Agents
Compare the documented AI agent templates available on XO Workspaces.
Each agent runs in an isolated XO workspace, but the interface, setup requirements, and best-fit work differ. Use the comparison below before creating a project.
Compare Agent Options
| Agent | Best fit | Primary interface | Initial setup |
|---|---|---|---|
| OpenClaw | A long-running messaging agent with skills and external integrations | OpenClaw dashboard and connected messaging apps | At least one model provider and one channel |
| Claude Code | Software development in a remote coding environment | Browser IDE and the claude terminal interface | Anthropic API key or Claude OAuth |
| Hermes | A messaging-first assistant with a dedicated monitoring dashboard | Hermes dashboard and connected messaging apps | Provider and channel recommended; both can be added later |
| XO Cowork | Chat-first work organized around persistent project files and memory | XO Cowork project view and optional channels | Provider can be added during project creation or onboarding |
| Antigravity | Google's agent-first coding platform — multi-step reasoning and multi-file edits | Browser IDE and the antigravity terminal interface | Google account sign-in on first launch |
If you still are not sure, follow the short Choose an Agent decision guide.
Explore the Agents
OpenClaw
Deploy a messaging agent, then add channels, models, skills, and integrations.
Claude Code
Run Claude Code from a browser-based editor inside an XO workspace.
Hermes
Run a messaging-first agent and monitor it from the Hermes dashboard.
XO Cowork
Work from a built-in chat and file view with persistent per-project context.
Antigravity
Run Google's agent-first coding platform with multi-step reasoning in a sandbox.
Ready to begin? Continue to Launch Your First Agent.