For councils, firms and care providers
Nonimo lets staff keep using ChatGPT, Claude, Copilot or Gemini — while client names, IDs and case numbers are masked before anything leaves each computer. And it gives you the evidence that the control was on.
IT rolls it out across the team like any other managed extension. It masks before sending, and can block what must never leave — a card number, an API key, a client file.
No kernel drivers. No traffic interception.
Installed on your infrastructure. It receives counts and categories — how many names, how many IDs, under which policy — and never the text itself. We could not read your documents if we wanted to.
Your data stays yours. Literally.
One page that shows the control was on: what was caught, in which categories, under which policy version. It is the document an auditor, a privacy officer or a board asks for.
Evidence, not promises.
Two to four weeks measuring real risk without changing anyone’s work. You see what would have left, by category — most organisations are surprised.
With real numbers on the table, you decide per category: observe, warn, mask or block. IT applies it centrally. Staff see explanations, not obstacles.
The panel produces the report. When someone asks “how do you control AI use?”, you have a better answer than a policy PDF.
No. The engine runs on each computer and has no code to connect to any server. What the compliance panel receives is counts and categories — never the text itself. The panel itself is installed on your infrastructure, so even that summary stays with you.
No. Staff keep using ChatGPT, Claude, Copilot or Gemini as they already do. Nonimo works before the text leaves: masked items are visible, explained in one sentence, and reversible with one click.
Nonimo does not use lists of names or an AI model to decide. It finds names by their position — the form label, the signature, the title — so an unusual surname is caught exactly like a common one. That behaviour is deliberate and testable in the live demo.
Yes. Policy is set centrally by IT: observe, warn, mask, clean or block, per data category. A pilot usually starts in observe mode, which measures risk for two weeks without changing anyone’s work.
Through your IT provider or internal IT: a browser extension rolled out by policy (Chrome, Edge) plus an optional desktop app. No kernel drivers, no network appliance, no traffic interception.