Compare a chatbot reply, a fixed script, a rule-based automation, an agent, and a human worker on the same everyday task. You’ll learn to recognize agency by looking for a goal, context, available actions, feedback, and some freedom to choose what happens next.

Turn a vague request like “handle my inbox” or “book a trip” into a goal an agent could actually pursue. You’ll separate the desired outcome, the context it needs, and the success signal that tells it whether it is done.

List the actions an agent can and cannot take in a situation, from sending a message to calling a tool or asking a human. You’ll reason about why the action space matters more than the agent’s words: it defines what the agent can change in the world.

Apply the previous explanations in a guided problem.

Trace an agent through the cycle of observing context, choosing an action, seeing the result, and deciding what to do next. You’ll use simple tasks to see how feedback creates retries, corrections, stopping, or escalation.

Decide where an agent needs boundaries: time, cost, data access, confidence, tool permissions, and approval for risky actions. You’ll distinguish useful autonomy from full human judgment by choosing when the agent should continue, stop, or hand off.

Apply the previous explanations in a guided problem.

Check your understanding with a short quiz.

Review this chapter with practice based on your mistakes.