The product leader who turned Claude Code into Anthropic's revenue engine just walked out three weeks before the company's IPO.

The Summary

  • Cat Wu, head of product for Claude Code at Anthropic, left the company on Friday, weeks before its November IPO
  • Claude Code became one of AI's most important products under Wu's leadership, giving Anthropic the viral hit it needed to compete with OpenAI
  • Wu joined in summer 2024, briefly jumped to Cursor with co-creator Boris Cherny, then returned to oversee the product's explosive growth
  • When asked about time off, Wu replied simply: "post agi!"

The Signal

Cat Wu's exit carries weight because Claude Code is the closest thing Anthropic has to a ChatGPT moment. The tool launched in research preview in February 2025 and became the viral product that finally gave Anthropic mainstream traction beyond enterprise API contracts. Before Claude Code, Anthropic was the safety-first alternative that developers respected but consumers didn't care about. After Claude Code, it became the company building tools people actually wanted to use.

The timing stings. Three weeks before an IPO is when product leaders typically stay put, smile for roadshow presentations, and wait for their equity to vest at a higher valuation. Walking away from that payday signals something. Either Wu believes her equity position is already strong enough, or she sees something coming that makes the lockup period not worth waiting through.

"When asked if she's taking time off, Wu said, 'post agi!'"

That's either a joke or the most revealing comment in the piece. If it's serious, it suggests Wu thinks the current generation of AI tools has already done what they're going to do. The next breakthrough won't come from product management at existing labs. It'll come from something else entirely.

The Cursor interlude matters too. Wu and Boris Cherny left for AI coding startup Cursor in summer 2025, then returned to Anthropic two weeks later. That's not a normal pattern. People don't bounce between competitors like that unless there's either extraordinary retention pressure or unclear strategic direction. Anthropic clearly wanted them back badly enough to make it happen. Now Wu's gone anyway.

Key pattern emerging:

  • Thariq Shihipar, another Claude Code team member, also left last month
  • The product's core team is fragmenting right as Anthropic tries to go public
  • IPO pressure and product momentum don't always point the same direction

This isn't a founder departure or a CTO leaving. It's the person who made Anthropic's most important product feel human and usable walking away. Product leaders leave when they've either shipped everything they came to ship, or when they see the roadmap turning into something they don't want to build. Given the IPO timing, the second option looks more likely.

The Implication

Watch who else leaves Anthropic between now and lockup expiration. Pre-IPO departures from key product people usually mean one of two things: they've already made enough money that the IPO upside doesn't matter, or they don't believe in the company's post-IPO direction. Either way, it's a signal about what life looks like inside AI labs once the fundraising narrative collides with public market expectations.

For anyone building in the agent space, the real lesson is in Wu's "post agi" comment. If the people who built the tools think we're already past the breakthrough moment, the next wave of value creation isn't in better models or slicker interfaces. It's in what gets built on top of them.

Sources

Business Insider Tech