# See it for yourself

> How to evaluate Holistic AI EndLayer: explore a populated demo console, start from your Microsoft tenant, and see how a typical trial runs.

The quickest way to judge Holistic AI EndLayer, the Guardian Agent Platform, is to look at it working. There are two good starting points, and neither needs a single device enrolled.

## Look at a populated console first

The demo console is the real EndLayer console running over an invented fleet, with a month of history, graded by the same rules a real fleet is. The fleet is fixed, so the device that is behind on patches today is still behind tomorrow, and nothing in it comes from any organisation's data.

In a demo you can see:

- **Shadow AI:** the AI tools, agents, MCP servers and AI websites nobody has ruled on yet.
- **Events:** prompts warned on, redacted or blocked, with masked evidence.
- **Governance:** the policy library, frameworks and kill switches.
- **FinOps and adoption:** token cost and usage by model, tool, project and person.
- **Reports:** the Executive, CISO, cost and framework reports.

[Book a demo](https://www.endlayer.ai/demo) and we will walk you through it, or give you access to explore on your own.

## Or start with your Microsoft tenant

If you manage devices with Intune, the [Microsoft integration](https://www.endlayer.ai/docs/connections) gives a real number about your real fleet before a single device is enrolled: how many managed devices EndLayer cannot yet see. It only reads, until you grant the separate permission to push the agent.

## How a trial usually runs

1. **Your organisation is set up:** Holistic AI creates your organisation and invites the administrator you name by email.
2. **Look around:** Use the demo console, your Intune inventory, or both, to agree what you want to learn.
3. **Enrol a handful of devices:** By hand in about ten minutes each with `endlayer enroll`, or with one enrolment secret through your device management tool. Start with devices whose owners know it is happening.
4. **Set the policy baselines:** This turns reporting into grading. Until it is done, most checks stay ungraded on purpose.
5. **Turn on prompt scanning at Monitor:** The events show what it would have warned on or blocked, before it interrupts anybody's work.
6. **Widen it:** Roll out to the fleet, then move the rules that have proved themselves to Warn, Redact or Block.

## What a good trial answers

By the end of a trial, you should be able to answer these questions about your own organisation:

- **Inventory:** which AI tools, agents and MCP servers are installed, and on which devices?
- **Access:** what have agents been given access to, and do any hold keys they should not?
- **Use:** who uses AI, on which projects, and what does it cost in tokens?
- **Policy:** which findings break your policy, and which need a decision?
- **Control:** what would the guardrails have stopped, and what did they stop once switched on?

## Before you enrol a device

- **Read the trust model:** see [trust model](https://www.endlayer.ai/docs/trust) and [what is collected](https://www.endlayer.ai/docs/collection).
- **Check privacy settings:** conversation content is off by default. See [privacy and retention](https://www.endlayer.ai/docs/privacy).
- **Tell people:** a short notice to the people whose devices take part makes a trial smoother.

## Getting in touch

[Contact us](https://www.endlayer.ai/contact) or speak to your account team at Holistic AI. If you already run EndLayer and something looks wrong on a device, `endlayer doctor` is the first thing to run.

  - [Book a demo](https://www.endlayer.ai/demo): See the console over a populated fleet.
  - [FAQ](https://www.endlayer.ai/docs/faq): Short answers to common evaluation questions.
  - [Roadmap](https://www.endlayer.ai/roadmap): What is available today, and what is planned.
  - [Trust](https://www.endlayer.ai/trust): Certifications, policies and subprocessors.

Source: https://www.endlayer.ai/docs/see-it · Updated 2026-11-20
