Centeye Lumen · AI governance and registry
Lumen finds every AI tool in use and what it can reach, gives each one an owner, and acts on your rules.
67% of people using AI on work devices used accounts that weren’t the company’s. Verizon 2026 Data Breach Investigations Report
Not ready for a trial? Get the free AI Exposure Check: what AI your clients use and what it can reach.
Part of the Centeye platform · People security
Your people already use ChatGPT, Claude, Gemini, Copilot and a dozen AI note-takers, and AI agents now work inside their browsers. Lumen is the governance and registry platform for all of it. It finds which AI tools are in use and by whom, shows what data goes out and which access is riskiest, and acts on your rules: coach, block or revoke. Every tool then lands in one register with an owner and a review date. AI stays an advantage, not a leak.
Most data leaving a company today doesn’t go out by email. It goes into a chat window, a browser extension, or an AI app someone connected to their mailbox with two clicks. AI governance means knowing every AI tool and agent in use, giving each one an owner and a review date in a single registry, setting what is allowed, and acting when the rules are broken. It is how you bring shadow AI under control without banning it.
Lumen is governance and protection in one: a register of every AI tool, policies people actually follow, and the power to cut off what breaks them, whether a person or an AI agent is doing the work.
Capabilities
Each capability with a real situation: what happens, and what Lumen does about it.
Sees when company data is pasted or uploaded into an AI tool, and how much, including through personal accounts. It never reads the content.
A recruiter pastes a large block of text into a free AI chatbot from a personal account.
Lumen records the size, the destination and the personal sign-in as a possible disclosure, without reading a word of it.
AI apps people connected to their mail and files, and old consents and sessions still open long after anyone used them, quietly reaching your data.
An AI note-taker someone tried once in March still has full access to their mailbox and calendar in October.
Lumen flags the access as unused for months and revokes it at your setting, checked at Microsoft or Google.
AI now works inside the browser. Lumen applies data-loss rules to each action, a paste, an upload, a sign-in with a personal account, so nothing slips out unnoticed.
An AI browser agent filling in a form tries to upload the customer list to a site outside your approved tools.
Lumen applies your data rules to that one action, blocks the upload, and lets the rest of the task carry on.
Cut off an AI site, a browser extension or an app’s access, and revoke one specific grant, on the rules you set. Every revoke can be undone.
A browser extension that promises AI summaries asks to read every page your staff open.
Lumen blocks the extension across the company and tells the people who installed it why.
Every AI tool in use, who uses it, what it can reach, who owns it, and when it is next reviewed. Your AI policy, written down and kept current.
Finance starts using a new AI spreadsheet assistant nobody told IT about.
Lumen adds it to the register with who uses it, what it can reach and an owner to approve it, and sets its review date.
When someone is about to break policy, Lumen tells them why and points them to the approved tool, so people learn instead of working around it.
Someone is about to upload a contract to a personal AI account.
Lumen shows a short note on the rule and a link to the approved AI tool, and the contract goes there instead.
In the console
Real screens from the Centeye console. The clients, people and data in them are made up.
Under the hood
How it works
Lumen doesn’t stop at an alert. Praxis links what it finds to the rest of the attack, then fixes it at the setting you choose.
Spots what is wrong, as it happens.
Links it to everything related, in one case.
On your dial: tells you, asks first, or handles it.
In action
Pick a case to see what happened, what Centeye saw, and what it did.
Wednesday, 3:12 pm
Someone in finance drags a payroll spreadsheet into a personal ChatGPT account to tidy it up.
Lumen sees 4 MB of spreadsheet data going to an AI service under a personal sign-in, and marks it as a possible disclosure.
The person sees a short note pointing to the company’s approved AI tool. Their manager gets a summary, and personal AI accounts are blocked for company data from then on.
Monday, 11:05 am
A sales rep tried an AI meeting note-taker in the spring and forgot about it. It still holds full access to their mailbox and calendar.
Lumen finds the grant unused for five months, far broader than the job needs, from a vendor that isn’t in the register.
The access is revoked on the rep’s OK and confirmed at Microsoft. The note-taker goes into the register as “not approved”.
Tuesday, 2:30 pm
An AI browser agent building a market report starts uploading the customer list to a free file-conversion site.
Lumen checks that one action against your data rules: customer data, going to a site that isn’t approved.
That upload is blocked, the agent finishes the rest of its task, and the person gets a short note on why.
Friday, 9:48 am
An email asks a manager to “enable the new AI assistant”. Clicking it grants an unknown app access to all their mail.
Lumen sees a new AI app with full mailbox access, from a vendor it has never seen, approved minutes after a suspicious email.
The grant is revoked and the app blocked for the company. Lighthouse pulls the email from every other inbox.
Better together
Identity, endpoint, email, AI use and agents are one picture to an attacker. Each link below is made automatically, in the same case.
Getting started
Connect Microsoft 365 or Google Workspace in under 10 minutes to see every connected AI app. Add the Chrome and Edge extension and the Windows sensor for the browser and laptop view.
