
What happened
NTT-AT will start offering "RelAi 窓口アシスタント" on October 15, a generative AI service that transcribes and summarizes counter conversations and drafts answers from internal manuals, FAQs and ordinances, with cited source documents.
Why it matters
The service is positioned to reduce the effort staff spend hunting for information and assembling replies, smoothing variation in service quality caused by differences in experience and knowledge, and it runs in a dedicated environment where input data is not used for AI training.
What to watch
The service starts on October 15, with deployment support and document preparation handled by NTT-AT's language-processing and AI specialists.
WHO IT HITSMunicipal counter staff and the administrators overseeing ward offices are the main audience, since the tool targets their day-to-day resident interactions and back-office record-keeping. NTT-AT's specialists also take on document tuning and ongoing updates, which may matter to the IT and operations teams that would otherwise handle that work.
Summaries like this, in your inbox every morning.
The service is built on findings from a counter-AI pilot conducted at Katsushika Ward. According to NTT-AT, feedback from municipal staff and issues in counter operations were reflected in the product, with improvements to the user interface and operating methods as well as added functions for real-world use. That makes the Katsushika work the evident origin of the commercial offering, though the body does not detail what specifically changed.
A notable feature set separates this from a generic chat tool. Voice recognition runs in real time and summarizes conversations, speaker separation uses NTT-AT's own high-directivity compact microphone "FR-1200", and personal information such as names and addresses is automatically masked in the transcript. Answers are generated from ward knowledge such as manuals, FAQs and ordinances, with the supporting document shown and the relevant passage highlighted, and the system also records and summarizes interactions and creates response records.
The commercial version adds functions aimed at two kinds of gaps the pilot may have surfaced. Translation addresses multilingual needs for communication with foreign residents, and a customer-harassment alert lets staff press a button on a smartphone to notify managers of possible harassment, who can then check the conversation record to understand the situation and the visitor's demands. Together with deployment in a dedicated environment not used for AI training and support for closed networks such as LGWAN, the package appears aimed at the security requirements municipal buyers typically impose.
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.
A UK AI small-cap screen highlighted Beeks Financial Cloud Group, whose Market Edge Intelligence platform appl…

A Columbia University estimate puts cumulative AI infrastructure investment and spending at around $9 trillion…

After Character.AI filed for bankruptcy this week, a legal filing shows Disney agreed to offer jobs to at leas…

The Telegraph reported that Microsoft withdrew from a UK data center build with British AI cloud startup Nscal…

Anthropic has onboarded six METR employees, giving them company hardware, office space and internal tools roug…

Anthropic's internal study shows junior new hires in technical departments fell from 27 percent to 9 percent s…
