Câu 6: AB-620: Designing and Building Integrated AI Agent Solutions in Copilot Studio

A team is developing an agent in Copilot Studio and wants to evaluate its behavior before expanding access. The team needs a repeatable, consistent method to test the agent using predefined interactions. You need to ensure that agent responses can be evaluated consistently across multiple test runs. What should you do?

1. A. Review evaluation scores from previous production sessions.
2. B. Enable telemetry and analytics to monitor intent matches, fallback usage, and response frequency.
3. C. Define test sets of prompts and expectations. — đáp án hiện tại
4. D. Change the evaluation method to a qualitative approach.
5. E. Publish the agent to additional channels for broader usage.