> ## Documentation Index
> Fetch the complete documentation index at: https://docs.verify-group.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Voice intelligence

> Trigger AI voice analysis on call sessions, interpret stress and deception scores, and combine signals into a composite participant fraud score.

## Scenarios

| Scenario                   | Use case                      |
| -------------------------- | ----------------------------- |
| `customerSupportComplaint` | FNOL / intake calls (default) |
| `fraudSignalDetection`     | SIU / fraud investigation     |
| `debtCollection`           | Collections calls             |

## Trigger analysis

```bash theme={null}
POST /api/v1/voice/sessions/{sessionId}/analyze
```

```json theme={null}
{"forceReanalysis": false}
```

## Result structure

```json theme={null}
{
  "analysisResults": {
    "primaryScore": 62,
    "recommendedDecision": "MANUAL_REVIEW",
    "headline": "Moderate stress detected",
    "keyScores": {"stress": 68, "deception": 45, "coaching": 31},
    "importantCallSignals": [{"type": "elevated_stress", "confidence": 0.78}]
  }
}
```

## Composite fraud score

```bash theme={null}
POST /api/v1/forensics/poi-fraud-check
```

```json theme={null}
{"claimId": "claim-uuid"}
```

Returns per-participant composite scores combining voice, KYC, and evidence signals.

<Tip>When a `SCENARIO_MISMATCH_NOT_SCORED` warning is returned, the backend retries automatically with the suggested scenario.</Tip>