Agent
A model given tools and a goal, left to loop — deciding what to do next and doing it — until the job is done or it gives up.
Also written: agentic.
The word is used loosely, which makes it worth pinning down. An agent is a model in a loop with the ability to act: it is given a goal and a set of tools (Letting a model call functions you define — search, read a file, send a request — and read back what they return.), it decides which tool to call, it sees the result, and it decides again.
What separates an agent from a chatbot is not intelligence but autonomy over the next step. Nobody approves each action. That is where the usefulness comes from, and all of the risk: an agent that misunderstands the goal will pursue the wrong thing energetically, and an agent with a long loop can spend a great deal of money before anyone notices.