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

# Live demo

> Try hexgate in your browser — define an agent, chat with it, watch policy decisions stream live.

Launch a disposable hexgate sandbox in your browser: define an agent in a
notebook, then chat with it in the dashboard Playground and watch policy
decisions stream. It's throwaway — everything vanishes when you're done.

<Card title="▶ Launch the live demo" icon="play" href="https://quang-le--hexgate-spawner-web.modal.run/">
  Opens your own isolated sandbox — no install, no login. Bring your OpenAI key.
</Card>

<Note>
  **Bring your own key (BYOK).** You paste your own OpenAI key into the notebook —
  it stays in your throwaway sandbox and is never logged.
</Note>

## What you do

<Steps>
  <Step title="Define your tools + agent">
    In the notebook, write your tools (`@tool`) and the agent
    (`create_agent(...)`) as real Python.
  </Step>

  <Step title="Add your key & start">
    Paste your OpenAI key and click **Start** — the agent goes live.
  </Step>

  <Step title="Open the Playground & chat">
    Click **Open the playground** — send a message and watch reasoning, tool
    calls, and **policy decisions** (allow / deny / approval) stream live. Edit
    the policy in the **Policies** tab and the next message picks it up.
  </Step>
</Steps>

## Run it locally

```bash theme={null}
make demo-notebook-build   # one-time: deps + marimo + dashboard build
make demo-notebook         # launch the whole world (one process)
```

Open `http://localhost:2718`, define your agent, add your key, click **Start**,
then open the Playground.