Type: url
Original: https://simonwillison.net/2026/Jul/21/cat-and-thariq/
Status: generated
Summary
An edited transcript of Simon Willison’s AI Engineer World’s Fair fireside chat with Cat Wu and Thariq Shihipar of Anthropic’s Claude Code team. They describe Claude Tag, a multiplayer Claude that lives in a team’s Slack and which they say lands 65% of their product engineering pull requests; an internal active-user and retention bar a feature must clear before it ships externally; and a multi-month move from human code review to fully automated review for the product’s outer layers, with code owners retained for its core. The prompting material is the densest part: an 80% reduction in the Claude Code system prompt, examples and “do not do X” lists becoming counterproductive on frontier models, and a separate system prompt per model.