VC Hot Takes, Vol. 1: AI, narrative, and the early bets nobody will defend out loud

Venture is drowning in consensus. Same AI thesis, same valuations nobody will defend out loud, same “we’re so back” energy on every timeline. So I went the other way and asked a group of investors I truly respect for one thing: the take they’d defend in a room full of people who disagree.

Some sent a single line. Some sent something closer to an essay. I’m running all of it, grouped by the three fault lines they kept landing on.

Common theme: Obviously founder narrative overlapping with AI.

The more subtle thread that weaves it all together: betting on a person before anyone else will, is where the edge still lives.

The funds racing to automate judgment, and the people betting that judgment is the whole job.

“Using AI agents to screen seed founders is like using a high-tech ouija board. The criteria everyone’s scraping for today would have screened out the three $30B+ companies in Founder Collective’s Fund I. Automated filters can analyze patterns, but rarely does consensus mint the next big breakout. Many of FC’s highest-conviction investments are in ‘weird’ or unsexy spaces, because the people building them have extraordinary motivation and are more likely than others to ‘see around corners’ in their industries. Seed investing is a people business, built on relationships and randomness. Don’t underestimate getting into the wild and really getting to witness the fire burning in an ‘atypical’ founder. Hand gestures, intonations and eye contact convey a lot.”

“Most AI “technical moats” are just temporary API arbitrage. The real moat is product taste, workflow ownership, and distribution.”

““Won’t the model just do this?” is the laziest question in venture. Claude may one-shot code that passes every test and eval, but someone still has to run it at 2am when a vendor breaks their API, a retry quietly double-charges a customer, or an agent reaches for data it should never touch. Writing software and operating software are not the same thing. One is a demo. The other is a company. The moat was never the code.”

“VC firms should be less focused on investing in AI/software startups and more focused on consuming AI/software as users themselves.”

“Pattern matching is just bias masquerading as data”

Couldn’t help adding my own spicy take here.

Pricing has detached from what companies are actually doing. Three reads on what that breaks.

“VC market in one word right now = kingmaking. VCs have shifted from funding proof to funding possibility. The more investors price startups off hypothetical endgames, the more fundraising becomes a contest of narrative gravity rather than business fundamentals, and the greater the risk of creating the next generation of zombie companies.”

“Heavy discounts on secondary sales are an inefficiency tax, a failure of process that can and should be avoided. When you practice narrative priming and early buyer activation long before a sale, you destroy the information asymmetry that comes with treating liquidity as a reactive, tactical event. Pre-warmed people who understand the story before the sale turn secondaries into a solvable data, storytelling, and relationship intelligence challenge, not a financial engineering exercise.”

“The more the VC grapevine and FOMO drive deal heat, the more founders value independent, first-principles conviction.”

Takes on why the least glamorous version of this work is the real one.

“Venture (especially early stage) is currently a low status job, and that’s a good thing!”

“Inspired by Lorne Michaels on producing, seed investing is an invisible art. If you’re any good at it, you leave no fingerprints.”

Two forward-looking bets on who comes out ahead, and neither is the consensus pick.

“VCs will make more money on college graduates than college dropouts in the next 10 years.”

“Founders win this vintage. Grow fast enough, dilute almost nothing.”

Line these up and they argue with each other. One side says AI is rewriting what this job even is or is becoming. Another says narrative and FOMO have detached pricing from anything real. Maria Palma says founders are winning the vintage. Ashley Mayer says the job has never been lower status. Maybe everyone is right.

But read them again and they keep walking through different doors into the same room. The loud part of this market is AI, narrative, and heat. The quiet part is taste, conviction, and the read on a person nobody else believes in yet. That second part is where almost every take here sits, even the ones that start somewhere else.

The edge in early venture was never the model, the round name, or the price. It is the read on a person. Everything loud in this market is a distraction from that. Everything quiet is the work.

More soon. What’s your hot take?