Quick answer
An AI agent uses an LLM to act. It calls tools, searches the web or writes code. It repeats until the job is done.
What it is
An AI agent uses an LLM to call tools. It runs each tool, reads the output, and goes again until the task is done.
Why it matters
Agents move AI from answering questions to doing work: research, writing, code changes, and any job that needs several steps.
How to use it
- Set a clear goal and a short list of tools.
- Keep a human in the loop for important tasks.
Examples
- A research agent reads the top web results and writes a brief.
- A support agent sorts tickets, drafts replies, and passes hard cases to humans.