Why Your AI Agent Fails in Production (And How to Catch It)

Why watch

A hands-on workshop demonstrating how to build production-grade evaluation pipelines for AI agents using DocsHound (LangGraph), OpenTelemetry, and OpenInference. The video covers standardizing traces across frameworks (ADK, LangGraph, CrewAI), implementing LLM-as-a-judge scoring, and tracking performance metrics. Designed for developers building AI agents who need reliable testing strategies beyond manual tweaking.