Entry
A work email and a free tier are enough, so nothing reaches procurement or security review.
Builders
Low-code tools let any team publish an agent wired into company systems, with no ticket to IT.
Visibility
Network logs show domains, licenses show seats and surveys show what people admit. None shows it all.
Security
The work moves to tools the blocklist doesn’t know yet, where the chart can’t see it.
Privacy
The reply is calmer, and the customer’s details sit with a provider you have no contract with.
IT
It works, so colleagues share it, and no inventory shows that it reads HR files.
AI governance
The answer becomes an email to every department head, and a list that’s stale by Friday.
Why it keeps happening
In a global study by KPMG and the University of Melbourne, 70% of employees using AI at work use free public tools, and 42% use tools their employer provides.1 Uploading company information to public AI tools was most common where generative AI was banned: 67%, against 33% where there was no policy.1 A ban removes your view, not the risk.
See what’s in use, not who’s using it
AI discovery lists the AI tools in use across the company. By default it shows usage in aggregate, keeps no prompts and blocks nothing.
See AI discovery and inventoryTurn each finding into an owned record
Add a finding to your AI inventory and give it an owner, or raise a use case request from it so your reviewers decide.
See AI intake and reviewBlock what you must. Leave a way to ask
Controls apply to tools in your AI inventory. When one blocks a tool, people can request access with a reason, and your team approves or rejects it.
See AI agent governanceWhat is shadow AI?
Shadow AI is AI that people use for work without their organization knowing about it or approving it, such as a free chatbot or an agent someone built in a low-code tool. Most of it is people reaching for what helps, especially when there’s no approved option. The risk is the data that goes with it, so governing it starts with seeing what’s in use.
Does Enzai monitor individual employees?
Not by default. AI discovery reports the AI services it recognizes in aggregate, without naming anyone. Admins can switch attribution on. If you plan to, agree it with employee representatives first.
Can it find agents people build themselves?
Yes, for Microsoft Copilot Studio. Enzai syncs the agents in your Power Platform environments every day and lists the ones nobody has registered. Registering an agent adds it to your AI inventory and sends the connectors it uses for review. Teams can also register agents through a use case request or the Enzai MCP server.
Should we just block AI tools we haven’t approved?
A block with no way back moves use somewhere you can’t see. In Enzai, you can publish a control in Observe mode first: nothing is blocked, and you see what it would have done. Switch to Enforce when you’re ready. Start with the tools that carry the most data, not the most users.
