# The console

> A tour of the Holistic AI EndLayer console: the Inbox, Events, Shadow AI, Governance, cost and adoption views, and the reports a review needs.

Holistic AI EndLayer, the Guardian Agent Platform, has one console for security, IT and finance teams. It updates live as devices report, so nobody needs to reload a page to see the latest state of the fleet.

## Where the day starts

| Area | What it shows |
|---|---|
| **Dashboard** | What the fleet is doing, now and over the period you choose |
| **Inbox** | What is waiting on a decision: software to remove, devices to admit, issues to fix, and anomalies |
| **Events** | What happened, newest first: prompts blocked, controls changed, devices gone quiet |
| **Shadow AI** | The AI nobody has ruled on yet: apps, agents, MCP servers and AI websites, each with a decision to make |
| **Attack Surface** | The fleet read the way an attacker would read it |

**Anomaly detection** sits in the Inbox. It flags the same person, agent or AI app warned or blocked many times in a short period, an agent with many refused tool calls, or a surge far above the usual rate.

## Control

| Area | What it shows |
|---|---|
| **Surfaces** | Every computer, server, phone and cloud account, its posture and when it last reported |
| **Monitoring** | Who uses AI, on which projects, through which apps, agents and MCP servers |
| **Governance** | The policy library, frameworks, device groups, exceptions, kill switches and version history |

### Frameworks

Governance includes **nine built-in frameworks**: SOC 2, EU AI Act, ISO/IEC 42001, NIST AI RMF, GDPR, PII and DLP, Guardian, Cloud, and Computer use and impersonation. Each has its own page showing coverage by control point, with results that have not been evaluated kept in their own column, and an export for auditors.

Framework mappings support a review. They show which signals give evidence for each control; they do not certify anything.

## Insights

| Area | What it shows |
|---|---|
| **FinOps** | Token cost by model, tool, project and person |
| **AI Adoption** | Who uses AI, how much, and how that is changing |
| **Value Quadrant** | Usage set against the issues each team or person carries |
| **People** | Each person's devices, tools, usage and findings |
| **Reports** | Ready-made reports for people without a login |

Three choices make the cost figure one you can quote:

- **Devices report tokens, never money:** EndLayer prices tokens against a model price table refreshed daily, so history can be priced again.
- **It is called token cost:** where a vendor's own bill is connected, it is shown beside the derived figure rather than in its place.
- **Rankings are by tokens:** a model with no known price would otherwise sink to the bottom, which is where unaccounted usage should never go.

## Reports

Reports cover 7, 30 or 90 days, as a PDF or as slides, and can be scheduled.

- **Executive:** the fleet and its risk, for leadership.
- **CISO:** findings, controls and evidence for the security team.
- **Cost and ROI:** token cost, seats and value.
- **AI adoption:** usage across the organisation.
- **Framework:** coverage for one framework, for an auditor or a review.

Every Monday, owners and administrators get the **Briefing** email: the past week at a glance, with what needs attention and a link to review each item. It can also be posted to a Slack channel.

## Connections and settings

**Connections** is where devices and cloud accounts are brought in: fleet install, Microsoft, Jamf, cloud providers and alert destinations. See [integrations](https://www.endlayer.ai/docs/connections).

**Settings** holds people and roles, single sign-on, notifications, retention and enrolment.

## What a review gets

| Item | What it gives a reviewer |
|---|---|
| **Frameworks** | Controls evidenced by real signals, with unevaluated results shown separately |
| **History** | What the policy said and when it changed, so a change in evidence can be told from a change in policy |
| **Reports** | A period of all of it, for people without a login |
| **Audit trail** | Every publish, removal, exception and kill switch, with who, when and the reason given |

Reading raw evidence is itself audited, and the record is written before anything is shown.

  - [See it for yourself](https://www.endlayer.ai/docs/see-it): Walk through a populated console.
  - [Govern](https://www.endlayer.ai/solutions/govern): Policy, frameworks and evidence in one place.

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