Phil

Asymmetric bets: capped downside, uncapped upside

Phil frames the core investing lesson that transfers to life: structure decisions so the most you can lose is small and fixed while the upside is many multiples larger. A $3M check can only lose $3M but might return $300M.

“Like, so if I invest in a company, let's say I put $3 million into a company, there is a possibility of it being $300 million, but the downside is capped at $3 million. I could only lose $3 million. I could gain $300 million.”

Steal thisSeek out bets where the loss is small and fixed but the win is 100x—not the linear one-in-one-out trade of an hourly job.

Power law: 10 companies return the whole fund

Phil explains VC portfolio math: out of hundreds of investments, almost all the returns come from roughly 10 companies. The same power law applies in life—a few relationships or bets drive nearly all the value, so you need a portfolio to find the outliers.

“So almost all of the returns are going to come from like 10 companies out of hundreds that we will have invested in.”

Steal thisIncrease surface area—say yes to more shots—because a tiny handful of outliers will drive nearly all the value.

This VC skipped a hire because Claude does the research

Phil says he uses Claude more than ChatGPT for company research—as a sparring partner—and that his firm declined to hire someone this year because they're leaning on Claude more than ever. Shaan jokes they effectively hired a person named Claude.

“Claude is just my partner. It's my sparring partner. Like, I'll— it's like having— like, we actually didn't hire somebody this year because we're using Claude more than ever.”

The Last Mile Problem: what makes a vertical AI app defensible

Phil's fund argues the moat against general models is the 'last mile': domain context and workflows (the terms and norms of healthcare, legal, accounting), deep integrations (Salesforce, EHRs), and compliance with human-in-the-loop. These are what let a vertical app produce trusted outcomes a generalist model won't.

“And basically we talk about what matters and it's what you just said, domain context. And workflows. So like if you've got the specific terms and norms of a given field, if it's healthcare, legal, accounting, you can produce more trusted outcomes. Um, and integrations. So like integrations into your Salesforce instance are valuable, um, or in health EHRs or whatever the case may be. And then there's also some like compliance and human-in-the-loop element that I think really matters.”

Steal thisBuild vertical AI on domain context, deep integrations, and compliance/human-in-the-loop—the 'last mile' a general model won't bother to cover.

Delphi.ai: make a talking digital clone of yourself

Phil recommends Delphi (delphi.ai), which lets you build a conversational digital clone—used by figures like Arnold Schwarzenegger, Lenny Rachitsky, and Brian Halligan for founder coaching. The company raised from Sequoia and is reportedly doing very well.

“Yeah, so you can make, it's delphi.ai. Um, you can basically make a digital clone of yourself. And they have them with like Arnold Schwarzenegger and other folks like that. Like you can, you can talk to them, basically do what kind of what you just said.”

Excite passed on buying Google for $750K to protect ad views

Phil cites a classic innovator's dilemma: early on, Larry and Sergey tried to sell Google to Excite for $750,000, but Excite refused because they wanted users to stay on the page seeing ads rather than getting instant answers. The same fear—that better answers cannibalize search ad revenue—is what people worried would sink Google in the AI era.

“So Excite. Had an— like, Google wanted to sell to Excite for $750,000 very early on. It was just Larry and Sergey wanted to sell this idea to them. And Excite said, um, we don't want them to get answers. We want them to stay on the page so we can serve up ads. And then, so obviously the rest is history there.”

Apple co-founder Ron Wayne sold his 10% for $800

Phil tells the cautionary tale of Ron Wayne, Apple's third co-founder, who held a 10% stake but thought it was worthless and was scared of being personally liable for a line of credit. He sold back to Jobs and Wozniak for $800 and later ran a stamp shop in Milpitas.

“Yeah, Ron Wayne was a co-founder of Apple with Jobs and Wozniak. And very early on he, he had a 10% stake in the company and he thought it wasn't gonna be worth anything and wanted to focus on other stuff. So he sold his stake back to them for $800, which is crazy.”

Masa's $20M Alibaba bet became $100B

Phil recounts SoftBank's Masayoshi Son's legendary bet: he invested $20 million into Alibaba early after meeting Jack Ma, and it grew into roughly $100 billion—one of the greatest venture returns ever.

“He invested in them very early on after he met Jack Ma. He put in $20 million and it turned into $100 billion.”

The Breaking Bad house Airbnb model that lost to a streamer

Phil and a friend flew to Albuquerque within hours to bid on the iconic Breaking Bad house listed at $400K, building a model around an $800/night Airbnb. They were willing to pay up to $900K, but it sold for over a million to streamer Adin Ross, who could monetize the audience in ways they couldn't.

“But basically we decided we could spend up to $900,000, and ultimately the house went for over a million. We don't know exactly how much, but it was sold to Adin Ross, who, uh, for those of you who don't know, who's a popular streamer. And, you know, for him, he's got a totally different revenue stream, which is like he can make a ton of money streaming this to this audience.”

Barry's Bootcamp for seniors—a franchiseable fitness club

Phil pitches a Barry's/SoulCycle-style group fitness class designed for old people: lower impact, emphasis on balance, joint work and even cognitive exercises, built as a social club. No killer brand exists for the demo, and the boutique-fitness playbook (F45, Solidcore) can simply be copied with the demographic swapped.

“And the idea is just a Barry's Bootcamp style class for old people. And you could franchise it.”

Steal thisCopy the boutique-fitness franchise blueprint (Barry's, F45) but design it for the 55+ demo—low-impact, balance and joint work, sold as a social third place.