The Azure Copilot Observability Agent autonomously investigates incidents by correlating logs, metrics, alerts, and ML anomalies. It surfaces root causes with visual charts and suggests next steps to resolve issues. The agent can also monitor AI agents in Microsoft Foundry, tracking Gen AI errors and token consumption. Users can write plain-language instructions to tailor the agent's autonomous behavior to their team's workflow. Microsoft Azure expert Matt McSpirit explains how to gain control of incident response at scale. The agent integrates with OpenTelemetry and native Azure signals for end-to-end full-stack tracing. It unifies telemetry from various sources into a consistent schema for a complete service health view. The agent learns about applications and environments to conduct detailed investigations. It provides comprehensive reports, including root cause analysis and visual evidence of findings. The Observability Agent can also be customized with natural language instructions to prioritize specific alerts and escalation rules.