Vocabulary

Why “Agentic” and “Autonomous” Are Not the Same Word

These two words get used as if they mean the same thing, and the confusion is not harmless. It changes what people expect a system to do, and what they think they need to govern.

“Agentic” describes a capability. A system is agentic if it can take multi-step actions toward a goal, calling tools, chaining decisions, adapting its plan based on what it observes along the way. Agentic is a description of what a system can do.

“Autonomous” describes a degree of oversight. A system is autonomous to the extent it operates without a human reviewing or approving each step. Autonomy is a description of how much a human is, or is not, in the loop.

A system can be highly agentic and still have very little autonomy, a coding agent that proposes a multi-step plan but requires approval before every file it modifies is agentic without being particularly autonomous. A system can also be minimally agentic and highly autonomous, a simple scheduled script that runs unattended every night is autonomous without being agentic at all.

The reason this matters for governance is that the two dimensions call for different controls. Agentic capability raises questions about what actions are possible. Autonomy raises questions about who is watching, and how often. Treating them as one axis instead of two is how gaps get missed, usually the gap between what a system can do and how closely anyone is actually checking it.