Noisegate: a differential privacy gateway that lets an untrusted LLM agent query sensitive data over MCP (Model Context Protocol), with a formal guarantee no individual's record can leak even if the agent is adversarial - enforcement lives in trusted code below the model, validated by a runnable attack gallery.
A local-first privacy engine for agentic AI. It finds secrets and personal data in your files, folders and databases — and it can decide, on its own, how much of that data an agent is allowed to have for the job it says it's doing. Nothing is uploaded. No daemon. No cloud. Zero network calls.
Governed AI agents for sensitive data and business-critical apps. Agents never hold credentials: a gateway checks every call against per-resource permissions, and each answer runs in a sealed container.
