When my agent is slow or wrong, where do I even look?
Microsoft Foundry adds agent tracing via Azure Monitor for debugging slow or incorrect agents
Microsoft Foundry now integrates with Azure Monitor Application Insights to provide distributed tracing for AI agents, exposing inputs, outputs, tool calls, latencies, and token counts per request. Developers can drill into individual spans to diagnose where an agent went wrong or slow. This is a developer-tooling tutorial, not a product announcement, but signals Microsoft's push to make agent observability a first-class concern in its AI platform.