OCR PDF
Planned support includes language selection, deskew, cleanup options, confidence warnings, and explicit cloud processing labels.
Phase gate blocked
OCR tools for scanned PDFs.
OCR is planned as a cloud workflow because scan cleanup, language packs, deskew, and searchable PDF output are quality-sensitive. The UI shows the required processing, error, export, and delete states, but production OCR launch stays blocked until Phase 1 KPI/support evidence passes.
OCR prototype gate: blocked
Evidence classification is missing. Production OCR navigation, launch copy, and mutating controls remain blocked until retained production KPI/support evidence completes every required Phase 1 gate.
- Phase 1 KPI gate artifact is missing or unreadable.
- Retained production KPI/support evidence has not been verified.
File
Searchable PDF job
Lease scan.pdf has been selected for OCR with searchable PDF output, text export, and visible cloud-processing disclosure.
Warnings
Review low-confidence pages
- Page 3Rotated receipt detectedDeskew before export
- Page 8Low contrast stampReview extracted text
- Page 12Table columns mergedExport text with warnings
Error state
If OCR fails, show the failed stage, provider boundary, retry action, support link, and delete controls for uploaded source files.
Searchable PDF
Planned support includes language selection, deskew, cleanup options, confidence warnings, and explicit cloud processing labels.
Scanned PDF to Word
Planned support includes language selection, deskew, cleanup options, confidence warnings, and explicit cloud processing labels.
Extract scanned text
Planned support includes language selection, deskew, cleanup options, confidence warnings, and explicit cloud processing labels.
OCR image to PDF
Planned support includes language selection, deskew, cleanup options, confidence warnings, and explicit cloud processing labels.
OCR API
Planned support includes language selection, deskew, cleanup options, confidence warnings, and explicit cloud processing labels.
OCR accuracy has limits
Handwriting, low contrast, rotated pages, mixed languages, and complex tables can produce errors. Users should review OCR output before relying on it.