Kayrix AI Privacy Gateway
Put privacy controls between your business and AI.
Kayrix is designed to connect privacy controls, provider governance and reviewable evidence in the path between your applications and local or cloud AI providers.
Private Early Access. Kayrix is in private alpha. Pilot scope and availability are agreed individually.
Three responsibilities. One controlled request path.
Apply privacy controls before forwarding.
Check supported inputs against configured privacy rules and transform detected identifiers before an allowed request continues to a provider. Detection coverage and policy settings matter; no detector should be treated as infallible.
Make provider choice explicit.
Keep the selection of configured providers and permitted models visible to the people responsible for the deployment. Local and cloud options should follow an agreed workflow, not an accidental application setting.
Keep evidence available for review.
Give IT and privacy professionals a way to examine the recorded policy context, processing metadata and integrity status for traffic handled by the gateway.
Where Kayrix fits
Your connected chat client or AI application sends a supported request to Kayrix. The gateway applies the configured privacy and provider controls before forwarding an allowed request. Transaction metadata and hashes support a separate evidence and review path.
Kayrix is the control layer in this architecture, not the language model itself. Its coverage applies to integrated traffic, not every AI application someone might use independently.
Local and cloud are deployment choices, not privacy shortcuts.
Evaluate an on-premises model, a configured cloud provider, or a mix of approved options. The right choice depends on your use case, infrastructure and data-handling requirements.
A privacy statement should lead to something you can inspect.
Explore how the evidence model connects recorded processing to policies and provider metadata—and where that evidence stops.
Built around shared responsibility
For business owners evaluating AI adoption, IT teams integrating it, and DPOs and security professionals reviewing the controls.
Start with a defined pilot.
Bring a use case and the questions you need answered. Begin with synthetic or otherwise explicitly approved test material, a defined provider configuration and agreed review criteria.