Microsoft Copilot Studio: Improvements to agent evaluations experience
Microsoft Copilot Studio is adding enhanced agent evaluation capabilities including richer explanations, reasoning traces, knowledge source citations, and run comparisons. These improvements help makers better understand and troubleshoot agent performance and quality issues.
Microsoft's description
Microsoft Copilot Studio is enhancing the Evaluations experience to help makers better understand agent quality and behavior. New capabilities include richer evaluation explanations, agent reasoning traces, cited knowledge sources, evaluation run comparison, support for larger datasets, customizable test generation, and dataset generation from knowledge sources. These improvements help makers identify quality issues more quickly, understand why evaluations succeed or fail, compare results across runs, and create higher quality evaluation datasets with less manual effort. Additional validation and guidance will help makers resolve configuration issues before running evaluations.