Lenny's Podcast
Head of Claude Code: What happens after coding is solved | Boris Cherny
with Boris Cherny, Head of Claude Code at Anthropic
19 Feb 2026
24 min read
1h 12m
TL;DR
Claude Code has solved software engineering in less than a year—Boris writes 100% AI-generated code and ships 10-30 pull requests daily without editing a single line by hand since November. The next frontier isn't coding but having AI identify what to build by analyzing user feedback and telemetry, fundamentally shifting software roles from "engineer" to "builder."
Boris Cherny is the Head of Claude Code at Anthropic, leading the team that built the AI coding assistant that now generates 4% of all GitHub commits globally. Previously, he worked at Meta/Instagram as one of the top most productive engineers and led code quality initiatives across Facebook, Instagram, and WhatsApp. He briefly left Anthropic to join Cursor but returned within two weeks after realizing his core motivation was Anthropic's safety-focused mission.
Takeaways
1
**Coding is officially solved** Claude Code now handles 100% of code generation for senior engineers like Boris, who ship 10-30 PRs daily without hand-editing. At Anthropic, the 4x team expansion coincided with 200% productivity gains per engineer—a transformation that took less than a year and is now accelerating industry-wide.
2
**AI ideation is the new bottleneck** With code generation solved, the constraint has shifted to deciding what to build. Boris uses Claude to scan feedback channels, bug reports, and telemetry to surface priorities and propose features, automating the product discovery process that humans previously owned alone.
3
**Underfunding teams forces better AI leverage** Boris's principle of intentionally staffing teams lightly forces engineers to maximize Claude's capabilities rather than add headcount. This scarcity creates intrinsic motivation to delegate more work to AI, delivering faster shipping velocity with fewer people than traditional approaches.