What is an Autonomous Agent?

The defining trait of autonomy is closing the loop between perception, decision, and action internally. A traditional script executes a fixed sequence regardless of what it observes; an autonomous agent evaluates the current state — an inbox, a database, a task queue — and decides which action best advances its goal, then adapts if that action doesn't produce the expected result. Autonomy exists on a spectrum, not as a binary. An agent can be autonomous within bounds — free to choose how to accomplish a task, but required to seek approval before taking a specific class of high-consequence action. This bounded autonomy is what makes agents usable in real business settings, where full independence would be too risky but full manual control would defeat the point of automating at all.