RelevX Redactor automatically detects and removes personal data from official documents. We built it for Santa Coloma de Gramenet City Council and it gave them back over 1,600 hours a year.
On the left, a document as it comes out. On the right, the same document ready to publish.
Original · 6 personal data items
Anonymised · ready to publish
Every public body that publishes documents ( judgments, contracts, minutes, resolutions ) is required to strip out the personal data they contain first. Both the GDPR and Spanish data protection law demand it. It is not optional and there are no shortcuts.
The problem is that doing it properly is slow. You have to read the whole document, find every name, every ID number, every address, every account number, and cover them one by one. In Santa Coloma that meant about 20 minutes per document. Multiplied by 5,000 documents a year, that is over 1,600 hours: the equivalent of one full-time person doing nothing but redacting data.
And the worst part is not the time. It is that a manual, repetitive process at that scale fails. One missed surname on page eleven of a thirty-page document is enough for that PDF to be published with personal data inside. In a public body that is not an oversight: it is a data breach that must be reported.
Three steps. The second is optional and is exactly what makes the result trustworthy.
One or two hundred. The application opens them, extracts the text and locates the personal data: names and surnames, national ID numbers, postal addresses, phone numbers, emails, IBANs and bank cards, number plates and case references.
The application flags what it has found and separates what it is sure about from what it is not. An IBAN has an unmistakable format; a proper name that is also a place name does not. It shows you that second list so you decide, and you can manually add anything you want covered even if it was not detected.
Every correction is remembered
The document comes out with the data genuinely removed, not hidden under a black box that can be lifted by copying the text. It also generates a log of what was anonymised in each file.
Most anonymisation tools upload your documents to a server. That changes the legal problem entirely.
And in any of the ways it may be written, not only the one you expect.
This is the baseline, not the limit. The set expands with whatever fields each organisation needs: a court, a council and a clinic do not handle the same data nor call it the same, and some information identifies someone in one context and not in another.
Every validation you make teaches it something. The system has memory and sharpens itself with use.
When you confirm a fragment was a name, or manually flag something it had missed, that correction is not lost when you close the document. The system keeps it and applies it to the next ones.
The practical effect shows quickly: the first documents from an organisation need more review, because the system does not yet know how you write, your case-file formats or how you name the parties. As more are processed, the list of doubts shortens and reviewing goes from being a task to being a glance.
That is the difference between a tool that always does the same thing and one that adapts to how you work. And it is also why the validation step is not a nuisance: every minute you spend correcting today is time you will not spend tomorrow.
You can work document by document or load a whole batch and process it at once. Each file in the batch keeps its own validation panel, and once all are reviewed they download together.
For a monthly publication of resolutions this changes the scale of the work: it is not anonymising one document fifty times, it is launching the batch, reviewing what the system flags as uncertain and downloading.
If you want to see it in action, at redactor.relevx.com you will find the product site, with more technical detail and the option to book a demo. It is available in Spanish, Catalan and English.
It is the first question everyone asks, and it is a fair one. The answer is that personal data is not always written the same way.
The same person can appear in very different ways within the same document. An ID number may or may not carry a hyphen. An address may be split across three lines.
That is why the system does not look for text strings: it interprets the document. It understands that a fragment is acting as a person's name from the role it plays in the sentence, even if it has never seen it before. And that is exactly why the review step exists: when the interpretation is not certain, it asks rather than deciding on its own.
The distinction sounds like a nuance and it is not: it decides whether the document is still subject to the GDPR or no longer is.
The data is replaced, but it can be undone
It is still personal data. Re-identification is possible, so the document remains within the scope of the GDPR.
The data disappears from the file
It is no longer personal data. With no way to recover it, the document falls outside the scope of the GDPR. That is what Redactor does.
Redactor does the second: the data disappears from the file. And this connects to the most repeated mistake in manual anonymisation, which is drawing a black box over the text in a PDF viewer. Visually it looks covered, but the text is still in the file: it can be recovered by selecting and copying it, or by opening the PDF with any extractor. Documents published by public bodies have leaked data exactly this way.
| By hand | Find and replace | RelevX Redactor | |
|---|---|---|---|
| Time per document | ~20 min | ~8 min | 2-3 s |
| Detects variants of the same name | Yes | No | Yes |
| Finds data you did not know was there | Sometimes | No | Yes |
| Removes the data from the file | Depends how it is done | Yes | Yes |
| Leaves a log of what was anonymised | No | No | Yes |
| Consistent at page 300 | No | Yes | Yes |
| Processes batches of documents | No | No | Yes |
The by-hand and find-and-replace times are those measured in the Santa Coloma de Gramenet City Council project on their own documents.
The fastest way to know whether it fits is to see it with one of your documents. Book a demo and we will try it on your case.
Redactor has its own site at redactor.relevx.com, also available in Catalan and English.