VisionLab

VisionLab reads images the way you would name the result: define an extraction task (a prompt plus the typed fields you want), point it at an invoice, menu, screenshot, or diagram, and it returns structured JSON with a confidence score, the evidence it used, and flags for any secrets or PII it spotted.

Every run is a real vision call on the live Nyquest relay — cost and tokens are measured, not estimated — and each extraction task exports as a portable Nyquest Vision Package.

Enterprise-grade underneath: database-enforced per-tenant isolation, durable tracked jobs that survive restarts, and image bytes in S3 object storage. A Pro workspace gated behind Nyquest Pro, enforced on both the frontend and the backend.