Agent Autonomy Levels rank how much an AI agent can decide and do alone.
It is like sending a kid to buy groceries. Low level texts carrot photos. High level buys cookies too.
It says how much an Agent may choose on its own. Teams use it for permissions and risk checks.
Agent
Autonomy levels show how far an Agent can work alone.
Human-in-the-loop
Lower autonomy means a human must approve key steps.
Agent Security
Higher autonomy needs stricter permissions and audits.