Verify a real photo against production-style policies
Pick a policy, use a sample image or upload your own, and see the verdict, confidence, failed criteria, and model reasoning returned by the same VLM pipeline used in production.
Pick a policy
Each policy is a separate set of rules the AI evaluates against.
Vehicle damage check
Flags visible body damage including dents, scratches, broken glass, and missing panels. Designed for fleet handover and rental return inspections.
Hypothetical policy modelled on fleet and rental return inspection flows.
Drop a photo here, or click to browse
JPEG, PNG, or WebP — up to 10MB
Results appear here
Pick a policy and a photo on the left. We'll show the verdict, confidence, and a breakdown of every criterion the AI evaluated.
Fast enough for rider flows
Designed for end-of-ride, delivery, and handover moments where seconds matter.
Structured API output
Every response includes a verdict, confidence, category, criteria IDs, and plain-language reasoning.
Demo images stay transient
Uploaded demo photos are analyzed in memory and are not written to customer verification history.
1 · Pick a policy
Each policy is a list of criteria the AI evaluates. Standard policies are single pass/fail. Advanced policies return per-criterion verdicts and a confidence score.
2 · Drop a photo
Use one of the sample images or drop your own. JPEG, PNG, or WebP up to 10MB. Photos are never stored.
3 · Read the result
You get the same JSON the production API returns: verdict, confidence, criteria breakdown, and the model's reasoning.