Japan starts adding task-running agents to its Gennai government platform
Japan’s Digital Agency is beginning a phased rollout of agent chat within Gennai, its AI platform for government staff. The programme covers approximately 180,000 employees, with a more advanced application-building environment targeted for December. New governance and security standards assign responsibilities to users, administrators and developers. An open-source release for local authorities and businesses is planned for around February 2027.
Artificial Intelligence··Evening
Agent chat starts a phased rollout
Japan’s Digital Agency is beginning to add task-running AI agents to Gennai, its generative AI platform for government employees. The rollout announced on October 20 covers approximately 180,000 staff. Agent Chat is being introduced to ministries and other government bodies in stages from that day. It allows employees to use natural-language conversation to create and run agents that perform several tasks. The agency identifies work involving multiple steps, large collections of documents and specialized investigation or analysis as intended uses. Its programme aims to let staff adapt tools to their own work rather than rely only on prebuilt applications.[1], [2]
An application-building workshop is planned for December
A second environment, Gennai Workshop, is intended for building more advanced agents and AI applications. The Digital Agency targets introduction by around December. Project Design’s account of Digital Minister Toshiharu Furukawa’s briefing describes an environment usable by staff without specialist knowledge. The official announcement describes functions corresponding to a coding agent: generating and modifying code, operating on files, executing commands and running tests in response to natural-language instructions. The two environments have different introduction schedules. Agent Chat begins its phased rollout now, while the workshop remains a planned addition. The agency links staff-led improvement of these tools with its administrative reform programme.[1], [2]
Safety standards assign responsibilities by role
The agency has also published Gennai governance, security and safety standards. They specify responsibilities and requirements separately for users, administrators and developers. Japan’s AI Safety Institute, which works on AI safety evaluation methods and standards, helped prepare them. The agency says more advanced functions and a wider range of executable operations require rules governing how government staff use agents. It also plans to release Agent Chat and Gennai Workshop around February 2027 as Gennai open-source software version 2.0. The announced open-source release would allow use, modification and redistribution by local authorities, private businesses and other users beyond the central government.[1], [2]
