PDF upload

Split a PDF into page images, upload to file service, optionally run extraction, then save rows into task-specific dataset slugs.

Tenant

Company (for slug)

Extraction model

Used only for per-page Run extraction. Normalization/classification are saved to datasets for manual expected-output editing later.

Enable preprocessing

Pipeline mirrors scripts/extract.py: server renders with MuPDF.js (PyMuPDF parity), then enhance → bw → trim → pad.

PDF file

Drag and drop your PDF here or click to browse