Or what they are feeding into them. Not the assistant you rolled out — the personal accounts, the browser extensions, the coding tools, and the agents someone wired into your file storage because it was quicker than raising a ticket.
Most organisations discover the scale of it only after something goes wrong. An assistant connected to a mailbox and a document store can move tens of thousands of files in an afternoon, and it looks like ordinary authorised traffic the whole way.
We deliver AI usage discovery, governance and control as a month-to-month service, built on Check Point. We show you what is actually running, then help you decide what to do about it.
A lightweight script runs through your existing management tooling at login. No agent, no endpoint install, no change control battle. It returns an inventory of the AI applications in use across the devices it touches.
30 days with the agent deployed to a scoped group, often 25 to 30 people to start. Monitor only. Nothing is blocked and nothing changes for your users.
We walk you through what your people are actually using, which accounts are personal rather than corporate, and where data is going. If it is not compelling, we stop there.
Policy build, user scoping, exception tuning and handover. Then governance enforced — blocking, redaction and justification prompts where you want them.
The discovery alone usually changes the conversation. It is the cheapest security assessment you will run this year, because it costs nothing and commits you to nothing.
Every AI application in use, by person and by team. Corporate and personal accounts shown separately, so you can see who signed in with a work address and who did not. It classifies what people are using AI for, not just which tool they opened.
Allow, block, log, or require a justification before use. Policy layers like a firewall rulebase, so you can be permissive where it is harmless and strict where it is not.
Prompts are inspected for intent rather than pattern-matched, so something commercially sensitive is caught even without an obvious identifier. Redaction happenson the device, before the prompt leaves — sensitive content never reaches the model.
Where staff have connected AI agents to your systems, you can permit reading while blocking create and delete on a per-tool basis. This is the control that matters most, and the one almost nobody has.
Coverage is Windows, macOS and Chrome, deployed through the management tooling you already run. Other operating systems are not currently supported.
Want the background first? Read Shadow AI: your workforce is already using it. Can you see it?
| Essentials | Enterprise | |
|---|---|---|
| Shadow AI discovery — every application in use | Yes | Yes |
| Corporate versus personal account detection | Yes | Yes |
| Usage reporting by person, team and use case | Yes | Yes |
| Access policy — allow, block, log | Yes | Yes |
| Justification prompts before use | Yes | Yes |
| Prompt inspection and on-device redaction | — | Yes |
| Intent-based data loss prevention | — | Yes |
| AI agent discovery and posture | — | Yes |
| Per-tool and per-operation agent controls | — | Yes |
| Connected-platform visibility for enterprise AI accounts | — | Yes |
| Executive and compliance reporting | — | Yes |
Both packages are priced per user, per month. Larger deployments pay less per user.
The platform covers licensing, your tenancy and full console access. What changes between these is how much of the running you hand to us.
| Co-managed | Managed | Fully managed | |
|---|---|---|---|
| Full console access | Yes | Yes | Yes |
| Monthly usage and exposure report | Yes | Yes | Yes |
| Platform kept current, tenancy managed | Yes | Yes | Yes |
| Policy changes | You | Us | Us |
| Review of new AI applications appearing | You | Us | Us |
| Detection triage and investigation | You | Us | Us |
| Quarterly executive briefing | — | Yes | Yes |
| Agent and MCP posture review | — | — | Yes |
| User coaching — we write and send the comms | — | — | Yes |
| Incident handling when something is found | — | — | Yes |
| Board-ready reporting pack | — | — | Yes |
| Named engineer | — | — | Yes |
Co-managed suits organisations with their own IT or security team who want the visibility and will act on it themselves. Ad-hoc help is available hourly in 15-minute increments, with no retainer and no minimum.Managed suits organisations who want the platform run for them but will make their own decisions about people and policy. Fully managed suits organisations with no internal security function, or with a board asking questions they need answered properly.
You can move between them at any time, in either direction, from one billing period to the next. Support is business hours with after-hours escalation.
Per user, per month, and the rate steps down as the deployment grows. The lower rate applies to every user, not only those above a threshold, and your count is set at the start of each billing period, so growth is picked up automatically.
Up to 500 users the service is month-to-month with no minimum term; above that it is offered on a 12-month term.Ask for pricing and we will send a firm quote for your user count and the way you want it run.
Getting the policy right at the start is what separates a platform that protects you from one that annoys your staff. Onboarding covers user scoping, policy build, exception tuning through the first weeks, integration with your management tooling, and handover.
It is a one-off charge set by deployment size, andcharged only if you proceed after the trial. If you decide not to continue, there is nothing to pay. We confirm the figure for your deployment before you commit.
We use it ourselves. This is deployed across our own business, not demonstrated from a slide. You keep full console access from day one — you see everything we see, and you decide how much of the running you hand over.
Pricing is quoted in AUD, exclusive of GST. Usage is calculated daily and invoiced monthly in arrears. Platform capability and the division of features between packages is determined by the vendor and may change. A written quotation is valid for 30 days.
A short conversation is usually enough to work out whether we can help, and what it would cost.