How Can Seed Investors Win in the Age of Consensus?

What we talked about.

Andy McLoughlin, Managing Partner at Uncork Capital, joins Sunny Ray to discuss how seed investors can find an edge when the market rewards consensus founders. Uncork, founded in 2004 by Jeff Clavier as SoftTech VC, now co-leads seed rounds of roughly $6 to $10 million as larger Series A firms move up market. McLoughlin recounts co-founding Huddle in 2006 and what competing against giant platforms taught him about timing and exits. He explains the common thread behind early checks in Postmates, Intercom and Pipedrive, namely relentless founders who did not fit the Silicon Valley mold. The conversation covers Uncork's concentrated fund construction, how AI agents are reshaping per seat software pricing, and why frontier tech bets such as Loft Orbital and fusion are more binary than enterprise software. McLoughlin also shares his view on payments infrastructure for agents, what AI native truly means, and where AI funding may be running ahead of itself, including robotics and world models. He closes with where founders can find him and the fund.

Uncork's Andy McLoughlin explains how seed investors back non-consensus founders before the big funds see what he sees.

The questions, and the answers.

What is Uncork Capital building, and what problem are you solving?

Uncork started as SoftTech VC, a seed fund Jeff Clavier founded 22 years ago. Today we co-lead seed rounds of $6 to $10 million, writing $3 to $7 million checks. Series A firms have died or become mega funds, so we back non-consensus founders one step earlier, before the big firms are ready.

What did building Huddle in 2006 teach you that shows up in how you evaluate founders today?

We had huge white space, but the big players eventually arrive. You either raise an obscene amount and go scorched earth, like Box and Dropbox, or sell early enough to make a good return. Today I ask what a team of crack engineers at an AI lab couldn't roll out in 60 days.

Was there a common signal behind your early checks in Intercom, Postmates and Pipedrive?

Many were European founders moving to San Francisco, the bravest and brightest willing to take on the most competitive place on earth. Some were relentless machines who ran through walls. Pipedrive's Estonian founders didn't fit the Silicon Valley mold, but their early metrics with almost zero spend were impressive.

Why size your fund around 35 concentrated companies?

With 35 companies you expect three to five breakouts that can each return the fund, plus some that return one or two times. Our LPs also like concentrated positions. Lately we've opened the aperture, occasionally buying one to three percent of a generational opportunity, like the nuclear fusion deal Jeff did.

How do you underwrite revenue models when AI agents replace the seat that software used to price on?

There's no single answer. The death of SaaS was massively overstated, and good products are still priced per seat. Our portfolio company Coda sells secure development environments, and for every human seat they probably sell five agent seats. Where outcomes are measurable, pricing on outcomes becomes possible.

How do you decide what earns a slot in your frontier tech allocation?

The 10 percent is not a hard rule and may grow with robotics, mobility and fusion bets. Enterprise software stays our backbone for now. We've done deep tech longer than almost any seed fund, Jeff seeded Fitbit, and these bets are more binary, like Loft Orbital, where a launch could blow up a deal.

What does AI native actually mean to you operationally?

The best AI native companies still have humans, but they've given each one an Iron Man suit through agents so one person does the work of many. That's becoming table stakes. What excites me more is what they build with the technology, because an efficient company making something nobody cares about is worth nothing.

Which part of the AI stack is overfunded and which is underfunded?

A year ago people called coding startups overfunded, then Cursor was acquired for $60 billion. Legal tech looks overfunded on today's market size, but new technology can blow market size up, as with taxis. Robotics and world models are drawing heavy cash, though they depend on many other things working.