Miku Hirano: The Cinnamon AI Founder Building Japan's AI Future From Inside Three Cabinets

The most-quoted thing Miku Hirano said on the Tech for Impact Summit 2026 Main Stage was not about artificial intelligence, and she had not planned to say it.

Two members of her sales team had missed their targets by similar margins — one man, one woman. She noticed that her own reaction to each had been different. About the man she thought: it is a shame he could not achieve the target. About the woman she thought: shouganai — it happens, like that.

“This means I had gender discrimination toward them, and also toward myself,” she told the room. “Probably I discriminated against myself when I compared Cinnamon AI to other male-founded tech companies.”

It is an unusual thing for a chief executive to say out loud, in front of investors, on a stage with a Bloomberg reporter moderating. It is also the shortest route into understanding how Hirano thinks: she debugs systems, including the one she is running inside her own head.

The engineer before the CEO

Hirano is a builder by training, not a manager who later acquired a technical vocabulary. She holds a master’s degree in artificial intelligence from the University of Tokyo, and she started her first company, Naked Technology, in the mobile era. It was acquired by mixi — at the time one of Japan’s dominant social platforms.

In 2012 she co-founded Cinnamon in Singapore with Hajime Hotta. The company that grew out of it, Cinnamon AI, does something narrow and difficult: it gets structured, usable information out of unstructured enterprise documents. Contracts, invoices, medical records, insurance claims, engineering specifications — the paper and PDF sediment that accumulates in every large organisation and that no database was designed to read. Cinnamon’s current framing for the problem is “Super RAG,” an approach to retrieval-augmented generation aimed squarely at documents that resist being retrieved from.

That is not a consumer AI story, which is precisely why it is a useful one. The bottleneck in most enterprise AI projects is not the model. It is that the organisation’s actual knowledge is trapped in formats the model cannot reach. Hirano has spent more than a decade on the unglamorous side of that problem.

AI as a structural reset, not a product category

The thread Hirano pulled hardest at T4IS 2026 was what AI does to the economics of starting a company.

Her argument was concrete. Founding a company in 2026 requires far less capital than founding one in 2016 did, because AI tooling collapses both product development and back-office overhead at the same time. The one-person or “seed-strapped” startup is no longer a compromise — it is a viable structural form.

Then she followed it to the uncomfortable conclusion, out loud, as the founder of a venture-backed company: “If I started a company right now, differently, I would do that. I would not need to raise money.”

She was speaking on a panel about the funding gap facing female founders, and the observation cuts both ways there. For founders whom traditional venture capital systematically underfunds, a world requiring less external capital is a route around the gatekeepers. It is also, she noted, a reduction in the leverage of the venture model itself — for everyone. The structural unfairness becomes cheaper to route around, and the returns available to investors who back underpriced founders early may compress along with it.

That is a genuinely awkward thing to say in a room full of allocators. She said it anyway. It is the same instinct as the sales-team story: follow the system to where it actually leads, not to where it flatters you.

Three cabinets, and the cascade from the top

Hirano’s second vantage point is inside the Japanese policy machine. She has served as a committee member on the top-level Cabinet councils under Prime Minister Fumio Kishida and Prime Minister Shigeru Ishiba — the New Capitalism councils — and she described continuing that service on the growth-strategy side under Prime Minister Sanae Takaichi.

Her account of what changed there is the most transferable policy lesson from the session. Under Kishida, the rule was that half of the members of top-level Cabinet councils had to be women. Ishiba continued it. Under Takaichi, she said, the proportion has come down somewhat — to roughly 30 to 40 percent — but that is still far above where Japan sat before 2021.

The number itself was not her point. The cascade was.

“Because if the top councils could do that, all the other councils got changed too,” she said.

A quota at the apex of a system does not stay at the apex. It propagates downward through every committee that takes its composition cues from the one above it. For anyone trying to move an institution that has no appetite for being moved, that is a more useful mechanism than exhortation: change the composition of the smallest, highest-leverage room, and the rest follows without a fight.

Outside the day job

Hirano sits on the board of JGC Holdings, the global plant-engineering group, as an external director — a role that puts an AI founder inside the governance of exactly the kind of heavy-industry business whose documents she has spent a decade trying to make legible.

She is also the founder of Kamui Whisky, a craft distillery on Rishiri, a small island off the northern tip of Hokkaido. It is not an obvious second act for an AI engineer, and she is not the first technologist to conclude that something slow and physical is the right counterweight to something fast and abstract.

Her recognitions include selection to the World Economic Forum’s 2022 class of Young Global Leaders, and the Veuve Clicquot Business Woman Award, New Generation prize.

Why this session belonged at T4IS

Hirano appeared on the Tech for Impact Summit 2026 Main Stage on the “Women Building the Future” panel, alongside Miwa Seki of MPower Partners and Rika Nakazawa of NTT, moderated by Lisa Du of Bloomberg News. The full panel recap covers the capital-allocation data the session opened with, the boardroom-power discussion, and Hirano’s own remarks in their original context.

The reason it sat naturally in the T4IS programme is that it refused to stay inside one discipline. Forty minutes moved from venture valuation data, to how corporate boards actually make decisions, to Cabinet-level quota policy, to an admission of internalised bias from a founder running an enterprise AI company. That is the shape of most real problems in impact technology. They do not resolve inside a single function, which is why the summit is built around conversations rather than lectures — and why the rooms are deliberately mixed across business, policy, and culture.

About Tech for Impact Summit

Tech for Impact Summit is an invitation-only executive gathering in Tokyo, convening leaders across business, policy, and culture to deploy high-impact technology against humanity’s most urgent crises. It is a partner event of SusHi Tech Tokyo. The next edition takes place May 18–19, 2027, and Miku Hirano spoke at the 2026 edition.

Explore the speaker roster or read our guide to SusHi Tech Tokyo 2027 to see how the wider innovation week fits together.

Attendance is by invitation. If you lead an organisation working at the intersection of technology and impact and would like to be considered for the 2027 gathering, you can register your interest — we review every submission personally.

Learn more at tech4impactsummit.com.