Your files never leave your device
This is a claim you can check in two minutes. Here is what is true, what is not, and how to confirm it.
What happens when you drop a file
- The browser reads the file into your device memory. There is no upload field, form or request carrying its content.
- PDFs: read with pdf.js and pdf-lib, which run as JavaScript in the tab.
- OCR: Tesseract compiled to WebAssembly, in a browser worker. The language models and engine are our own files (extrair.pt/ocr/), not a third party’s.
- We store nothing. Closing the tab erases everything. There are no accounts and no file database.
What the site downloads (and when)
To process without a server, the site downloads the engine code. The first time you use an OCR tool that is about 9 MB (WebAssembly engine plus Portuguese and English models), from extrair.pt, kept in your browser cache. After that, processing makes no requests.
How to verify
- Open developer tools (F12) and the Network tab.
- Use a tool once so the engine is cached, then clear the list.
- Drop a file. The list should stay empty during processing.
Each tool also has a “Network requests” panel that counts requests the page makes while processing. It shows the page’s own requests; internal worker requests only show in the browser tools, so the steps above are the test that counts.
Visit analytics
We only load Google Tag Manager if you say yes in the notice. If you say no (or ignore it), the site does not contact Google. Even with “yes”, the content of your files is never sent: analytics events do not include it.
Limits, stated plainly
- Browser extensions, malware or a compromised device can see what is on screen. That is outside our control.
- OCR and field detection make mistakes. Always confirm important values (amounts, IBAN, NIF).
- We need recent browsers (WebAssembly and modern JavaScript).
Open-source software used
| Component | Licence | Used for |
|---|---|---|
| pdf.js | Apache-2.0 | Reading PDFs, text and pages |
| pdf-lib | MIT | Merge, split, rotate pages; build PDFs |
| tesseract.js + tesseract.js-core | Apache-2.0 | OCR in WebAssembly |
| tessdata_fast (por, eng) | Apache-2.0 | Tesseract language models |
| jsQR | Apache-2.0 | Reading QR codes |
| fflate | MIT | ZIP and XLSX |
| three.js | MIT | 3D animations |