Agent

An AI system that can autonomously take a sequence of actions to complete a goal, rather than just producing a single response. It decides what to do next based on observations from previous steps.

Example: A coding agent that reads failing tests, writes a fix, runs the tests, and iterates until they pass — without you guiding each step.

In practice: The word is used loosely. Some people call any tool-calling LLM an agent; others reserve it for systems with genuine planning loops and long-horizon goals. Context matters when you see the term.