A scanned PDF is a stack of photographs. OCR renders each page, reads it with a vision model, and writes an invisible text layer underneath the image — so the document looks identical but is now searchable, selectable and ready to convert to Word.
This takes a second on the first visit, then it is cached.
Drop in the scan.
Choose a searchable PDF or plain text.
Run OCR — it works page by page, so you can watch it go.
Very good on clean scans and printed text, including tables and multiple columns. Bad photographs, heavy skew and handwriting are harder, and the text layer will have mistakes in it.
Identical. The original image stays on top; the recovered text sits invisibly underneath it, which is how searchable scans have always worked.
Not this one. OCR PDF is part of Baker, at $10 a month — along with the other conversion, compression and security tools, and the AI ones. Signing, editing, watermarking, merging, splitting, rotating, organising and cropping are free, three documents a day up to ten pages each, with no account.