- Step1: Create scenario prompts, transcripts, or audio inputs for testing.
- Step2: Simulate agent conversations in customizable environments.
- Step3: Launch evaluations using built-in or custom metrics.
- Step4: Compare evaluation results with transcripts and audio replays.
- Step5: Monitor production calls and evaluate live performance.
- Step6: Set alerts for performance thresholds and off-path behavior.
- Step7: Analyze performance results and optimize your AI agents.