What
Agora is
AI can now pass the University of Tokyo entrance exam at the top of the class and win gold at the International Math Olympiad. The ability to "solve a given problem that has a fixed answer," as in conventional exams, will increasingly be taken over by AI.
What becomes important, then, is the ability to "pose your own questions." Just as entrepreneurs do: finding, for yourself, a problem worth tackling that has no predetermined answer — and deciding how to solve it and whom to involve, then carrying it through.
Using AI deeply not as a "tool for looking up answers" but as a "partner to think with," and — starting from your own interests — creating something new that AI alone could not produce, and bringing it into the world. We built a place to have exactly that experience.
Ask AI carelessly, and it answers carelessly. Ask it to "give me a good idea," and what comes back tends to be a reheated assortment of things heard somewhere before — the same safe, average answer anyone would get.
But an answer anyone can obtain has no value. Anything truly new lies only beyond that point.
—— "That's ultimately a mashup of this and that. Rebuild it from scratch."
—— "That conclusion is too clean — if it were wrong, where would it break? Construct a counterexample."
Rather than applying a memorized formula, you derive the formula itself: questioning from the roots and pushing AI to the very limit where it stalls. In the end, the human steps one pace beyond, reaching new ground.
An example
of Agora
This program itself was born from one dialogue with AI
1. AI's first answer was a clever-looking "mashup"
Yamada, Founding Partner of Roppongi Venture Capital, asked AI: "I've reviewed my own rather ordinary incubation-program plan, but its differentiation from existing services is weak and the design can hardly be called innovative. As a generative AI, you have learned from all manner of things in the world and have the power to generate new concepts. Could you take your time and think through a program design unlike any before?"
AI came back with "Five Disruptive Principles." Buzzwords lined up; it looked new. But Yamada felt: "These look like the methods of famous startup programs I've seen somewhere, just bundled together and wrapped in buzzwords. Could you think from a more fundamental level?" Rephrased and resubmitted, the substance was the same. Yamada pushed it back three times.
2. "Derive it yourself, like a formula — think from zero" — AI starts to move like a mathematician
So Yamada changed how he asked. "Instead of answering by researching and combining existing things, could you think from first principles, from zero? Show me the thinking process itself — as if you were deriving a formula for what is needed to reach the goal."
At that, AI began to think in terms of "structure," like a different entity. First, it rewrote "success" as a single equation.
P(success) = P(E1) × P(E2|E1) × P(E3|E2) × P(E4|E3) × P(E5|E4)
E1–E5 are gates you clear in order — E1: realizing "I can do this too"; E2: building; E3: shipping; E4: persevering; E5: fitting with society. Because it's a product, if any one term is zero, the whole thing is zero.
From there, AI narrowed in on the crux: "Most of Japan's talented young people are not stuck from E2 onward. It's the very first gate, E1, that is nearly zero — because they head to medical school or a big company before ever realizing 'I can do this too.'"
It then broke E1 down further. Why can't they realize it? (a) They define themselves only by their "deviation-score rank"; (b) there is no one taking on challenges nearby; (c) the cost of failure feels too high to take the first step. And it went on to derive how to loosen each of these.
3. "Isn't this too clean? Doubt yourself." — AI breaks its own answer
Here, too, Yamada didn't stop. "It feels a little too neatly wrapped up. Would you doubt it yourself, once?"
And AI began to dismantle its own conclusion. "Is the order of E1-first really correct? For both Zuckerberg and Jobs, the order was reversed — they built and shipped first, and only realized 'I'm an entrepreneur' afterward." "And does 'teaching' geniuses even work? What changes them is not lectures, but a word in the hallway, or a paper they happened to read late at night."
—— It took the answer it had just produced, broke it itself, and moved beyond it.
4. And so, something that doesn't yet exist was born
Before we knew it, ideas incomparable to that first clever-looking answer had emerged, one after another. What truly mattered was less the content of the answers than the "way" of getting there — the very stance of not "receiving" answers from AI, but doubting, breaking, making it derive from zero, and doubting again.
Attempts to "question deeply with AI" do exist in the world. But where Agora is new is what comes after. Exceptionally talented middle and high schoolers use AI deeply — not as a "tool for looking up answers" but as a "partner to think with" — and, from their own interests, create something that does not yet exist, which AI alone could not produce, and bring it into the world. The thinking process itself is released as a "work," and, if it goes well, can lead to compensation. Students who used to be on the learning and consuming side move to the side that creates and ships — this form does not yet exist in the world.
—— And this very program, "Agora," was itself born from this dialogue. Agora is a place to acquire that "way of engaging with AI" alongside peers of the same intensity, and to keep creating things that do not yet exist.
Read the full dialogue →What you
gain at Agora
Train how to think in the AI era, and leave behind a "work" as proof
In an era when AI answers anything, what and how should a person think? You confront that question not in the abstract, but by actually teaming up with AI to build something new. The "power to think deeply rather than use AI shallowly" that you train here is a cutting-edge weapon that works whatever path you take next. And the thinking and works produced here remain as your own track record, one that cannot be measured by test scores — strong material for telling "what I thought about and what I put into the world," for instance when applying to universities abroad.
Your intellectual curiosity, fully satisfied
Together with exceptionally talented peers, you push AI to its limits, watch the moment it still gets stuck, and step beyond it yourself. This kind of intellectual grappling is rare to experience. Moreover, you can debate directly with the adults who run and accompany Agora — a finance professional who majored in mathematics at Kyoto University, and an entrepreneur who founded and sold several companies in his teens — and widen the range of your thinking.
See the team →A rehearsal for entrepreneurship, with nothing to lose
The time-consuming work — planning, operations, editing, sales — is handled by the host, Yamada (and you're more than welcome to take part in these too). So you can spend your short time on the most interesting part: thinking. Unlike starting alone, you divide the work with peers and experienced people, so things take shape far more easily than going solo — and even if you fail, there's no real damage.
The upside if it works
The work you create could become a note article, a book, or a business. Rather than selling your time by the hour, if it hits, it can lead to uncapped upside. It becomes a far more meaningful "challenge" than a top student earning money through high-paying tutoring. Experiencing, while still a student, that your own mind itself becomes value — that pays off both in applications to universities abroad and well beyond.
How
Agora runs
Agora is not a "course" where you gather every week for a fixed term. Once you become a member, you can also choose to take part only when a theme looks interesting — joining for two or three hours and turning it into a work. There is no graduation, either.
So even when you're busy with school, clubs, cram school, or exams, you can keep at it without strain. Interaction is primarily online. Participation is free, and admission is selective, in small numbers.
How to
take part
For Cohort 1, we are proceeding with a very small, selected group (we are not recruiting widely; it is centered on invitations and referrals).
Even so, if you feel "this is where I want to think," please get in touch — adding a line about the question you most want to dig into right now, or something you've lost track of time working on.
An application form is coming soon.