Governance Hub
A live-style view of what Demarc AI records at the boundary between your people, your AI, and the outside world. Data below is a realistic sample — download it and explore.
What this means, plainly
In the last 30 days, your people used AI — times. In — of those, nothing sensitive was involved and work flowed normally. The gateway quietly caught — pieces of sensitive information — Social Security numbers, account numbers, patient identifiers — and masked them before they could travel anywhere they shouldn't. It also stopped — actions that policy doesn't permit, like an AI agent attempting to email a customer list externally. Every one of these events is recorded in an audit trail you can hand to a reviewer as-is.
Compliance posture
What each model backend is seeing (30 days)
Classification-aware routing sends each request to an approved model based on data sensitivity. Sensitive material never leaves the local models — that's the demarcation line at work.
| Model backend | Runs | Sensitivity tier | Requests | Redactions | Blocks | Avg response |
|---|
Recent request traces
| Time | User / Role | Channel | Policy triggered | Action |
|---|
Detections by category (30 days)
Watch it work — on your own document
Drop in a document and watch Demarc AI find and mask sensitive information in real time, then explain what it did and hand you the report. Your file is processed entirely in your browser and never uploaded anywhere.
Your document, protected
Tell us what you really think
You're one of a small group testing this early. Your critique — especially what didn't work — decides what we build next. Two minutes, no wrong answers.
Demand signals
Engagement measured on this device. Counters live in the visitor's browser; to aggregate across all visitors, set BEACON_URL at the top of the script (see the companion guide).