An introduction
Know what AI is running across your organisation — and whether it matches policy.
AI tooling arrives through people, not procurement. endlayer measures what is actually installed and configured across your estate, grades it against the policy you set, and keeps the evidence — from software you can run entirely inside your own environment.
Eleven short pages. If you want the whole thing afterwards, the full documentation is forty-seven.
What it is
What endlayer measures across an estate, which layers report today, and what it does with the measurement.
OverviewSecurity review
Every promise that is structural rather than configurable, what is collected, and the longer list of what is deliberately never read.
Trust modelHow it is deployed
One machine in ten minutes, a fleet with no human at any of them, or one read-only connection to your Microsoft tenant.
DeploymentHow it is built, and why that matters to the people who have to approve it
Measure from the endpoint, without becoming a risk yourself
The agent runs unprivileged, as the person whose machine it is. It never elevates, never executes the tools it finds, and never opens a connection to anything it discovers. Where it cannot measure something it reports that, rather than guessing.
Trust modelSeparate the facts from the verdict
Devices report what is true. Your policy decides what that means, and grading happens where the policy lives. A control nobody has written a policy for is reported and left ungraded — never quietly passed.
How it worksCollect the least that answers the question
Metadata by default: what is installed, how it is configured, which models, how many tokens. Conversation content is off, and turning it on takes a stated lawful basis, an administrator capability and an audit trail on every read.
Privacy and retentionEverything on this site
What endlayer is
Security and privacy
Deploying it