Ben Holmes of Warp demonstrated how teams can build self-improving artificial intelligence agents. Automated judge agents evaluate execution logs against strict rubrics, find hidden inefficiencies, and submit code changes to update agent instructions.
FireBrief
FireBrief