Most people think redacting a document means drawing a black box over some text. It does not. The name is still there, sitting in the file, fully selectable by anyone who opens it in the right application. Courts have leaked witness identities this way. Government agencies have published records where the redactions came off with a single mouse drag. The problem is older and more widespread than most people realise.
DokMine is a document tool that takes a different approach. Hand it a PDF, a Word file, an Excel workbook, a PowerPoint deck, or a scanned image, and it finds the personal data inside: emails, phone numbers, names, URLs, IDs, credit card numbers, IBANs, IP addresses. Then it either extracts that list for you, or rewrites the file with those values removed and hands it back in its original format. A .docx comes back as a .docx. A .pdf comes back as a .pdf. Nothing gets flattened to plain text.
What makes this interesting is the depth. DokMine does not just scan the visible body text. It looks inside tracked changes, speaker notes, hidden spreadsheet sheets, formula caches, embedded screenshots, and document metadata. It runs OCR on images inside documents so text that exists only as pixels does not slip through. That is the part most tools skip entirely.
Key Features of DokMine
Format-Preserving Redaction
When DokMine rewrites a file, the structure survives intact. A Word document keeps its styles, tables, headers, and numbered lists. An Excel workbook keeps its formulas, sheet structure, and number formats. A PDF keeps its selectable text on pages that were not touched. You get a working document back, not a printout of one.
OCR on Embedded Images
A screenshot pasted into a slide deck is just pixels. A scanned signature page inside a PDF has no text layer at all. DokMine runs OCR over these images, detects personal data in the recognised text, and blacks out the matching regions directly in the image data. There is no recoverable overlay. The pixels change. This applies to standalone image files too: PNG, JPG, TIFF, BMP, WEBP are all processed the same way.
Extraction Mode
Before you redact anything, you can run extraction on a file or a whole batch. DokMine returns a grouped list of everything it found: emails, URLs, phone numbers, names, card numbers, each itemised per file. This is useful for auditing a document set before it goes out, checking what a folder of old invoices actually contains, or establishing which files hold personal data at all before deciding what to do with them.
Tokenised Redaction
Instead of blacking everything out, you can replace matched values with consistent typed tokens: [EMAIL_1], [PHONE_2], [NAME_1]. The same value maps to the same token throughout the entire document. A contract still reads as a contract. You can still tell that two mentions referred to the same person; you just cannot identify who that person is. This is particularly useful for turning a real client case study into an anonymised one without rewriting it from scratch.
How It Helped Sara Clean Up a Client Report at 9 PM
Sara works in compliance at a mid-sized consulting firm. At around 9 PM on a Tuesday, she was preparing a project debrief to share with a new client. The document was a 34-page Word file compiled over several months by multiple team members. It contained references to a previous client: a name in the header on every page, an email address in a comment left by a colleague, a phone number buried in a footnote on page 22, and a scanned letter embedded on the last page.
She uploaded the file to DokMine and ran extraction first. The results came back grouped: three email addresses, two phone numbers, one person name, and a note that the embedded image had been OCR’d and contained a match. She could see exactly what the document exposed before touching anything.
She then ran redaction with token mode enabled. The file came back as a .docx in under a minute. The header now read [NAME_1] on every page, the comment had been rewritten, the footnote was clean, and the scanned letter had the personal details blacked out in the image itself. She opened it in Word, checked the notes pane and the header, and sent it. The whole process took about four minutes.
How DokMine Processes Your Files
Step 1: Upload
Drag a single file or a batch onto the workbench. Supported formats include PDF, DOCX, XLSX, PPTX, EML, and standalone images. Files are used only to produce your result and are removed from the server after processing. Download links expire. Nothing reaches any analytics provider.
Step 2: Choose Your Detectors
Select which entity types you want to target: emails, phone numbers, URLs, person names, credit card numbers, IBANs, IP addresses, or some combination. Card candidates are tested against a Luhn checksum; IBANs are validated against their mod-97 check digits before being reported, which cuts down on false positives.
Step 3: Extract, Then Redact
Run extraction first to see what the file contains. Then run redaction on the same selection. Choose blackout for a clean visual result, or typed tokens to keep the document readable. Download the returned file and review it yourself before sharing: DokMine flags images it could not read rather than silently passing them as checked, so you always know where a human eye is still needed.
Who Should Try DokMine
If you regularly share documents outside your organisation, prepare files for publication, or deal with data-protection requirements in any capacity, DokMine is worth testing. It is free to start with no card required. The free tier removes files after processing, so there is no commitment involved in finding out what your documents actually contain. Try it at dokmine.com and run extraction on something you were about to send out or upload into an LLM like ChatGPT/Claude/Gemini.
