# EnterSense · voice

Product stage: production. Developer API: intent resolution available.

Product stage supplied by founder; this API resolves supplied transcripts, not audio.

## Input boundary

Acquisition adapter: client_transcript. The application supplies the input. The resolver does not access audio, cameras or the body. Execute emitted intents only through explicitly registered application handlers.

## Actions and customization

Use GET /api/v1/actions and GET /api/v1/templates. POST /api/v1/validate checks rules; POST /api/v1/resolve produces intents. Customize match, action payload, enabled and minConfidence. Confidence thresholds require an input confidence score. All matching enabled rules resolve in supplied order.

## Voice opens home

```json
{
  "input": {
    "confidence": null,
    "sense": "voice",
    "transcript": "open home"
  },
  "rules": [
    {
      "action": {
        "target": "/",
        "type": "navigation.open"
      },
      "enabled": true,
      "id": "home",
      "match": "open home",
      "minConfidence": null
    }
  ]
}
```

## Roadmap

Planning targets, not release promises. Each sensing release depends on research, consent, evaluation and operational readiness. Production voice, beta motion and university smell research are founder-reported product stages; the developer API documents its own implemented scope.

- Q4 2026: Build with voice. Explore motion. (Current delivery).
- Q1 2027: EnterSmell · research to evaluation (Proposed research milestone).
- Q2 2027: EnterTaste · first research interfaces (Proposed, validation dependent).
- Q3 2027: Sight and touch · expand deliberately (Proposed, validation dependent).

## Errors and privacy

Consult the generated OpenAPI for status codes and limits. Validation failures are structured errors; rate limiting is 429. The resolver is stateless and does not persist submitted input. Infrastructure request metadata may be retained. No medical use or PHI service is offered.
