This guide takes a multi-page PDF from upload to a flattened copy. It assumes the service and app are running and the models are already downloaded.

​
Check the file first

Redact accepts PDFs of at most 25 MB and 25 pages. Larger files are rejected at upload. If yours is bigger, split it and redact each part. See file limits.

​
Plan for the time

Pages are analysed one after another, not in parallel. With models cached, expect roughly 20–30 seconds of OCR per page plus a few seconds of reasoning, so a 25-page PDF takes 10 minutes or more. The screen shows Reading page N of M locally… as it goes. Each page has its own time limit of 180 seconds in the service. A page that exceeds it fails the analysis. If you don’t need suggestions for every page, select Continue without local analysis at any point and mark the pages by hand.

​
Analyse and review

  1. Upload the PDF, for example public/test-files/redact/statement-multipage.pdf.
  2. Choose Private or Confidential, add a direction if you want, and select Prepare for review.
  3. When review opens, go through every page. Done stays disabled until each suggestion on each page is accepted or rejected.
  4. Add manual marks with Add redaction where a field was missed.
  5. Select Done, then Save copy.
The saved file is a new PDF whose pages are images. No text from the source PDF is copied into it. See About flattened copies.

​
If analysis fails part-way

Select Try again to rerun, or Back to change settings. If the error says the analyzer is busy, a previous analysis is still holding the service; see Troubleshooting.