THRSP1148 Theater Thu 19 Nov 14:50 PST
Leverage recursive self-improvement to train better AI agents
Edo Segal
Session description
Microsoft's catalogue description
Recursive self-improvement is a closely scrutinized concept on the AI frontier. But for people building agents, it also presents compelling possibilities: giving agents the opportunity to train against synthetic customers, fix their own tools when they fail, and proactively improve. Napster CTO Edo Segal breaks down how to build AI agents with RSI capabilities responsibly, how RSI can improve agent performance, and what Agentic RSI means for future deployment of AI agents.