> ## Documentation Index
> Fetch the complete documentation index at: https://docs.getbifrost.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# How Edge works

> The Bifrost Edge experience from a user's point of view: one browser sign-in, an always-on menu-bar agent, and every AI request routed through Bifrost automatically.

Bifrost Edge is designed to be invisible. After a one-time sign-in, users keep using the AI tools they already have - Claude Desktop, ChatGPT, Cursor, coding agents in the terminal - and Edge quietly routes that traffic through your Bifrost in the background. There is no proxy to configure, no base URL to change, and nothing to remember.

```mermaid theme={null}
sequenceDiagram
    participant U as User
    participant E as Edge agent
    participant B as Bifrost
    U->>E: Sign in once in the browser (SSO)
    B-->>E: Identity
    Note over E: Agent connects and turns on
    U->>E: Uses any AI app as normal
    E->>B: Routes the AI request
    B-->>E: Governed response
    E-->>U: Response in the app
```

## One sign-in

The first time Edge runs, the user signs in through their browser using your organization's existing single sign-on. That sign-in links the machine to the user and syncs all policies assigned to them. No API keys are copied or pasted, and nothing sensitive lives in the app itself.

## An always-on menu-bar agent

Once signed in, Edge lives in the menu bar (macOS) or system tray (Windows and Linux). From there a user can see whether they are connected, which key is active, and turn routing on or off. Most people set it once and never think about it again.

<Frame>
  <img src="https://mintcdn.com/bifrost/Rlyk_CaVXv2CLvLo/media/edge/edge-agent.png?fit=max&auto=format&n=Rlyk_CaVXv2CLvLo&q=85&s=5e12e7ae7ccfc8d97caec5e12927ec56" alt="Bifrost Edge menu-bar status and controls" width="720" height="515" data-path="media/edge/edge-agent.png" />
</Frame>

<CardGroup cols={2}>
  <Card title="Connection status" icon="signal">
    A clear indicator shows when AI traffic is being governed, and surfaces a warning if something needs attention.
  </Card>

  <Card title="Key selection" icon="key">
    Users with more than one virtual key can pick which one to use, with budget visible at a glance.
  </Card>
</CardGroup>

## Every app, automatically

Because Edge routes traffic at the machine level, it covers the AI surfaces people actually use without any per-app setup:

<CardGroup cols={3}>
  <Card title="Desktop apps" icon="display">
    Claude Desktop, the ChatGPT app, Cursor, and other desktop AI clients.
  </Card>

  <Card title="AI in the browser" icon="globe">
    ChatGPT on the web and other browser-based AI surfaces.
  </Card>

  <Card title="Coding agents" icon="terminal">
    Claude Code, Codex, and similar agents in the terminal and IDE.
  </Card>
</CardGroup>

The result is that governance follows the user instead of waiting for them to opt in. See the full list on the [Supported applications](/edge/supported-applications) page.

<Frame>
  <img src="https://mintcdn.com/bifrost/60LFLR_6YbjQbw-C/media/edge/edge-app-usage.png?fit=max&auto=format&n=60LFLR_6YbjQbw-C&q=85&s=4431b646f700f4d484066c5a20504ade" alt="Bifrost Edge showing AI traffic routed and governed" width="3732" height="2792" data-path="media/edge/edge-app-usage.png" />
</Frame>

***

## Next steps

* Control which apps are allowed in [Govern AI apps](/edge/app-governance).
* Control MCP servers in [Govern MCP servers](/edge/mcp-governance).
* Roll Edge out to your fleet in [Deploy with MDM](/edge/deployment-mdm).