The Future of Coding Agents with Sasha Rush (Cursor/Cornell)

Sasha Rush, a researcher at Cursor and professor at Cornell, describes building coding agents and how they are changing everyday software work. The episode explains why code is an especially productive setting for agents, how much recent progress comes from better ways of using models rather than better models alone, and how Cursor approaches training coding models. It covers long-running agents, context limits, bug finding and the balance between autonomy and human oversight, as well as the broader shift in software engineering.