# Agent

> AI that doesn't just answer, but takes steps to finish a task: using tools and working through several stages on its own.

**Category:** Core concept

**An agent goes beyond a single reply.** You give it a goal, and it **plans and carries out the steps**, looking things up, using tools, checking its work, to get there. It is the difference between an assistant who **answers a question** and one who goes off and **completes the errand**.

Agents are where a lot of AI is heading. You will meet them properly later in the course; for now, just know that **"answering" and "acting" are two different modes**.

## At work

Instead of answering 'how do I chase these overdue invoices', an agent could read the list, draft a reminder for each one, and line them up for your okay. It does the legwork, you keep the final say.

## Often confused with

A plain chatbot or assistant. An assistant answers your question; an agent goes and does the task, step by step, and only comes back when it is stuck or done.

## Related terms

- Model

## See it in a lesson

[What is AI?](/en/learn/foundations/what-is-ai)

## Further reading

- [Building effective agents (Anthropic)](https://www.anthropic.com/engineering/building-effective-agents): Anthropic's own guide to what agents are and when to use them.
- [What are AI agents? (IBM)](https://www.ibm.com/think/topics/ai-agents): A clear explainer of how agents plan and use tools.
- [Intelligent agent (Wikipedia)](https://en.wikipedia.org/wiki/Intelligent_agent): The wider computer-science idea behind the term.

---

Last updated: 24/07/2026  
Source: Rollo Academy glossary