Full apps expose a gap in AI coding skill

A new benchmark tests AI agents on building complete, multi-feature web apps rather than single coding tasks.

Also today

Today's top five AI stories, one line each, at 7am Eastern. Free. One-click unsubscribe.