Finding what to redact is the other half. Draw boxes on a live preview with undo, or hand Bindery a list — literal strings or regular expressions — and let it search every page; it suggests emails, phone numbers and identifiers, and the on-device AI model (or your own API key, sending excerpts only) can propose names and addresses, each as a box you can remove before applying. A redaction log downloads with the result. Then finish the job the way a careful professional would: strip metadata, run the copy-and-search test in another viewer, and have a second person check anything high-stakes. Nothing is uploaded at any step.
How it works
- Drop the PDF and open Redact; draw boxes or add a search list
- Review every suggested box, remove the wrong ones, apply
- Strip metadata, then copy-and-search the output to prove the text is gone
Questions people ask
Why not just draw a black box in a markup tool?
Because the box is decoration. Markup and white-out change what the page looks like, not what it contains; the text object stays in the file and can be selected, extracted or searched. Bindery’s Markup tool warns about this and points you to Redact for anything that must truly disappear.
What happens to the rest of a redacted page?
It becomes an image. Rasterizing is what guarantees the covered text is gone, and the cost is that the other text on that page is no longer selectable. Pages without boxes are untouched, so keep boxes to the pages that need them.
How do I prove the redaction worked?
Open the output in a different viewer, press Ctrl+A, copy into a plain-text editor and search for a word you covered. Try a text extractor. Check comments, attachments and metadata. If anything survives, do not send the file. For legal or safeguarding work, have a second person repeat the test.