Scott, Dillon, and Matt discuss whether code written by AI deserves extra doubt. They use the Rust rewrite of TypeScript as a test case, then compare maintenance risks, small trials, and the cost of dependency upgrades. Plus, a return to design system problems and the question of when to stop debating a change.

Scott, Dillon, and Matt discuss how to get support for technical changes. They cover trust, demos, early feedback, and deadlines. They also compare browsers after Arc and explain why testing only in Chrome can hide problems.

Scott, Dillon, and Matt recap a hard week for the internet: two critical PHP vulnerabilities, the eMudhra certificate incident, and a GitLab flaw that was there for 10 years. They also talk about the end of Arc, the REA reverse-engineering skill, and the new wave of personal agents, including Claude Duet. Plus, Dillon gets a new keyboard, and two hosts are on call during the recording.