For managing partners, practice administrators and compliance officers
Today it’s up to each individual employee, not the firm.
A memo banning ChatGPT doesn’t make it go away. With Nonimo, your staff keeps Copilot, Gemini, Claude and ChatGPT, while Social Security numbers, patient IDs and client names get swapped for placeholders on their own computer before a prompt goes out. Pilot it with one practice group first.
The problem with a ban
In July 2025, Propeller Insights polled 1,000 working U.S. adults who use AI on the job, on behalf of WalkMe. Of those, 78% admitted using AI tools their employer had not signed off on (WalkMe survey, SAP News, August 2025).
A no-AI policy doesn’t stop the drafting and the summarizing. It pushes them onto free accounts and home laptops, where your firm has no settings, no visibility and no say. The smarter move is to say yes and put the guardrails up first, so a client’s SSN never reaches a chatbot in the first place.
Pilot program
Your admin or outside MSP force-installs the browser extension in Chrome and Edge for just those users, from the console they manage the fleet with today. You won’t touch the firewall, and there are no drivers, proxies or root certs involved.
Folks who prefer hitting one key to clean up a document can add the Windows or Mac app too.
After that it runs by itself. Inside ChatGPT, Copilot and the rest, a name, an SSN or a bank account number becomes a label the moment before submit. No hotkey to memorize, no new step in anybody’s routine.
Every workstation running the desktop app can export a signed usage report for whatever stretch of dates you need. It counts redactions by category, never carries the underlying text and never points to an individual. nonimo.ai/verify tells any reader whether the file is authentic and unchanged.
Something concrete for an outside auditor, your privacy officer or the partners’ meeting.
Once the pilot group is comfortable, extend that same policy firm-wide.
At their desks
Hardly a thing. Same browser tab, same ChatGPT, Copilot, Gemini or Claude login as yesterday. By the time someone hits Enter, the extension has already put placeholders where the client specifics were, and it shows each one it changed. Clicking a swap restores the original. If your developers call the OpenAI API from in-house apps, Nonimo Gateway handles that traffic on your own servers.
Pick your field to see the IDs that show up in its documents, what Nonimo hides before an AI tool sees them, and the reason for each.
Before you write the policy
Yes. Wherever the Nonimo desktop app is installed, pull a signed usage report covering whichever weeks you pick. It lists how many items were redacted, category by category. It holds no prompt text and doesn’t name the employee behind anything. Run it through nonimo.ai/verify and anyone, your auditor included, sees whether it is authentic and untouched.
Nope. People stay in the ChatGPT, Copilot, Gemini or Claude window they already know. Whatever got redacted is highlighted for them, and if they need it back, it’s one click to undo.
Not if IT pushed it by policy. Chrome and Edge lock a force-installed extension in place, so the employee at the keyboard has no remove or disable switch for it; that is how managed browsers behave everywhere. The desktop app is opt-in, for anybody who likes it. Team-wide coverage comes from the policy-deployed extension.
Yes, for the setup. Whoever runs your IT, an internal admin or an outside MSP, deploys it through browser policy just like other company-managed extensions. If nobody does that for you today, ask and we’ll point you to someone who can.
They don’t. Redaction happens locally, so ChatGPT or Copilot receives the cleaned-up prompt, the client specifics remain on that machine, and Nonimo never gets a copy of anything your people write.