How It Works
Four steps to validate your use cases
From persona setup to automated scoring, everything you need to ship confident usecase agents.
How the automated evaluation loop works end to end.
- A persona AI places an outbound call to your usecase's inbound number.
- Your usecase agent handles the conversation using its configured prompt and tools.
- The call is recorded and transcribed in real time.
- An evaluator AI reads the transcript and scores it against scenario success criteria.
- Results appear in the Reports tab with pass/fail status and improvement notes.
Phase 01
Create Personas
Define simulated callers with demographics, tone, personality, and background. Each persona gets a dedicated evaluation phone number for live AI-to-AI test calls.
Phase 02
Define Scenarios
Describe call situations, opening lines, and success criteria. Use the AI assistant to draft scenarios or build them manually.
Phase 03
Run Live Tests
Pick a usecase, select personas and scenarios, and launch a test run. Each combination places a real outbound call from the persona to your usecase agent.
Phase 04
Review Reports
See pass/fail status, scores, transcripts, and improvement suggestions for every persona × scenario combination in your run.
Demo Video