loading
loading
An AI that takes actions autonomously to complete a goal over multiple steps.
Unlike a chatbot that replies once and waits, an agent observes its environment, decides which tools to call, and keeps acting until the task is done or it gets stuck. Agents can write files, run code, browse the web, and call APIs. The loop of observe-think-act is what separates an agent from a simple language model call.
Plainly
Think of Agent as a named building block in a big LEGO app. An AI that takes actions autonomously to complete a goal over multiple steps.
In practice
Use it when you are mapping how the app is structured or explaining a feature to a teammate or agent. In practice, define the owner, input, output, and failure mode before you rely on it.