Prepare documents for AI and stay in control of your data
You paste a document into an AI, and inside there's a client's name, an email, an IBAN. It's already gone: you can't pull it back.
Find and mask names, emails, IBANs and API keys on your device. Detection can miss things, so check and correct the result, then copy the prepared text into the AI you use.
A .dmproj.json file saved from here: the text, the dictionary and your corrections come back.
Paste the AI reply: the placeholders this tab knows go back to the real values.
NewSeveral documents, one matter
Work on contracts and attachments while keeping each document's review. Confirm matches between people and save the whole matter to pick it up later.
See how
- Add the documents of the matter, one after the other.
- Review each one: what the engine found, what you cover or free by hand.
- If the same name appears in more than one document, it is you who confirms whether it is the same person: nothing is linked on its own.
- Save the whole matter as a file and resume it later, revisions included.
Processing happens in your browser: your documents are not uploaded to our servers, and no account is required.
Protect your professional documents in seconds
NDAs, source code, financial reports and CVs: pick a template and watch the engine mask every sensitive detail.
How it works
Three steps. Your data never travels: only neutral tags do.
Don't take our word for it
Three ways to verify that your texts do not leave your device. The first one is already running.
It is counting for real, while you work. Anyone who uploads your files to a server cannot show this counter: it would light up.
This is not a promise: your browser enforces it: a strict Content-Security-Policy blocks connections to other servers. The only exception is the optional contact form, which the program uses only when you press Send.
Specialized Tools
Frequently asked questions
Do you run AI on my documents?
No document is ever sent to an AI service. Some tools use recognition models (to read the text inside scans) that run inside your browser, with the model files hosted on our own domain: no image and no text leaves the device. If you then choose to paste the result into ChatGPT, Claude or Gemini, that exchange happens directly between you and that provider, outside Secure Local Suite.
Is my text uploaded or stored anywhere?
No. There is no server and no database: everything runs in your browser's memory and is wiped when you close the page.
Does it work offline?
Yes. Once the page has loaded you can disconnect: sanitization and restore run entirely on your device.
How do I restore the AI's answer?
Open “Got an AI response?”, paste the reply and click Restore: every tag ([NAME_1], [IBAN_1]…) is replaced with its original value. The mapping lives in memory only, for the current session.
What data does it detect?
Names, companies, emails, phone numbers, Italian tax codes and VAT numbers (checksum-validated), IBANs, payment cards, IP addresses, amounts, API keys and passwords. Profiles let you mask only what matters in your context.
Why tags like [NAME_1] instead of asterisks?
Because AI models work better with consistent placeholders: the same person is always [NAME_1], so the answer keeps its meaning and can be restored exactly.
How do I clear everything?
Just close or reload the page: everything lives in memory, so nothing survives. For an explicit wipe there's “Reset session” in the footer, which clears the text, the result and the mapping dictionary, after that you can no longer restore an AI response from that session. If you only want to process a new text while keeping restore available, use “New text” instead.
Is it really free?
Yes: free, no account, no limits. PDF files can be uploaded directly; Word is not supported yet. If you need it, or company features, tell us through the contact form.