AI Agent Adoption Rate in Japan: What Public Surveys Show

The figures in this article were checked on October 2, 2026, against materials published by Japan's Ministry of Internal Affairs and Communications (MIC), the Information-technology Promotion Agency, Japan (IPA) and Japan's Small and Medium Enterprise Agency. Survey periods and respondents are given in the text. Check the sources at the end for the latest editions.

As of October 2026, few public statistics measure the adoption rate of AI agents alone. Most surveys ask about generative AI or AI in general. Among them, the IPA survey asks about AI agents by department. This article looks separately at AI agent figures and at the generative AI figures that underpin them.

Before the numbers: how survey definitions differ

AI usage rates change depending on who was asked, what was asked and when. The three surveys used in this article differ as follows.

The MIC survey selects respondents from an online research company's panel. It also covers only people working at companies that have already started digitalization. So it cannot be read as the share for Japanese companies as a whole. Whenever you use any of these figures, state who was surveyed alongside it.

Corporate generative AI use in the MIC White Paper

According to MIC's 2026 White Paper on Information and Communications, 86.4% of respondents in Japan said their company uses generative AI in some part of its work. The previous year's survey showed 55.2%, a large increase.

A combined 68.9% said their policy was to use generative AI actively or in limited areas, up from 49.7% the year before. By company size, the figure was 74.0% for large companies and 58.1% for small and medium-sized ones.

Meanwhile, 27.0% said their organization was not working on transforming operations with generative AI. The White Paper notes that this share is noticeably higher than in the United States, Germany and China, and especially high among small and medium-sized companies. This suggests that at many companies, use stays at the level of individual effort.

By type of task, the most common use was help with meeting minutes and emails, at about 70%. Sales came next at about 60%, followed by internal help desks at about 50%. Among concerns, security issues such as leaks of internal information were cited most often.

In the same White Paper's survey of individuals, 58.8% of people in Japan said they use (or have used) generative AI services at work or in daily life. The previous year's figure was 26.7%. It appears that workers have started using it ahead of company policy.

DX and AI efforts in the IPA survey

In IPA's DX Trends 2026, 42.3% of companies said they had adopted AI (including generative AI) and 15.7% said they were trialing it, for a total of 58.0%. By headcount, 34.6% of companies with 100 or fewer employees said they were interested in AI but had no plans yet. For generative AI specifically, the share of companies that had adopted it rose from 22.6% in fiscal 2024 to 44.0% in fiscal 2025.

Generative AI is used mainly by individuals. While 63.3% said individuals use it for work, 11.9% said it is built into departmental business processes and 18.6% said it is built into company-wide services.

The most common use was summarizing, translating and proofreading documents and audio, at 82.5%, followed by writing documents and reports at 80.5% and searching, collecting and analyzing information and reporting at 77.0%. By contrast, planning support for production, logistics and service delivery stood at 6.0%, and enhancing the company's own products and services at 10.9%. IPA concludes that AI results are mostly inward-facing, such as efficiency gains, and that outward-facing effects such as higher customer satisfaction or sales remain limited.

AI agent use is still low in every department

IPA also asked companies that had adopted, were trialing or were considering generative AI about AI agent use by department. The highest was IT systems at 19.4%. General affairs and legal, and R&D, were at 10.1%, sales (including customer service) at 8.1%, and the lowest, distribution, at 2.4%. IPA concludes that AI agent use remains low in every department.

AI use at small and medium-sized companies in the SME White Paper

The summary of the 2026 White Paper on Small and Medium Enterprises says about 30% of small and medium-sized companies have worked on using AI. In the Teikoku Databank survey, 30.3% of all responses (5,645) said they had worked on using AI as a labor-saving investment since 2019. Back-office functions and sales, marketing and customer service are named as the areas using it.

When the 3,888 companies not using AI were asked why, the top reason was 'we cannot picture which work to use it for,' at 63.4%. Next came 'we lack people to drive its use' at 40.0% and 'we have no internal rules or guidelines' at 26.2%. For public support available to small and medium-sized companies, see AI adoption for SMEs.

How to compare with other countries

The MIC White Paper asks the same questions in the United States, Germany and China. The share using generative AI in some part of their work was 86.4% in Japan, 90.9% in the United States, 91.6% in Germany and 98.1% in China. The White Paper says the gap between Japan and the other three countries has narrowed. On the share with a set policy for using generative AI, Japan rose from the previous year but remained lower than the other three. Overall, Japan tends to have more 'not adopted' or 'don't know' answers, even by type of task.

Note that Japan had 515 respondents and each of the other three countries had 309. Respondents were managers and above selected from each country's panel. When reading country differences, it is more useful for your own company to focus on whether organizations have structured efforts and on differences by type of task, rather than on whether AI is used at all.

Cautions when reading market-size figures

Articles on AI agents often show market-size figures next to adoption rates. Market size is a different thing from adoption rate. It is an estimate of total sales or spending, not of how many companies use it.

• Check the source: even the market-size charts in the MIC White Paper are based on data from private research firms
• Check the scope: for example, the AI platform market cited in the White Paper uses its own definition that includes software supporting the AI lifecycle
• Check whether it is actual or forecast: figures for future years are forecasts and change when assumptions change

A large market does not mean AI will work for your company. For adoption decisions, figures on which tasks companies of a similar industry and size use it for are more useful.

How to choose figures for your own evaluation

When explaining internally, how AI is used task by task is more useful than how high the usage rate is. In the MIC White Paper, the most common use is help with minutes and emails. In the IPA survey, it is summarizing, translating and proofreading documents. AI agent use is still low, but IT system departments are ahead.

For which tasks to start with, see AI agent use cases. The basics of AI agents are explained in What is an AI employee. On Employee Store, you can compare AI agents built by developers by job, supported tools and price, and adopt them.