Yohei Nakajima came to venture as a builder. He has made tools as a hobby since high school without ever holding an engineering job, and he began sharing them during COVID. A lot of no-code deal flow followed, then Web3 deal flow after he sold his own NFT art, and now he does it deliberately with AI tools.
Yohei is General Partner and co-founder of Untapped Capital, a small firm on the frontier of applying AI to venture capital. His route there began about ten years ago at TechStars, helping spin up the Disney Accelerator, and runs back to a college thesis on angel investment organisations.
In this episode of Going Solo, we discussed:
Building in public is a marketing strategy a VC can repeat for decades. Most VCs end up marketing a history of what they’ve done, and that only lasts so long. Sharing no-code tools brought Yohei no-code deal flow, selling his own NFT art brought Web3 deal flow, and with AI that approach became intentional.
Yohei explains why he starts from the assumption that venture could be fully automated and then the parts a firm keeps human says a lot about their uniqueness, with decision-making scoring the last thing Untapped plans to automate.
“Is this a good investment?” is a horrible question to ask AI. Yohei asks for the hypotheses that would make an investment more or less successful, plus the existing proof points for each, so the decision stays with the human. Automatically filtering out non-fits is a good use of AI, while fuzzy ranking of possible fits is the trap, because everybody ends up looking at the same things.
Early-stage investing is a discovery game. In the game behind Yohei’s shared discovery paradox paper, players do best when they share clues and spread their guesses across different boxes, because players who do not coordinate all pick the same box and lower the group’s chance of finding the prize. Broadly, alpha comes from discovery and beta from estimation, which is why venture needs many small firms alongside the large ones, and Yohei is extending the research through an autonomous research lab called Distributed Discovery.
Pitches have never been more polished, and a lot of it is AI, so reviewing unpolished signal matters more today. Yohei reads founders’ X reply tabs and the posts that drew engagement for the raw interactions that give personality signals, and he prefers a one to two page memo as an initial pitch.
We also got into whether a fully automated VC trading secondaries is possible (Yohei thinks it is), two ways to build an AI-ready firm, and what stays the same in ten years: human connection, family and education.
Hope you enjoy it.
DG
