An AI chat your team can use with real client data

Every message is masked before it reaches the model. Names, tax codes, IBANs and other identifiers become labels such as PERSON_1, the model answers on the labels, and your browser puts the real values back into the reply.

The map between labels and real values stays on our servers in the EU for the session and is deleted when it ends.
02

Your people already use AI with client files

Teams use AI because it saves them hours, and when the company gives them no approved tool they use their own accounts. The Microsoft and LinkedIn Work Trend Index found that 78% of people who use AI at work bring their own tools. In those accounts the company cannot see which documents go out, to which provider, or for how long they are kept.

03

What it costs when client data goes into a personal AI account

The text sent to a chatbot leaves the company with all the names and numbers it contains. Cyberhaven measured that 34.8% of the corporate data employees put into AI tools in 2025 was sensitive, and IBM's Cost of a Data Breach 2025 reports that one organisation in five had a breach caused by shadow AI, with an average of $670,000 more per breach where shadow AI was high. For a law firm, an accountant or an HR team the client's name in a prompt is also a question of professional secrecy and of the GDPR, because the provider becomes a recipient of the data.

Sources

All the figures, with the method behind each one →

04

How the secure chat works

  1. You write as usual, with real names and numbers, or paste a passage from a document.
  2. Before the message leaves, the Redactic engine finds names, organisations, addresses, tax codes, IBANs and other identifiers, including those recognised by their check digits, and replaces each with a label.
  3. The model receives only the labelled text and answers on it.
  4. Your browser shows the answer with the real values back in place.

The conversation is not saved: each visit opens a new session, and the map between labels and values expires with it.

05

What the administrator controls

06

Where it fits

Built for law firms that keep client confidentiality, payroll offices and accountants, HR teams that handle CVs and payslips, psychologists and expert witnesses and estate agents with tenant files.

FAQ

Frequently asked questions.

Does the AI provider see my clients' names?+

The provider receives the message with labels in place of the names and identifiers that Redactic detects. The real values return only in your browser.

Is the masked text anonymous?+

It is pseudonymised. The labels hide who the text is about, and the GDPR still treats the text as personal data for whoever holds the map, which is why the map stays on our side and expires with the session.

Are conversations stored?+

No. Each visit opens a new session, there is no conversation history, and the map between labels and values is deleted when the session expires.

Can we choose the models?+

Yes. The administrator chooses the models the team can use and can restrict the list to providers hosted in the EU or with zero data retention.

Bring AI into the company with the data protected

Tell us how many people would use it, which documents they work on and which rules your sector follows. We prepare the workspace, the models and the agreement for your organisation.