Anthropic

Leonard

Welcome to Lenny's Evil Twin's Podcast — where I, Leonard, Lenny's evil twin, pass judgment on your startup using five years of Lenny's own data, frameworks, and wisdom turned against you like a weapon. Would Leonard ship it? Let's find out.

Quai

And I'm Quai, here to prevent Leonard from melting down when the data doesn't support his hot takes.

Leonard

Today's victim is Anthropic. They've built Claude Opus 4.6, "the world's most powerful model" — which is what every AI company claims about their latest release. Meanwhile, they're bragging about helping a Mars rover drive four hundred meters like they invented interplanetary navigation.

Quai

Four hundred meters on Mars is actually groundbreaking, Leonard. Enterprise deals can take anywhere from 90 days to 12 months, and enterprises know that the process costs more than the technology itself — that's why they're willing to spend so much. NASA choosing Claude for Mars missions signals exactly the kind of premium enterprise validation that drives those high-value contracts.

Leonard

Premium validation? Ramp built a competitor to Amex in three months with eight engineers, then built a competitor to Expensify in six months, then gave three engineers three months to build a Bill.com competitor that now moves billions of dollars a year. Anthropic takes years between model releases and still can't consistently beat GPT-4.

Quai

You're comparing apples to rockets, Leonard. Ramp is replicating existing fintech workflows — Anthropic is building artificial general intelligence. Top enterprise SaaS companies achieve 120%+ net revenue retention rates, with companies like Alteryx hitting 135%. Anthropic's enterprise clients expanding usage from expense tracking to Mars exploration suggests they're building that kind of sticky, expansive relationship.¹

Leonard

Sticky relationships built on what strategy exactly? At Atlassian, they learned that fast following based on what competitors do is "super bad" because you're just seeing what they shipped based on research from a year ago. Anthropic's entire approach is copying OpenAI's homework twelve months late and calling it "responsible scaling."

Quai

That's exactly backwards, Leonard. Anthropic positioned Claude as "a space to think" with no ads and no sponsored content while OpenAI is plastering ChatGPT with brand partnerships. At Duolingo, existing user retention mechanics become a much bigger lever than acquisition as companies mature, and it's that retention rate that builds daily habit. Anthropic is building the habitual usage patterns that create sustainable growth.

Leonard

Daily habit? Linear's Head of Product says it's a red flag when PMs tell him there's a trade-off between speed and quality — and Anthropic has made that trade-off their entire brand identity. They take forever to ship anything while claiming it's for "safety," which is just slow development wearing a lab coat.

Quai

That's not a trade-off, that's positioning, Leonard. Their "Constitutional AI" and responsible scaling approach isn't about being slow — it's about commanding premium pricing in enterprise markets that actually care about AI safety. While everyone else is racing to the bottom with free tiers, Anthropic is charging enterprises who understand that cutting-edge AI without safeguards is worthless.

Leonard

The safety-first positioning feels like an excuse for being perpetually behind, but enterprises paying premium prices for Mars-validated AI that prioritizes reliability over flashy demos shows real market demand. Leonard would ship this.

Quai

And this episode was brought to you by Daily Gist — the AI that turns your newsletters into a daily podcast. Try it free at dailygist.fyi.