Question

A Business Solutions Architect has deployed a custom agent built on the Microsoft Foundry Agents service into production. The operations team needs to trace individual request latency, correlate failed tool calls with downstream API dependencies, and set alerts on p95 response time so they can proactively tune the agent. Which monitoring approach best meets these requirements?

Accepted Answer

Integrate the Foundry Agents service with Azure Monitor / Application Insights to capture distributed traces, dependency telemetry, and configure alerts on latency percentiles.

Answer

Rely on the Copilot Studio analytics dashboard to review aggregate session counts and topic-level engagement trends.

Answer

Export daily conversation transcripts to Dataverse and build a Power BI report that summarizes total message volume per day.

Answer

Enable Microsoft Purview audit logging to record who invoked the agent and when for compliance review.