Docmind
Docmind is capture: intelligent document processing. A partner sends a business document (PDF or image). Docpier classifies it, extracts typed fields with confidence, and returns JSON your product can file, display, or pass downstream.
Typical buyers: ERPs, DMS, banks, fiduciaries, and portals that already own the customer relationship and need data — not another UI.
What the API does
| For a product manager | For an architect |
|---|---|
| Replace manual typing of invoice headers, VAT, and counterparties. | Multipart POST → 202 + documentId. Processing is asynchronous (OCR, classify, extract). |
| Support Swiss and EU commercial paper: invoices, receipts, reminders, bank paper, policies. | Result is a typed documentType discriminator on data[]. Branch in your domain, not on filename. |
| Operators correct mistakes; those corrections train quality. | Batch feedback POST with the full corrected resource. |
You do not poll. Docpier POSTs PROCESSING_FINALISED or PROCESSING_FAILED to your webhook. Then you GET.
MIME: PDF, JPEG, PNG, TIFF. Path version is /v1 only. Scope: docmind.capture (Keycloak still issues idp.capture).
A Docmind workspace is not a Docbooks workspace. Enable capture and booking if you also post bookings.