When the people who built ChatGPT start their own venture fund, you pay attention. Not because founder-turned-VC stories are inherently compelling (they’re not always), but because these particular founders understand something most investors don’t: what AI models will actually do next.
Zero Shot, a new $100 million fund backed by OpenAI alumni, just closed its first $20 million and has already written checks to at least three startups. The fund’s five founding partners read like a who’s who of early OpenAI: Evan Morikawa (former head of applied engineering for DALL-E and ChatGPT), Andrew Mayne (OpenAI’s original prompt engineer and host of The OpenAI podcast), Shawn Jain (engineer, researcher, and founder of the GenAI startup Synthefy), Kelly Kovacs (previously a partner at 01A), and Brett Rounsaville (formerly of Twitter and Disney, now CEO at Mayne’s consulting firm Interdimensional).
The fund’s genesis is refreshingly honest. These founders spent years getting asked to consult for other VCs on AI technology and advising founder friends launching startups. Eventually, they decided to cut out the middleman.
“Maybe we should do our own fund, because we think we have a pretty good sense of where things are headed, and we have this great access to people who we think are incredible builders,” Mayne told TechCrunch, recalling the decision.
Knowing What Not to Fund
Here’s where Zero Shot gets interesting. Instead of chasing every AI trend, the founders are actually willing to say no to entire categories of startups. That’s rare in venture, where FOMO typically wins.
Mayne is bearish on most “vibe coding” platforms because he believes model makers with their engineering expertise will quickly make such tools feel redundant. Morikawa dismisses many robotics startups focused on embodied AI training data, calling the current efforts “a lot of hoping and praying” that researchers will solve problems that aren’t close to solvable yet.
The skepticism extends to digital twins. Mayne has actually run due diligence on several of these startups, even building reasoning models to test their assumptions. His conclusion? Standard LLMs work just as well.
This isn’t contrarianism for its own sake. It’s pattern recognition born from being inside the room where these models were built.
“There is a real skill in knowing how to predict where these models will be going next, because it’s extremely not obvious. It’s not linear,” Morikawa said.
Early Bets Worth Watching
Zero Shot has already backed three startups, two of which are public. The first is Worktrace AI, founded by early OpenAI product manager Angela Jiang. The startup builds AI-based management software to help enterprises identify and automate repetitive tasks. Worktrace raised a $10 million seed from notable backers including Mira Murati and OpenAI’s own fund.
The second is Foundry Robotics, working on next-generation AI-enhanced factory robotics. It recently closed a $13.5 million seed led by Khosla Ventures.
A third investment remains in stealth.
The portfolio moves tell you something about the fund’s thesis: they’re not betting on AI as a narrow productivity layer or a chatbot wrapper. They’re backing startups trying to solve concrete problems in robotics, enterprise automation, and infrastructure where having actual model expertise matters.
The Advisor Bench
Zero Shot has also assembled an advisory board that reads like an all-star team of former OpenAI leadership: Diane Yoon (former head of people), Steve Dowling (ex-head of communications at both OpenAI and Apple), and Luke Miller (former product leader at OpenAI). These advisors get carried interest, so they’re genuinely aligned with the fund’s returns.
What’s notable is that none of this feels like a vanity project. These aren’t celebrities trading on their former titles. Morikawa is still deeply involved in robotics at Generalist. Mayne is actively consulting through Interdimensional. Jain founded his own company. They’re not parking their expertise; they’re compounding it through investing.
The real question is whether insider knowledge actually translates to better business outcomes in venture. Pattern recognition doesn’t always pay off, and being right about AI trends doesn’t guarantee picking winners. But when you’re actively dismissing categories of startups that less informed VCs are still funding, you’re at least betting differently. And in venture, different is often more valuable than right.
