Use LLMs without exposing sensitive data
Privnexa automatically detects and replaces sensitive data before the request reaches the provider.
Get started
Documentation
Try it now — the chat runs in your browser
No data leaves your device.
You've used your 10 free messages
Create a free account to keep chatting.
Create an account
The first load downloads the model and caches it.
Cliente → [Privnexa] → [PNX_EMAIL_1], [PNX_IBAN_1] → LLM Provider
Provider → [Privnexa] → mario.rossi@example.com, IT60X054… → Cliente
See what the provider receives
Type anything. Nothing leaves this page.
Detection runs entirely in your browser, including the neural name model.
Works with any provider
Keep your provider. Change what it sees.
OpenAIAnthropic Claude
Google GeminiMistral AI
Meta LlamaGroq
DeepSeekTogether AI
OpenRouterRequesty
Azure OpenAI
Your own endpoint
Provider names are trademarks of their respective owners, listed for compatibility only.
Why this matters under the GDPR
- Pseudonymisation —
- Data minimisation —
- Transfers outside the EU —
What Privnexa does
Pricing