AI Agent Debugging with Galileo | Multi-Agent Observability Demo Pt. 1

Description

See how Galileo's agent observability platform helps teams debug complex multi-agent workflows in minutes. This demo with our lead AI Engineer, Pratyusha Duvvuri, walks through instrumenting a 4-agent customer ops desk, visualizing agent execution flows, and identifying tool ordering bugs using Galileo's agent graph view. In this comprehensive demo, you'll learn: - One-minute instrumentation with @log decorator - Automatic session aggregation across 7 customer scenarios - Agent Graph for bird's-eye workflow visualization - Trace-level debugging for individual agent executions - Identifying tool call ordering issues visually - Debugging multi-agent workflows without complex logging - Context adherence metrics for agent performance Galileo provides the only platform with 8 purpose-built agent eval metrics, automatic agent visualization, and session-level metrics. Perfect for AI engineers building multi-agent systems, customer support automation, and complex agentic workflows. 🔗 Try Galileo: https://app.galileo.ai/sign-up?utm_medium=organic&utm_source=youtube 📚 Docs: https://v2docs.galileo.ai/ 0:00 - Introduction: Multi-Agent Debugging Challenge 0:27 - Demo Series Overview: Favorite Platform Features 0:42 - Use Case: AI Customer Ops Desk 1:48 - Four-Agent Architecture Overview 3:48 - Galileo Instrumentation Strategy 4:09 - Live Code Demo: One-Minute Setup 4:42 - Running 7 Customer Scenarios Simultaneously 5:36 - Agent Graph: The Bird's-Eye View 6:12 - System Metrics Overview 7:09 - Trace View: Individual Session Debugging 8:12 - Root Cause: Incorrect Tool Order 8:42 - Why Agent Graph Matters for Debugging 8:51 - Preview: Deeper Use Cases in Future Sessions