AI-DLP endpoint layer
An optional SentriKat add-on that stops secrets and personal data from leaking into AI assistants — powered by Contextia, the open-source, on-device AI data-loss-prevention engine.
What AI-DLP catches
Catches secrets & PII before they leave
API keys, tokens, credentials, source code and personal data are detected in-line, on the endpoint, before a prompt ever reaches an external AI assistant.
Works with the assistants your teams already use
ChatGPT, Microsoft Copilot, Claude and other web/desktop assistants — the layer sits at the endpoint, not inside any one vendor.
On-device, zero-network by design
Detection runs locally. No prompt content is sent anywhere to be scanned. Telemetry is aggregated and secret-free by design — a redaction never contains the value it redacted.
Open-source engine you can inspect
The detection engine is Contextia, MIT-licensed and public. You audit exactly what runs on your endpoints — no black box.
How it fits into SentriKat
AI-DLP is an optional add-on, like the Compliance Pack. When it's on, SentriKat pushes Contextia's policy to your endpoints and surfaces aggregated, secret-free telemetry in the console. When it's off, it simply isn't there — no nagging, no locked features.
Activation happens in your customer portal. Want it on your plan? Talk to [email protected].
Contextia is a separate open-source project (MIT). SentriKat packages it as a managed endpoint layer for regulated teams. Learn more at contextia.dev.