Autonomy Levels

Workers create drafts but never execute any actions. All proposed actions remain as drafts until a human manually copies and executes them outside the system.

Best for: Initial testing, learning worker behavior, high-risk operations.

Workers generate proposals that appear in your Inbox. Each proposal requires explicit approval before execution. This is the recommended setting for most use cases.

Best for: Customer communication, business operations, routine tasks with oversight.

Workers automatically execute proposed actions within configured safety limits. Actions are still logged and can be reviewed after the fact.

Best for: Trusted, routine tasks; internal notifications; research and monitoring.

Note: Auto-execute is only available on the Business plan and may be limited by role autonomy ceilings.

• Documentation home
• Getting started guide