AI governance is a new recurring line: billed per device like the rest of your stack, deployed from your RMM, ticketed in your PSA, and backed by evidence your clients' auditors accept. One console runs every client.
Customer lists, contracts and patient records go into chatbots with no data agreement, no audit trail and no way to get the data back. Proctored sees it and can require work accounts per app.
You own the endpoint, so you own the question "how did that leave the building?" Proctored answers it before it's asked: the device, the user, the data type, the rule, the decision.
ISO/IEC 42001 and NIST AI RMF sit next to HIPAA, PCI DSS and CMMC in what your clients get sent. Proctored maps its controls to all eight and produces the evidence monthly.
Proctored is sold to service providers per device. There is no hardware, no per-client setup fee and nothing to install on the client's network, so the margin is yours to set and the cost to operate stays flat as you add clients.
One installer per platform for all your clients, signed and notarized on macOS. Push it with the client ID; devices enrol and report within a minute. Re-imaged machines resume their own record, so there are never duplicates.
Your RMM script monitor runs governit-agent check and reads the exit code: protected, degraded with the failing checks named, or not running. Ready-made monitors for NinjaOne, Datto RMM, N-able and ConnectWise RMM.
Alerts open tickets in ConnectWise PSA, Autotask or HaloPSA with per-client company mapping, at the severity floor you choose. When the alert resolves, Proctored posts the note back and the ticket closes. Slack, Teams, email and webhooks too, scoped platform-wide, per provider or per client.
A printable AI Governance Report per client with usage, protection, shadow AI, coverage, policy changes and control evidence, frozen with a digest. Plus a sealed evidence bundle and CSV for the client's own auditor.
The template applies the right rules: HIPAA and PHI for a clinic, PCI for a credit union, CUI for a defence supplier. Approve the AI tools the client uses; block the rest.
Your time: about an hourDevices enrol from the RMM. The Applications page shows which apps are inspected from real traffic. Staff see a panel with your name on it and a "check before you send" box.
Your time: the RMM job, then minutesUsage, blocks, shadow AI, personal accounts, coverage and control evidence, frozen with a digest. Walk the client through it; adjust two rules; approve an exception or two.
Your time: the quarterly-review slot you already haveBrowser plug-ins see one window. Enterprise security platforms were not built to be sold per client by a service provider. Proctored is an agent built for the channel.
| Capability | Proctored | Browser-extension tools | Enterprise security platforms |
|---|---|---|---|
| Desktop AI apps and CLI tools inspected for contentClaude Desktop, ChatGPT app, Claude Code, Codex | ✓ | ✗ block or allow only | ◐ if proxied |
| MCP servers inventoried, blocked, and tool calls inspected | ✓ | ✗ | ◐ discovery |
| Employee exception requests with time-boxed, device-bound approvals | ✓ | ✗ | ◐ |
| Coverage evidence per app and regression alerts | ✓ | ✗ | ✗ |
| USB, clipboard, data at rest and private-app access in the same agent | ✓ | ✗ | ◐ separate products |
| Three-tier tenancy, white label, PSA ticketing, RMM health probe | ✓ | ◐ | ✗ enterprise licensing |
| Priced per device per month for service providers | ✓ | ✓ | ✗ per user, enterprise |
“We can see which AI tools your staff use, and whether they're on a work account or a personal one.”
“Sensitive data is stopped on the laptop before it reaches the AI, not discovered afterwards.”
“You get a report your auditor can use, with the evidence sealed so no one can quietly change it.”
Proctored doesn't hook the kernel or inject into browsers. It governs AI services only and runs as a protected system service with a small user-session component. It coexists with EDR, RMM and backup agents; allow-list the Proctored binary in your EDR as you would any new tool.
Yes, with client admin or client viewer roles that see only their own client. You keep the partner view across all of them. Single sign-on with Entra ID or Google Workspace works at your level and at the client's.
Your name and help contact appear on the device panel, the block page and every report. Each service provider has its own organization key, branding and webhooks.
Per client you choose fail open (AI traffic continues, the failure is logged and alerted) or fail closed (blocked until the agent is healthy). Your RMM probe and the console both report degraded enforcement.
Events belong to the client's tenant inside your service-provider tenant; each tier sees only its own. Proctored hosts the console; it can also run on a server you control. Retention is set per client with automatic purge.
Creating the client and applying a template takes an hour. The RMM push does the rest; devices report within a minute of installing. The first report arrives on the first of the next month.
You'll get a service-provider tenant with your branding, installers for your RMM, and a thirty-minute walkthrough on a real device.