In plain English
An agent does not just answer a single question, it pursues a goal. Give it an objective and it breaks the work into steps, decides what to do next, calls external tools such as a search engine, an API, or a code interpreter, checks its own progress, and adjusts. The language model is the reasoning engine; the agent is the loop around it that lets the model act with little human input.
Why it matters
Because an agent takes real actions, the stakes are higher than with a chatbot. A wrong step can send an email, change a record, or spend money, which is why clear limits and human oversight matter.
A worked example
You ask an agent to research three competitors and produce a comparison table. It searches the web, opens several pages, pulls out the details, and assembles the table without you guiding each step.
Common confusion
An agent is not the same as a chatbot. A chatbot replies one message at a time, while an agent plans across many steps and takes actions, which also makes its mistakes harder to spot.