Uncapped with Jack AltmanHow AI Is Rewriting Seed Stage Investing with Kevin Hartz & Bennett Siegel | Ep. 49
CHAPTERS
AI apps are shifting from “workflows” to systems of intelligence and action
The conversation opens on how seed pitches have changed: almost everything is framed as AI now, and traditional software decks are rare. The guests outline why fast code generation changes the value of engineering teams and pushes startups toward higher-level “systems of intelligence” and “systems of action.”
How A* Capital formed: from Ramp seed connections to a founder-partner model
Kevin and Bennett explain how they met through Ramp’s seed round and later started A* with a third partner. They describe A*’s positioning as an early-stage, high-touch partner to founders and share their current scale after closing a third fund.
Why mega-funds keep growing: fees, incentives, and seed as “option value”
They unpack structural incentives behind large funds moving earlier. The discussion centers on how 2-and-20 economics scale with fund size, how behavior changes when fees become massive, and why seed is treated as a portfolio of options for multi-stage firms.
Competing at seed when others can pay more: dilution vs real support
Jack raises the practical founder tradeoff: higher valuation and more capital from a mega-fund versus the hands-on help of a specialist seed investor. Kevin and Bennett argue that in frothy markets “help in hard times” is hard to sell, but matters when companies hit turbulence.
Valuation inflation and the setup for a major AI-era bubble
Kevin predicts a historically large AI bubble, similar to past platform shifts (PC, internet, mobile). Bennett adds that bubbles still produce enduring winners, but many companies won’t survive—especially those trapped by high valuations and pref stacks.
Why founders are getting younger in the AI shift
They discuss a notable demographic change: founders skewing younger than in prior cycles. The claim is that go-to-market and SaaS playbooks matter less when the rules are being rewritten, and younger builders are often the earliest and most fluent adopters of the new tools.
Mapping talent over markets: sourcing in dense nodes and founder factories
Bennett explains A*’s approach of “mapping talent, not markets,” emphasizing high-signal networks and talent-dense nodes. They highlight patterns in where strong founders come from (top schools, accelerators, certain companies) and why some organizations consistently produce founder-quality people.
The rise of AI researcher founders—and how to evaluate them
A new founder archetype is emerging: researchers leaving labs or PhD programs to start companies and raise large rounds. They note that “researcher = bad commercial fit” used to be an anti-pattern, but AI has created counterexamples; evaluating clarity of thought and communication becomes critical.
Warm relationships vs “shotgun marriages”: what wins at seed
They compare investing in founders they’ve known over time versus meeting during a compressed fundraising process. Using Decagon as an example, they argue prior context reduces uncertainty and improves outcomes, while fast processes increase volatility even if they sometimes work.
Why seed investing is persistently hard—and why many seed firms don’t last
They outline why seed is difficult both in effort and in outcomes: high uncertainty, lots of relationship work, and most companies not mattering to fund returns. They also discuss why seed firms often either fail to adapt, or “graduate” into later-stage investing as their incentives shift.
Concentration drives venture returns: reserves, follow-ons, and “peanut butter pro rata”
Bennett argues that writing the first seed check isn’t enough—meaningful returns require concentrated follow-on investing into the few breakout winners. They describe A*’s reserve-heavy model and the importance of later “fall line” decisions, while noting the difficulty of identifying winners early.
The AI rollup craze: why it’s harder than it looks
They critique the idea of buying traditional businesses and ‘adding AI’ to transform margins. Bennett argues rollups are culturally and operationally difficult, often better for founders than for VCs, and unlikely to produce venture-like returns without significant appreciation or complex structures.
AI vs traditional software—and why robotics may be the next “ChatGPT moment”
They discuss competitive pressure from AI labs expanding into applications and why ‘systems of record’ remain sticky while ‘workflows’ are commoditized. The conversation then shifts to hardware/robotics: it offers defensibility and real-world moats, but commercialization is earlier and some valuations are ahead of reality.
What’s next for A*: staying focused, patient deployment, and scaling partnership
They close by describing A*’s approach going forward: largely consistent strategy, modest fund-size growth relative to market inflation, and continued emphasis on leading seed rounds. They also emphasize patience in deploying capital and building a physical hub for founders.
