PromptKit FrançaisRequest the beta
Menu

PromptKit for teams

People paste client data, credentials and internal code into AI chats every day. PromptKit stops it in the browser, before the message leaves, and you set the rules once for everyone.

Contact us about your teamRead the security page

The problem

Sensitive data ends up in AI chats

An API key in a stack trace, a customer's e-mail in a draft, an IBAN in a spreadsheet: once sent to ChatGPT, Claude or Gemini, it is out of your hands.

Banning the tools rarely works

People keep using them on the side. A guard rail in the browser lets them keep working while the riskiest items stay inside.

Personal data has rules

Under the GDPR, sending personal data to a third-party service needs a reason and safeguards. PromptKit helps people avoid sending it by mistake. It does not make an organization compliant on its own.

Deploy it once

Managed policy

PromptKit reads its settings from the browser's managed policy: Chrome and Edge, through Group Policy, Intune or any tool that writes browser policies. You can check the result in chrome://policy.

Your words and your rules

Deploy keywords (project or client names) and custom rules, for example a client code pattern. They apply to everyone.

Minimum actions

Set the least strict action for each category. People can go stricter, never looser.

Locks

Force protection on for every site, and remove Send anyway and the pause.

Fail closed

If a check cannot run, the message is blocked instead of going through.

Games and files

Turn the waiting games off, or require file checks.

An example

A policy for Chrome

On Linux, in /etc/opt/chrome/policies/managed/promptkit.json. On Windows, the same structure is deployed by Group Policy or Intune.

{
  "3rdparty": {
    "extensions": {
      "<EXTENSION_ID>": {
        "blockFrom": "medium",
        "allowOverride": false,
        "failClosed": true,
        "keywords": ["Project Orion"],
        "customRules": [
          { "id": "client", "label": "Client code", "pattern": "CLI-\\d{6}", "severity": "high", "token": "CLIENT" }
        ]
      }
    }
  }
}

What stays private

Nothing is reported back

PromptKit never sends anything to the administrator, or to us. There is no dashboard of what people typed, because nothing leaves the browser.

Easy to check

Two permissions: storage and three sites. Your security team can verify that it makes no network request, in a few minutes, with the steps on the security page.

Talk to us

Tell us about your team

How many people, which browsers, and what you need to keep out of AI chats. We answer within a few working days. Write to [à compléter: e-mail de contact].

Today

Everyone can use PromptKit for free, with the limits of the free plan. A Pro license is personal. For a team, contact us before buying.

Request the beta