AI glossary
AI agent
A system in which a language model uses tools in a loop — deciding, acting, checking the result — to reach a goal with little step-by-step guidance.
A chatbot answers; an agent does. Given a goal, the model decides which tool to use (search the web, read a file, run code, call an API), looks at the result and decides the next step, repeating until the task is done or it needs your input.
Agents are powerful for open-ended work like coding or research, but they cost more and their mistakes can compound, so they need clear limits and checkpoints. Read what an AI agent is for the full picture.
Example: You ask a coding agent to fix a failing test. It reads the error, opens the files involved, changes the code, runs the tests again and repeats until they pass; then it shows you the change to review.
In practice
- Start with tasks whose result you can check easily.
- Give it only the permissions it needs and require approval for anything irreversible: sending, paying or deleting.
- Set limits on time, steps or spending.
To see how they are designed, continue with agent design patterns.
