
What happened
The Digital Agency published a video walkthrough of its 行政実務用RAG apps on the ガバメントAI 源内 platform, including 文書管理質問応答-RAG, which answered that information posted on websites or social media counts as an administrative document.
Why it matters
Because answers also show the file name and page number referenced, staff can check the source themselves, easing the load on internal help desks and reducing reliance on a few knowledgeable people, though the agency warns AI answers may contain errors and final checks must be done by the user.
Summaries like this, every morning.
The video is framed around a familiar administrative pain point: manuals, guidelines and reports are numerous and lengthy, so confirming a procedure or its basis takes time and knowledge ends up concentrated in a few people. The Digital Agency positions RAG as the fix, since registering those documents as external data lets the AI locate, organize, compare and summarize the relevant material from a natural-language question.
Two examples illustrate the range. EASY質問応答-RAG covers the manual and FAQ for EASY, the electronic system for managing administrative documents, and walks a user through drafting a decision document; its answers display the referenced file name and page number so the user can compare against the original. 文書管理質問応答-RAG, built for internal Digital Agency use, applies the agency's own document-management rules and, in the shown exchange, concluded that information posted on websites or social media is an administrative document, then described how it should be recorded in the file management ledger, including an indefinite retention period and the website as storage location.
The agency presents self-service answers as a way to reduce the burden on inquiry desks and to prevent knowledge from depending on specific staff. It also describes an arrangement in which ministries and agencies request development and supply the documents the AI should reference, with the resulting apps generally published on ガバメントAI 源内 when they are expected to serve multiple organizations. Alongside the walkthrough, the agency points readers to separate explainers on SEABIS travel and reimbursement procedures and on legal research with Lawsy.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
OpenAI announced GPT-6 Cyber on November 5, 2026, a model built for cybersecurity research that finds vulnerab…

Japan's AI Law, in force since September 1, 2025, carries no penalties; the country instead leans on soft-law…

Shimmochi runs a Claude Code "commander" agent that rebuilds his task list from GitHub and Notion, decides the…

Mikami of Canly's store-acquisition team used aidlc-workflows' bugfix scope on a Google Business Profile CSV b…

Hitachi Energy is expanding its partnership with Celonis, the process-intelligence leader, to scale AI use acr…

Spesia added an "AI consultation on data and measures" chat to Flowgent, its Amazon operations analysis system
