Galileo Instrumentation for LangChain Apps

Introduction

In Monitoring Agentic AI Applications you instrumented and observed an agentic AI travel planner. That application is a Flask API wrapping a LangGraph workflow, where each node (coordinator, flight specialist, hotel specialist, activity specialist, and synthesizer) calls an LLM through LangChain’s ChatOpenAI.

In this follow-on workshop, you will take that same application and add Galileo instrumentation to it. Rather than re-instrumenting from scratch, you’ll attach a single Galileo callback to the LangGraph workflow so every agent’s LLM call is captured. This lets you compare and validate AI trace visibility across tools using a workload you already understand.