Ex-OpenAI Operators Launch $100M Zero Shot Fund to Back AI Startups That Survive the Next Model Release
TL;DR
A group of OpenAI alumni led by ex-applied engineering chief Evan Morikawa have quietly built Zero Shot, a new seed-stage venture firm targeting $100 million for its debut fund. The partners are already cutting checks into AI startups like Worktrace AI and Foundry Robotics, and they are using their insider knowledge of frontier model roadmaps to deliberately avoid categories they believe the next generation of foundation models will simply absorb.
Key Takeaways
Insider information is the entire pitch. Zero Shot's edge is not network or check size, it is roadmap visibility. When five people who actually shipped GPT-era systems tell an LP they will not touch vibe coding wrappers, digital twins, or egocentric-video robotics training, that is not a vibes call, it is a bet that the model providers themselves are about to commoditize those layers.
$100M is a Goldilocks number for an alumni fund. Big enough to lead seed rounds and own real ownership, small enough that the partners can return the fund on two or three winners. Compare with Conviction's much larger raises and Pear's mega vehicles, and Zero Shot looks deliberately undersized, which is the right move for a first-time GP team.
Five GPs is unusual and risky. Most successful seed funds run on one to three decision-makers. Five partners means more sourcing surface area, but it also means slower decisions and dilution of carry. The team will need a clear lead-partner model to avoid the committee dynamics that have killed plenty of operator funds.
The OpenAI diaspora is now a fund category, not a coincidence. Between Conviction, Halcyon, Adept alumni vehicles, and now Zero Shot, ex-frontier-lab employees have become a recognizable LP segment of their own. Expect more, and expect LPs to start asking pointed questions about who actually has differentiated insight versus who just has a logo on the resume.
Fund Overview
Fund Name: Zero Shot Fund I
Fund Size: $100 million target (first close completed, amount undisclosed)
Stage: Seed
Check Size: Not publicly disclosed, consistent with a $100M seed fund this likely lands in the $1M to $4M range with reserves
Geography: US-centric, San Francisco Bay Area focus
Focus: AI applications and infrastructure where the partners believe foundation model trajectories will not erase the moat
Key LPs: Not disclosed
Why This Fund Matters
The most interesting thing about Zero Shot is not the OpenAI pedigree, it is the negative thesis. Most seed funds raise capital by promising to find the next big thing. Zero Shot is raising capital by promising to avoid the obvious things, specifically the AI wrappers and tooling categories where the model providers themselves are quietly building the same product into their next release. That is a much harder thesis to fake, and it requires exactly the kind of ground-truth visibility that ex-frontier-lab employees have and almost no one else does.
This matters because the AI seed market in 2026 is drowning in undifferentiated picks-and-shovels plays. Almost every general-purpose seed fund has loaded up on coding agents, RAG vendors, and vertical AI copilots. If Morikawa and the team are right that meaningful chunks of that portfolio are about to be commoditized into a free OpenAI feature within 12 months, the rest of the market is sitting on a wave of writedowns that has not been marked yet.
The portfolio choices so far back the thesis. Worktrace AI, founded by ex-OpenAI PM Angela Jiang, is a workflow automation play that lives above the model layer. Foundry Robotics, which raised a $13.5M seed led by Khosla Ventures, is hardware-plus-AI where the moat is physical integration rather than prompt engineering. Both are categories where simply releasing a better GPT does not eat the company.
For LPs, the question is whether five operators with no prior GP track record can actually convert insider visibility into top-decile returns. Operator funds have a mixed history. The diaspora from PayPal worked, the diaspora from Google Brain mostly did not. The OpenAI cohort is still unproven as investors, and Zero Shot is one of the bigger live experiments in that bet.
The Team
Evan Morikawa led applied engineering at OpenAI, the team responsible for taking research models and making them production-ready for ChatGPT and the API. He is the closest thing the fund has to a household name in AI engineering circles. Andrew Mayne ran prompt engineering at OpenAI and built a public profile as host of the OpenAI podcast, which gives the fund inbound founder reach that few operator vehicles can match. Shawn Jain came from research and engineering, and brings the technical credibility for deep diligence. Kelly Kovacs was previously at 01A, the operating partner organization built around Sam Altman, which gives the team a pre-existing investor muscle. Brett Rounsaville rounds out the GP set with operator experience from Twitter and Disney.
The advisory board is where it gets interesting. The fund has pulled in OpenAI's former head of people, the former head of communications who also did time at Apple, and a former product leader. That advisor stack reads like a reference list designed to win deals rather than a passive cap-table garnish.
Early Portfolio
Worktrace AI, an enterprise task management and automation startup founded by ex-OpenAI PM Angela Jiang. Foundry Robotics, AI-enhanced factory robotics with $13.5 million in seed funding led by Khosla Ventures. One additional stealth-stage company has reportedly received a check.
What This Means for Founders
If you are building in AI and you can credibly explain why your company will not be obsolete when GPT-6 ships, Zero Shot is a near-perfect first call. The partners can actually evaluate your technical premise against unpublished frontier-lab information, which means diligence will be fast and high-signal. That is a rare advantage in a market where most VCs are running pattern-matching against TechCrunch headlines.
The flip side is that if your wedge is a thin layer on top of a frontier model, expect a fast no. Zero Shot has explicitly told the press that vibe coding platforms, digital twin startups, and egocentric-video robotics training are categories it will not touch. Founders in those spaces should not waste pitch slots here.
Fund Momentum Take
This is one of the more intellectually honest fund pitches we have seen this year. Zero Shot is essentially saying out loud what every GP at a top firm whispers privately, which is that a meaningful fraction of the AI seed portfolio of 2024 and 2025 is going to get vaporized by the next two model releases. Building a fund around that observation is a real edge, not a marketing line.
Our concerns are structural. Five GPs without prior investing track records is a lot of mouths around the table, and the partnership math gets ugly fast if even one of them turns out to be a poor picker. We would also like to see at least one experienced GP on the team, ideally someone who has actually marked up a fund through a full cycle, to anchor the portfolio construction discipline that operator funds typically lack.
Net net, Zero Shot is the kind of fund LPs should take a hard look at if they want differentiated AI exposure that is not just another a16z co-investment. We would put it in the swing-for-the-fences bucket of an emerging manager allocation. The thesis is sharp, the team has real edge, and the worst case is that the partners learn investing on someone else's $100M. The best case is that they call the next set of category resets correctly and return three or four times.
Frequently Asked Questions
What stage does Zero Shot invest at?
Seed stage, with the fund explicitly framed around being one of the first institutional checks into AI startups.
Who are the partners?
Evan Morikawa, Andrew Mayne, Shawn Jain, Kelly Kovacs, and Brett Rounsaville. The first three are OpenAI alumni from applied engineering, prompt engineering, and research. Kovacs comes from 01A, Rounsaville from Twitter and Disney.
What categories will the fund avoid?
The partners have publicly flagged vibe coding platforms, digital twin startups, and egocentric-video robotics training as categories they believe foundation model providers will commoditize.
How does Zero Shot compare to Conviction?
Conviction is larger, more solo-GP-driven under Sarah Guo, and more focused on infrastructure. Zero Shot is smaller, more application-layer, and built around a partnership of operators rather than a single brand-name GP.
Is the fund accepting cold inbound from founders?
The fund is actively writing checks and Andrew Mayne's public profile gives it strong inbound. Founders in non-excluded categories should expect responsive review.
Have a fund closing to announce? Submit your fund here.
Need help raising capital? Check out our Fundraising Advisory services.