Skip to content
BrainRoad BrainRoad
Documentation Menu

OpenClaw and Hermes

BrainRoad hosts two sister AI helper runtimes. Here is what each one is, how they differ, and how to switch between them.

On this page

Two Runtimes, Your Choice

BrainRoad hosts two open-source AI agent runtimes for every account:

  • OpenClaw, an MIT-licensed agent runtime with its own chat interface, channel adapters, sessions, skills, and configuration screens.
  • Hermes, an MIT-licensed agent runtime from NousResearch, with its own chat, sessions, plugins, and configuration screens.

Both auto-start when your account is provisioned, both work from the same Business Brain, and both route outside-world actions through the same review Inbox. Run both, or stop the one you do not use. You switch between them with the runtime switcher in the dashboard sidebar.

BrainRoad does not rebuild the runtimes’ own screens. From the Helper page you open the runtime’s native interface in a new tab: “Open OpenClaw chat” or “Open Hermes chat” for conversation, and the shell buttons for the dashboard Console. Channels, sessions, skills, and runtime configuration stay inside the runtime’s own interface.

What Differs Between Them

A few dashboard features are OpenClaw-only today:

  • Company (/dashboard/company): the workspace where your OpenClaw helper can operate with hired engineer agents. Dimmed when Hermes is selected.
  • Browser (/dashboard/browser): the live view of the sandboxed browser running alongside your OpenClaw helper. Hermes has no browser page.
  • The managed work mailbox is provisioned through the OpenClaw path after the helper names itself; Hermes does not mint a separate mailbox. Both helpers share the account’s review Inbox either way.

Everything Brain-related and review-related works the same for both.

Messaging Channels

Each runtime ships its own channel adapters, configured inside the runtime’s own interface, not in the BrainRoad dashboard. Connecting a channel such as Telegram, Discord, Slack, WhatsApp, or Signal requires your own bot token or account credentials with that provider, and each channel keeps its provider’s own rules. BrainRoad does not connect any messaging channel automatically.

One honest caveat: OpenClaw lists an iMessage channel, but it requires a bridge running on a local Mac. BrainRoad’s hosted workspaces are Linux, so iMessage is effectively not available on BrainRoad.

A helper with a live channel connection is kept running so it never misses a message. See starting, stopping, and sleep.

AI Models and Keys

  • Trial key included. Every new account gets a BrainRoad-managed OpenRouter trial key, already configured, so your helper works before you add any key of your own.
  • Bring your own key later. The dashboard supports keys for Anthropic, OpenAI, Google, OpenRouter, xAI, and more (see Choosing a Provider), added on the Helper page’s API Keys tab. AI provider usage may be billed separately from your subscription.

Which Should You Use?

There is no wrong answer. Both are capable general-purpose helpers working from the same Brain. Try each one’s chat and keep the interface you like. If you want the Company workspace or the watchable browser, use OpenClaw. Many users simply leave both running.