Watch a prompt
get intercepted.
From the moment a user starts typing, Ventrin is scanning locally — stripping names, IDs, and secrets before anything leaves the browser.
Three steps,
one control layer.
Your team uses AI as normal.
Ventrin runs invisibly as a Chrome extension. Users open ChatGPT, Claude, Gemini or Copilot and type as they always would — no training required. The protection operates entirely behind the scenes.
Sensitive data is stripped on-device.
Before the prompt is sent, Ventrin's local model scans the input and redacts names, dates of birth, NHS numbers, passport IDs, API keys and more — replacing them with labelled placeholders like [NAME] and [AWS_KEY]. Nothing sensitive leaves the browser.
The sanitised prompt is sent safely.
The cleaned prompt reaches the AI tool. Your team gets useful output — without ever exposing client details, patient records or infrastructure credentials. Admins see a full event log in the dashboard.
Let Your Team Use AI Safely.
Install in two minutes. No IT change required.
Manifest V3 · Local detection · No prompt storage by default.