Anthropic's Boris Cherny: Why Coding Is Solved, and What Comes Next

Summary

Boris Cherny, creator of Claude Code at Anthropic, explains how AI coding agents have changed his work: he no longer writes code by hand and instead runs many parallel agents and scheduled loops. He argues coding is effectively solved for his own codebase, predicts teams will become more cross-disciplinary, and discusses how AI may reshape software moats and startups. He also covers MCP and computer use for knowledge work, and closes with upcoming product directions around loops, batch agents, and computer use.

• Claude Code began as a bet on model progress ahead of product-market fit.
• Boris says the model writes 100% of his code and he ships dozens of PRs daily.
• Loops and sub-agents are becoming central to managing parallel agent work.
• He expects more cross-disciplinary generalists on software teams.
• AI may erode switching costs and process-power moats while network effects and scale remain durable.
• He sees AI making it easier for startups to disrupt large incumbents.
• MCP and computer use are presented as ways to give agents access to knowledge-work tools.
• Upcoming product areas include loops, batch, Claude Design, and computer use.