Shadow AI
Shadow AI gives you visibility into the AI tools your workforce actually uses — sanctioned or not — across browsers, desktops, and the terminal, so you can govern AI usage instead of guessing at it.
What it does
- Discover usage of AI tools (ChatGPT, Claude, Microsoft Copilot, Gemini, GitHub Copilot, Cursor, and many more) via the browser extension, desktop agent, and coding-agent telemetry.
- Attribute activity to people — see who is using which tools, how often, and through which surface (web, desktop app, CLI).
- Enrich identities from your directory (Okta / Microsoft Entra ID) so usage can be sliced by department, job title, and manager.
- Govern — sanction, tolerate, or forbid tools, classify them by risk, and review what your workforce is adopting.
Privacy by default. Shadow AI captures usage metadata only — apps, sessions, counts, models, file metadata, and device health. Prompt and response content is never read or stored. See Data & Privacy.
How it works
Shadow AI collects signals from lightweight clients you deploy to your fleet, plus optional telemetry from coding agents. Everything is attributed to a user and device, classified against an AI-tool registry, and surfaced in the dashboard.
Get started
Deploy clients
Roll out the browser extension and/or desktop agent to your fleet, and optionally wire up coding-agent telemetry. See Quickstart and Deploying Agents & Extensions.
Enrich identities (optional)
Connect Okta or Entra ID so usage can be sliced by department, title, and manager. See Directory Integration.
Review & govern
Read the dashboard, then sanction, tolerate, or forbid tools by policy. See Understanding the Dashboard and AI Tools, Registry & Policy.