Every scan, every screenshot, into text. Offline. Once.

Yes — the full source is MIT at github.com/bensblueprints/textract-ocr-mvp, always. $19 buys the signed installer, 1-click setup and updates instead of npm i && npm start.

It's Tesseract's accuracy — excellent on clean scans, screenshots and printed documents, mediocre on handwriting or heavily skewed photos. It's not a cloud vision model, and we're not claiming it is. For sensitive paperwork the trade-off is worth it: nothing gets uploaded.

Real. Each page is rebuilt as the original image plus an invisible text layer positioned from OCR word bounding boxes — and the test suite round-trips the output through a PDF parser to prove the text layer extracts correctly, not just eyeballs it in a reader.

For OCR, yes. But Acrobat is a full PDF editor — forms, signatures, redaction, page editing. Textract does one job (getting text out of images and scans, offline) and does it for the price of one month of Acrobat.

Only if you download an additional OCR language, which is clearly surfaced with a progress UI. The OCR engine runs from bundled local files — never a CDN — and there's no telemetry, no uploads, no account.