Microsoft Copilot Studio - Automatic Evaluation from the Test Pane

We are announcing the ability to automatically evaluate each user query and agent response during testing using the General Quality Grader in Microsoft Copilot Studio. This feature will reach general availability on March 31, 2026.

How does this affect me?
Users will be able to enable an opt-in toggle in the Test Pane to automatically evaluate every interaction while testing their agent. Evaluations will run in the background and provide real-time quality feedback without requiring manual execution. This will allow users to identify and address response quality issues earlier in the build process before publishing.

What action do I need to take?
This message is for awareness, and no action is required.