb0-1cbd-4991-a4bc-6d5e112e6a0b","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI AgentがEndpointでActionする限り、","quote_start":0,"quote_end":29,"text_sha256":"beb241037d2dd8b3b3870951b0bb4ca43cf3f5aa49ce721c07c9be86eef460cf","block_sha256":"beb241037d2dd8b3b3870951b0bb4ca43cf3f5aa49ce721c07c9be86eef460cf","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_cd1dfc68-a150-4075-a78a-fece212cb34e"},{"id":"occ_faf5238dce795fcbc5a151e5","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_e47e539e-17e7-443a-a6d0-5b70ed2ca9ab","section_id":"sec_a2f3d9bd-ece3-43aa-b63d-4493f9bdc80a","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":5,"end":12,"exact":"Agentic","quote":"DDOGはAgentic Production。","quote_start":0,"quote_end":24,"text_sha256":"7ed910f20e8aec10fa928a1428b093bfa6637335d213b9e25a8afd935f656557","block_sha256":"7ed910f20e8aec10fa928a1428b093bfa6637335d213b9e25a8afd935f656557","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_e47e539e-17e7-443a-a6d0-5b70ed2ca9ab"},{"id":"occ_7e3b379d55f6115873885f3c","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_ebc05fb9-710f-4a64-835a-0e39d9c6fcad","section_id":"sec_6e1d781c-d3c6-4296-a768-52834a764fb7","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agentは最終的に何かを実行する。","quote_start":0,"quote_end":21,"text_sha256":"6bab50d092eb912569f97ccb573e447b958d6461837cb1bbcb9eef2fce937cdf","block_sha256":"6bab50d092eb912569f97ccb573e447b958d6461837cb1bbcb9eef2fce937cdf","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_ebc05fb9-710f-4a64-835a-0e39d9c6fcad"},{"id":"occ_d07923dbe54ff54b0f8060c8","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_eec8453d-f504-4e77-828c-607d5fb97a52","section_id":"sec_6e1d781c-d3c6-4296-a768-52834a764fb7","layer":"body","character_id":null,"count":2,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":49,"end":57,"exact":"AI Agent","quote":"CrowdStrikeは2026年9月にFalcon Guardianを発表し、EndpointをAI Agent SecurityのControl PointとしてAIDRをRuntime Enforcementまで広げた。CrowdStrikeはFalconが数億Device規模へ展開されているとしており、こ","quote_start":0,"quote_end":157,"text_sha256":"e369a34934d8667517bfc10d70ae2984f7567c311adafc53a87b4cdd4b867865","block_sha256":"e369a34934d8667517bfc10d70ae2984f7567c311adafc53a87b4cdd4b867865","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_eec8453d-f504-4e77-828c-607d5fb97a52"},{"id":"occ_87a6c59df348df39476e6c94","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_f34d35f5-b97c-440a-8351-7557a29fe61c","section_id":"sec_a2f3d9bd-ece3-43aa-b63d-4493f9bdc80a","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":4,"end":12,"exact":"AI Agent","quote":"どちらもAI Agent普及の構造的Beneficiaryになり得る。","quote_start":0,"quote_end":35,"text_sha256":"214dc679386032e983eb85d4e46d88a6f3902e6c648beb1ff8ab9d000fc96dc5","block_sha256":"214dc679386032e983eb85d4e46d88a6f3902e6c648beb1ff8ab9d000fc96dc5","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_f34d35f5-b97c-440a-8351-7557a29fe61c"},{"id":"occ_718217e616e052359d1b0c4f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a","block_id":"blk_f6b5d443-08af-4263-99c9-2155a4e48b5f","section_id":"sec_80ef111a-87f8-49d3-bda1-4c5e1a7a3b1c","layer":"body","character_id":null,"count":1,"matched_aliases":["AI Agent"],"evidence":{"text_basis":"markdown","start":0,"end":8,"exact":"AI Agent","quote":"AI Agent関連Dealは増えているものの、Management自身、全社Revenueへまだ目立って現れる規模ではないと説明している。","quote_start":0,"quote_end":70,"text_sha256":"40ca719ea66954a7b00632e23c6de04c8b4c19ea6e4858b46943f08e82425d1d","block_sha256":"40ca719ea66954a7b00632e23c6de04c8b4c19ea6e4858b46943f08e82425d1d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604/#blk_f6b5d443-08af-4263-99c9-2155a4e48b5f"},{"id":"occ_a48a78a42512d232fbc7fcc2","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f76185ef-dd8b-44e4-a82a-7b8cc8225b9a","work_id":"wrk_72f5f141-da8e-4f5b-a337-f46180c37f2e","block_id":"blk_00959d86-6ea2-4299-b2ae-3690d938c987","section_id":"sec_ae6ae385-cd5e-4d97-aa0a-e5ca233f04bf","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":3,"end":10,"exact":"Agentic","quote":"さらにAgentic AIではLatencyの価値が上がる。","quote_start":0,"quote_end":30,"text_sha256":"1de2591d9ab95c68c47fcc31c0f7ed69cd98eaf2f9d7c86e31128dc7b988dadc","block_sha256":"1de2591d9ab95c68c47fcc31c0f7ed69cd98eaf2f9d7c86e31128dc7b988dadc","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f76185ef-dd8b-44e4-a82a-7b8cc8225b9a/#blk_00959d86-6ea2-4299-b2ae-3690d938c987"},{"id":"occ_0ba6aec74216c7d3574e5552","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_f76185ef-dd8b-44e4-a82a-7b8cc8225b9a","work_id":"wrk_72f5f141-da8e-4f5b-a337-f46180c37f2e","block_id":"blk_5d78f72f-7f4f-4502-9498-1e454589eb2c","section_id":"sec_554b0216-d987-4463-bfde-4307c2ed3bc6","layer":"body","character_id":null,"count":1,"matched_aliases":["Agentic"],"evidence":{"text_basis":"markdown","start":0,"end":7,"exact":"Agentic","quote":"Agentic AIの普及によって、","quote_start":0,"quote_end":18,"text_sha256":"2fa979992c6067dcaaca911ab1032703d3cfb92a65dff51edfa6fbc8cdd2f540","block_sha256":"2fa979992c6067dcaaca911ab1032703d3cfb92a65dff51edfa6fbc8cdd2f540","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_f76185ef-dd8b-44e4-a82a-7b8cc8225b9a/#blk_5d78f72f-7f4f-4502-9498-1e454589eb2c"},{"id":"occ_6ab199e2ac2f739ab9bfd5af","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280","work_id":"wrk_a85448e1-149a-46bd-b7e9-d66795012c12","block_id":"blk_09bb534f-fa5a-492a-8648-2d95d8c26596","section_id":"sec_b887b3da-932f-40fb-9485-f19270b880c9","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"なぜならAIエージェントは、単に質問に答えるだけではないからだ。メールを読む。予定を管理する。資料を作る。検索する。買い物をする。予約をする。返信する。画面を見て、アプリを操作し、ファイルを扱い、個人の生活や仕事の文脈を理解","quote_start":0,"quote_end":112,"text_sha256":"d497102a53e367293c66f73f4a2f9610b97be3715d5ed1d403fc9d103bada3c9","block_sha256":"d497102a53e367293c66f73f4a2f9610b97be3715d5ed1d403fc9d103bada3c9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280/#blk_09bb534f-fa5a-492a-8648-2d95d8c26596"},{"id":"occ_96ce7324bbe156035995efee","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280","work_id":"wrk_a85448e1-149a-46bd-b7e9-d66795012c12","block_id":"blk_0b529f4d-7bac-430b-8ad3-7683704f6394","section_id":"sec_e012475b-c9f9-4af9-a3a5-58ad32f1d09a","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":75,"end":81,"exact":"エージェント","quote":"の段階に留まるなら、各国は米国クラウドやグローバルAIサービスを利用するだけでも一定程度は足りる。言語モデルとエージェント機能であれば、オープンウェイトモデルを自国データで調整し、自国クラウドやオンプレ環境で動かすことも比較的現実的である。行政文書、企業内データ、医療・金融・法務の機密情報を国外に出さずに処理するという意","quote_start":20,"quote_end":181,"text_sha256":"06045f639c49722a5f9d94d9602fa1fe2b62361129fef60d38d280da20e44097","block_sha256":"06045f639c49722a5f9d94d9602fa1fe2b62361129fef60d38d280da20e44097","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280/#blk_0b529f4d-7bac-430b-8ad3-7683704f6394"},{"id":"occ_b85718c5a85c8a1f9c3f489b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280","work_id":"wrk_a85448e1-149a-46bd-b7e9-d66795012c12","block_id":"blk_229741e2-6bc3-4601-9013-6fbacdd89849","section_id":"sec_1f7bad82-68b6-433f-9648-e18c2aa4f26c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":44,"end":50,"exact":"エージェント","quote":"メモリは単なる部材ではない。AI文明の記憶である。  \nCPUは単なる演算装置ではない。エージェントの指揮者である。  \nスマホやPCは単なる消費財ではない。個人がAIを持つための資本財になる。","quote_start":0,"quote_end":97,"text_sha256":"de5465bf094e957e1229185a580b5aedf05821508d64ae999fa1eefe736a92af","block_sha256":"de5465bf094e957e1229185a580b5aedf05821508d64ae999fa1eefe736a92af","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280/#blk_229741e2-6bc3-4601-9013-6fbacdd89849"},{"id":"occ_4892068c0ebeb14e9c7735bf","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280","work_id":"wrk_a85448e1-149a-46bd-b7e9-d66795012c12","block_id":"blk_2b59fcc2-a9bc-4c14-b51f-a8aea8ae38a6","section_id":"sec_ec66ad9a-44c0-4ef8-9def-eac1566f1e92","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":75,"end":81,"exact":"エージェント","quote":"Iワークステーションは、開発、動画編集、画像生成、投資分析、設計、研究、資料作成、営業、経理、ローカルLLM、エージェント実行に使える。高価でも、仕事で回収できるなら買われる。","quote_start":20,"quote_end":108,"text_sha256":"a492ccefbd1161ac1387ba3cdc523e7f0bed8c6de9f8c9914af4696c7a666638","block_sha256":"a492ccefbd1161ac1387ba3cdc523e7f0bed8c6de9f8c9914af4696c7a666638","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280/#blk_2b59fcc2-a9bc-4c14-b51f-a8aea8ae38a6"},{"id":"occ_e532d899ab2acad88f0fbe82","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280","work_id":"wrk_a85448e1-149a-46bd-b7e9-d66795012c12","block_id":"blk_2d563161-08ea-4212-a3ee-d5a8153b48bd","section_id":"sec_7bb945fe-0c5c-4899-af97-3f98fc9a4bc4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":34,"end":40,"exact":"エージェント","quote":"これは、コンピューティングの中心が「人間が操作するアプリ」から「AIエージェントが実行する環境」へ移るということだ。","quote_start":0,"quote_end":58,"text_sha256":"ece8ef2bc0fd9830da3751fd176d12f835b49217b9063ed035e95eaac08a35ee","block_sha256":"ece8ef2bc0fd9830da3751fd176d12f835b49217b9063ed035e95eaac08a35ee","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280/#blk_2d563161-08ea-4212-a3ee-d5a8153b48bd"},{"id":"occ_ae4feb7cf21c41d9d3c9bd86","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280","work_id":"wrk_a85448e1-149a-46bd-b7e9-d66795012c12","block_id":"blk_2d9b4bec-aa92-414c-a416-5dbbe86a02a9","section_id":"sec_7bb945fe-0c5c-4899-af97-3f98fc9a4bc4","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":30,"end":36,"exact":"エージェント","quote":"MicrosoftはWindowsとCopilot、企業向けエージェント、セキュアな実行環境を結びつけようとしている。AppleはApple Intelligence、App Intents、オンデバイスモデルによって、iPhoneやMac上で個人コンテキストを扱うAIを","quote_start":0,"quote_end":136,"text_sha256":"3ca4f5dae44fffd38b4a6d7fec17f7a5d090a5d1a4a7b4d9a149222b698c1d67","block_sha256":"3ca4f5dae44fffd38b4a6d7fec17f7a5d090a5d1a4a7b4d9a149222b698c1d67","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280/#blk_2d9b4bec-aa92-414c-a416-5dbbe86a02a9"},{"id":"occ_9018b6cfcbb4623f3a56fc61","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280","work_id":"wrk_a85448e1-149a-46bd-b7e9-d66795012c12","block_id":"blk_3558420a-4eab-45e9-9abf-ce732104c019","section_id":"sec_7bb945fe-0c5c-4899-af97-3f98fc9a4bc4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":58,"end":64,"exact":"エージェント","quote":"コンピュータでは、人間がアプリを開き、ボタンを押し、ファイルを操作していた。これからは、人間が意図を伝え、AIエージェントがアプリ、ファイル、ブラウザ、API、データベース、OS機能を呼び出して作業を完了する。","quote_start":3,"quote_end":108,"text_sha256":"903ca2448c6d3ca2de7a311c767bd8febd1c36f18b671c51399d3f31795a6aa1","block_sha256":"903ca2448c6d3ca2de7a311c767bd8febd1c36f18b671c51399d3f31795a6aa1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280/#blk_3558420a-4eab-45e9-9abf-ce732104c019"},{"id":"occ_5dd4370d8ef50d50cacf9c85","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280","work_id":"wrk_a85448e1-149a-46bd-b7e9-d66795012c12","block_id":"blk_644a29b1-0dfa-4090-99f9-f24a6a9a3d7d","section_id":"sec_7bb945fe-0c5c-4899-af97-3f98fc9a4bc4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":31,"end":37,"exact":"エージェント","quote":"各社は単にAI機能を追加しているわけではない。OSレベルでAIエージェントを動かす環境を作ろうとしている。","quote_start":0,"quote_end":53,"text_sha256":"843931dfcc01dddfdbdabd043b664964e92fc36c9583d970d372d15e5ab33015","block_sha256":"843931dfcc01dddfdbdabd043b664964e92fc36c9583d970d372d15e5ab33015","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280/#blk_644a29b1-0dfa-4090-99f9-f24a6a9a3d7d"},{"id":"occ_ae4f00d222faa814065dc1e0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280","work_id":"wrk_a85448e1-149a-46bd-b7e9-d66795012c12","block_id":"blk_760362cd-e367-408e-8ac9-7f438264965a","section_id":"sec_1f7bad82-68b6-433f-9648-e18c2aa4f26c","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":43,"end":49,"exact":"エージェント","quote":"AIが半導体需要を増やす。半導体とメモリが高くなる。スマホやPCが高くなる。だが、AIエージェントを使うにはスマホやPCが必要になる。人々は高いデバイスを買う。AIエージェントが仕事、学習、創作、投資、事務、営業、生活管理を助ける。デバイス代を回収できる人は、さらに高性能な端末を買う。","quote_start":0,"quote_end":143,"text_sha256":"1b92273fe9f1e03098ba4143dc3b70219bf80b6b9405bb1d74b89bfbf2ea9ee9","block_sha256":"1b92273fe9f1e03098ba4143dc3b70219bf80b6b9405bb1d74b89bfbf2ea9ee9","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280/#blk_760362cd-e367-408e-8ac9-7f438264965a"},{"id":"occ_cba3b27fd39e3ffebc5b8ecd","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280","work_id":"wrk_a85448e1-149a-46bd-b7e9-d66795012c12","block_id":"blk_7c3e4576-5090-41d1-b8dc-47b24d3e8e29","section_id":"sec_b887b3da-932f-40fb-9485-f19270b880c9","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":117,"end":123,"exact":"エージェント","quote":"買うのではなく、人間がAIサブスクに接続し、そのサブスク収益でAIデータセンターが拡張される。さらに企業はAIエージェントやロボットで利益を出し、その利益でGPU、メモリ、ストレージ、電力、冷却、ネットワークを買う。","quote_start":62,"quote_end":170,"text_sha256":"a79dab950df99cb8b61055a31d7987eb774fa3cef58ebdec8f930cb4dfd450f1","block_sha256":"a79dab950df99cb8b61055a31d7987eb774fa3cef58ebdec8f930cb4dfd450f1","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280/#blk_7c3e4576-5090-41d1-b8dc-47b24d3e8e29"},{"id":"occ_62bf2215d4f340c6d2c2c177","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280","work_id":"wrk_a85448e1-149a-46bd-b7e9-d66795012c12","block_id":"blk_aaf62dcc-fbcd-426b-a5f9-63c05d8c692f","section_id":"sec_b887b3da-932f-40fb-9485-f19270b880c9","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":7,"end":13,"exact":"エージェント","quote":"なぜなら、AIエージェントは単なる道具ではなく、判断、記憶、作業、生活管理、学習、創作、仕事の補助者になるからだ。自分のAIを持てる人は、自分の文脈を自分の手元に置ける。クラウドAIに接続するしかない人は、自分の文脈を外部サー","quote_start":0,"quote_end":113,"text_sha256":"28c234f012f684143658b032ceed798cc029a6b7350174475e3fc11fd759c8fe","block_sha256":"28c234f012f684143658b032ceed798cc029a6b7350174475e3fc11fd759c8fe","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280/#blk_aaf62dcc-fbcd-426b-a5f9-63c05d8c692f"},{"id":"occ_aab53b16e37a213e211d1b44","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280","work_id":"wrk_a85448e1-149a-46bd-b7e9-d66795012c12","block_id":"blk_c2535afe-bf9a-4896-921c-87488e91a44a","section_id":"sec_7bb945fe-0c5c-4899-af97-3f98fc9a4bc4","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":20,"end":26,"exact":"エージェント","quote":"ウィンドウやファイルを管理するものから、エージェントの権限、記憶、ツール呼び出し、サンドボックス、監査、セキュリティ、ローカル推論、クラウド補完を管理するものへ変わる。","quote_start":0,"quote_end":84,"text_sha256":"b7bd22032d1bf369ed1342705f1fac2291157b70b27d5f85837687cbe302c99d","block_sha256":"b7bd22032d1bf369ed1342705f1fac2291157b70b27d5f85837687cbe302c99d","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280/#blk_c2535afe-bf9a-4896-921c-87488e91a44a"},{"id":"occ_21e6fed6d3b55ac7fa50c04f","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280","work_id":"wrk_a85448e1-149a-46bd-b7e9-d66795012c12","block_id":"blk_dd510bad-2d3b-4e0e-a84d-ddd59ca37e1e","section_id":"sec_ec66ad9a-44c0-4ef8-9def-eac1566f1e92","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":53,"end":59,"exact":"エージェント","quote":"スマホはすでに、連絡、決済、本人確認、銀行、行政、仕事、求人、学校、医療、災害情報の入口である。ここにAIエージェントが加われば、スマホは個人AIを常駐させる端末になる。","quote_start":0,"quote_end":85,"text_sha256":"68b339a0a5edc8ce323204a46137a19a0bc028abb093645ae89669fcacc75174","block_sha256":"68b339a0a5edc8ce323204a46137a19a0bc028abb093645ae89669fcacc75174","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280/#blk_dd510bad-2d3b-4e0e-a84d-ddd59ca37e1e"},{"id":"occ_e84ec3a6c12c3e486bcd5866","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280","work_id":"wrk_a85448e1-149a-46bd-b7e9-d66795012c12","block_id":"blk_f7b38b8b-1738-438b-90b1-a1ef63bd0ced","section_id":"sec_2243dc29-cfe0-40c0-ad44-cef4e0ba05e5","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":214,"end":220,"exact":"エージェント","quote":"エッジAIのような成熟・特殊プロセスで効いてくる。DatadogやCrowdStrike、Rubrikは、AIエージェントが増えるほど必要になる監視、セキュリティ、復旧のレイヤーを担う。","quote_start":159,"quote_end":252,"text_sha256":"11ed345a65a1d583f6c557c2684353b3af1be580d29656a0e1744f3f8e51e977","block_sha256":"11ed345a65a1d583f6c557c2684353b3af1be580d29656a0e1744f3f8e51e977","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280/#blk_f7b38b8b-1738-438b-90b1-a1ef63bd0ced"},{"id":"occ_e33cfdd1af5618f05744bf16","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280","work_id":"wrk_a85448e1-149a-46bd-b7e9-d66795012c12","block_id":"blk_fc8d69f3-ea19-4d50-a117-f19ae35837bd","section_id":"sec_bca4afb9-2317-4226-87d8-e5b780d21173","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":19,"end":25,"exact":"エージェント","quote":"高価格端末では、オンデバイスAI、個人エージェント、長期記憶、ローカル処理、セキュリティ、アプリ横断操作が使える。一方、低価格端末では、最低限の通信、決済、SNS、クラウドAIの軽量利用に限られる。","quote_start":0,"quote_end":99,"text_sha256":"6bf55729efd18b2e4b88df73824b2a06b60ddba2361f3a9543405b2357dbc1b3","block_sha256":"6bf55729efd18b2e4b88df73824b2a06b60ddba2361f3a9543405b2357dbc1b3","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280/#blk_fc8d69f3-ea19-4d50-a117-f19ae35837bd"},{"id":"occ_42688c8d42bbfe36f26a94d0","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_fd5eeabd-4433-491f-b463-c900cf908355","work_id":"wrk_d47ffc56-09dd-4f57-a300-be2346278872","block_id":"blk_1ef3abbd-79eb-4266-ab25-3501c57de500","section_id":"sec_394b80f9-5ff5-4ffb-845a-5fac25f71418","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":53,"end":59,"exact":"エージェント","quote":"AI推論では、モデル本体だけでなく、KVキャッシュ、RAG用データ、ベクトルデータベース、生成結果、ログ、エージェントの作業状態などを保存し、繰り返し呼び出す必要がある。","quote_start":0,"quote_end":85,"text_sha256":"af9360306e0d0293e528f115a2899fae15fc6d11d6d64ae070d2ce3320efb92e","block_sha256":"af9360306e0d0293e528f115a2899fae15fc6d11d6d64ae070d2ce3320efb92e","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_fd5eeabd-4433-491f-b463-c900cf908355/#blk_1ef3abbd-79eb-4266-ab25-3501c57de500"},{"id":"occ_525570da357e0c679b372b8b","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ffbdfa80-0e8b-40b5-8765-ba1e5b977d47","work_id":"wrk_75c66a96-f27d-4786-b253-3d6d4f826277","block_id":"blk_0a132a93-ec6a-45a8-bdb7-3855f23b5111","section_id":"sec_e1c16d8a-fc5b-4a8d-8728-7e797a8dfcab","layer":"body","character_id":null,"count":2,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"一般的なAIエージェントがブラウザ、ファイル、API、コード実行環境などのデジタルツールを使うのに対して、ロボットエージェントは腕、手、脚、車輪、カメラ、触覚センサーなどの物理的なツールを使います。","quote_start":0,"quote_end":99,"text_sha256":"f85e4dcd868b7af32f9ffc3c628f9834a5e00284d3d1c3f19312f17a3d9e7b50","block_sha256":"f85e4dcd868b7af32f9ffc3c628f9834a5e00284d3d1c3f19312f17a3d9e7b50","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ffbdfa80-0e8b-40b5-8765-ba1e5b977d47/#blk_0a132a93-ec6a-45a8-bdb7-3855f23b5111"},{"id":"occ_d58e11fcf1e13a9c7fb26461","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ffbdfa80-0e8b-40b5-8765-ba1e5b977d47","work_id":"wrk_75c66a96-f27d-4786-b253-3d6d4f826277","block_id":"blk_5c1bfa70-87fd-4183-9802-cdfe5fae748f","section_id":"sec_80ea3fd9-54df-4928-af22-5adb7a8b3d7c","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":13,"end":19,"exact":"エージェント","quote":"> **WLA的なロボットエージェントを構築できる部品を、ほぼ一通り揃えつつある**","quote_start":0,"quote_end":42,"text_sha256":"6ce1932c0aca1ec0e9a956d3d014f7c03204aac05ffc722e5399a1bed6e5f8ed","block_sha256":"6ce1932c0aca1ec0e9a956d3d014f7c03204aac05ffc722e5399a1bed6e5f8ed","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ffbdfa80-0e8b-40b5-8765-ba1e5b977d47/#blk_5c1bfa70-87fd-4183-9802-cdfe5fae748f"},{"id":"occ_8782add01cb307cad81370bc","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ffbdfa80-0e8b-40b5-8765-ba1e5b977d47","work_id":"wrk_75c66a96-f27d-4786-b253-3d6d4f826277","block_id":"blk_a04762e3-6346-42f2-ba8a-f365cf29084e","section_id":"sec_e1c16d8a-fc5b-4a8d-8728-7e797a8dfcab","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":6,"end":12,"exact":"エージェント","quote":"> **AIエージェントのロボット版**","quote_start":0,"quote_end":20,"text_sha256":"3bb3e0d015dc2ebcd8d1bebbce7b9a43d33e3e494ae2d63c7a5b36a875406f40","block_sha256":"3bb3e0d015dc2ebcd8d1bebbce7b9a43d33e3e494ae2d63c7a5b36a875406f40","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ffbdfa80-0e8b-40b5-8765-ba1e5b977d47/#blk_a04762e3-6346-42f2-ba8a-f365cf29084e"},{"id":"occ_025cc87d3d73cada35a2e628","topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","revision_id":"rev_ffbdfa80-0e8b-40b5-8765-ba1e5b977d47","work_id":"wrk_75c66a96-f27d-4786-b253-3d6d4f826277","block_id":"blk_cfca9a2c-efbc-4c66-9a0c-5a1b0c18b1e7","section_id":"sec_a4ce543a-4827-4394-a30e-fe308ab499db","layer":"body","character_id":null,"count":1,"matched_aliases":["エージェント"],"evidence":{"text_basis":"markdown","start":26,"end":32,"exact":"エージェント","quote":"これはまさに、現在デジタル空間で発展している**AIエージェントに、物理的な身体と世界モデルを与えたもの**だといえるでしょう。","quote_start":0,"quote_end":64,"text_sha256":"041825574866afef8bd3ce32143c3464302d948818f2aba8d0075003e5162289","block_sha256":"041825574866afef8bd3ce32143c3464302d948818f2aba8d0075003e5162289","offset_unit":"unicode_code_point"},"detection_method":"deterministic_literal","dictionary_revision":"topics-20260909-v1","review_status":"automatic_match","role":"unassessed","href":"articles/rev_ffbdfa80-0e8b-40b5-8765-ba1e5b977d47/#blk_cfca9a2c-efbc-4c66-9a0c-5a1b0c18b1e7"}],"timeline":{"dated":[{"date":{"precision":"instant","search_end":"2025-01-29","search_start":"2025-01-28","timezone":null,"value":"2025-01-28 19:49:31"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_9236b1753eac5adf92ed518c"],"revision_id":"rev_d8e8cb66-632c-4170-afae-461911315dbd","work_id":"wrk_7ab4bb8d-7296-4037-b30a-70a60e28196a"},{"date":{"precision":"instant","search_end":"2025-07-26","search_start":"2025-07-25","timezone":null,"value":"2025-07-25 14:34:32"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_dfd903b48b2c1889bf50cc8b","occ_a8c70bd5a99d3e762acef342","occ_bbb3a06acf4a782bb8e7d093","occ_32993da78bda64618935ec0f","occ_8afe3615a363c81e175e6ebb"],"revision_id":"rev_f17ef624-e9fd-49a4-b061-966093172de4","work_id":"wrk_168418fa-2b60-4053-acfe-b86f6898d033"},{"date":{"precision":"instant","search_end":"2026-01-09","search_start":"2026-01-08","timezone":null,"value":"2026-01-08 08:36:32"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_339e06f3af4a7e7d3c011787","occ_fff1b113b3baa16e293d6501","occ_9b653480921bcbc8aaec6e47"],"revision_id":"rev_0ee327e2-ecfe-42e6-b337-fce8abbe7b71","work_id":"wrk_e5060e37-5e03-4669-88d5-6a5989ca3dcb"},{"date":{"precision":"instant","search_end":"2026-01-24","search_start":"2026-01-23","timezone":null,"value":"2026-01-23 23:03:48"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_bc971794bb747581647472d9","occ_25e9090212bc8cbf89a59c5d","occ_dd3a4ad271f60014d46f5617","occ_d71582fd11d24183f4369a8d"],"revision_id":"rev_9ae4d311-bf43-4bb1-a7b5-36a5649efc7c","work_id":"wrk_9cfc75d7-7e8e-4397-add6-437f60011804"},{"date":{"precision":"instant","search_end":"2026-02-16","search_start":"2026-02-15","timezone":null,"value":"2026-02-15 15:56:00"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_780d781dadc708b65786e067"],"revision_id":"rev_b8a0343f-f76e-47f4-8c47-4536da944126","work_id":"wrk_0eee3d56-e636-4917-98f6-cddcb3cca3f5"},{"date":{"precision":"instant","search_end":"2026-04-08","search_start":"2026-04-07","timezone":null,"value":"2026-04-07 20:24:17"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_5d68c30e9d7e164256732601","occ_fd181c779c07afe3020edf2d","occ_3168a6ef1bfa5ad35855b05c","occ_4334cb0d6c194bc3af0bbeb8","occ_c01cd4e18ad9482ae8a110c0","occ_89e3416dd9f9da568ef6cd45","occ_5eadba9eaf030da57803de81","occ_ac5861385577edd9e4872be8","occ_7d6a21d5af4aef6906c988b8","occ_cb6b719d129224c2e9b4957c","occ_891f9fc1733a69be6eede449"],"revision_id":"rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b","work_id":"wrk_b4bdeaca-dc31-448c-8f47-03f9e39c14b0"},{"date":{"precision":"instant","search_end":"2026-04-08","search_start":"2026-04-07","timezone":null,"value":"2026-04-07 22:44:58"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_e2f7506c3dabbfe84ad7f8e8","occ_e75a7b49a9c04c62c2d9b527","occ_999cc56b6fc2dd020f09c5e9","occ_bf10ff43daa71fd522df1545","occ_b92b8b27dd16d478ac2a98ec","occ_df055e8b36a55f7e2220206f","occ_bbdb47a6dcc2887388115552","occ_1c3105840c18954fd9a8b7e5","occ_caf31525b24e9c633f42e723","occ_02c76b2afaeda6ca44c8dfcb","occ_7c94f85aaef7bac39db8357f","occ_e74a4eb9e3191f0694b4488a","occ_a23abf5b17bf36bf4224f6dd","occ_cf2cd2d5ec9a3ba7f313ec9d","occ_b81878b79abcfc208840aeca","occ_35193add0569a4833d8d8c1f"],"revision_id":"rev_b228986a-43e7-4e2b-8b16-758283d367bc","work_id":"wrk_b193a2ff-ca61-4efc-864e-0778162ac852"},{"date":{"precision":"instant","search_end":"2026-04-17","search_start":"2026-04-16","timezone":null,"value":"2026-04-16 00:46:36"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_f7838da8a5568a2cc54e7098","occ_6b15083b4205f0147d307813","occ_5040e1b0ce4f173a187bf0af"],"revision_id":"rev_9af30ed5-ec26-4204-9f65-58324ee5fba5","work_id":"wrk_c56e9825-8401-406e-aaae-662904613d26"},{"date":{"precision":"instant","search_end":"2026-04-18","search_start":"2026-04-17","timezone":null,"value":"2026-04-17 17:39:04"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_2054138096ee129c7c5f2e22","occ_058d27d479eb8d99f347a78c","occ_50e4a2342627d5a5b052a7e0","occ_392a35fa87408ba3516b4146","occ_eb1b197e5416664029736a4a","occ_d0060cd7bcd7efa23017fc90","occ_32ec034ce4b971354c1a0fb4","occ_e8b4eb1e79834aa96c58fcc6","occ_a78fed6220e8f3e84d680ea9","occ_80619066f9cb5acca544d33b","occ_dc9bdd2308a6b2136227bc9a","occ_7eeb6c68112a49eaed985240","occ_3a5031232f3ee44bd09968b1","occ_288b3857358e8388c5e38d22","occ_5c09e6a98631689ac1816ce1","occ_93ef4039e1401d52e2e2ef90","occ_077d9b0063ad89ce86f0ed58","occ_fa0c722f0755f14086194807","occ_45fb43c8144df0cd6bec7b8a","occ_381b440a4cb5de45e2a6950f","occ_581abd19614e3b9bcd4adb37","occ_c9a223a4279832a5e1e0c81e","occ_11bc737a8a461d84c27e5dff","occ_03e766ea2e86014f0704ea9b","occ_07ff66391d74596a94d4ac19","occ_b174415d2741221f4d22f1f4","occ_90e599a51f44f0efb885301c","occ_db9016d374e79a7809c9cf3b","occ_894e089007b8847390af4500","occ_43d6fbafee3511f8979257ee","occ_4580b04e1bf2696d83cab473","occ_a758996de23f050ecc5435f8","occ_b2d24e377efcd0a88293d9de","occ_b596c94556d7e77d8f4b4dd3","occ_21d13c94c7bd1ad6bf910e6a","occ_bc2caa731ac4b9464aa195d1","occ_f6e9eccb86a66605caaa24df","occ_f5aa88f808fff6f9926c9c04","occ_8c9f80d0b3a6731c12db6a3b","occ_fe99d9eb3d1ace49c3e1fa2e","occ_70085e6af534544e16ce275a","occ_9462a1f84a6a101404a51754","occ_65ba63ae3fd3f6b3956e9124","occ_266fab017e953c4a19a3ed88","occ_d5d9d52610b758b13593badf","occ_a01f489797419f1766566036","occ_083ca625ba4fd9836c842a50","occ_57cea19eec247e7978bdf69c"],"revision_id":"rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","work_id":"wrk_897da4b4-3324-4db0-b370-04a77562fa7c"},{"date":{"precision":"instant","search_end":"2026-04-21","search_start":"2026-04-20","timezone":null,"value":"2026-04-20 18:17:43"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_746f3f88335bd169e19551e6","occ_08e38a6c1c114f782d29976c","occ_8ed270cd2569c168a275b32a","occ_b93a7fe69765daeba7e56bce","occ_c3b7a77446fd84546864164a","occ_17e1c852d21d239be67f6e45","occ_2b92daf1fb816daa559f61f2"],"revision_id":"rev_57810a5f-315e-49a7-ac85-72cc27e35eff","work_id":"wrk_fc529d17-dbc8-4405-b2ca-a48accef7253"},{"date":{"precision":"instant","search_end":"2026-04-23","search_start":"2026-04-22","timezone":null,"value":"2026-04-22 05:22:27"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_fbf7802aa52bdc06c1d51f19","occ_3699df96bea6ded85d9d1399","occ_4db807a347470c4acc3702f5"],"revision_id":"rev_7cf42fcc-1956-4694-a0b6-5eaa9be5c284","work_id":"wrk_cda7c85e-9714-493f-a66d-ac6f36454630"},{"date":{"precision":"instant","search_end":"2026-04-25","search_start":"2026-04-24","timezone":null,"value":"2026-04-24 01:44:34"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_f066c968bb4bf85b29202d0a","occ_1086387e12530f56f9db5a09","occ_656d6e2d2d77666b7c4249fe","occ_74cd7800d2cbbd89909d44b1","occ_3ebce07b4121be0ceb2571e3","occ_793772215d671881b4a41b5f","occ_dc00f16ec79de4509b34fcaa","occ_0bf6340db4faae5d1baa3854","occ_9ffcf44b1c8d3d4147199936","occ_861315cbbd7baf95a7c99484","occ_15b3123e4469104660074690","occ_c0a3dfb44e7fdc060b995f74","occ_ce9e681a0fc7df1e0162dd4f","occ_c3cb3e81e5c30b403469e403","occ_fea3ad93fa20499952e2e7d7","occ_86d3b218458c7c6273f90d7b","occ_92e9c48b65a62e1a3d7e3fd6","occ_1d0e3c45b87775b638d29cea","occ_8f61cf342172eef9e2229226","occ_da682c087cacce36d7a32d19","occ_71fb308cd83453c0e20900ea","occ_f71a239b4c45a2e69c2ea622","occ_68bca3d1af8a078d7aa8a4df","occ_305a2ee15490af70b4b912f1","occ_668d39e186d10ce0bc86b5f1","occ_8ab0a052259b7dd52df4a607","occ_581ad7342c98745919656d4a","occ_7add1f1376169ab3f751a2a5","occ_51c421326f8d84a5af4fba0c"],"revision_id":"rev_f487aa08-2568-42ad-85cb-11c258b5b71d","work_id":"wrk_a5567405-5ef5-4c25-abcd-63351a43cdfd"},{"date":{"precision":"instant","search_end":"2026-05-01","search_start":"2026-04-30","timezone":null,"value":"2026-04-30 05:15:52"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_837ac728993b59501d277a57","occ_bef24913c4e971730c59cbb1","occ_a1e8ab120b3d4f032609aeb7","occ_97b68d3d7a487774d5d10b30","occ_ce57bb5abc53429ad1ff9260","occ_de2ad23cf002142685ac601b","occ_89fd0302feb3dfdbd058aeee","occ_f7937cb05e90dd17414571b7","occ_ccfb503593e9d5818d731286","occ_9d1f37ce6dac7b7e92393fc1","occ_f9ec0630db80c19a3c7c5255","occ_8fb3449112af98c1e4ddd298","occ_ac95be298a1a7e9fcb4ec06c","occ_06f7b4f34bab12fbd26a477b","occ_deaf2fd93f0ffda5f2cd0291","occ_09306594ef1d2fccc0b075b4","occ_4af0e646e02ed9fae15c51bc","occ_85044c6e04a694b5e5fee18b","occ_f79b76dcfe821e3aa987e7a3","occ_3decd8a7784e73462f55ca16","occ_583b9610c4f7f19abff451b1","occ_51b6657de8c8348b7d7e8628","occ_fe8d7f3bba277228bbd56cc5"],"revision_id":"rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","work_id":"wrk_05a2b7de-6391-4b00-b331-625ad6b4aa6d"},{"date":{"precision":"instant","search_end":"2026-05-08","search_start":"2026-05-07","timezone":null,"value":"2026-05-07 22:10:59"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_0aa35f6892b98cb73bd216cc","occ_b3336d60b88db9e5cca7613a","occ_76202a99c2f37112e09830fe","occ_44e3fcd7ac43c56c00daa9d6","occ_198e86728ad58eeb3a286066","occ_861cb236d94919c253bdf064","occ_1b74ee7c05e97d25c4db2e30","occ_992e4940b1cfa986a031492a","occ_450d58b8ad0c328fb8e9a601","occ_82d36688e180f53e5ceffb71","occ_2deb2ded47f34e765dd99349","occ_fb74ecc94016cc88cafe31b7","occ_f6ff6ff2709aadc2d4c04dbf","occ_4ec5c88c828c6e38060e7a88","occ_cceb687bc3caa9f4e3ce38ef","occ_961196891fbb00c87fb6d86e","occ_3ff2ce0bbc383102cac9974e","occ_f340d4d99c46d6d917441b11","occ_ecb455370d23a2d74b6cdf16","occ_99833a4aba3ece22468045bf","occ_7233f3c6f8b91b18eb8732af","occ_b18d3ffac03aa535854b53be","occ_a01847dd3c7b2fc4ce1ebc08","occ_782429a7b732533896c0cf74","occ_8590b3f457022fcbd88e2166","occ_d66f23e98c331ff54f64cfed","occ_b612dfc8feb0be1ad490ab7b","occ_ab33193476fba0bf71b4b097","occ_b38f0460795da4e5a40db753","occ_20483a3791ac16b69004e1b0","occ_f74cc974e33e9878ce66a471","occ_a6eb63390b36f6e204745056","occ_c5bbfd7f1c384295c648b6ec","occ_fe5c9d0b7c54763576df9609","occ_8a25edf78b977fcc46ac0e07","occ_01104bc609e4a4a5b6f00cc8","occ_aa49785114658ef53cd608cc","occ_a5252a23b84395eded61ca7f","occ_bd766a4cc42456804a8a6308","occ_bd86373f0e253e5662d43929","occ_4e162b2a1a6ad6d7ec3fbb2c","occ_3dd3660aa1ff02c4825158e8","occ_3f92a4cf308e07644b47043e","occ_4c4280d990ba0db84c5c8351","occ_e08210059a9c42a3cc47152f","occ_18f787cc663527f177d7f7c1","occ_d58713b4e1f9099114c0eee1","occ_1bfcf2c59c5570304ff0675d","occ_58d4405ed79b64a54225c4e3","occ_2fa327f9de434f3d8a2fb9ac","occ_54641cbf7ba1ae720b9e729f","occ_6f7fc68e61fb95944a22ff2c","occ_638c91efff84d682fca58949","occ_cdf2d420e2f838bcc843a75e","occ_fef578f32d890fcb41540f9d","occ_1e6e13e725121c8f1f9a485b","occ_e5685d7c126f75cbdfd36e56","occ_e0b912d7935085acb1665644","occ_abd99423a0bc730910b9b0c4","occ_f2c3d97bf010ad17357a606b","occ_1c2000941082ce5d0f1098cc","occ_662d913cf67f1cc4286bbef5","occ_1315dc4fd1bc17f6792496fd"],"revision_id":"rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","work_id":"wrk_dea71d50-dc48-48c0-a352-3461dbce4132"},{"date":{"precision":"instant","search_end":"2026-05-09","search_start":"2026-05-08","timezone":null,"value":"2026-05-08 09:09:09"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_f980ec67e41b45a9750cbba3","occ_68a966e8470362248063f408","occ_fcaa0f0961ebc03c0557eb8b","occ_65449f2519fafb68b97b0e93","occ_30f6371de9c86d2e7abb4043","occ_310ca66696c36ec02809400f","occ_df26b30e3438f1d45b35fa4e","occ_ccaba9a91c6327b1ce5f4a12","occ_f034dff429fc1c287091dd10","occ_67140877c542a7bf5b85aa72","occ_a283d60060433f6ee232b087","occ_e6a7cb191376ed7ce93f597e","occ_ce9179b29065c0ae92dbc27c","occ_9a102ce7c163df708e53e6d7","occ_77f82fc25e0e5c9cf7e636f5","occ_d04e02562b76110b1f083a0a","occ_904ea9311eaf57a8e6f5f7ab","occ_7694591f65d93830ee79753c","occ_4dc51dd5e3c4fc9605517ddc","occ_a08eb4d37eafa4f6c8358ce8","occ_bb692c2022d245027e82e84a","occ_0544ec48e49e51df41f887fb","occ_18b327eaf6df216902f5e701","occ_ef3395771253fd475da3561e","occ_72d63b184a24e916b1038f23","occ_81bb77161e0d8c9109cb18ed","occ_060196bc2612cf8d99fc555c","occ_daa36cd558cd08b3acd9ecac","occ_9722ac1ee501bb7c9e680f13","occ_d4d539b6d8397f9fdf97afae","occ_6c915d33a98df0fa6735a57d","occ_d13297393a816a44f66cef14","occ_c9f998f33e5732261a83d53e","occ_40a17608ea204915b18d22cb","occ_0f66aae0cc8406ff479b863f","occ_8142df7791394d069d75b4d2","occ_9ff93a05dd66432a4a46de8a","occ_d436499469b4851696420975","occ_3394344b92add63b39c2ba36","occ_2882cb6a6a4ee77a819b76cc","occ_cd3e4819b84c821dd7c4ebfd","occ_3d43d65d603d7203daae4508","occ_52dad15e797ceef49540ade3","occ_8f389f3af51afb486a2ae198","occ_e211019e26bc33b073770c0a","occ_9e7640c36753fb617d70309e","occ_fb3b94c7a3721747b998ec1a","occ_4982743b21e9d96eed439fb1","occ_068b8155295d461030cc5f3f","occ_a023210ccb63bd0fc183889b","occ_c194dca1cccb1c69d1475219","occ_6b71ab237a298b554d1cf83e","occ_90928df314ac26cfbb8d0952","occ_bf3246d1b1e5c404d7d006eb","occ_e87c37347859eac81b4c786b","occ_cc860d13dc10ec346b025b1c","occ_11b4238d1eae89920666dfca","occ_404ed4c7e20b516c18582c83","occ_29b64c7c397cd8afc12779f5","occ_71eca70a030e4d5febdc2ec0","occ_ce92b1e102000d95e056c5a4","occ_1268dac483027e7bbd05e281","occ_a07ac8f016bf5dddb5436bb3","occ_a2842a6771e27f765023527f","occ_ce2773cb5581f527377b23f5","occ_b28735c9127309c2e0038707","occ_360494029b6cf3fa45488e34","occ_89cc166f7a2aee8432e5eb19","occ_07b8e4acd1c461f147ab7f11","occ_086764acbd3099c987be0282","occ_aa51106dfb48f5173cb6a67e","occ_69e9bc558caed7077c1f75ab"],"revision_id":"rev_6cb909ec-1291-4cbb-a768-f72123829d08","work_id":"wrk_72dcc676-6c79-475a-9f86-c4f174e32af2"},{"date":{"precision":"instant","search_end":"2026-05-11","search_start":"2026-05-10","timezone":null,"value":"2026-05-10 19:19:43"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_99a4da3135f267aca62b803e","occ_34bf3347ab98c3bb2041299c","occ_801c18ed69ded4de17c9cd98","occ_1c038ca01dc8e8057d1e3388","occ_a15bba190308a7669552329d","occ_441507bdaf3f77200c53efd0","occ_7cd41fc4a7319107466b5a8a","occ_4143656a9867b1449ddd9a00","occ_48413337f770deb56a75bde8","occ_da380f1fafc4281b32c569b5","occ_d31bb2de70570c74c5491087","occ_43aecb1dbd040d9d255cec4f","occ_4e9f36bb7940da7a5899ffe6","occ_16a400cd1a07f8bf11377903"],"revision_id":"rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","work_id":"wrk_a3440338-a420-4877-947f-8193ae66700c"},{"date":{"precision":"instant","search_end":"2026-05-11","search_start":"2026-05-10","timezone":null,"value":"2026-05-10 03:03:48"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_95a37f23e0ef9932d06235a4","occ_fd94742ff8e0192f116a370e","occ_18ac44bd1135432a02781e12","occ_646bb56276971c16e6510cb8","occ_4d259acb3d91bf21443a5f36","occ_24492af95d824dcfaf84942f","occ_f49ef30f3e4cfead0682b570","occ_d6a0b58ab08672958df3c210","occ_a13f62e08903983072354833","occ_9027612b3530adc6b0d36159","occ_4aca9ba11585c5234593ccad","occ_0b4b29593e92a74853cb0baa","occ_4a689c4ff43b55bb7d94ffe6","occ_27620229cee384edf7c981a0"],"revision_id":"rev_98f85497-697b-4917-aabc-df1485131d9b","work_id":"wrk_23480985-b0bd-47ea-a46d-716d46f4d122"},{"date":{"precision":"instant","search_end":"2026-05-14","search_start":"2026-05-13","timezone":null,"value":"2026-05-13 17:54:34"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_7afe5d20f34268919ba4e3d2","occ_a628838f1c2357cd364ef6a7","occ_9652f0a57c58e47f8aa036db"],"revision_id":"rev_6abb0df2-b5a4-4f1c-8f48-a346d6d0f247","work_id":"wrk_37d80f4d-ab1e-4385-b7ef-e7ae09b765e3"},{"date":{"precision":"instant","search_end":"2026-05-15","search_start":"2026-05-14","timezone":null,"value":"2026-05-14 04:17:48"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_446e2fed8fbf7d01a0b269f5","occ_17ec2e93277aaa31591d0e25","occ_1fb0a52fb26ea143b631b17b","occ_e809e751acd93fad941b7341","occ_406815a8ca1c551cc9248ff0","occ_671ac9ed098fbbdbb9eccc14","occ_e8cc7d3cd29505d4595a8517","occ_3d63bc3218b4f00a019454c0","occ_eadb075caef088c397ee30ea","occ_5bf5c9be3d8ff1b437a59dd1"],"revision_id":"rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f","work_id":"wrk_f7d6728c-f58d-4c77-958d-67d9910c0d5d"},{"date":{"precision":"instant","search_end":"2026-05-16","search_start":"2026-05-15","timezone":null,"value":"2026-05-15 18:50:18"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_ec5156d56afb5241213790f3","occ_5e87a9ab261dfb9d5483a73b","occ_f68fbc373d70156b5b203df5","occ_39a1e254e708e0200cb067c0"],"revision_id":"rev_da6063d5-59b2-4519-9c42-45d8f4f206c8","work_id":"wrk_ef613b57-c920-49a8-bb03-b7fdb149eeca"},{"date":{"precision":"instant","search_end":"2026-05-16","search_start":"2026-05-15","timezone":null,"value":"2026-05-15 20:04:40"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_41f9f77f7eff2cf8b5812ec6","occ_e3ece3729e33050684875599","occ_8351ea1b463b8fa282d5b9ff","occ_b3269688fcc79dc627434093","occ_7db2f0b1fcb319c89b6dfe02","occ_bde1a7d9fd1d9575a4aa61fc","occ_ac2b55f2430bb1f820689e75","occ_474af06e866d374874d5298d"],"revision_id":"rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e","work_id":"wrk_65ea8146-0c98-40e2-b084-2dacaa4f1bf1"},{"date":{"precision":"instant","search_end":"2026-05-18","search_start":"2026-05-17","timezone":null,"value":"2026-05-17 07:41:05"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_de979046205d4814134fa852","occ_1ff48827094d014345de2393","occ_2e0854b09a472b77ab5aad4c"],"revision_id":"rev_62acaca2-fec9-4724-be08-6ebd3c061f93","work_id":"wrk_d43a6ffd-ef17-4cd2-96fa-06f7d800821a"},{"date":{"precision":"instant","search_end":"2026-05-23","search_start":"2026-05-22","timezone":null,"value":"2026-05-22 22:10:50"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_d5ea94aec6f6160055ab7434","occ_31d164adc314ca35d22c4e0f","occ_e7e9496303228839b6ab737b","occ_c1c439708f864aed63761244","occ_6553047511d97702b495ddc6","occ_ac0b9ef00465de2d4b5a71a7","occ_d49d1119fb9a5e527df8a048","occ_6f4c27fffbced4e29d3bc18f","occ_a701f21e6e3314f21aaa505c","occ_43d9013da1c54dee2cd24c61","occ_fcba8c117e13e61f0f85795e"],"revision_id":"rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","work_id":"wrk_95f3adb2-9ef8-4c60-8a3b-5ded0268dac4"},{"date":{"precision":"instant","search_end":"2026-05-24","search_start":"2026-05-23","timezone":null,"value":"2026-05-23 08:23:34"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_11af537e40249aa395ccc978","occ_fa23991f746a69d3ec78df0c"],"revision_id":"rev_49a8bdbd-8ee8-4c28-8af4-da1787c47955","work_id":"wrk_cb25df93-468d-4abf-9350-ba1bcf0d3b49"},{"date":{"precision":"instant","search_end":"2026-05-28","search_start":"2026-05-27","timezone":null,"value":"2026-05-27 20:02:55"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_76647c0b3f55939defe09258"],"revision_id":"rev_3621f98d-f1b9-4e97-9c8a-8b259cfd31c3","work_id":"wrk_68ec8173-3ea7-4526-bfea-020e82eae4c6"},{"date":{"precision":"day","search_end":"2026-06-02","search_start":"2026-06-01","timezone":null,"value":"2026-06-01"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_f7076030d1ffc6727b1838e2","occ_fe1d8a5634834d36926cb15b","occ_442ab0afe4fb27bd60557ddb","occ_c4e05c360b8598483a63a7c4","occ_473b58a84eb7b7fa1ee07dd4","occ_51acd73d5afaceb06d65e934","occ_284511d990901a38e51f2090","occ_a24b429d1ff891c73db870f5","occ_7e7afc8c876371971cee6bd3","occ_b969bb07b88ea316b624c630","occ_85b86aaa2339714e54096258","occ_523431a58d92b7fe0d932164","occ_25d853079d0501efd681536d","occ_d21f22566130403cca5d3040","occ_58a616266fbdb879262c67b7","occ_39772f79d5c0a38e756b1061","occ_e36966e1741280aceb982d5c","occ_b7d6766caf574a4c6a8f35e4","occ_1dbebd9fe42d0ab17e64ca49","occ_e40cc05930a51b9cbf5fc759","occ_dcd99c789b5fddbff32a6abc","occ_60c0856245d6b7475162aa8a","occ_d2164e446c99fa1ae36bedef","occ_45f687798fc4d032e330f218","occ_ed8b8897d57b57ada201ffe2","occ_077f49506821390f3267a7ff","occ_f876ca25cfe0da53810d38e2","occ_4bc88aa0eefe7073b80938f4","occ_252cf58f12d5aa526a1566a3","occ_8665af2fa94044903d8c6968","occ_3fff98a2a89f35d55461a49a","occ_33b4822568c6bf6532afca53","occ_c654eb1d8eb71b1d79f76a83","occ_ddbb47536c25a454e4bce161","occ_c126c575436e45810d26506f","occ_9015e0768e5fb0893f1d1fc5","occ_e3e62f0fbf46e17104b8b76f","occ_fa28cead51af9b2f8894d26b","occ_384b482a78e82eda64231bbf","occ_3249e95341ec0df9d52bf391","occ_b9dadafea110fe653c9a91f1","occ_88d602b527cde8df1f3f6524","occ_6982df901eef98f80c9ab07c","occ_34f1bdfc312607ff723aa4f1","occ_e026cc6d2bd51ccf188900f6","occ_0d9b873dc05b136175a6b471","occ_3a791c53d52e084b3dd9d6ee","occ_368bfdc7dd72316d8a69af35","occ_bcf97e1c480a9b89d8bb195d","occ_62c3ec4262d8033ee78375e4","occ_8af31c2de194d8021fd2fc46","occ_afb9ca4cd78f0c8b510400fb","occ_2dabf86e335661e1c6fc8951","occ_b32ecf104881debb7f17e910"],"revision_id":"rev_90fb7b96-f10b-43df-a612-0c23e8a50740","work_id":"wrk_0c4c3acf-0221-4d96-87a4-771da7ba1e3a"},{"date":{"precision":"instant","search_end":"2026-06-02","search_start":"2026-06-01","timezone":null,"value":"2026-06-01 20:25:08"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_17212ec48ef70288fab04b69"],"revision_id":"rev_a4008bc2-ae2c-4bd7-8615-0b3c4e0936ff","work_id":"wrk_9bc2084d-1dcc-40ef-9cc9-a53fe61434c0"},{"date":{"precision":"day","search_end":"2026-06-03","search_start":"2026-06-02","timezone":null,"value":"2026-06-02"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_5c823a13c9f34c3de72e735f","occ_8347ec2b73052574907954b4","occ_0aa50c89bab6a0c869ab51d5","occ_23ff166a83bbf7bb01b128aa","occ_6a48896fa91da1fcf26203d3","occ_e625be87ae09f7e7679f5c18","occ_e5a34e3ca68ddb075c1c21d8","occ_95020926ba01fae58ee1bb69"],"revision_id":"rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a","work_id":"wrk_dcf18d88-5123-4a4f-aa9b-564dc767f048"},{"date":{"precision":"instant","search_end":"2026-06-06","search_start":"2026-06-05","timezone":null,"value":"2026-06-05 21:38:17"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_955cae9d940bbd9be4b92a9e","occ_e19c3413b636592d2f796d5c","occ_44c108e7189c13652260ec8d","occ_1f4ca6eb400d70264c5974a9","occ_44d52cf21b609497206fd812","occ_f2d35b889a03c4133c51ce32","occ_593c78a718419be98be61d54","occ_0476d34a1044582db5575f27","occ_c09537bd232616e8d21eec92","occ_564f84064a208ee783bb6cf8","occ_5e05990d871f0803cbd4540e","occ_c9102b56e047ae4808a1e6e5","occ_a2b72a097b618b2e1dd6fea8","occ_021058504842d7f0d45a6dfa","occ_48090803b3b91a2d7b8f5d21","occ_797d9737db146a55085757e1","occ_db9943f1203d500573a7fb12","occ_0598ac3813e16d704a11ee43","occ_acf4bb05621446a63578df3f","occ_85a9e7326c89e93eb952b09f"],"revision_id":"rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","work_id":"wrk_e8dd57e9-048c-4852-81d1-2a029a08ac21"},{"date":{"precision":"day","search_end":"2026-06-06","search_start":"2026-06-05","timezone":null,"value":"2026-06-05"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_ecebfed5895ca147ff0a5951","occ_786669a5ef449d36583d7edb","occ_e0a6eb2be656e27eba0fd68c","occ_dc0570029c87f05727c2c2fb","occ_5fec097117c372ff4d80fc3c"],"revision_id":"rev_5cb295dd-df59-4d28-9799-42065461c569","work_id":"wrk_5380fdec-35cc-4092-a93e-530c9abca5c9"},{"date":{"precision":"instant","search_end":"2026-06-06","search_start":"2026-06-05","timezone":null,"value":"2026-06-05 17:19:35"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_0ab6c3f031ae044a93a101d7","occ_96163b40b36afd75ed4dc6fb","occ_eb259a4b472a29b7264da06e","occ_6f20ae62411fe32606008c0d","occ_ee8b5741b81e55dd09b0c690"],"revision_id":"rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a","work_id":"wrk_c22379bc-7081-4f8f-a3e9-4418f450cf96"},{"date":{"precision":"day","search_end":"2026-06-06","search_start":"2026-06-05","timezone":null,"value":"2026-06-05"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_71f90c016382a005ad69f8da","occ_c979ee0e2a5efdb36121afec","occ_d78fbc2f7791589eefd63ae6","occ_a59325df529059eb740e26ac","occ_3296471888d271858b848a0d","occ_64ef4c6c0210fdc4a8935584","occ_1bee24f95a1b72f0123c026c","occ_f26ac20436042a6d3f940a9f","occ_68f37c58327074a90414e4ef","occ_0979be2bade8cc6e853695d9","occ_55fbf0f7d88e52153f8c890b","occ_f38a7d5f420f294e4ac10cb9","occ_785870bad8f837c8a6867c02","occ_02e850b6384b532d074e13a6","occ_af6da19fc8750213542595a7"],"revision_id":"rev_c1530075-b9e3-4144-b8ce-f6e480c778fa","work_id":"wrk_50367f10-59d7-41cc-990d-213cec7e7f79"},{"date":{"precision":"instant","search_end":"2026-06-08","search_start":"2026-06-07","timezone":null,"value":"2026-06-07 17:15:14"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_260347df1679f90ccd197e5c"],"revision_id":"rev_15381314-c653-4ce9-9b25-d8495cdbf4d1","work_id":"wrk_220c8e7e-eccc-4357-89f5-9a6b78a08a7c"},{"date":{"precision":"instant","search_end":"2026-06-11","search_start":"2026-06-10","timezone":"+09:00","value":"2026-06-10T12:32:57.000+09:00"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_e490c9326d2d728ebd1973e4","occ_1d24d29db694901110baa9a3","occ_ee2fbb2b58433c756cdd7b2d","occ_7e3e040d5c3b56574eeb0941"],"revision_id":"rev_11f7264b-855b-41d8-b760-87dbfa630e9f","work_id":"wrk_739d4c20-62f1-49ee-9121-4689ad38c0d3"},{"date":{"precision":"instant","search_end":"2026-06-11","search_start":"2026-06-10","timezone":null,"value":"2026-06-10 12:32:57"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_525570da357e0c679b372b8b","occ_d58e11fcf1e13a9c7fb26461","occ_8782add01cb307cad81370bc","occ_025cc87d3d73cada35a2e628"],"revision_id":"rev_ffbdfa80-0e8b-40b5-8765-ba1e5b977d47","work_id":"wrk_75c66a96-f27d-4786-b253-3d6d4f826277"},{"date":{"precision":"day","search_end":"2026-06-12","search_start":"2026-06-11","timezone":null,"value":"2026-06-11"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_faf5d35c4f600f03b25e045c","occ_b3369afef9123dfe56f18b5e","occ_ff0da06e2fc6bf799ce0e4f5","occ_56624c691df2b3cbfa787811","occ_7062f2be944ea1b20836dd1f","occ_fdc8124b7580ac1094be89da","occ_5fd07d28ae1ddf033c852651","occ_3111daaaf9d82c0df977d0f8","occ_c863c831a9ea80b585b07066","occ_f411995da0a92539dcb0c4e1","occ_e4c8b0950e255b2a34d43847","occ_bc510a03de969a9428a39231","occ_28dec8caa88ab2b5745e1c79","occ_1c6457338b96f5657c7ccccc","occ_e1e3aae6b41f040f7c34226b","occ_9c591ef5f1505012a4d396aa","occ_3508b8774263b79b7db89fea"],"revision_id":"rev_a3e10132-4b92-414d-8744-acbeb6f0b294","work_id":"wrk_ab79e5be-0326-461e-868c-ec7e75b2a90e"},{"date":{"precision":"day","search_end":"2026-06-12","search_start":"2026-06-11","timezone":null,"value":"2026-06-11"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_d16640426401de828a5ebc99"],"revision_id":"rev_afc88027-978f-428a-a597-dd5d129dc441","work_id":"wrk_9fae1367-ebbb-47c5-ad7b-3700deb207cd"},{"date":{"precision":"day","search_end":"2026-06-14","search_start":"2026-06-13","timezone":null,"value":"2026-06-13"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_5fe6866232c64d38a8cfc955"],"revision_id":"rev_a749fcdf-8155-46e9-a466-83e867886a08","work_id":"wrk_e146f08d-b638-466d-b368-146816bcc837"},{"date":{"precision":"instant","search_end":"2026-06-14","search_start":"2026-06-13","timezone":null,"value":"2026-06-13 20:40:43"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_a4f0eda42fe3d1d16b9fbdcf","occ_e3e3b7fd6a12c603451f3006","occ_b359cb9786b6f0727e89f74a","occ_3c0831fe32d576d2d3223559","occ_e4557177e8953b0bb3be2a8d"],"revision_id":"rev_ca87695f-4cf9-40a1-bc61-6a123a71ffe6","work_id":"wrk_92c272f3-ada3-4a0c-815e-d33c175ffbe0"},{"date":{"precision":"day","search_end":"2026-06-14","search_start":"2026-06-13","timezone":null,"value":"2026-06-13"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_9ed80af5f7946b7df18e8dbe","occ_3a80f403fd45e2c755520856","occ_a7d2915aa6a1786b68e922a9","occ_b4d1d646665db69d4703d58f","occ_c2cf281772ea5a2816d5f444","occ_4f8159cdacc3c8297fb245ba"],"revision_id":"rev_dbe69fb8-f76d-44e6-b111-aa27e548a204","work_id":"wrk_0252d3aa-e602-4a99-abf9-0b9ec2b5796a"},{"date":{"precision":"day","search_end":"2026-06-17","search_start":"2026-06-16","timezone":null,"value":"2026-06-16"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_86b6c16436b1d9c6fe89c08d"],"revision_id":"rev_78657238-b40f-4a7c-b670-325df3688f39","work_id":"wrk_70fbeeef-5587-4e6a-a017-b7e323d0b279"},{"date":{"precision":"day","search_end":"2026-06-17","search_start":"2026-06-16","timezone":null,"value":"2026-06-16"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_0852122354854a15b403f677","occ_68f42be88da27b7d6bce6fa6","occ_7efdba82004fb8de946c6d00","occ_dd8d8b9f652ef9c6bd909ac0","occ_6b77ca4492bef769b3206c50"],"revision_id":"rev_d47a57d8-f3a6-4505-9ea6-c03ae4416616","work_id":"wrk_0ebc3696-2c90-4588-8ed4-2ea51e51433d"},{"date":{"precision":"instant","search_end":"2026-06-19","search_start":"2026-06-18","timezone":null,"value":"2026-06-18 17:28:27"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_36eb816a01b278fd55481854","occ_1cd04e07424490905ad795cf","occ_350bb7aadbeeb01e8784c0c2","occ_924219fc9d5c36f939650f58","occ_92a11f7079061e25048aa42a","occ_55eccd8cbf93e8e41f56df29","occ_eaa6caa89441d9fe0ba37ef3","occ_673e5f0d2ed3258a45304109","occ_e0f1c4b998ec9215b10ad3fa","occ_2d05e1b8f7fbb6a60e1483fc","occ_6cf5729e4f5461eef4b243f4","occ_2bea30037b924c2a4f4b5e69","occ_d1c5f0d8cbf6c206d13e0ec0","occ_81d4e8cb66d54abd51e2fc49","occ_ba81cfc9d14e0f5923eefd23","occ_e3f3318fa63d5a0ea2600591","occ_fa60b3dfeee84d4f7458edf4","occ_a12c6fc2de90d02d6c939053","occ_a1dfb35658f13240384e2871","occ_093a79970b193b9d8ac783f0","occ_8c1f6edc2b191544e9f1a7c4","occ_a33384e6aaee680640c75ab9"],"revision_id":"rev_12eaaf24-14fe-4445-936b-b81ef492d19d","work_id":"wrk_cd2e4a6c-2b36-4815-ba7a-e21b688d91cb"},{"date":{"precision":"instant","search_end":"2026-06-19","search_start":"2026-06-18","timezone":null,"value":"2026-06-18 18:53:28"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_49c5cec346c9c9716828ee00","occ_ef032a16a535bc33a9497396","occ_b8982e65473f1cbcb92f376e","occ_2c2146d31d45d3e5a584d991","occ_376036b328d468a3d3556229","occ_35e1edd4d07cd7fa8821f2b3","occ_d957eb73144b2f3ffd17bd6b","occ_5471f8a84235309f6d6a4a5d","occ_ab84957f2b61c98c4b6f2b72","occ_f3784ab7cfddfa51e24c188d"],"revision_id":"rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b","work_id":"wrk_bbe3569f-7530-42b2-8b5f-75a35e4b1cba"},{"date":{"precision":"day","search_end":"2026-06-19","search_start":"2026-06-18","timezone":null,"value":"2026-06-18"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_a88f09f2712b6d4f63fbabf5","occ_00d2804984d8ed8956d7eacb","occ_70da3141e7c4ab431668adac","occ_d56a56afdd8f6e78e33fe936","occ_f6f952cdd20db9e0a0622055"],"revision_id":"rev_92ff5158-6c34-4526-ab02-c8db8332951f","work_id":"wrk_ea89805e-b45e-4a9c-b22a-29b168da02ed"},{"date":{"precision":"day","search_end":"2026-06-19","search_start":"2026-06-18","timezone":null,"value":"2026-06-18"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_086c6eb5a7e8af79ca08f885","occ_8b1b4ba1116578af13dd8742","occ_fc4ae6a331c6edf3511d5c35","occ_b28585fc7b042b9cd3c8132c","occ_b6e8249983e7e4c817cb206e","occ_243272d393ac707bcd43bf55","occ_85781ef9f153bf791237c5c6","occ_1bf7108cad43542229d1b27a","occ_9c853a840ad6f51353629b6e","occ_d30502028150ac7cd58eae5b","occ_f080e65067466c95b1b3f712","occ_1875cfb1b329f61123d5b821","occ_2a3ea04f815772b1c259b954","occ_a2dcc7ff2989f211b4909588","occ_4a6a06d3047f74cc6adff32a","occ_c45012b7b4856915b8a9c9df","occ_31e40acc76cd702a81c3e2c7"],"revision_id":"rev_cf107e9a-1286-4df3-bd1f-adb77c070354","work_id":"wrk_a1c546ea-ef0c-48c4-aa48-b5bb2fd37499"},{"date":{"precision":"instant","search_end":"2026-06-21","search_start":"2026-06-20","timezone":null,"value":"2026-06-20 02:24:26"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_6ab199e2ac2f739ab9bfd5af","occ_96ce7324bbe156035995efee","occ_b85718c5a85c8a1f9c3f489b","occ_4892068c0ebeb14e9c7735bf","occ_e532d899ab2acad88f0fbe82","occ_ae4feb7cf21c41d9d3c9bd86","occ_9018b6cfcbb4623f3a56fc61","occ_5dd4370d8ef50d50cacf9c85","occ_ae4f00d222faa814065dc1e0","occ_cba3b27fd39e3ffebc5b8ecd","occ_62bf2215d4f340c6d2c2c177","occ_aab53b16e37a213e211d1b44","occ_21e6fed6d3b55ac7fa50c04f","occ_e84ec3a6c12c3e486bcd5866","occ_e33cfdd1af5618f05744bf16"],"revision_id":"rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280","work_id":"wrk_a85448e1-149a-46bd-b7e9-d66795012c12"},{"date":{"precision":"day","search_end":"2026-06-23","search_start":"2026-06-22","timezone":null,"value":"2026-06-22"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_e29f62dc1769462cf5643fdb","occ_a525d50669eab7436fed7dbf","occ_d7ab5308382f6de58f1c514b","occ_527792e565e1a781bf246523","occ_e36782a50a7b42d261c58d0a","occ_bb9ce1ffd979ce25474be6f6","occ_2aa039933596d3ba8fbd5661","occ_dcc97b8d9b6691bcf9cca116","occ_de433c7a3df4a4d85729a20e","occ_2dbf2cec6c1265365cf7e849","occ_97bfd519d27316b67eef6587","occ_7c09a5ae0065878d7c37642e","occ_f2454af84815cd9eff9be1ea"],"revision_id":"rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","work_id":"wrk_22490a2d-59cf-4ab7-9288-4f1e6c0f5a39"},{"date":{"precision":"day","search_end":"2026-06-23","search_start":"2026-06-22","timezone":null,"value":"2026-06-22"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_35132489f6720b59bb97c8fc","occ_653ea002ac513be72c2ba500","occ_93a27da8c87e4b632fcd696b","occ_8174950fc6935fbb24256b22","occ_9208e85e5deb3fb6be377071","occ_985afecc29d19fb0aa89ddc0","occ_8354a61b97972228cedeae9a","occ_3f7ce6c08d5e63ab5d91c89e","occ_82ac75b8fb2e8de4ae12778d","occ_61bcaa4b3e680dd0b58cc67a","occ_0301d199d102459f602a28b9","occ_ead187f497b0fbd1c65c9511","occ_582bc76564c5895b02017022","occ_f949a2fb36db2306c5de06d0","occ_9c317687facbddd0b2062415","occ_550ffabe8ded8170f3b61fcf","occ_99f7df3d9678a1653b0848e9","occ_7ce1f283be45888843de0296","occ_ba981449f5bada7ea478da6d","occ_a6037412f864df398b541564","occ_92ae87b25548510f9dafdac9","occ_a75d18837e53ac20db8a8542","occ_55b35b2f057e3651c8261df0","occ_5b0634ad3ad1f1906b58aa8d","occ_fedee72ec73e04f0dc654cb5","occ_b052c5213221d72d5e4004ac","occ_19364ee4141973cc7ffdbc8a"],"revision_id":"rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","work_id":"wrk_e0315bdc-338d-4bd7-8701-a2aa52fe1efe"},{"date":{"precision":"instant","search_end":"2026-06-23","search_start":"2026-06-22","timezone":"+09:00","value":"2026-06-22T20:54:37.000+09:00"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_d5b7b13463a8024497d1a93d","occ_46007cc5ce0852f6dd82d624","occ_0fd436af7ecff046782b5b55","occ_56018daf2b692a2a94060045","occ_59199f258197c0c1266fe50f","occ_9b41d5d6d653e6b04dffeed0","occ_1332091fe05fe174f8869d1d","occ_6a9c9295f5a95bdb911c4004","occ_31f1f494e1622d0513cd1c45","occ_e36760fbb6816aea2f5e21ca","occ_56ab23dbdcc9e2cd97bcccd2","occ_2fecae83cc27924cd4d5065a","occ_a0b23f95537708e8bc3e87e2","occ_2c1eeb32ef2e75c74bf01d23","occ_d222320633f5fadb809f42f2","occ_b78319805f33a358fa82d5a4","occ_2741fb5c0d3c04822984b371","occ_e4c4325f08f3eae43ee32f5d","occ_6c55b0ab4f5efa02c1707980","occ_4a5bef90bdd17759bc1c0b48","occ_a3e78283f185b3c9a96cd5af","occ_66c57b6eac2cee396c49686c","occ_511f22792f9655e51369dead","occ_20715c6057b95459ba756346","occ_e314c71e263799b6da97010c","occ_bf82dd346036d7dd48658157","occ_a37acc0f38d9d48b4e357ee2","occ_7c698bbb30761fdb84368519","occ_2c8d7161c29f34e87dae06a9","occ_7709d4bd48503c56f90c9336","occ_1b4da103719bcc9c50edc37b","occ_43f32e10c4013a15fa2acb75","occ_673aba27119c0c100c0fb8ab","occ_ae7fa94a308983fcfdb978b0","occ_b069891508e0a17183835400","occ_1bbc996efa7474c07badc722","occ_4ad75b39d3d6c03a2039b407","occ_eeb4b570ca4494b231750ef3","occ_eb04eb269b2fe4b1be7b5b51","occ_45bb68e59c77c2cbcfe47dee","occ_447d083e3ccece715d54c967","occ_c7e25c95a3a86332c813b9f3","occ_4565bc67c1b6769a60f84536","occ_782a46389ee382070cb4f9c8","occ_84b88b44f04b28583444c48f","occ_6fc3e00c43a8b4da4516237d","occ_d1fc550ff98cf6c9d643a1fc"],"revision_id":"rev_ddc53551-0396-485b-a72e-8b3d5126293f","work_id":"wrk_c352474c-1be9-45a6-bbba-7a95e219168d"},{"date":{"precision":"instant","search_end":"2026-06-24","search_start":"2026-06-23","timezone":null,"value":"2026-06-23 13:51:47"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_e5dcb9e0e000499d115b2d1a","occ_c16b8178122847b907deb8a3","occ_b8eaf7d264b664511188d586"],"revision_id":"rev_977092a4-68df-47ac-8de2-4e66567c8a95","work_id":"wrk_27b36bee-da31-4942-a5ea-33f6f77ddee3"},{"date":{"precision":"instant","search_end":"2026-06-25","search_start":"2026-06-24","timezone":null,"value":"2026-06-24 17:41:44"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_dc29634f5c2176afd7683a20"],"revision_id":"rev_ae55c104-d140-4064-adf1-726fae29a7f3","work_id":"wrk_a1470532-dd67-430b-91ce-2205bc4c54e0"},{"date":{"precision":"instant","search_end":"2026-06-26","search_start":"2026-06-25","timezone":null,"value":"2026-06-25 21:55:11"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_ad38695eb88abbf52c629988","occ_9366e0b4a4babf8bf7000df3","occ_496460fc8c6f8d25b672b136","occ_88921c6e4493e73529569c40","occ_5c44cf0d4a3f91040f859189","occ_498293ff79fdcbfaa7f2317f","occ_a773665a3e8348fe1a8e3634","occ_cc28d49687f091986b7be863"],"revision_id":"rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8","work_id":"wrk_8f40984a-d96e-441d-8ace-83f20958435a"},{"date":{"precision":"instant","search_end":"2026-06-26","search_start":"2026-06-25","timezone":"+09:00","value":"2026-06-25T17:41:03.000+09:00"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_57c31a0b4b275262afd9384b","occ_48b3b679cb9a5d244f1671cb"],"revision_id":"rev_abe6b561-f51c-4549-a97f-aab2b725492e","work_id":"wrk_f31c79fd-b257-498a-9ee1-d344ac94ffc7"},{"date":{"precision":"instant","search_end":"2026-06-27","search_start":"2026-06-26","timezone":null,"value":"2026-06-26 16:31:58"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_0e4524debd558d709f2808af","occ_5b68e4757a9f2f3e3cb44f6f","occ_8a5b3a3f1f9f87a432a5c9b1"],"revision_id":"rev_aea0fd30-a496-42e0-975e-3de7b0cd01ab","work_id":"wrk_4efede78-d9d9-4a37-854a-51a334476949"},{"date":{"precision":"instant","search_end":"2026-06-29","search_start":"2026-06-28","timezone":null,"value":"2026-06-28 19:15:09"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_40b1aaa33dfd1bb0150ef630"],"revision_id":"rev_53605de4-06cc-4e0b-9bc9-21f393408c0a","work_id":"wrk_006833f6-f225-48d3-8754-46fe28d6bd8e"},{"date":{"precision":"instant","search_end":"2026-07-03","search_start":"2026-07-02","timezone":null,"value":"2026-07-02 11:59:12"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_71328541c4f82a6af6979a5b","occ_c9e9e10816f9f23c844507ab","occ_42f2ad8b3271844754ba2620","occ_d786aabc2b08ccf81f02a6a7"],"revision_id":"rev_54b154f3-ea9f-45e9-b38b-a16c54ca3e2b","work_id":"wrk_52a8c79a-5c56-4f34-a515-1b1dcb4c0477"},{"date":{"precision":"instant","search_end":"2026-07-04","search_start":"2026-07-03","timezone":null,"value":"2026-07-03 06:12:13"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_50cb278a35e5fc7f672d5a80","occ_d634bc5d2e8388a4185c0971","occ_29ce7b37c060dca9f2c24be6"],"revision_id":"rev_c8b19a4c-93eb-4695-8f38-c8166d22cbd9","work_id":"wrk_f4bf8d6d-6894-4267-93c4-c8e1e447fab3"},{"date":{"precision":"instant","search_end":"2026-07-05","search_start":"2026-07-04","timezone":null,"value":"2026-07-04 02:28:11"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_74cc7beca37d527b7bc1fb67"],"revision_id":"rev_44c73f5c-647d-467e-8aa2-527eb6352a8a","work_id":"wrk_45578dda-8dcf-452e-909d-3f4c17ea166c"},{"date":{"precision":"instant","search_end":"2026-07-05","search_start":"2026-07-04","timezone":"+09:00","value":"2026-07-04T02:28:11.000+09:00"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_57d9b4b71d3c4159211737d4"],"revision_id":"rev_a30fdebb-6a5f-4491-8480-bc05cbb21f96","work_id":"wrk_97fc23be-d3e1-4a2a-986e-b33ae2fd0697"},{"date":{"precision":"instant","search_end":"2026-07-16","search_start":"2026-07-15","timezone":null,"value":"2026-07-15 21:42:22"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_ce16737cb7be64073712dc62","occ_91735fabae4b275a50785aba","occ_2f94c8c8bb97ccfa21119287","occ_3047387b0bb9fed2c717db04","occ_393e8b362641729e88977fc1"],"revision_id":"rev_f3a5d7c9-6dfa-4c38-8bf2-a37063801332","work_id":"wrk_d32481f4-3a5a-40d8-a817-86b115c8c4f4"},{"date":{"precision":"instant","search_end":"2026-07-21","search_start":"2026-07-20","timezone":null,"value":"2026-07-20 07:19:08"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_cbbda3dbbb9f9f6cd7f37019","occ_f504fb92929129aa1214edcc","occ_cb952a290ab820122772b36e","occ_72a6b4e414fe69f9e79f0763","occ_14d7151f3f468124df54706b","occ_a0f23ff4f6febfacbce07bb6","occ_62b88fdff99d1b1b1774f6f3","occ_9b6459974bd50312dce537bc","occ_dae589609e52ad02f1dd2692","occ_807671ed5677784ce1f06618","occ_2014325810467fd4006771b8","occ_0eb3720383e343e0e313b29d","occ_00fcef7a59d96b9cd9af39b0","occ_9e1ad46c6e2294b76c1fb9fe","occ_4819d69820386dcf6393ce7f","occ_2d8aabb8859b2a688e7c3e85","occ_84937f81f73887d3b095b8c9","occ_8ed9fe2189f6fd719755919f"],"revision_id":"rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","work_id":"wrk_2e3fa847-c032-4040-9284-2106f69cf66c"},{"date":{"precision":"instant","search_end":"2026-07-25","search_start":"2026-07-24","timezone":null,"value":"2026-07-24 06:23:02"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_56c87317e89c1f67f358c0b5"],"revision_id":"rev_2ed1383f-5c73-49bd-bd5b-f19deda1814c","work_id":"wrk_209cb693-e035-4d91-b959-d2a9b4df1de3"},{"date":{"precision":"instant","search_end":"2026-07-26","search_start":"2026-07-25","timezone":null,"value":"2026-07-25 22:39:55"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_f3d835eedb96fa210f6d9300","occ_d8178f3fc4b0c50773ddcc6e","occ_115b900fa5e042d14b5e92e2","occ_7ac80fd789510d2fd0b6854f"],"revision_id":"rev_00217178-ae5e-4e98-87ca-c35048af1ca4","work_id":"wrk_6f50f82d-878e-437d-81af-1073e3707af7"},{"date":{"precision":"instant","search_end":"2026-07-27","search_start":"2026-07-26","timezone":null,"value":"2026-07-26 00:17:26"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_52eae49a4f02ffd6dfd372e6","occ_47508df3bf1a9d02bdc3a6b6","occ_e6c60bbd64a7c24df1e1b474"],"revision_id":"rev_57995821-c471-42b5-ae3a-f989b7bd06f5","work_id":"wrk_ffa6d69b-69a7-4979-a51d-6b89aa7b0ff5"},{"date":{"precision":"instant","search_end":"2026-07-30","search_start":"2026-07-29","timezone":null,"value":"2026-07-29 23:58:25"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_42688c8d42bbfe36f26a94d0"],"revision_id":"rev_fd5eeabd-4433-491f-b463-c900cf908355","work_id":"wrk_d47ffc56-09dd-4f57-a300-be2346278872"},{"date":{"precision":"instant","search_end":"2026-07-31","search_start":"2026-07-30","timezone":null,"value":"2026-07-30 19:28:20"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_edeb5ed6c75582cca7bde973"],"revision_id":"rev_120b7405-aefc-4d5d-a68b-9713613e3e1f","work_id":"wrk_91a0ac80-8870-4ec9-b8a6-4081acad5334"},{"date":{"precision":"instant","search_end":"2026-07-31","search_start":"2026-07-30","timezone":null,"value":"2026-07-30 19:33:36"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_b6e0b171e1685765fc54ae11","occ_0552eb7e97e0da53b9fc549f","occ_d6b752b08ee7d4a34d07251f","occ_f19005d2358a9e61de32583e","occ_4859aca31cd43de9e75da2a3","occ_6e303f99a6f32c8d86caeab3","occ_9f0f18edc77639131f1a20dd","occ_071f5d283e00246eaa48c8b1"],"revision_id":"rev_716e29e3-ca33-4e0a-89d8-a4c146c91139","work_id":"wrk_6b2d4c96-c0d4-4d57-8d9e-9077ea538a79"},{"date":{"precision":"instant","search_end":"2026-07-31","search_start":"2026-07-30","timezone":null,"value":"2026-07-30 03:22:51"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_428f1e8b6271cf89f066a32e","occ_81c1664aa255a484cb16b4b7"],"revision_id":"rev_9c10c6f9-4055-4591-a368-74112c253a08","work_id":"wrk_a91b4c62-c6f7-43f1-8bc0-9224e28ea11a"},{"date":{"precision":"instant","search_end":"2026-08-01","search_start":"2026-07-31","timezone":"+09:00","value":"2026-07-31T11:07:53.000+09:00"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_97940617f3834f6e22a96111","occ_8199871809a532f45e40a830","occ_0a11f804c3e0ac476f32eb87","occ_b7ca902f3d4bbac974a8c0c8","occ_e645e768ffa607d6615cda54","occ_08964ceb6f539619cfd0b015","occ_4826ef00418292094966e14d","occ_285d6133609e85e581f3592b","occ_41531284c1ae6c2c22cacec5","occ_eec1aa36cf280511c693f93c","occ_8d49a44833f4873365a070f8","occ_8501d3b762f91e5ab63ccda1","occ_4dabe415b256956e16d5e5f0","occ_5e60d7a7fc3defd2e5527a09","occ_f10c0b27b7d6a3736210e8cd","occ_6f129a76a9c244e11293d6c8","occ_74fcd5d190a7388eb2074a02","occ_f71f7b8b71f89df8c7c72999"],"revision_id":"rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","work_id":"wrk_3d2e988e-cac0-4000-b005-9292af7ac61f"},{"date":{"precision":"instant","search_end":"2026-08-01","search_start":"2026-07-31","timezone":null,"value":"2026-07-31 07:30:01"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_857ab7649c35db908bc974e9","occ_dc18fc779e8138ffed2f3ff2","occ_5352c8dab880c4306d4726b4","occ_b7ec460c5dc854b438a272f7","occ_ab6329c11fb828ff50eaf0b2","occ_deeab118c5ed36fd866c8fa8","occ_3566de1ac03c958ad19d3844"],"revision_id":"rev_94f3d3d6-6b52-4f98-89a0-318fe428f5cb","work_id":"wrk_43737b34-e6d4-44d7-b140-1cd6aacb15fb"},{"date":{"precision":"instant","search_end":"2026-08-02","search_start":"2026-08-01","timezone":null,"value":"2026-08-01 15:36:29"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_3c90e7022a58316dd2b11cc6","occ_6bd2233a8c3160198eb6fea6","occ_af325729afc27bb13f9c38ab","occ_107db8af0b93fa86f27d6257","occ_228b62ae8e46f9e1ff080a42","occ_cb0051811fe2efc7c45e7c9f","occ_47e3dc4f6f76399978d1e81f","occ_9e87d5cc99fcf0f104670209","occ_9b20bbe4448897105ab227fe","occ_438abc6d90fb5249e8ca1a39","occ_b75b66c35935fccbbff5b286","occ_00055aa2719762edc150d35a","occ_5b87b264daab123e0e1d1e5e","occ_31474963b2835ec55e7829dd","occ_4216ce70a66feed283b1a836","occ_f482f22ed4782c028827d8ad","occ_eaf1662476d0965a69ebd630","occ_7a9b38fc617754cbb0cb2c04","occ_9d68d5bab75b0bcc3ceba988","occ_5ba9f054d03721a4e27ba768","occ_2668668a70b13abebdf248dc","occ_9967cfd45c24085ba429dd12","occ_d4fb6b5e5d910df150909c46","occ_1c39854d157f71ff05f1db1b","occ_07314a3a09c21e444772ea0a","occ_f06875b811944da3b38bb11d"],"revision_id":"rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","work_id":"wrk_eaab3954-2ccc-46a2-902f-fbbd679471b9"},{"date":{"precision":"day","search_end":"2026-08-04","search_start":"2026-08-03","timezone":null,"value":"2026-08-03"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_b2b30c5ecda8a4c60348101f","occ_9e9d0b9f254db8529f773893","occ_48f3ba659280115fd07d3992","occ_2b59b1dfc68e9212906011d8","occ_2a6f88bd7201eaee69900aa6","occ_bdd2e82e125fab99c43e1acf","occ_c7d85eb12242ddf5498d63f0","occ_0d184c88b71f560fe6ac183b","occ_c7086fa45e4fa582e6cd43bd","occ_a216e5f241bdd1c46ba2abf0","occ_cc93a2acfa8d2b2fc1b7e814","occ_566d9156b08bd83b6521ca1d","occ_bd10489b64b8031e1e34f5c6","occ_daa17445f1d5fc256d2d2150","occ_0d21f6685c9676e4118fa5c0","occ_fe0627828ba7531e1079bbd7","occ_47d895906d7fec1e50cb5575","occ_961513d8338a576336b170f8","occ_f1604385194446b87b55f97a","occ_b29349e3ceffc3b701f695f9"],"revision_id":"rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","work_id":"wrk_77df700f-3c41-42ae-ad85-1b2c4392d920"},{"date":{"precision":"instant","search_end":"2026-08-04","search_start":"2026-08-03","timezone":null,"value":"2026-08-03 16:28:45"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_21ec342b6c14b34e3ec98658"],"revision_id":"rev_d4189e74-ba90-4c4f-8495-9a2fccea5a2a","work_id":"wrk_ba0d2481-72cf-412b-92be-570c7d9c07e4"},{"date":{"precision":"instant","search_end":"2026-08-07","search_start":"2026-08-06","timezone":null,"value":"2026-08-06 05:35:18"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_d22a93fe249d3f300919ee71","occ_e4ae926793e32630f907f6e6","occ_87ce44186993797747290f8f","occ_0cdc991b6944d508e9b3f0fc","occ_9a6683fc7e24ae4d5ed730df","occ_f15e4ef5cc637c21d0da495c","occ_1b0dd79fbb6965e7919f1b95","occ_f657bde02da6e0de0595dafa","occ_c13fa2d5f34257ec336db5a9","occ_b97a045e67a8d7590e0fa57e","occ_d084accc3b64123cd73d8645"],"revision_id":"rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa","work_id":"wrk_82866dd9-73ac-495c-82f9-5873005352f3"},{"date":{"precision":"day","search_end":"2026-08-09","search_start":"2026-08-08","timezone":null,"value":"2026-08-08"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_f9f96935eed0fe222ae41acd"],"revision_id":"rev_c89081f7-eb04-4049-a4a4-68cbe681a3f5","work_id":"wrk_08ecb116-f4c7-4c99-9574-e485a5adb41f"},{"date":{"precision":"instant","search_end":"2026-08-19","search_start":"2026-08-18","timezone":null,"value":"2026-08-18 13:47:59"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_a7a8faf88ea741d81e36b288","occ_3e4f592a3e3edc62aae8afc4"],"revision_id":"rev_e1c0a212-5ad0-4fbc-81f6-60d2df49d753","work_id":"wrk_88c25067-a100-4eee-96f7-180120d2ac61"},{"date":{"precision":"instant","search_end":"2026-08-22","search_start":"2026-08-21","timezone":null,"value":"2026-08-21 14:45:06"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_7071d581e3e873b769d9ed79","occ_305902ce79ea6886510e4871","occ_726a31f786a43c0ac4af7c6e","occ_db89d7b6059b2d659aa8dcc5"],"revision_id":"rev_5bf9454f-9956-42b9-a29c-ae99782c33c0","work_id":"wrk_37743298-dcff-4252-a0d0-d94538fc2a68"},{"date":{"precision":"instant","search_end":"2026-08-23","search_start":"2026-08-22","timezone":null,"value":"2026-08-22 21:48:26"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_641c8ae9ba58d2235872377b","occ_f799aed8cd0f44d215fabebb","occ_5aea4232391d4fb51f1cfaf6","occ_945538d184c9cba5bcad56f1","occ_430414f055c905998ff6ea9f","occ_8dbd0c82a238708472edfe25","occ_7e68d34de0165694d5ca1640","occ_befcbf5757dd49d67bff6e08"],"revision_id":"rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e","work_id":"wrk_078157a8-8d73-48b6-b0ad-972882b9971f"},{"date":{"precision":"instant","search_end":"2026-08-24","search_start":"2026-08-23","timezone":null,"value":"2026-08-23 20:19:37"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_a48a78a42512d232fbc7fcc2","occ_0ba6aec74216c7d3574e5552"],"revision_id":"rev_f76185ef-dd8b-44e4-a82a-7b8cc8225b9a","work_id":"wrk_72f5f141-da8e-4f5b-a337-f46180c37f2e"},{"date":{"precision":"instant","search_end":"2026-08-26","search_start":"2026-08-25","timezone":null,"value":"2026-08-25 18:35:04"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_47ef544fbc7ffbccceb2030a"],"revision_id":"rev_e2c369be-a992-44a9-bbc1-81ed4928e89f","work_id":"wrk_d2f61eaa-5ee5-4d1f-b70d-7356a79ea61d"},{"date":{"precision":"instant","search_end":"2026-08-30","search_start":"2026-08-29","timezone":null,"value":"2026-08-29 20:33:34"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_a9d580c56dc8fa35309d5f3d","occ_7c960daa15e321651df7a999"],"revision_id":"rev_5f44b55a-c37f-42cf-ac49-298da914a87a","work_id":"wrk_57e07914-940b-4727-b215-7d77a2ac51c0"},{"date":{"precision":"instant","search_end":"2026-09-02","search_start":"2026-09-01","timezone":null,"value":"2026-09-01 14:51:02"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_d113e815a6ed5608a0346067"],"revision_id":"rev_20f6e19f-c172-4e63-83ab-04729ac80c82","work_id":"wrk_ae883f23-4159-4408-941d-addc931dbce4"},{"date":{"precision":"instant","search_end":"2026-09-05","search_start":"2026-09-04","timezone":null,"value":"2026-09-04 22:47:02"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_71bdcc5e031811a2696c337e","occ_60278044afdab30387620b31","occ_fc0fdf6b2077af0ac93e29c8","occ_bb3f52c4bedcc9eaa0f1b771","occ_06c5ea77946343c98997454b"],"revision_id":"rev_73f07a40-b588-4fa6-bf2b-2880462e214c","work_id":"wrk_83edd5a9-4c70-4085-ba5b-602f9d005928"},{"date":{"precision":"instant","search_end":"2026-09-07","search_start":"2026-09-06","timezone":null,"value":"2026-09-06 02:50:18"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_a6aa1f07100b404ea293ac40","occ_842fa91db9738661708fbd69","occ_fde479fbcfab603b14c8c9f7","occ_1f3d1b17fe2c2abc1b2f940f","occ_aff6beb651432112ffeb6bd5","occ_140133cc44be1caa68c3c42a","occ_84942d2777c75ae7b9fba04e","occ_a4df2a12aeea4a63bca79329","occ_b41e6298f3d88b27162193f3","occ_f05f770eae59be978b8933c1","occ_21dbc6ecf4cae950e83de17f","occ_a22222f03a097f6cc22f3b5e"],"revision_id":"rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","work_id":"wrk_7b4fd6ec-e3a4-4a64-b21f-d3e9dbb0ed11"},{"date":{"precision":"instant","search_end":"2026-09-07","search_start":"2026-09-06","timezone":null,"value":"2026-09-06 02:46:20"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_a13686462103c57e7f51f946","occ_6d9b2422bd887c9ec56a2777","occ_97d7a5f3887018082454ec67","occ_8c8e7a01b4ab35fabb89a54e","occ_236a90399815a8314534ca43","occ_dd3663cffa87c144dc7ca1b5","occ_acfb9197e6be2912dd5eff4e","occ_2595d236341b48d49194f6a6","occ_1e8006ccef4039e56efeb4c9","occ_ac60ddd8ee180aad00b25dbf"],"revision_id":"rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42","work_id":"wrk_675d7d8c-da38-4285-a483-5126d06bc835"},{"date":{"precision":"instant","search_end":"2026-09-07","search_start":"2026-09-06","timezone":null,"value":"2026-09-06 14:30:30"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_3c6c7bed04b0d05ca49aa037","occ_bf634e963d4b81fe03b579f6","occ_c912c2324333f472e4228bf0","occ_9fdfa6fda4b514bf03e5abe0","occ_054ee3905e74836867c40ca2","occ_9cad052a00e90065f76e65a5","occ_918dc07b87b81db7553b38cd","occ_d3fa3b06c62cf77a8eefe3e8"],"revision_id":"rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646","work_id":"wrk_4c390ee8-19a9-47a2-baff-32ac4e8e79ad"},{"date":{"precision":"instant","search_end":"2026-09-09","search_start":"2026-09-08","timezone":"+09:00","value":"2026-09-08T18:38:54.000+09:00"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_b2e55705d545205e3c511524","occ_951090f827a66d636c0bc990","occ_892672ea40d986fb3f4c3668","occ_c8361009cabf8548178152cf","occ_d368f8e851cc6756bdb1206b","occ_29d8531f962e43436d325651","occ_0927ffa859454142dc57d386","occ_242ff9ae5dcbc0a996318b62","occ_9ff1c9b1d32e6aac880d09ef","occ_e346c79df2dbe331bbb07b4b","occ_8fd1a87009043b36e32b4b39","occ_ea26b2f87182f5118daad36c","occ_d1fb78ce970040adcb1056ee","occ_cf172916f6d7f46a70781842","occ_c3bc283d1b21168554a401e4","occ_264f781e4c3c5a0e730cfcb1","occ_e133584641ee743c07e62cc4","occ_26f4f48b8b7d33e6359c03ac","occ_9e233bfbe6cbbca1fd8aa2d6","occ_3add396e0478701f429ec05a","occ_86a5200b1bcfe646a7b72ff5"],"revision_id":"rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","work_id":"wrk_0bf6ec28-040e-4940-a9e0-45eb805889bd"},{"date":{"precision":"instant","search_end":"2026-09-10","search_start":"2026-09-09","timezone":"+09:00","value":"2026-09-09T16:07:22.000+09:00"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_68f3ca9998b4bc16c722fe7b","occ_916ac3539e7390a7498ecda0"],"revision_id":"rev_c832f985-caf8-4ac7-bf58-f785e8782922","work_id":"wrk_37471fb9-72db-4a02-a976-c42f743def51"},{"date":{"precision":"instant","search_end":"2026-09-11","search_start":"2026-09-10","timezone":"+09:00","value":"2026-09-10T04:12:13.000+09:00"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_28114c06d30cd08d5d60565b","occ_f71449d8a9e1f1f6233c35e4","occ_7b20076f11d60ce98a7bdafc","occ_33b88553bc5e6f53661dd306","occ_4214df484ce76f04582f4846"],"revision_id":"rev_eaa7004e-c9e4-4bc4-9718-2e518decec70","work_id":"wrk_dfcbc670-103a-4931-8d23-d5f51671c61b"},{"date":{"precision":"instant","search_end":"2026-09-12","search_start":"2026-09-11","timezone":"+09:00","value":"2026-09-11T20:34:18.000+09:00"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_7cc54ff4370ca70eb968b584","occ_6b1dae22f3807c4af08bc1d6","occ_a65fe2867b60142509506dc8","occ_67b86dda66bb55dfa7f35d35"],"revision_id":"rev_0f7c606d-558f-4edb-99eb-4609a9fb0a95","work_id":"wrk_958dfab7-4e11-4e38-9d35-db36e1f1b2f7"},{"date":{"precision":"instant","search_end":"2026-09-14","search_start":"2026-09-13","timezone":"+09:00","value":"2026-09-13T19:26:49.000+09:00"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_87a91703efcb6bfdad9324e0","occ_fdc965970117cf944e09d529"],"revision_id":"rev_01c4b1d1-a854-4507-a2d7-bcf6b40fafd5","work_id":"wrk_dbe273d0-7a27-4a33-8051-93c6d0c21198"},{"date":{"precision":"instant","search_end":"2026-09-14","search_start":"2026-09-13","timezone":"+09:00","value":"2026-09-13T02:06:24.000+09:00"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_791f82086be44996e5a91c3e","occ_24030b03ff1fcc0481c18ffa","occ_25ba7398fa19f64e665e3816","occ_adfa506e822ec3a8a00e3950","occ_44ff2d0871339fc27f2b72bb","occ_213f7066612218d66a4e2c39","occ_b25e13adae8f77e7a204e90d","occ_7697307a36365c9f2479986e","occ_2a73d6e14e31d8dd19b714dc","occ_b951bbc036807a3c0d0d66b8"],"revision_id":"rev_c950677f-d85d-461a-bf3d-314cf6b08a37","work_id":"wrk_a1c11bb3-0dd1-46bc-9f3c-ec06fe1df011"},{"date":{"precision":"instant","search_end":"2026-09-14","search_start":"2026-09-13","timezone":"+09:00","value":"2026-09-13T08:33:21.000+09:00"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_3c2e811f54184c4bfea8934e","occ_6e9805971f2266ea7365848e","occ_9053a00fcf969f0d18121cf7","occ_0e32fb552776503681a783d5","occ_bc80cd0daddb82970ce865f7","occ_cd74394e379214986790e5f4","occ_04e16713a92c6be0acc2ae3f","occ_0a2c21caa879af5276c893d5","occ_df1641ce1d669af1edc5281f","occ_95aff9849d0fe35f05c24e89","occ_dda14b4ef8d99314ee806414","occ_293c991f1016f087bbd8cb04","occ_af413f415ada3ebc757433d1","occ_5a453586681717669b0007e5","occ_659b52a4204a18bb370e6626","occ_8ccd2886257ad8bee752abb1","occ_852a996f917bbf28baf791b2","occ_af701fb0603ac97834146181","occ_96faa0e4028255cbee038b21","occ_7a0a1e3b9a2fa4bc8b61ad31","occ_bf04dcb7a0f002f2ad65784f","occ_6a0b659f187212b698d43c02","occ_9939615a2041df715d6a938b","occ_8d18f0ffb8078d371d1006d7","occ_48b9b2320befd35952f75e83","occ_8d8d03c54c7e40005fde7eb3","occ_af082f28165196aee21fa808","occ_5ddaf6051e421755cd1e6f60","occ_4b544798d2658405d89bf220","occ_3451e8ac5ad091fec8281b9f","occ_f7c9800f3780c8a79c4502d7","occ_531a8c37fc6c3c9d28c5a9b5","occ_eda00bc4d0ef1630932e139b","occ_beb5167c9b41c5daa811ca93","occ_faf5238dce795fcbc5a151e5","occ_7e3b379d55f6115873885f3c","occ_d07923dbe54ff54b0f8060c8","occ_87a6c59df348df39476e6c94","occ_718217e616e052359d1b0c4f"],"revision_id":"rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","work_id":"wrk_ee8fdd8f-49bb-41f3-a20f-4cfb4669fe9a"},{"date":{"precision":"instant","search_end":"2026-09-16","search_start":"2026-09-15","timezone":"+09:00","value":"2026-09-15T01:29:06.000+09:00"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_64f2e52139cca0049a09a1d5"],"revision_id":"rev_bf97dda8-b55b-4e4d-b1cf-b775474af9b2","work_id":"wrk_72fede57-1e40-4a22-81c2-67b2328ba457"},{"date":{"precision":"instant","search_end":"2026-09-22","search_start":"2026-09-21","timezone":"+09:00","value":"2026-09-21T14:06:22.000+09:00"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_30cef7c63806f2db97554024"],"revision_id":"rev_f45cfd6f-625d-4e49-b987-25f79284501d","work_id":"wrk_28a55046-0758-4fb4-b7b9-5ec8fa7681a7"},{"date":{"precision":"instant","search_end":"2026-09-24","search_start":"2026-09-23","timezone":"+09:00","value":"2026-09-23T06:22:02.000+09:00"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_b26059060da4a21f3d913624","occ_0a8495fd831d048c99a93d97","occ_d5e834500471d26ea0f0c2b6","occ_d3f8a3d546e789a6a160bc3e","occ_ed003de2cd4e01e50a1a14ae","occ_ba5a673e071bdc2d5ce78422","occ_70e2ec026ed8a10aeb9049a7","occ_29e79378b18a3c3346e30d0e","occ_8b9fbb33c7cea16e2e65732e","occ_97185306b75579b71ab7a86b","occ_53249cc02380a29d419660ea","occ_f0b7bb617e4ef07455d4908f","occ_545f4e3858ca58910d6abba3","occ_7716e84a72131c6eb031642c","occ_9a9868c56ccd513807e87a81","occ_0b71671fbfb861b1c932fa1d","occ_7fb8f15d39356469d03218cb","occ_455d7c6d294be0ff04819844","occ_37594ba1eaaaf279efc728fb","occ_8f9cfbea354ac132c3a7b408","occ_ff3dde1240e6607a2aeb16a6","occ_b474d766d85f43b3cb03695c","occ_9d04d46a9e541aaadbb6217c"],"revision_id":"rev_2e5b712c-b120-4b45-91dd-33e6758666be","work_id":"wrk_c9d1ffd9-bc82-481f-b5ce-3dbca5dfe331"},{"date":{"precision":"instant","search_end":"2026-09-24","search_start":"2026-09-23","timezone":"+09:00","value":"2026-09-23T18:34:18.000+09:00"},"date_kind":"source_date","is_distribution_representative":true,"occurrence_ids":["occ_d598ca1df6e7cebe537586a2","occ_31da7960e520617565f80bbe","occ_da7301e76cc5a17e47c13a0d","occ_b16e27b735dfa55792dde41b","occ_ba15b15a8da0b67eedb132db","occ_8d5e8d271a33e0aba12ae7ee","occ_50237c3d6174af64c3a8c3aa","occ_e037bc58023cb136079b52e8","occ_39962f08f600b8666f5d1517"],"revision_id":"rev_e5e24c6f-5887-4f16-b03e-081da5340883","work_id":"wrk_470cca01-eadf-46c3-bd50-cd79e3b1e197"}],"earliest_revision_ids":["rev_d8e8cb66-632c-4170-afae-461911315dbd"],"interpretation":"Oldest candidates by the selected metadata axis within this Snapshot, using currently saved bodies; not author-history first use.","undated":[]},"aggregates_by_layer":{"bibliography":{"month":{"cells":[{"denominator":2,"mention_count":0,"period":"2022-07","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2022-12","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-01","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-02","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-06","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-11","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-04","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-10","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":9,"mention_count":0,"period":"2025-01","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":6,"mention_count":0,"period":"2025-02","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":4,"mention_count":0,"period":"2025-03","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2025-05","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2025-06","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":4,"mention_count":0,"period":"2025-07","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-08","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":6,"mention_count":0,"period":"2025-09","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":12,"mention_count":0,"period":"2025-10","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-11","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-12","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":7,"mention_count":0,"period":"2026-01","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2026-02","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":3,"mention_count":0,"period":"2026-03","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":15,"mention_count":0,"period":"2026-04","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":18,"mention_count":0,"period":"2026-05","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":50,"mention_count":0,"period":"2026-06","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":51,"mention_count":0,"period":"2026-07","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":40,"mention_count":0,"period":"2026-08","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":30,"mention_count":0,"period":"2026-09","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":0,"mention_count":0,"period":null,"revision_ids":[],"section_count":0,"share":null,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0}],"periods":["2022-07","2022-12","2023-01","2023-02","2023-06","2023-11","2024-04","2024-10","2025-01","2025-02","2025-03","2025-05","2025-06","2025-07","2025-08","2025-09","2025-10","2025-11","2025-12","2026-01","2026-02","2026-03","2026-04","2026-05","2026-06","2026-07","2026-08","2026-09"],"undated_or_insufficient_precision_work_count":0,"unknown_date_work_count":0},"quarter":{"cells":[{"denominator":2,"mention_count":0,"period":"2022-Q3","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2022-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2023-Q1","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":19,"mention_count":0,"period":"2025-Q1","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":12,"mention_count":0,"period":"2025-Q3","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":16,"mention_count":0,"period":"2025-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":11,"mention_count":0,"period":"2026-Q1","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":83,"mention_count":0,"period":"2026-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":121,"mention_count":0,"period":"2026-Q3","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":0,"mention_count":0,"period":null,"revision_ids":[],"section_count":0,"share":null,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0}],"periods":["2022-Q3","2022-Q4","2023-Q1","2023-Q2","2023-Q4","2024-Q2","2024-Q4","2025-Q1","2025-Q2","2025-Q3","2025-Q4","2026-Q1","2026-Q2","2026-Q3"],"undated_or_insufficient_precision_work_count":0,"unknown_date_work_count":0},"year":{"cells":[{"denominator":3,"mention_count":0,"period":"2022","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":4,"mention_count":0,"period":"2023","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2024","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":49,"mention_count":0,"period":"2025","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":215,"mention_count":0,"period":"2026","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":0,"mention_count":0,"period":null,"revision_ids":[],"section_count":0,"share":null,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0}],"periods":["2022","2023","2024","2025","2026"],"undated_or_insufficient_precision_work_count":0,"unknown_date_work_count":0}},"body":{"month":{"cells":[{"denominator":2,"mention_count":0,"period":"2022-07","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2022-12","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-01","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-02","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-06","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-11","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-04","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-10","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":9,"mention_count":1,"period":"2025-01","revision_ids":["rev_d8e8cb66-632c-4170-afae-461911315dbd"],"section_count":1,"share":0.1111111111111111,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":1},{"denominator":6,"mention_count":0,"period":"2025-02","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":4,"mention_count":0,"period":"2025-03","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2025-05","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2025-06","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":4,"mention_count":11,"period":"2025-07","revision_ids":["rev_f17ef624-e9fd-49a4-b061-966093172de4"],"section_count":2,"share":0.25,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":1},{"denominator":2,"mention_count":0,"period":"2025-08","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":6,"mention_count":0,"period":"2025-09","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":12,"mention_count":0,"period":"2025-10","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-11","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-12","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":7,"mention_count":7,"period":"2026-01","revision_ids":["rev_0ee327e2-ecfe-42e6-b337-fce8abbe7b71","rev_9ae4d311-bf43-4bb1-a7b5-36a5649efc7c"],"section_count":5,"share":0.2857142857142857,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":2},{"denominator":1,"mention_count":1,"period":"2026-02","revision_ids":["rev_b8a0343f-f76e-47f4-8c47-4536da944126"],"section_count":1,"share":1.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":1},{"denominator":3,"mention_count":0,"period":"2026-03","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":15,"mention_count":169,"period":"2026-04","revision_ids":["rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","rev_57810a5f-315e-49a7-ac85-72cc27e35eff","rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b","rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","rev_7cf42fcc-1956-4694-a0b6-5eaa9be5c284","rev_9af30ed5-ec26-4204-9f65-58324ee5fba5","rev_b228986a-43e7-4e2b-8b16-758283d367bc","rev_f487aa08-2568-42ad-85cb-11c258b5b71d"],"section_count":92,"share":0.5333333333333333,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":8},{"denominator":18,"mention_count":230,"period":"2026-05","revision_ids":["rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f","rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","rev_3621f98d-f1b9-4e97-9c8a-8b259cfd31c3","rev_49a8bdbd-8ee8-4c28-8af4-da1787c47955","rev_62acaca2-fec9-4724-be08-6ebd3c061f93","rev_6abb0df2-b5a4-4f1c-8f48-a346d6d0f247","rev_6cb909ec-1291-4cbb-a768-f72123829d08","rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","rev_98f85497-697b-4917-aabc-df1485131d9b","rev_da6063d5-59b2-4519-9c42-45d8f4f206c8","rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e"],"section_count":132,"share":0.6666666666666666,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":12},{"denominator":50,"mention_count":375,"period":"2026-06","revision_ids":["rev_11f7264b-855b-41d8-b760-87dbfa630e9f","rev_12eaaf24-14fe-4445-936b-b81ef492d19d","rev_15381314-c653-4ce9-9b25-d8495cdbf4d1","rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","rev_53605de4-06cc-4e0b-9bc9-21f393408c0a","rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b","rev_5cb295dd-df59-4d28-9799-42065461c569","rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","rev_78657238-b40f-4a7c-b670-325df3688f39","rev_90fb7b96-f10b-43df-a612-0c23e8a50740","rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a","rev_92ff5158-6c34-4526-ab02-c8db8332951f","rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8","rev_977092a4-68df-47ac-8de2-4e66567c8a95","rev_a3e10132-4b92-414d-8744-acbeb6f0b294","rev_a4008bc2-ae2c-4bd7-8615-0b3c4e0936ff","rev_a749fcdf-8155-46e9-a466-83e867886a08","rev_abe6b561-f51c-4549-a97f-aab2b725492e","rev_ae55c104-d140-4064-adf1-726fae29a7f3","rev_aea0fd30-a496-42e0-975e-3de7b0cd01ab","rev_afc88027-978f-428a-a597-dd5d129dc441","rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a","rev_c1530075-b9e3-4144-b8ce-f6e480c778fa","rev_ca87695f-4cf9-40a1-bc61-6a123a71ffe6","rev_cf107e9a-1286-4df3-bd1f-adb77c070354","rev_d47a57d8-f3a6-4505-9ea6-c03ae4416616","rev_dbe69fb8-f76d-44e6-b111-aa27e548a204","rev_ddc53551-0396-485b-a72e-8b3d5126293f","rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280","rev_ffbdfa80-0e8b-40b5-8765-ba1e5b977d47"],"section_count":175,"share":0.62,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":31},{"denominator":51,"mention_count":79,"period":"2026-07","revision_ids":["rev_00217178-ae5e-4e98-87ca-c35048af1ca4","rev_120b7405-aefc-4d5d-a68b-9713613e3e1f","rev_2ed1383f-5c73-49bd-bd5b-f19deda1814c","rev_44c73f5c-647d-467e-8aa2-527eb6352a8a","rev_54b154f3-ea9f-45e9-b38b-a16c54ca3e2b","rev_57995821-c471-42b5-ae3a-f989b7bd06f5","rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","rev_716e29e3-ca33-4e0a-89d8-a4c146c91139","rev_94f3d3d6-6b52-4f98-89a0-318fe428f5cb","rev_9c10c6f9-4055-4591-a368-74112c253a08","rev_a30fdebb-6a5f-4491-8480-bc05cbb21f96","rev_c8b19a4c-93eb-4695-8f38-c8166d22cbd9","rev_f3a5d7c9-6dfa-4c38-8bf2-a37063801332","rev_fd5eeabd-4433-491f-b463-c900cf908355"],"section_count":55,"share":0.29411764705882354,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":15},{"denominator":40,"mention_count":85,"period":"2026-08","revision_ids":["rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e","rev_5bf9454f-9956-42b9-a29c-ae99782c33c0","rev_5f44b55a-c37f-42cf-ac49-298da914a87a","rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa","rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","rev_c89081f7-eb04-4049-a4a4-68cbe681a3f5","rev_d4189e74-ba90-4c4f-8495-9a2fccea5a2a","rev_e1c0a212-5ad0-4fbc-81f6-60d2df49d753","rev_e2c369be-a992-44a9-bbc1-81ed4928e89f","rev_f76185ef-dd8b-44e4-a82a-7b8cc8225b9a"],"section_count":60,"share":0.275,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":11},{"denominator":30,"mention_count":159,"period":"2026-09","revision_ids":["rev_01c4b1d1-a854-4507-a2d7-bcf6b40fafd5","rev_0f7c606d-558f-4edb-99eb-4609a9fb0a95","rev_20f6e19f-c172-4e63-83ab-04729ac80c82","rev_2e5b712c-b120-4b45-91dd-33e6758666be","rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","rev_73f07a40-b588-4fa6-bf2b-2880462e214c","rev_bf97dda8-b55b-4e4d-b1cf-b775474af9b2","rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42","rev_c832f985-caf8-4ac7-bf58-f785e8782922","rev_c950677f-d85d-461a-bf3d-314cf6b08a37","rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646","rev_e5e24c6f-5887-4f16-b03e-081da5340883","rev_eaa7004e-c9e4-4bc4-9718-2e518decec70","rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","rev_f45cfd6f-625d-4e49-b987-25f79284501d","rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604"],"section_count":115,"share":0.5333333333333333,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":16},{"denominator":0,"mention_count":0,"period":null,"revision_ids":[],"section_count":0,"share":null,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0}],"periods":["2022-07","2022-12","2023-01","2023-02","2023-06","2023-11","2024-04","2024-10","2025-01","2025-02","2025-03","2025-05","2025-06","2025-07","2025-08","2025-09","2025-10","2025-11","2025-12","2026-01","2026-02","2026-03","2026-04","2026-05","2026-06","2026-07","2026-08","2026-09"],"undated_or_insufficient_precision_work_count":0,"unknown_date_work_count":0},"quarter":{"cells":[{"denominator":2,"mention_count":0,"period":"2022-Q3","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2022-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2023-Q1","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":19,"mention_count":1,"period":"2025-Q1","revision_ids":["rev_d8e8cb66-632c-4170-afae-461911315dbd"],"section_count":1,"share":0.05263157894736842,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":1},{"denominator":2,"mention_count":0,"period":"2025-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":12,"mention_count":11,"period":"2025-Q3","revision_ids":["rev_f17ef624-e9fd-49a4-b061-966093172de4"],"section_count":2,"share":0.08333333333333333,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":1},{"denominator":16,"mention_count":0,"period":"2025-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":11,"mention_count":8,"period":"2026-Q1","revision_ids":["rev_0ee327e2-ecfe-42e6-b337-fce8abbe7b71","rev_9ae4d311-bf43-4bb1-a7b5-36a5649efc7c","rev_b8a0343f-f76e-47f4-8c47-4536da944126"],"section_count":6,"share":0.2727272727272727,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":3},{"denominator":83,"mention_count":774,"period":"2026-Q2","revision_ids":["rev_11f7264b-855b-41d8-b760-87dbfa630e9f","rev_12eaaf24-14fe-4445-936b-b81ef492d19d","rev_15381314-c653-4ce9-9b25-d8495cdbf4d1","rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f","rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","rev_3621f98d-f1b9-4e97-9c8a-8b259cfd31c3","rev_49a8bdbd-8ee8-4c28-8af4-da1787c47955","rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","rev_53605de4-06cc-4e0b-9bc9-21f393408c0a","rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b","rev_57810a5f-315e-49a7-ac85-72cc27e35eff","rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b","rev_5cb295dd-df59-4d28-9799-42065461c569","rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","rev_62acaca2-fec9-4724-be08-6ebd3c061f93","rev_6abb0df2-b5a4-4f1c-8f48-a346d6d0f247","rev_6cb909ec-1291-4cbb-a768-f72123829d08","rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","rev_78657238-b40f-4a7c-b670-325df3688f39","rev_7cf42fcc-1956-4694-a0b6-5eaa9be5c284","rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","rev_90fb7b96-f10b-43df-a612-0c23e8a50740","rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a","rev_92ff5158-6c34-4526-ab02-c8db8332951f","rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8","rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","rev_977092a4-68df-47ac-8de2-4e66567c8a95","rev_98f85497-697b-4917-aabc-df1485131d9b","rev_9af30ed5-ec26-4204-9f65-58324ee5fba5","rev_a3e10132-4b92-414d-8744-acbeb6f0b294","rev_a4008bc2-ae2c-4bd7-8615-0b3c4e0936ff","rev_a749fcdf-8155-46e9-a466-83e867886a08","rev_abe6b561-f51c-4549-a97f-aab2b725492e","rev_ae55c104-d140-4064-adf1-726fae29a7f3","rev_aea0fd30-a496-42e0-975e-3de7b0cd01ab","rev_afc88027-978f-428a-a597-dd5d129dc441","rev_b228986a-43e7-4e2b-8b16-758283d367bc","rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a","rev_c1530075-b9e3-4144-b8ce-f6e480c778fa","rev_ca87695f-4cf9-40a1-bc61-6a123a71ffe6","rev_cf107e9a-1286-4df3-bd1f-adb77c070354","rev_d47a57d8-f3a6-4505-9ea6-c03ae4416616","rev_da6063d5-59b2-4519-9c42-45d8f4f206c8","rev_dbe69fb8-f76d-44e6-b111-aa27e548a204","rev_ddc53551-0396-485b-a72e-8b3d5126293f","rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e","rev_f487aa08-2568-42ad-85cb-11c258b5b71d","rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280","rev_ffbdfa80-0e8b-40b5-8765-ba1e5b977d47"],"section_count":399,"share":0.6144578313253012,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":51},{"denominator":121,"mention_count":323,"period":"2026-Q3","revision_ids":["rev_00217178-ae5e-4e98-87ca-c35048af1ca4","rev_01c4b1d1-a854-4507-a2d7-bcf6b40fafd5","rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e","rev_0f7c606d-558f-4edb-99eb-4609a9fb0a95","rev_120b7405-aefc-4d5d-a68b-9713613e3e1f","rev_20f6e19f-c172-4e63-83ab-04729ac80c82","rev_2e5b712c-b120-4b45-91dd-33e6758666be","rev_2ed1383f-5c73-49bd-bd5b-f19deda1814c","rev_44c73f5c-647d-467e-8aa2-527eb6352a8a","rev_54b154f3-ea9f-45e9-b38b-a16c54ca3e2b","rev_57995821-c471-42b5-ae3a-f989b7bd06f5","rev_5bf9454f-9956-42b9-a29c-ae99782c33c0","rev_5f44b55a-c37f-42cf-ac49-298da914a87a","rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","rev_716e29e3-ca33-4e0a-89d8-a4c146c91139","rev_73f07a40-b588-4fa6-bf2b-2880462e214c","rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa","rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","rev_94f3d3d6-6b52-4f98-89a0-318fe428f5cb","rev_9c10c6f9-4055-4591-a368-74112c253a08","rev_a30fdebb-6a5f-4491-8480-bc05cbb21f96","rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","rev_bf97dda8-b55b-4e4d-b1cf-b775474af9b2","rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42","rev_c832f985-caf8-4ac7-bf58-f785e8782922","rev_c89081f7-eb04-4049-a4a4-68cbe681a3f5","rev_c8b19a4c-93eb-4695-8f38-c8166d22cbd9","rev_c950677f-d85d-461a-bf3d-314cf6b08a37","rev_d4189e74-ba90-4c4f-8495-9a2fccea5a2a","rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646","rev_e1c0a212-5ad0-4fbc-81f6-60d2df49d753","rev_e2c369be-a992-44a9-bbc1-81ed4928e89f","rev_e5e24c6f-5887-4f16-b03e-081da5340883","rev_eaa7004e-c9e4-4bc4-9718-2e518decec70","rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","rev_f3a5d7c9-6dfa-4c38-8bf2-a37063801332","rev_f45cfd6f-625d-4e49-b987-25f79284501d","rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","rev_f76185ef-dd8b-44e4-a82a-7b8cc8225b9a","rev_fd5eeabd-4433-491f-b463-c900cf908355"],"section_count":230,"share":0.34710743801652894,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":42},{"denominator":0,"mention_count":0,"period":null,"revision_ids":[],"section_count":0,"share":null,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0}],"periods":["2022-Q3","2022-Q4","2023-Q1","2023-Q2","2023-Q4","2024-Q2","2024-Q4","2025-Q1","2025-Q2","2025-Q3","2025-Q4","2026-Q1","2026-Q2","2026-Q3"],"undated_or_insufficient_precision_work_count":0,"unknown_date_work_count":0},"year":{"cells":[{"denominator":3,"mention_count":0,"period":"2022","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":4,"mention_count":0,"period":"2023","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2024","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":49,"mention_count":12,"period":"2025","revision_ids":["rev_d8e8cb66-632c-4170-afae-461911315dbd","rev_f17ef624-e9fd-49a4-b061-966093172de4"],"section_count":3,"share":0.04081632653061224,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":2},{"denominator":215,"mention_count":1105,"period":"2026","revision_ids":["rev_00217178-ae5e-4e98-87ca-c35048af1ca4","rev_01c4b1d1-a854-4507-a2d7-bcf6b40fafd5","rev_0d613fb6-d01b-48f0-8a64-b54a137ad32e","rev_0ee327e2-ecfe-42e6-b337-fce8abbe7b71","rev_0f7c606d-558f-4edb-99eb-4609a9fb0a95","rev_11f7264b-855b-41d8-b760-87dbfa630e9f","rev_120b7405-aefc-4d5d-a68b-9713613e3e1f","rev_12eaaf24-14fe-4445-936b-b81ef492d19d","rev_15381314-c653-4ce9-9b25-d8495cdbf4d1","rev_17f0f08f-aee1-41a1-8c07-82badf2b6a4f","rev_19d3384e-c9c8-4b8b-95ae-a78bbc4dfb8d","rev_20f6e19f-c172-4e63-83ab-04729ac80c82","rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","rev_2e5b712c-b120-4b45-91dd-33e6758666be","rev_2ed1383f-5c73-49bd-bd5b-f19deda1814c","rev_3621f98d-f1b9-4e97-9c8a-8b259cfd31c3","rev_44c73f5c-647d-467e-8aa2-527eb6352a8a","rev_49a8bdbd-8ee8-4c28-8af4-da1787c47955","rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","rev_53605de4-06cc-4e0b-9bc9-21f393408c0a","rev_54b154f3-ea9f-45e9-b38b-a16c54ca3e2b","rev_5603783d-df89-4ab1-ba77-c8a24ace4a7b","rev_57810a5f-315e-49a7-ac85-72cc27e35eff","rev_57995821-c471-42b5-ae3a-f989b7bd06f5","rev_584d62bb-c9f7-4d1f-b637-8fac93976f1b","rev_5bf9454f-9956-42b9-a29c-ae99782c33c0","rev_5cb295dd-df59-4d28-9799-42065461c569","rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","rev_5f44b55a-c37f-42cf-ac49-298da914a87a","rev_62acaca2-fec9-4724-be08-6ebd3c061f93","rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","rev_6abb0df2-b5a4-4f1c-8f48-a346d6d0f247","rev_6cb909ec-1291-4cbb-a768-f72123829d08","rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","rev_716e29e3-ca33-4e0a-89d8-a4c146c91139","rev_73f07a40-b588-4fa6-bf2b-2880462e214c","rev_749a1ec0-4ee6-4025-8b2d-a10bb57af52e","rev_78657238-b40f-4a7c-b670-325df3688f39","rev_7cf42fcc-1956-4694-a0b6-5eaa9be5c284","rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","rev_8eb03ce9-cad9-4ba7-91d6-90dffd2302fa","rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","rev_90fb7b96-f10b-43df-a612-0c23e8a50740","rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a","rev_92ff5158-6c34-4526-ab02-c8db8332951f","rev_9369ec68-2723-4ee0-81f8-c835d3d59ac8","rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","rev_94f3d3d6-6b52-4f98-89a0-318fe428f5cb","rev_977092a4-68df-47ac-8de2-4e66567c8a95","rev_98f85497-697b-4917-aabc-df1485131d9b","rev_9ae4d311-bf43-4bb1-a7b5-36a5649efc7c","rev_9af30ed5-ec26-4204-9f65-58324ee5fba5","rev_9c10c6f9-4055-4591-a368-74112c253a08","rev_a30fdebb-6a5f-4491-8480-bc05cbb21f96","rev_a3e10132-4b92-414d-8744-acbeb6f0b294","rev_a4008bc2-ae2c-4bd7-8615-0b3c4e0936ff","rev_a749fcdf-8155-46e9-a466-83e867886a08","rev_abe6b561-f51c-4549-a97f-aab2b725492e","rev_ae55c104-d140-4064-adf1-726fae29a7f3","rev_aea0fd30-a496-42e0-975e-3de7b0cd01ab","rev_afc88027-978f-428a-a597-dd5d129dc441","rev_b228986a-43e7-4e2b-8b16-758283d367bc","rev_b8a0343f-f76e-47f4-8c47-4536da944126","rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","rev_be83f7eb-e1d1-464a-ab3f-1463976bb30a","rev_bf97dda8-b55b-4e4d-b1cf-b775474af9b2","rev_c1530075-b9e3-4144-b8ce-f6e480c778fa","rev_c52b79dd-ede0-41f3-8d00-4ec44801ef42","rev_c832f985-caf8-4ac7-bf58-f785e8782922","rev_c89081f7-eb04-4049-a4a4-68cbe681a3f5","rev_c8b19a4c-93eb-4695-8f38-c8166d22cbd9","rev_c950677f-d85d-461a-bf3d-314cf6b08a37","rev_ca87695f-4cf9-40a1-bc61-6a123a71ffe6","rev_cf107e9a-1286-4df3-bd1f-adb77c070354","rev_d4189e74-ba90-4c4f-8495-9a2fccea5a2a","rev_d47a57d8-f3a6-4505-9ea6-c03ae4416616","rev_da6063d5-59b2-4519-9c42-45d8f4f206c8","rev_dbe69fb8-f76d-44e6-b111-aa27e548a204","rev_ddc53551-0396-485b-a72e-8b3d5126293f","rev_df2ff134-3d7b-48d0-82b5-c341fa1dc646","rev_e1c0a212-5ad0-4fbc-81f6-60d2df49d753","rev_e2c369be-a992-44a9-bbc1-81ed4928e89f","rev_e5e24c6f-5887-4f16-b03e-081da5340883","rev_eaa7004e-c9e4-4bc4-9718-2e518decec70","rev_ead2aeb8-4b54-444a-83a1-bbc8187c760e","rev_eb92ff59-a729-4aea-ac3f-9cbb667b2dd4","rev_f3a5d7c9-6dfa-4c38-8bf2-a37063801332","rev_f45cfd6f-625d-4e49-b987-25f79284501d","rev_f487aa08-2568-42ad-85cb-11c258b5b71d","rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604","rev_f76185ef-dd8b-44e4-a82a-7b8cc8225b9a","rev_fd4d1c88-70bf-4d6a-88e1-27b35ef0d280","rev_fd5eeabd-4433-491f-b463-c900cf908355","rev_ffbdfa80-0e8b-40b5-8765-ba1e5b977d47"],"section_count":635,"share":0.44651162790697674,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":96},{"denominator":0,"mention_count":0,"period":null,"revision_ids":[],"section_count":0,"share":null,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0}],"periods":["2022","2023","2024","2025","2026"],"undated_or_insufficient_precision_work_count":0,"unknown_date_work_count":0}},"caption":{"month":{"cells":[{"denominator":2,"mention_count":0,"period":"2022-07","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2022-12","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-01","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-02","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-06","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-11","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-04","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-10","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":9,"mention_count":0,"period":"2025-01","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":6,"mention_count":0,"period":"2025-02","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":4,"mention_count":0,"period":"2025-03","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2025-05","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2025-06","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":4,"mention_count":0,"period":"2025-07","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-08","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":6,"mention_count":0,"period":"2025-09","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":12,"mention_count":0,"period":"2025-10","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-11","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-12","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":7,"mention_count":0,"period":"2026-01","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2026-02","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":3,"mention_count":0,"period":"2026-03","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":15,"mention_count":0,"period":"2026-04","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":18,"mention_count":0,"period":"2026-05","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":50,"mention_count":0,"period":"2026-06","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":51,"mention_count":0,"period":"2026-07","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":40,"mention_count":0,"period":"2026-08","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":30,"mention_count":1,"period":"2026-09","revision_ids":["rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604"],"section_count":1,"share":0.03333333333333333,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":1},{"denominator":0,"mention_count":0,"period":null,"revision_ids":[],"section_count":0,"share":null,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0}],"periods":["2022-07","2022-12","2023-01","2023-02","2023-06","2023-11","2024-04","2024-10","2025-01","2025-02","2025-03","2025-05","2025-06","2025-07","2025-08","2025-09","2025-10","2025-11","2025-12","2026-01","2026-02","2026-03","2026-04","2026-05","2026-06","2026-07","2026-08","2026-09"],"undated_or_insufficient_precision_work_count":0,"unknown_date_work_count":0},"quarter":{"cells":[{"denominator":2,"mention_count":0,"period":"2022-Q3","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2022-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2023-Q1","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":19,"mention_count":0,"period":"2025-Q1","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":12,"mention_count":0,"period":"2025-Q3","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":16,"mention_count":0,"period":"2025-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":11,"mention_count":0,"period":"2026-Q1","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":83,"mention_count":0,"period":"2026-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":121,"mention_count":1,"period":"2026-Q3","revision_ids":["rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604"],"section_count":1,"share":0.008264462809917356,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":1},{"denominator":0,"mention_count":0,"period":null,"revision_ids":[],"section_count":0,"share":null,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0}],"periods":["2022-Q3","2022-Q4","2023-Q1","2023-Q2","2023-Q4","2024-Q2","2024-Q4","2025-Q1","2025-Q2","2025-Q3","2025-Q4","2026-Q1","2026-Q2","2026-Q3"],"undated_or_insufficient_precision_work_count":0,"unknown_date_work_count":0},"year":{"cells":[{"denominator":3,"mention_count":0,"period":"2022","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":4,"mention_count":0,"period":"2023","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2024","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":49,"mention_count":0,"period":"2025","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":215,"mention_count":1,"period":"2026","revision_ids":["rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604"],"section_count":1,"share":0.004651162790697674,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":1},{"denominator":0,"mention_count":0,"period":null,"revision_ids":[],"section_count":0,"share":null,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0}],"periods":["2022","2023","2024","2025","2026"],"undated_or_insufficient_precision_work_count":0,"unknown_date_work_count":0}},"character":{"month":{"cells":[{"denominator":2,"mention_count":0,"period":"2022-07","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2022-12","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-01","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-02","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-06","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-11","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-04","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-10","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":9,"mention_count":0,"period":"2025-01","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":6,"mention_count":0,"period":"2025-02","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":4,"mention_count":0,"period":"2025-03","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2025-05","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2025-06","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":4,"mention_count":0,"period":"2025-07","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-08","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":6,"mention_count":0,"period":"2025-09","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":12,"mention_count":0,"period":"2025-10","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-11","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-12","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":7,"mention_count":0,"period":"2026-01","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2026-02","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":3,"mention_count":0,"period":"2026-03","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":15,"mention_count":0,"period":"2026-04","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":18,"mention_count":0,"period":"2026-05","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":50,"mention_count":6,"period":"2026-06","revision_ids":["rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","rev_dbe69fb8-f76d-44e6-b111-aa27e548a204","rev_ddc53551-0396-485b-a72e-8b3d5126293f"],"section_count":4,"share":0.08,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":4},{"denominator":51,"mention_count":2,"period":"2026-07","revision_ids":["rev_17972011-3c15-4377-a2ea-f36af8fa6c2f","rev_65271768-1b9c-4fbd-a486-d503b37eaf9a"],"section_count":2,"share":0.0392156862745098,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":2},{"denominator":40,"mention_count":1,"period":"2026-08","revision_ids":["rev_bcca5c5e-94f7-4bf9-b249-187454f09a22"],"section_count":1,"share":0.025,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":1},{"denominator":30,"mention_count":0,"period":"2026-09","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":0,"mention_count":0,"period":null,"revision_ids":[],"section_count":0,"share":null,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0}],"periods":["2022-07","2022-12","2023-01","2023-02","2023-06","2023-11","2024-04","2024-10","2025-01","2025-02","2025-03","2025-05","2025-06","2025-07","2025-08","2025-09","2025-10","2025-11","2025-12","2026-01","2026-02","2026-03","2026-04","2026-05","2026-06","2026-07","2026-08","2026-09"],"undated_or_insufficient_precision_work_count":0,"unknown_date_work_count":0},"quarter":{"cells":[{"denominator":2,"mention_count":0,"period":"2022-Q3","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2022-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2023-Q1","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":19,"mention_count":0,"period":"2025-Q1","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":12,"mention_count":0,"period":"2025-Q3","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":16,"mention_count":0,"period":"2025-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":11,"mention_count":0,"period":"2026-Q1","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":83,"mention_count":6,"period":"2026-Q2","revision_ids":["rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","rev_dbe69fb8-f76d-44e6-b111-aa27e548a204","rev_ddc53551-0396-485b-a72e-8b3d5126293f"],"section_count":4,"share":0.04819277108433735,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":4},{"denominator":121,"mention_count":3,"period":"2026-Q3","revision_ids":["rev_17972011-3c15-4377-a2ea-f36af8fa6c2f","rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","rev_bcca5c5e-94f7-4bf9-b249-187454f09a22"],"section_count":3,"share":0.024793388429752067,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":3},{"denominator":0,"mention_count":0,"period":null,"revision_ids":[],"section_count":0,"share":null,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0}],"periods":["2022-Q3","2022-Q4","2023-Q1","2023-Q2","2023-Q4","2024-Q2","2024-Q4","2025-Q1","2025-Q2","2025-Q3","2025-Q4","2026-Q1","2026-Q2","2026-Q3"],"undated_or_insufficient_precision_work_count":0,"unknown_date_work_count":0},"year":{"cells":[{"denominator":3,"mention_count":0,"period":"2022","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":4,"mention_count":0,"period":"2023","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2024","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":49,"mention_count":0,"period":"2025","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":215,"mention_count":9,"period":"2026","revision_ids":["rev_17972011-3c15-4377-a2ea-f36af8fa6c2f","rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","rev_65271768-1b9c-4fbd-a486-d503b37eaf9a","rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","rev_bcca5c5e-94f7-4bf9-b249-187454f09a22","rev_dbe69fb8-f76d-44e6-b111-aa27e548a204","rev_ddc53551-0396-485b-a72e-8b3d5126293f"],"section_count":7,"share":0.03255813953488372,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":7},{"denominator":0,"mention_count":0,"period":null,"revision_ids":[],"section_count":0,"share":null,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0}],"periods":["2022","2023","2024","2025","2026"],"undated_or_insufficient_precision_work_count":0,"unknown_date_work_count":0}},"code":{"month":{"cells":[{"denominator":2,"mention_count":0,"period":"2022-07","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2022-12","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-01","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-02","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-06","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-11","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-04","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-10","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":9,"mention_count":0,"period":"2025-01","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":6,"mention_count":0,"period":"2025-02","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":4,"mention_count":0,"period":"2025-03","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2025-05","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2025-06","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":4,"mention_count":0,"period":"2025-07","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-08","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":6,"mention_count":0,"period":"2025-09","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":12,"mention_count":0,"period":"2025-10","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-11","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-12","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":7,"mention_count":0,"period":"2026-01","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2026-02","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":3,"mention_count":0,"period":"2026-03","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":15,"mention_count":0,"period":"2026-04","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":18,"mention_count":28,"period":"2026-05","revision_ids":["rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","rev_3621f98d-f1b9-4e97-9c8a-8b259cfd31c3","rev_6cb909ec-1291-4cbb-a768-f72123829d08","rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4"],"section_count":26,"share":0.2222222222222222,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":4},{"denominator":50,"mention_count":18,"period":"2026-06","revision_ids":["rev_5cb295dd-df59-4d28-9799-42065461c569","rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a","rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","rev_cf107e9a-1286-4df3-bd1f-adb77c070354","rev_ddc53551-0396-485b-a72e-8b3d5126293f"],"section_count":15,"share":0.12,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":6},{"denominator":51,"mention_count":5,"period":"2026-07","revision_ids":["rev_120b7405-aefc-4d5d-a68b-9713613e3e1f","rev_716e29e3-ca33-4e0a-89d8-a4c146c91139"],"section_count":4,"share":0.0392156862745098,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":2},{"denominator":40,"mention_count":4,"period":"2026-08","revision_ids":["rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","rev_d4189e74-ba90-4c4f-8495-9a2fccea5a2a"],"section_count":3,"share":0.05,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":2},{"denominator":30,"mention_count":0,"period":"2026-09","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":0,"mention_count":0,"period":null,"revision_ids":[],"section_count":0,"share":null,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0}],"periods":["2022-07","2022-12","2023-01","2023-02","2023-06","2023-11","2024-04","2024-10","2025-01","2025-02","2025-03","2025-05","2025-06","2025-07","2025-08","2025-09","2025-10","2025-11","2025-12","2026-01","2026-02","2026-03","2026-04","2026-05","2026-06","2026-07","2026-08","2026-09"],"undated_or_insufficient_precision_work_count":0,"unknown_date_work_count":0},"quarter":{"cells":[{"denominator":2,"mention_count":0,"period":"2022-Q3","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2022-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2023-Q1","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":19,"mention_count":0,"period":"2025-Q1","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":12,"mention_count":0,"period":"2025-Q3","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":16,"mention_count":0,"period":"2025-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":11,"mention_count":0,"period":"2026-Q1","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":83,"mention_count":46,"period":"2026-Q2","revision_ids":["rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","rev_3621f98d-f1b9-4e97-9c8a-8b259cfd31c3","rev_5cb295dd-df59-4d28-9799-42065461c569","rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","rev_6cb909ec-1291-4cbb-a768-f72123829d08","rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a","rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","rev_cf107e9a-1286-4df3-bd1f-adb77c070354","rev_ddc53551-0396-485b-a72e-8b3d5126293f"],"section_count":41,"share":0.12048192771084337,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":10},{"denominator":121,"mention_count":9,"period":"2026-Q3","revision_ids":["rev_120b7405-aefc-4d5d-a68b-9713613e3e1f","rev_716e29e3-ca33-4e0a-89d8-a4c146c91139","rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","rev_d4189e74-ba90-4c4f-8495-9a2fccea5a2a"],"section_count":7,"share":0.03305785123966942,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":4},{"denominator":0,"mention_count":0,"period":null,"revision_ids":[],"section_count":0,"share":null,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0}],"periods":["2022-Q3","2022-Q4","2023-Q1","2023-Q2","2023-Q4","2024-Q2","2024-Q4","2025-Q1","2025-Q2","2025-Q3","2025-Q4","2026-Q1","2026-Q2","2026-Q3"],"undated_or_insufficient_precision_work_count":0,"unknown_date_work_count":0},"year":{"cells":[{"denominator":3,"mention_count":0,"period":"2022","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":4,"mention_count":0,"period":"2023","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2024","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":49,"mention_count":0,"period":"2025","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":215,"mention_count":55,"period":"2026","revision_ids":["rev_120b7405-aefc-4d5d-a68b-9713613e3e1f","rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","rev_3621f98d-f1b9-4e97-9c8a-8b259cfd31c3","rev_5cb295dd-df59-4d28-9799-42065461c569","rev_5dbe090a-744f-4659-8ee9-8fbc7195e238","rev_6cb909ec-1291-4cbb-a768-f72123829d08","rev_716e29e3-ca33-4e0a-89d8-a4c146c91139","rev_8f8ffd4e-00d7-47a6-9906-ce0fa86b3835","rev_91468364-fe9d-4ab5-a6dc-f203e89bc64a","rev_93e994ee-fc71-4a2c-868a-e0b16e4b39b4","rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","rev_cf107e9a-1286-4df3-bd1f-adb77c070354","rev_d4189e74-ba90-4c4f-8495-9a2fccea5a2a","rev_ddc53551-0396-485b-a72e-8b3d5126293f"],"section_count":48,"share":0.06511627906976744,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":14},{"denominator":0,"mention_count":0,"period":null,"revision_ids":[],"section_count":0,"share":null,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0}],"periods":["2022","2023","2024","2025","2026"],"undated_or_insufficient_precision_work_count":0,"unknown_date_work_count":0}},"math":{"month":{"cells":[{"denominator":2,"mention_count":0,"period":"2022-07","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2022-12","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-01","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-02","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-06","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-11","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-04","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-10","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":9,"mention_count":0,"period":"2025-01","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":6,"mention_count":0,"period":"2025-02","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":4,"mention_count":0,"period":"2025-03","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2025-05","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2025-06","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":4,"mention_count":0,"period":"2025-07","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-08","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":6,"mention_count":0,"period":"2025-09","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":12,"mention_count":0,"period":"2025-10","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-11","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-12","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":7,"mention_count":0,"period":"2026-01","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2026-02","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":3,"mention_count":0,"period":"2026-03","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":15,"mention_count":0,"period":"2026-04","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":18,"mention_count":0,"period":"2026-05","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":50,"mention_count":0,"period":"2026-06","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":51,"mention_count":0,"period":"2026-07","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":40,"mention_count":0,"period":"2026-08","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":30,"mention_count":0,"period":"2026-09","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":0,"mention_count":0,"period":null,"revision_ids":[],"section_count":0,"share":null,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0}],"periods":["2022-07","2022-12","2023-01","2023-02","2023-06","2023-11","2024-04","2024-10","2025-01","2025-02","2025-03","2025-05","2025-06","2025-07","2025-08","2025-09","2025-10","2025-11","2025-12","2026-01","2026-02","2026-03","2026-04","2026-05","2026-06","2026-07","2026-08","2026-09"],"undated_or_insufficient_precision_work_count":0,"unknown_date_work_count":0},"quarter":{"cells":[{"denominator":2,"mention_count":0,"period":"2022-Q3","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2022-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2023-Q1","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":19,"mention_count":0,"period":"2025-Q1","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":12,"mention_count":0,"period":"2025-Q3","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":16,"mention_count":0,"period":"2025-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":11,"mention_count":0,"period":"2026-Q1","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":83,"mention_count":0,"period":"2026-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":121,"mention_count":0,"period":"2026-Q3","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":0,"mention_count":0,"period":null,"revision_ids":[],"section_count":0,"share":null,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0}],"periods":["2022-Q3","2022-Q4","2023-Q1","2023-Q2","2023-Q4","2024-Q2","2024-Q4","2025-Q1","2025-Q2","2025-Q3","2025-Q4","2026-Q1","2026-Q2","2026-Q3"],"undated_or_insufficient_precision_work_count":0,"unknown_date_work_count":0},"year":{"cells":[{"denominator":3,"mention_count":0,"period":"2022","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":4,"mention_count":0,"period":"2023","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2024","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":49,"mention_count":0,"period":"2025","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":215,"mention_count":0,"period":"2026","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":0,"mention_count":0,"period":null,"revision_ids":[],"section_count":0,"share":null,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0}],"periods":["2022","2023","2024","2025","2026"],"undated_or_insufficient_precision_work_count":0,"unknown_date_work_count":0}},"title":{"month":{"cells":[{"denominator":2,"mention_count":0,"period":"2022-07","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2022-12","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-01","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-02","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-06","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-11","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-04","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-10","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":9,"mention_count":0,"period":"2025-01","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":6,"mention_count":0,"period":"2025-02","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":4,"mention_count":0,"period":"2025-03","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2025-05","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2025-06","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":4,"mention_count":0,"period":"2025-07","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-08","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":6,"mention_count":0,"period":"2025-09","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":12,"mention_count":0,"period":"2025-10","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-11","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-12","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":7,"mention_count":0,"period":"2026-01","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2026-02","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":3,"mention_count":0,"period":"2026-03","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":15,"mention_count":1,"period":"2026-04","revision_ids":["rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6"],"section_count":0,"share":0.06666666666666667,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":1},{"denominator":18,"mention_count":2,"period":"2026-05","revision_ids":["rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608"],"section_count":0,"share":0.1111111111111111,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":2},{"denominator":50,"mention_count":3,"period":"2026-06","revision_ids":["rev_90fb7b96-f10b-43df-a612-0c23e8a50740","rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","rev_ddc53551-0396-485b-a72e-8b3d5126293f"],"section_count":0,"share":0.06,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":3},{"denominator":51,"mention_count":1,"period":"2026-07","revision_ids":["rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982"],"section_count":0,"share":0.0196078431372549,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":1},{"denominator":40,"mention_count":0,"period":"2026-08","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":30,"mention_count":2,"period":"2026-09","revision_ids":["rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604"],"section_count":0,"share":0.06666666666666667,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":2},{"denominator":0,"mention_count":0,"period":null,"revision_ids":[],"section_count":0,"share":null,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0}],"periods":["2022-07","2022-12","2023-01","2023-02","2023-06","2023-11","2024-04","2024-10","2025-01","2025-02","2025-03","2025-05","2025-06","2025-07","2025-08","2025-09","2025-10","2025-11","2025-12","2026-01","2026-02","2026-03","2026-04","2026-05","2026-06","2026-07","2026-08","2026-09"],"undated_or_insufficient_precision_work_count":0,"unknown_date_work_count":0},"quarter":{"cells":[{"denominator":2,"mention_count":0,"period":"2022-Q3","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2022-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2023-Q1","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2023-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":1,"mention_count":0,"period":"2024-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":19,"mention_count":0,"period":"2025-Q1","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2025-Q2","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":12,"mention_count":0,"period":"2025-Q3","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":16,"mention_count":0,"period":"2025-Q4","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":11,"mention_count":0,"period":"2026-Q1","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":83,"mention_count":6,"period":"2026-Q2","revision_ids":["rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","rev_90fb7b96-f10b-43df-a612-0c23e8a50740","rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","rev_ddc53551-0396-485b-a72e-8b3d5126293f"],"section_count":0,"share":0.07228915662650602,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":6},{"denominator":121,"mention_count":3,"period":"2026-Q3","revision_ids":["rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604"],"section_count":0,"share":0.024793388429752067,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":3},{"denominator":0,"mention_count":0,"period":null,"revision_ids":[],"section_count":0,"share":null,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0}],"periods":["2022-Q3","2022-Q4","2023-Q1","2023-Q2","2023-Q4","2024-Q2","2024-Q4","2025-Q1","2025-Q2","2025-Q3","2025-Q4","2026-Q1","2026-Q2","2026-Q3"],"undated_or_insufficient_precision_work_count":0,"unknown_date_work_count":0},"year":{"cells":[{"denominator":3,"mention_count":0,"period":"2022","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":4,"mention_count":0,"period":"2023","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":2,"mention_count":0,"period":"2024","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":49,"mention_count":0,"period":"2025","revision_ids":[],"section_count":0,"share":0.0,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0},{"denominator":215,"mention_count":9,"period":"2026","revision_ids":["rev_2cf94359-bc71-4b68-8fd3-9c93c45ac3eb","rev_513cc5f0-bfa3-418a-9ecc-cd67eca8bae6","rev_66a2ef4a-c9ce-47fa-ab34-8488c1c4a0f1","rev_6e3e8ec6-6ae2-401c-8fc1-8950e3d50982","rev_8c9a8f7a-e00e-4b6d-b58d-ba3e89e02608","rev_90fb7b96-f10b-43df-a612-0c23e8a50740","rev_bb9a9bde-af83-422a-9558-c75e6cb1da78","rev_ddc53551-0396-485b-a72e-8b3d5126293f","rev_f70f17fc-ae46-4fff-8f02-0ab2953c4604"],"section_count":0,"share":0.04186046511627907,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":9},{"denominator":0,"mention_count":0,"period":null,"revision_ids":[],"section_count":0,"share":null,"topic_id":"top_1b05f039-7a18-4428-98ec-ee653655374e","work_count":0}],"periods":["2022","2023","2024","2025","2026"],"undated_or_insufficient_precision_work_count":0,"unknown_date_work_count":0}}},"layers":{"body":"本文","character":"絶ノイア・Sil-Kathnaの章","title":"記事タイトル","caption":"画像の代替テキスト・キャプション","bibliography":"参考文献","code":"コード・テキスト図","math":"数式"}}