Building Self-Learning Loops for Your Agent — Fuad Ali, Arize AI

Why watch

Fuad Ali demonstrates how to build self-learning loops for AI agents using the OpenAI Agents SDK instrumented with Arize. The tutorial walks through converting a production failure (a shopping assistant with broken price filtering) into a complete improvement cycle: trace collection, code investigation, fix proposal, and automated testing against failed cases. Shows how agents can consume telemetry data alongside application code to diagnose issues, using Arize Signal for grouping failures and opening investigations.