Skip to content
BrainRoad BrainRoad
Documentation Menu

Starting & Stopping Your Agent

How to start, stop, and restart your agent from the dashboard.

On this page

Start and Stop from the Dashboard

Every agent has Start and Stop buttons on its detail page in the dashboard. These controls are straightforward — click Stop to shut your agent down, click Start to bring it back up.

Starting Your Agent

Click Start on a stopped agent. You will see a boot sequence animation with progress messages as the container comes online. This typically takes 10 to 20 seconds. Once the status badge turns green and shows Running, your agent is live and processing messages.

Stopping Your Agent

Click Stop on a running agent. The shutdown sequence scales your agent’s container down to zero. A brief shutdown animation plays while the process completes.

Your data is safe. Stopping an agent does not delete anything. Your files, conversation history, configuration, messaging connections, and API keys all persist on dedicated storage. When you start the agent again, everything is exactly where you left it.

A stopped agent does not consume compute resources and does not process incoming messages. Any messages sent to your agent’s connected channels while it is stopped will be waiting when it comes back online.

When to Restart

Restarting your agent (stop then start) is useful in a few situations:

  • Your agent seems stuck — if the agent is not responding to messages or the Activity tab shows no output, a restart clears the runtime state. See Agent Not Responding for more troubleshooting steps.
  • After manual config changes — if you edited configuration directly through the Console, a restart ensures all changes are loaded cleanly.
  • Wizard session conflicts — if the setup wizard reports a session is already running, restarting the agent clears the stuck session.

Most configuration changes made through the dashboard or wizard take effect immediately without a restart. You only need to restart for edge cases.

Billing While Stopped

Stopped agents do not count toward active compute on your plan. Storage for your agent’s persistent data is always maintained regardless of whether the agent is running.