The demographic narrative is evident when you stroll through a mid-sized Japanese city on a weekday morning. Elementary school hallways are half full. People who are well past the age at which most nations consider retirement work in municipal offices. This is no longer a forecast. It’s only Tuesday.
In this context, Agent, a company founded by Koji Shinomiya, has been operating for years, positioning itself between education support, AI deployment, and what he somewhat earnestly calls “solving social problems that happen to be sitting right in front of him.” More than 4,000 elementary schools in Japan currently have the company’s support for ICT implementation; this figure may seem abstract until you take into account the practical requirements.
Giving teachers tablets and leaving is not enough. Chief Digital Officers, advisors whose job it is to explain not only how a tool functions but also why it matters within a much larger institutional picture, are placed by agents inside education boards. Shinomiya appears to be adamant about this distinction. He contends that leadership devoid of digital literacy dooms implementation before it even begins. It’s the kind of observation that, while seemingly obvious in retrospect, explains why so many attempts at digital transformation go unnoticed.
There is a similar but different squeeze on universities. While the number of Japanese 18-year-olds continues to decline toward record lows, international enrollment reached a record above 336,000 in 2024, rising more than 20 percent year over year. AI tools are increasingly bridging the gap between institutions’ shrinking domestic base and growing global expectations. Meiji’s chatbots answering thousands of questions from students every day, Keio’s AI-based admissions screening, and the NTT West and Ritsumeikan partnership reaching about 50,000 current students and 400,000 alumni are no longer pilot programs. Whether or not that change was formally announced, they are turning into infrastructure.
It’s worth taking a moment to consider Shinomiya’s perspective on AI. He rejects the worried narrative that human thought is being undermined by automation. Rather, he presents it almost as a question of mental capacity, relieving people of monotonous cognitive work so they can focus on judgment, empathy, and motivation—things that a machine is still unable to convincingly mimic. In a classroom, this means that while a teacher concentrates on identifying which students appear to be quietly brilliant or disengaged, software manages practice problems and performance tracking. It’s an upbeat framing that somehow doesn’t sound forced.
It remains to be seen if Japan can truly realize this vision. The nation has an odd attitude toward change; when survival is at stake, it can be surprisingly adaptive in some situations and resistant in others. That kind of pressure could be caused by labor shortages. Analysts predict that within ten years, the workforce gap will grow to the point where automation becomes more of a necessity rather than a competitive advantage.
This distinction seems to be well understood by the agent. Instead of being designed as a universal export-ready platform, its model is purposefully localized, end-to-end, and woven into the particular institutions it serves. Shinomiya acknowledges this, pointing out that without significant adaptation, something successful in Japan might not necessarily translate to South Korea or China. Even though it restricts the simple narrative of global scalability, that candor is welcome.
Observing this from the outside, it seems that the most beneficial inventions seldom begin as ambitious concepts. They begin as particular solutions to particular, uncomfortable issues. There is no shortage of those in Japan right now. It’s still unclear if resolving them locally—school board by school board, university by university—ends up being a constraint or, ultimately, the whole point.
