AI Agent Observability

Elad Eizner, Sr. Solutions Architect, AWS | Tal Rosenstein, Sr. TAM, AWS

AWS Summit Tel Aviv 2026 · AWS AI Cloud Ops, Infrastructure & Security · 41 min · 16 pages

AI agents in production face critical observability challenges including execution tracking, performance monitoring, and debugging across distributed systems. This session covers current industry pain points and demonstrates AWS’s solution approach. You will see a real-life implementation demonstrating how to build comprehensive observability for applications built with Amazon Bedrock AgentCore and Strands. Using Amazon Bedrock AgentCore Observability, we’ll leverage OpenTelemetry traces, structured logging, and CloudWatch dashboards to demonstrate bottleneck identification and root cause analysis techniques. Learn how to evaluate agent behavior patterns and performance metrics. Through practical use cases, gain implementation strategies for scaling observability in enterprise environments and improving production reliability.

Generated automatically from the recording's transcript. It's meant to replace re-watching, not the recording itself — verify against the video at the timestamp before quoting formally. Slides belong to the speakers and organisers. Produced with recording-to-pdf.