Adopt

AI Agent Use Cases by Business Function, from Japanese Public Sources

・ Employee Store Operations

Summary

Which tasks are governments and companies using AI agents and generative AI for? This article sorts, by business function, the uses described in Japanese government white papers and case studies published by Japan's Ministry of Internal Affairs and Communications (MIC). It also covers what to watch for when reading case studies and how to apply them to your own company.

Find AI employees

The examples in this article come from public documents checked on October 2, 2026. Each example and its results are described only as far as the source document states. Adoption conditions differ by organization, so check the sources at the end for details.

A search for AI agent use cases turns up plenty of information. Some of it, however, has unclear sources or leaves out the conditions. This article covers only examples found in white papers and MIC's published materials.

Reading case studies: check the source and the conditions

How to read a case study depends on what stage it describes. Even the examples in white papers mix full-scale operations, proof-of-concept trials and future plans. The White Paper on Information and Communications notes that, due to space and other constraints, it cannot cover every example.

Read case studies differently by stage

What stage is the example at

In full-scale operationEasy to use as a referenceCheck the time, cost and team it took
At the proof-of-concept stageUse as a reference with conditionsProduction may not give the same result
A future planUse only as a sense of directionNot yet a result
  • Source: a public document, a company announcement or a news report. White papers also quote company announcements and news reports
  • Conditions: size of the organization, tools used, cost, and time from planning to adoption
  • How impact was measured: actual measurement, a survey or an estimate

Measurement methods differ even within the same document. For example, in the Kosai City case, the result that the work of drafting answers for the city council dropped to one-third comes from a staff survey. In the Yamagata City case, the figure is a projection that simply converts the number of consultations handled by generative AI into labor costs. Neither is the same as measured work time.

Uses described in government public documents

According to MIC's White Paper on Information and Communications (2026 edition), Japan's Digital Agency plans to start a large-scale trial of “Gennai,” a generative AI environment for government employees, for staff at all ministries and agencies in fiscal 2026, and to begin full use in fiscal 2027.

How local governments use AI can be read in the local government DX case studies on MIC's Regional DX Portal. Most aim to make staff work more efficient. The cases classified as generative AI use are as follows.

  • Tobetsu Town, Hokkaido: creates meeting minutes by combining a transcription tool with generative AI summaries. The town says work that used to take about four times the meeting length dropped to one-quarter
  • Kosai City, Shizuoka: introduced an app that drafts specifications for IT procurement, cutting drafting time to about one-fifth. It also uses purpose-specific chats for tasks such as drafting answers for the city council
  • Sanyo-Onoda City, Yamaguchi: co-developed with a private company a generative AI app that runs on the government-only network (LGWAN). The app references city ordinances and council minutes to draft answers and other documents
  • Nisshin City, Aichi: uses generative AI to write greetings and social media posts and to create Excel macros. The city built its own prompt collection and case collection so staff can learn how to ask step by step
  • Yamagata City, Yamagata: runs a 24-hour LINE consultation service that combines generative AI with specialist staff

In its “Guidebook on AI Use and Adoption in Local Governments: Adoption Procedures” (4th edition, published December 16, 2025), MIC presents how to proceed with adoption and examples of generative AI use. Examples cited in the white paper also include services for residents.

  • Chiba Prefecture: a generative AI chatbot guides people to the right welfare consultation desk based on what they type. At the desk, consultation audio is transcribed with personal information masked automatically, and staff check it before it is used for summaries
  • Nishiawakura Village, Okayama: feeds residents' opinions gathered at workshops into generative AI, which summarizes trends, strengths and weaknesses by field for use in discussions
  • Nagasaki Prefecture: when users enter their schedule and places they want to visit, generative AI suggests a model sightseeing route

Kosai City's purpose-specific chats are built with My GPTs, ChatGPT's customization feature. OpenAI plans to retire custom GPTs on December 11, 2026 (February 11, 2027 for Enterprise customers approved for an extension). The roadmap of the tools in use is also one of the conditions to consider when reading a case study.

Find AI employees

How companies use AI, according to white papers

The White Paper on Information and Communications (2026 edition) groups AI use cases at Japanese companies by business area. Examples involving AI agents include the following.

  • Rohto Pharmaceutical: announced it would start a proof-of-concept trial in December 2025 that combines several AI agents, such as for procurement and inventory management, aiming to optimize its supply chain
  • Daikin Industries: developed in-house a “skilled technician AI agent” that checks work videos for missed steps in inspecting and repairing air conditioners. Together with Hitachi, it has also developed and introduced a “failure diagnosis AI agent” that answers whether an equipment anomaly is a failure
  • Kirin Holdings: developed an “AI board member” trained on past meeting minutes and other records, which people submitting proposals to management meetings use to test their ideas in advance
  • Chugai Pharmaceutical: is rethinking how research itself is done using generative AI, centered on “AI drug discovery”
  • Panasonic Holdings: automates and speeds up design evaluation and shape decisions with “design AI” that combines simulation technology and AI

The White Paper on Small and Medium Enterprises (2026 edition) also includes small business examples using generative AI: Opto Science, which imports and sells optical equipment in Shinjuku, Tokyo, and Okada Kenma, which machines and assembles parts in Kanazawa, Ishikawa. Their details are covered in the breakdown by function below.

By function: sales, accounting, inquiries and development

Sales

Meiji Yasuda Life Insurance provides an AI-powered “digital secretary” to nearly all employees, supporting tasks such as writing meeting records. The white paper notes that the company envisions AI agents helping create proposals based on information from past meetings. At Opto Science, sales staff use generative AI to summarize overseas research papers and translate English manuals, filling gaps in technical knowledge. The company also uses it to choose products for ads and draft taglines, and says it always fact-checks the information generative AI provides.

Accounting and back office

In December 2025, RAKUS began offering a beta expense reimbursement feature in which an AI agent creates expense data when the user simply selects receipts, and also automates entries such as account codes. It also has a feature that checks whether the resulting claim follows internal rules. At Okada Kenma, a non-engineer senior managing director built apps by conversing with generative AI and combined inspection records, inventory, labor management, costing and more into one management system.

Inquiries and customer support

Gen-AX, a SoftBank subsidiary, is developing “X-Ghost,” which handles phone inquiries 24 hours a day, 365 days a year, and goes as far as looking up and registering data in systems. It combines an AI that handles the call with an AI that monitors speech and system activity. AEON has introduced an “AI assistant” for employees, trained on store operation manuals, at some stores. Among local governments, Kobe City is running a trial in which a voice bot answers tax-related calls and transfers them to staff when it cannot answer.

Development

NTT DATA says it is working on “AI-native development,” in which one AI writes requirements documents, another AI writes the code, and people focus on review and overall management. ITOCHU Techno-Solutions offers “Acsim,” in which AI supports creating business flows and documents for requirements definition. The white paper says prototyping requirements that used to take more than a month can be cut to an hour.

Usage patterns by business function

Sales

  • Drafting records and materials
  • Research and translation
  • Proposal writing still at the planning stage

Accounting and back office

  • Creating expense data
  • Checking against internal rules
  • Building in-house apps

Inquiries

  • 24-hour first response
  • Hand off to people when it cannot answer
  • Manual search for employees

Development

  • Support for requirements definition
  • Writing code
  • People review and manage

Across functions, a common pattern is AI handling drafts or first responses while people keep checks and decisions. The range of work you can hand to AI agents is set out in What AI agents can do. Use in sales is covered in detail in AI agents for sales.

Applying case studies to your company

Copying a case study as is will not necessarily give the same result. Write down how your conditions differ before you try it.

Applying a case study to your company
  1. 1Pick 1 taskTime-consuming work with set steps
  2. 2Find a similar caseSame task, similar scale
  3. 3Write down the differencesData, tools, team, cost
  4. 4Test smallLimit the scope and period
  5. 5Measure and decideJudge on accuracy and impact
  1. Pick one task: choose work that takes time and has fairly fixed steps
  2. Find a similar case: look for cases in the same task with a similar scale and team
  3. Write down the differences: compare available data, tools, internal team and budget
  4. Test small: limit the trial to specific departments and a set period
  5. Measure and decide: compare work time and output accuracy before and after, then decide whether to continue

Some descriptions also help with how to proceed. Tobetsu Town formed a small task force of staff who were already proficient with ChatGPT to test Microsoft 365 Copilot. In interviews for the White Paper on Information and Communications, Chugai Pharmaceutical named strong commitment from senior management as one reason its AI use was going relatively well.

The Tobetsu Town and Kosai City cases also list the time from planning to adoption and the annual cost. Compare figures like these before you run your own trial. How to measure is explained in How to evaluate AI agents.

Browse by function on Employee Store

Employee Store is a marketplace where companies can adopt AI agents (AI employees) built by developers. Categories are sales support, recruiting and scouting, social media management, customer support, back office, marketing, and development and data. You can find AI employees to adopt in the category closest to the work in these case studies. AI agents on offer are listed in AI employees. Prices are a one-time purchase, a monthly fee, or a setup fee plus a monthly fee.

FAQ

Where can I find AI agent adoption case studies?
Japanese government sources include MIC's White Paper on Information and Communications, the Small and Medium Enterprise Agency's White Paper on Small and Medium Enterprises, and the local government DX case studies on MIC's Regional DX Portal. For each case, check whether it is in full-scale operation, a trial or a plan.
Are local government cases useful for private companies?
Cases for work shared with the private sector, such as meeting minutes, document drafts and handling inquiries, are useful. However, local governments sometimes work under their own conditions, such as a government-only network. Write down the differences in conditions before applying them.
If I use the same tool as a case study, will I get the same results?
Not necessarily. Results depend on the data you handle, your work procedures and how familiar users are with the tool. Run a trial with a limited scope and period, and decide based on numbers you measure yourself.

About the author

Employee Store OperationsThe operations team behind Employee Store, a marketplace for AI agents. We check tool features and pricing against official sources and list them at the end of each article. If you spot an error, please let us know via the contact form.

Sources

Related articles

Ask AI

Ask AI if it fits your work.

Use your usual AI to explore what Employee Store offers and what to check before buying.

Ask ChatGPTAsk ClaudeAsk GeminiAsk Perplexity

Opens an external AI service. Confirm pricing and deliverables on the listing page.