Read
Turn a scanned PDF into text you can search, quote and annotate
That rare old textbook, the library scan you have had for years, the conference proceedings only available as photocopies — they are dead files until the photographs become characters. This turns them into living documents you can search, quote, annotate, and think with.
That gap between "I have the book" and "I can work with the book" is where most archive material, older papers and library scans sit. Getting across it is easy to do badly and hard to notice when you have.
Auditable page by page, not one accuracy number
The failure mode of bad recognition is not obvious garbage. It is a page that came out 70% right, whose errors you meet three weeks later inside a quotation you trusted.
- Every page gets its own record, and pages are flagged when the result looks wrong.
- The check is on the shape of the result, not on finding expected words — so it behaves the same on a language it has never seen.
- You can re-run recognition on just the pages that came out badly, instead of redoing the book.
- Search it
- Highlight it
- Ask about it
- Quotes land on the page
Recognise once, then every AI feature can read it
Recognition is not a one-off export. MindMax keeps the text on your computer, matched to the book by its contents, and everything you do with it afterwards reads that same text — so the credits are spent once, not every time you come back.
- Ask the scan a question, and the quotes in the answer open on the page image they came from.
- Run an AI reading pass over the whole book: a guide to its structure and notes in the margins, exactly as with a born-digital PDF.
- It joins your knowledge base: searched alongside everything else in the folder, and cited in answers that span several documents.
- Translate the whole book from the recognised text, with nothing to recognise a second time.
- Kept on your computer with a record for every page, matched to the PDF by its contents rather than its name — so renaming or moving the file does not lose the work, and the PDF itself is never modified. Highlights and notes work as normal, and reopening the book next month re-runs nothing.
- Ask itQuotes land on the page image.
- AI reading & notesGuides and margin notes, like any PDF.
- Knowledge baseSearched and cited with the rest of the folder.
- TranslateThe whole book, from the same text.
From photographs to something you can quote
Open the scan
It opens as images. Nothing has been processed and nothing has been spent.
Start recognition when you decide it is worth it
On this document, not on your folder.
It runs in batches and reports as it goes
Progress is kept per page, so closing the app does not throw away completed work.
Check the per-page ledger
Clean pages and flagged pages, listed. This is the step that stops a bad scan becoming a bad quotation.
Re-run the bad pages only
Fix the ten pages that failed instead of the four hundred that did not.
Use it like any other document
Search, highlight, annotate, and ask questions that come back with quotes linked to the page.
Where it gets hard
- Quality is set by the scan
Skewed pages, tight gutters, bleed-through from the reverse side and low-resolution photocopies all cost accuracy. No amount of processing recovers what the scan did not capture.
- Handwriting and marginalia are unreliable
Treat anything handwritten as a hint, not as text.
- Dense tables and multi-column layouts
The hardest case. Reading order is often wrong even when the characters are right.
- Long books take a while
It runs in batches and survives interruption, but a four-hundred-page scan is not instant and is not cheap.
Common questions
Does MindMax OCR my scanned PDFs automatically?
No. Recognition is started by you, on the document you choose. Running it silently on everything would spend your allowance on files you never intended to process.
Does OCR change my original PDF?
No. The PDF you scanned is not modified. The recognised text is kept by MindMax on your computer, matched to the file by its contents rather than its name, so renaming or moving the PDF does not lose it.
How do I know whether the recognition is any good?
Quality is reported per page rather than as a single number for the document, so pages that came out badly are visible up front instead of surfacing later as a strange quote.