Upload a contract, a payslip or a scanned file. Redactic finds the personal data, you check each item on the page, and you download a file that is masked with labels or blacked out for good.
Each value becomes a label of its category, numbered the same way in the whole document, so "Marco Rossi" and "Rossi" both point to ⟦PERSON_1⟧. You can send the masked text to an AI tool and put the real values back afterwards in your browser. This is pseudonymisation under GDPR article 4(5).
The values are removed from the file and a black box takes their place. There is no map to bring them back. Use it for documents you share with a court, a client or the other party.
Nothing leaves the console until you have seen it. In the review page you can:
When the review matches what was detected, "Apply and download" hands back the first result without running the job again.
A machine learning model running on our servers in the EU finds names, organisations and addresses. Rules that check the control digits find tax codes, IBANs, identity numbers and VAT numbers, for example the Italian codice fiscale, the German Steuer-IdNr, the French NIR, the Spanish DNI, the Dutch BSN and the Polish PESEL. You pick up to four countries for the document, and you can add a list of names that must always be masked.
Unredacted content is never sent to a third-party AI service for detection or redaction.
PDF, scanned PDF and images (the text is read with OCR), Word, Excel and CSV (column by column, the table keeps its structure), plain text and Markdown. A PDF, Word or Excel file comes back in the same format, with the layout kept.
Run an AI task on the masked document, or open the secure chat. Under each message you can open "What the AI sees" and read the exact text the model received, with labels in place of the names. The answer comes back with the real values restored in your browser. If you prefer your own AI tool, copy the masked text and use it there.
Every job comes with a redaction certificate that lists the categories found and the hashes of the original and the produced file. Documents are processed in the EU and never stored. The data processing agreement is signed when you subscribe, and the console is available in 14 languages.
Yes. The review page shows the document as it will be downloaded, and every item can be kept, changed or added before you apply.
Mask replaces values with labels that you can restore later, which is useful before an AI task. Black out removes them from the file for good.
Yes. Scanned pages and images are read with OCR, and the values found are masked or blacked out on the page.
No. Documents are never stored and never used to train AI models.
No. Masked data is pseudonymised and remains personal data under the GDPR. Review the output and the remaining context before you share it.
Upload a file and look at the review page. You will see each item Redactic found before anything is downloaded.