Fleet AIDevelopers
Fleet AI API
Bring vehicle-risk analysis into your existing application. Send supported telemetry from your backend and retain the evidence behind each result.
Keep your workflow.
Add maintenance context.
Partner access uses a provisioned API key. Requests are scoped to the partner identity; customer browser sessions are not a substitute for partner credentials.
Keep keys on your server. Never embed a long-lived API key in a public website, a shared screenshot, or a client-side application.
- Predictions
- Submit vehicle samples and inspect risk, confidence, diagnosis, and model availability.
- History and fleet review
- Retrieve stored prediction history and a partner-scoped fleet view.
- Event delivery
- Use configured webhooks or a scoped event stream. Browser streams require a short-lived ticket issued by your backend.
Start with a small integration.
| Method | Endpoint | Purpose |
|---|---|---|
| POST | /api/partner/predict | Analyze samples for one vehicle. |
| POST | /api/partner/predict/batch | Submit a group of vehicles. |
| GET | /api/partner/status?vehicleId= | Check model service availability. |
| GET | /api/partner/fleet | Review recent fleet predictions. |
| POST | /api/partner/webhooks | Register an event destination. |
Requests require authorized partner access. See the reference for parameters, error handling, and integration constraints.
Handle uncertainty as part of the result.
A successful HTTP response does not mean a vehicle is safe or that a breakdown has been confirmed. Handle insufficient data, unavailable sensors, and fallback predictions explicitly.
Synthetic validation is not field validation. Confirmed inspections and maintenance outcomes are needed to evaluate prediction quality in your operating environment.
Open integration documentation