# AI Agent

> A program that uses a model to work toward a goal. It picks each step.

**Field:** Artificial Intelligence  
**Category:** AI Agents  
**Also known as:** Agent, LLM Agent, Autonomous Agent

## Think of it like

Think of it like a helper you give a job to. It works out the steps and does them.

## Definition

An AI agent is a program that uses a language model to reach a goal by taking actions, not just by writing text. It decides what to do next, uses tools such as search, code or files, and looks at what happened. Then it decides again, step by step, until the goal is met. A chatbot answers one message at a time, while an agent can work for many steps on its own. How far it may act is set by the people who run it.

## Example

Jack asks an agent to find why a test fails. It reads the code, runs the test, edits a file and runs the test again until it passes.

## Why it matters

An agent does the work, not just the talk. A model alone can only answer.

## Common mistake

Giving an agent a vague goal and full access. Clear goals and tight limits make agents safer and better.

## Three principles

- **Goal, not a message.** It works toward an outcome.
- **Acts with tools.** Not just text.
- **Checks and repeats.** Step by step.

## Related terms

- [Agent Loop](https://learn.benibix.com/term/ai/agent-loop)
- [Tool Use (Function Calling)](https://learn.benibix.com/term/ai/tool-use)
- [AI Workflow](https://learn.benibix.com/term/ai/ai-workflow)
- [Agentic Coding](https://learn.benibix.com/term/ai/agentic-coding)
- [Large Language Model (LLM)](https://learn.benibix.com/term/ai/large-language-model)
- [Voice Agent](https://learn.benibix.com/term/ai/voice-agent)
- [Indirect Prompt Injection](https://learn.benibix.com/term/ai/indirect-prompt-injection)

## Told apart from

- [AI Agent vs AI Workflow](https://learn.benibix.com/compare/ai-agent-vs-ai-workflow): People often call any system that uses a language model an agent. In an AI workflow, a developer fixes the steps in code, and the model fills in each one. In an AI agent, the model picks its own next step, again and again, until the goal is met. Workflows are cheaper and easier to test. Agents handle tasks whose steps cannot be known in advance.
- [Chatbot vs AI Agent](https://learn.benibix.com/compare/chatbot-vs-ai-agent): People often think a chatbot and an AI agent are the same. A chatbot talks: you ask, it answers, and it waits for you. An agent works toward a goal on its own. It plans steps, uses tools, checks the results, and keeps going until the job is done. Many products now mix both, but the test is whether it acts or only replies.
- [AI Copilot vs AI Agent](https://learn.benibix.com/compare/ai-copilot-vs-ai-agent): People think an AI copilot and an AI agent are the same helper. They are not. A copilot sits in your app and suggests the next step, and you say yes or no to each one. An agent takes a goal and does the steps by itself, using tools, and comes back when it is done. With a copilot you drive. With an agent you check the result.

## Taught in

- [Agentic Coding and AGENTS.md](https://learn.benibix.com/learn/ai-working-with-ai/agentic-coding-and-agents-md): Prompt Engineering, 14/16
- [Deep Research and AI Tutors](https://learn.benibix.com/learn/ai-use-case/deep-research-and-ai-tutors): AI Assistants and Chatbots, 10/13

![A goal goes into an agent made of a language model and tools. The agent acts on files, the web and apps, and reports back when the goal is met.](https://learn.benibix.com/diagrams/ag-ai-agent.svg)

---

Source: https://learn.benibix.com/term/ai/ai-agent
Author: Behnia Zarei (a developer in Vienna, Austria, who builds and teaches data, AI and business intelligence at BENIBIX)
Publisher: BENIBIX Learn (https://learn.benibix.com/)
Last updated: 2026-11-19
Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Quote freely with attribution.
