
What happened
Glavis Architects joined a Fujieda City demonstration under the Ministry of Internal Affairs and Communications' trusted-AI program; SoftBank builds the knowledge-search system pairing SB Intuitions' domestic AI "Sarashina" with RAG for counter services.
Why it matters
The trial tests whether domestic generative AI is useful for administrative work, while Glavis handles organizing and systematizing government knowledge, drawing on its Public AI Knowledge Research Group.
What to watch
The aim of a nationwide shared framework hinges on whether the knowledge rules carry across municipalities; the body gives no date or funding figure, so watch what the demonstration finds.
WHO IT HITSThis lands on municipal government staff who must check laws, notices and manuals to make case-by-case decisions at service counters, and on vendors building domestic AI for the public sector.
Summaries like this, in your inbox every morning.
The trial sits inside the Ministry of Internal Affairs and Communications' "Development and Demonstration Project for Trustworthy AI in the Public Sector." SoftBank acts as the service developer, Fujieda City serves as the user organization and demonstration field, and SB Intuitions supplies the domestic generative AI "Sarashina." Glavis Architects' role is narrower than the system-building itself: it brings the thinking of the "Public AI Knowledge Base" examined by the Public AI Knowledge Research Group, whose chairman is Osamu Sudo, professor emeritus at the University of Tokyo, and whose senior adviser is Jiro Kokuryo, professor emeritus at Keio University.
The body frames the problem as one of reference material, not model power. Counter staff must check laws, notices and manuals and make decisions suited to each case, so administrative knowledge has to be correct, current, legitimate and continuously updated for AI to use it well. Glavis says it will examine how to organize that material so generative AI can reference it properly while keeping up with institutional changes.
Looking further out, the release notes that municipalities expect falling populations and shrinking staff, and it argues that systematizing administrative knowledge and making it available to staff through generative AI could ease workload and stabilize service quality. The stated long-term goal is not a separate system per municipality but a shared way of organizing knowledge that can be used across local governments. Whether that cross-municipality idea holds up likely depends on what the Fujieda demonstration actually shows.
For example, today's edition would include:
AI-summarized, only the topics you pick: one digest a day via Email, LINE, or Slack.
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.
Applied Materials plans to invest US$5b in India over the next decade, building a 140 acre semiconductor resea…

The Motley Fool's comparison concluded ASML is the better buy over Qualcomm, citing its EUV machine monopoly a…

Microsoft launched a redesigned Copilot super app combining chat, coding tools, and a new Autopilot agentic fe…

In a Sept. 27, 2026 Motley Fool video, Rick Orford compared GE Vernova and Bloom Energy as two routes to the A…

In a CNBC interview, Berkshire CEO Greg Abel said energy scarcity is holding back AI, and data centers are a "…

Investor's Business Daily says AI memory chip stocks Micron Technology (MU), Sandisk (SNDK), SK Hynix (SKHY)…
