OCR and document extraction

Clear source images and an explicit output format improve the quality of the organized result.

Supported input

Use common images and scanned PDFs. Keep multi-page documents in order, and ask for uncertain fields to be marked when pages are skewed, blurry, or handwritten.

State the output

Request plain text, a paragraph-preserving Word file, CSV, or field-based JSON. For invoices and forms, list the fields you need.

Review points

Names, numbers, decimal points, dates, and table columns are prone to recognition errors. Sample key pages and request a low-confidence field list before downloading.