For ChatGPT, Claude and Gemini · browser and desktop
Let your team use AI. Keep client data out of it.
Pretzel checks every prompt before it's sent and catches personal data, passwords and code. It offers to remove them, so people can keep working. Detection runs on the device.
No proxy or network changes. No credit card.
chatgpt.comdemo
Hi Dana. What can I help with?
This is the real warning your team sees. Try the buttons.
On-device detectionPrompts are checked before they leave
Browser and desktopChrome extension and a Windows / Mac app
Rules per teamDifferent policies for Legal, Finance, Eng
The prompt stays privateAdmins see the rule and site, not the text
Set it up in an afternoon. Check on it once a week.
01InstallAdd the Chrome extension or push it with Chrome Enterprise. For desktop AI apps, install Pretzel Desktop. People sign in with their work account.
02Choose what to protectStart from a preset for your industry, or ask the assistant in plain words. Review every change before it goes live.
03Publish, then check inPublishing sends the policy to every device within minutes. The overview shows coverage, catches and anything that needs you.
Four parts, one policy
Tour the productFor employeesBrowser extensionChecks prompts in ChatGPT, Claude and Gemini and offers a safe way to send.For employeesDesktop appCovers AI desktop apps. Lives in the tray and asks before anything risky is sent.For ITAdmin consolePolicies, people, activity and publishing, on one screen you check weekly.For ITAssistantAsk for a change in plain words. It adds divisions, rules and members, and shows the diff first.