Head of Claude Code: What happens after coding is solved | Boris Cherny
- Source
- lennysnewsletter.com
- Author
- Lenny Rachitsky
- Date
Context
Delegating all code writing can still leave a person responsible for inspecting the result. Boris Cherny, head of Claude Code at Anthropic, says the agent was writing all of his code at the time of the interview, while he continued reading it. He describes automatic AI review followed by human review for the team's changes. The distinction matters when interpreting claims that coding is solved: his account separates generating an implementation from checking its correctness and safety before others depend on it.
- AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
“100% of my code is written by quad code. I have not edited a single line by hand since November.”
Boris Cherny
“You want to under-resource things a little bit at the start.”
Boris Cherny
“In a year or two, it's not going to matter. Coding is largely solved.”
Boris Cherny
“The title software engineer is going to start to go away.”
Boris Cherny
“Coding now is describing what you want, not writing actual code.”
Lenny Rachitsky
- 00:00Introduction to Boris and Claude Code
- 03:45Why Boris briefly left Anthropic for Cursor
- 05:35One year of Claude Code
- 08:41The origin story of Claude Code
- 13:29How fast AI is transforming software development
- 15:01The importance of experimentation in AI innovation
- 16:17Boris's current coding workflow (100% AI-written)
- 17:32The next frontier
- 22:24The downside of rapid innovation
- 24:02Principles for the Claude Code team
- 26:48Why you should give engineers unlimited tokens
- 27:55Will coding skills still matter in the future?
- 32:15The printing press analogy for AI's impact
- 36:01Which roles will AI transform next?
- 40:41Tips for succeeding in the AI era
- 44:37Poll: Which roles are enjoying their jobs more with AI
- 46:32The principle of latent demand in product development
- 51:53How Cowork was built in just 10 days
- 54:04The three layers of AI safety at Anthropic
- 59:35Anxiety when AI agents aren't working
- 01:02:25Boris's Ukrainian roots
- 01:03:21Advice for building AI products
- 01:08:38Pro tips for using Claude Code effectively
- 01:11:16Thoughts on Codex
- 01:12:13Boris's post-AGI plans
- 01:14:02Lightning round and final thoughts
podcastThe 100-person AI lab that became Anthropic and Google's secret weapon | Edwin Chen (Surge AI)
podcastWhy humans are AI’s biggest bottleneck (and what’s coming in 2026) | Alexander Embiricos (OpenAI Codex Product Lead)
podcastThe coming AI security crisis (and what to do about it) | Sander Schulhoff
podcastWhy most AI products fail: Lessons from 50+ AI deployments at OpenAI, Google, and Amazon
Checking sign-in…
Loading comments…