Lawyers & notaries module · live
This is not one more PDF tool. It is an assistant that reads and structures legal documents: it makes a scanned deed searchable, pulls the schedule of exhibits out of an opponent's file, and picks out the dates and amounts in a contract.
Drop in a deed, a judgment or a case file
PDF or image — 500 pages, 50 MB maximum
Case file deleted 2 h after your last action
| Feature | Problem it solves | What you get |
|---|---|---|
| Deeds and exhibits made searchable | Scanned deeds nobody can search: the PDF holds nothing but a picture of the page. | Conversion to a searchable PDF — a keyword search runs through 500 pages at once. |
| Schedule of exhibits | Going through an opponent's exhibits takes hours. | The list of exhibits pulled out on its own — number, date, title — into Excel or JSON. |
| Timeline and dates | Rebuilding by hand the sequence of events in a contract. | Key dates and cited amounts detected, gathered into a summary you can click through. |
| Anonymisation and redaction | Protecting personal data: the GDPR, and professional privilege. | Names, addresses, account numbers and sensitive data spotted by the OCR, redacted for good. |
Documents are processed on our own server, in Germany, and deleted two hours after your last action — or straight away, if you ask. No third-party artificial intelligence service is called, and nothing you upload trains a model.
A badly scanned notarial deed, a stamp, an initialled margin: that is where ordinary character recognition gives up. Tesseract reads first; on the pages where it says itself that it read badly, PaddleOCR takes over. On microfilm views from 1840, it finds twelve fields out of fourteen where the first finds one.
The original page is kept exactly as it is. That matters for line numbering in pleadings: a citation points at a line, and that line must not move.
You drop the PDF in; the module lays an invisible text layer over every page, word by word, without touching the image. You get back a PDF where Ctrl+F lands on the right spot in the scan, and where the page is exactly the one you uploaded.
No, and that is the whole point. The text and the pixels concerned are removed from the file, not covered with a black rectangle: nothing is left to select, to copy or to find under the mask. A rectangle laid on top would only have hidden the screen.
Up to 500 pages in a single upload, read in one pass. Depending on how dense the pages are, a full file takes three to twenty minutes; progress is shown throughout, and you can close the tab and come back.
Two engines rather than one. Tesseract reads first; on the pages where it says itself that it read badly, PaddleOCR takes over. On microfilm views from 1840, the second finds twelve fields out of fourteen where the first finds one.
The schedule of exhibits — number, date, title —, the dates and amounts cited, and full-text search across the whole thing. All of it exports as a searchable PDF, a redacted PDF, CSV for a spreadsheet or JSON for the detail.
Case files are processed on our own machines, in Frankfurt, and deleted two hours after your last action — or immediately, if you ask. No third-party artificial intelligence service is called, and nothing you upload trains a model.
The site is hosted in Frankfurt. Germany applies the GDPR and adds its own federal act, the BDSG, within a legal tradition unusually attentive to privacy; its supervisory authorities — one per state, plus a federal one — are reckoned among the most demanding in Europe, hosting providers included. That is the framework we chose for your files.
The geomatics module turns PDF or image maps into GIS files. The accounting module reads your receipts and invoices, picks out the taxes and redoes the arithmetic. All three run on the same infrastructure and under the same rules.