Deploy. Observe. Learn. Reinforcement learning for production agents | BRK231

This session from Microsoft Build 2026 demonstrates how to use fine-tuning and reinforcement learning on Microsoft Foundry to improve production AI agents. It covers when fine-tuning reduces cost and latency versus when RL delivers deeper performance gains, with practical demos including a retail customer service agent scenario. The content is targeted at developers and ML engineers building production agent systems who want to understand evaluation techniques, grader configuration, and safe redeployment.