Live demo

Use AI. Without exposing customer data.

Use your own text to see how AVGfilter intercepts personal data before ChatGPT, Claude or another AI model can access it.

Try it yourself

From readable input to a safe AI request.

  • Personal data detected locally
  • Only tokens reach the AI model
  • A readable answer for the employee

Privacy filter, live

This online demo sends your input to a filtering service. Use the sample, not real confidential data. The browser extension and Doc Scanner work locally.

  1. Received
  2. Tokenized
  3. AI processing
  4. Restored
  5. Completed
01

Input

text from the employee

Detected personal data

Run the filter to see which data gets replaced.

02

Tokenized version

sent to the model

No tokenization performed yet.

03

What the model receives

literal payload

Nothing sent yet.

Detected data is replaced with tokens. The remaining text stays visible to the AI service.

04

Answer restored

readable for the employee

The answer is made readable again after processing.

Without protection no text reaches the modelready

Founding Pilot

Test AVGfilter in your own working environment.

We are taking a limited number of organisations into the Founding Pilot. Tell us where you use AI; we will review the application and contact you personally.

  • Test with your own workflows
  • Personal setup and guidance
  • No automatic account activation
Which AI tools do you use?

This short check keeps automated submissions out.