FollowAgents AI AGENTS DIRECTORY
Home / AI Agent Glossary / AI Agent
Core Concepts

AI Agent

An AI agent is a software system that uses a language model to decide what actions to take, then calls tools to carry them out, in pursuit of a goal, with limited direct human instruction at each step.

User goalToolsEnvironmentMemoryAI Agent
An agent sits between a goal, its tools, the environment, and memory.

An AI agent pairs a language model with the ability to act. Instead of just answering a question in a single turn, it holds a goal, decides what to do next, calls tools or APIs to gather information or make changes, and reacts to what those tools return — repeating that cycle until the goal is met or it needs to check in with a person.

The term covers a wide range of systems, from a coding assistant that reads a repository, runs tests, and edits files on its own, to a customer-support bot that looks up an order and issues a refund without a human clicking 'approve' each time. What makes something an agent, rather than a plain chatbot, is that it takes multiple independent actions toward a goal rather than producing one response per user turn.

Agents differ enormously in how much they're allowed to do unsupervised — some pause for approval before every consequential action, others run for long stretches on their own. FollowAgents' catalog covers many of these systems, mostly open-source, GitHub-hosted coding and automation agents reviewed under the FARS methodology.

How it works

Most agents share four ingredients, shown in the diagram: a goal (the task the user or a trigger gave it), tools it can call (shell commands, APIs, file edits, web search), the environment it observes the results in, and memory to carry context across steps. The agent repeatedly decides on an action, executes it through a tool, and reads the result back — the mechanism behind that cycle is covered in detail under agent-loop.

Example

A repository-maintenance agent given the goal 'fix the failing CI build' might read the error log (environment), search the codebase for the failing test (tool call), edit a file (tool call), rerun the test suite (tool call), and only stop once the suite passes — without a human specifying each individual step.

How it differs

An AI agent is often confused with a chatbot: a chatbot produces one reply per message and stops, while an agent takes a goal and keeps acting — calling tools, checking results, and deciding what to do next — until the goal is met or it hits a checkpoint. See agentic-ai for how that distinction plays out along a spectrum rather than a hard line.

Common misconceptions

Often assumed: An AI agent is just a chatbot with a friendlier name.
Actually: A chatbot answers one turn at a time; an agent independently chains multiple actions and tool calls toward a goal without a new prompt for each step.
Often assumed: If something is called an 'agent,' it must run fully unsupervised.
Actually: Most production agents include guardrails or human checkpoints somewhere in the loop — being an agent is about taking multi-step action, not about the absence of oversight.

FAQ

What is an AI agent in simple terms?
It's a system that uses a language model to decide what to do, then uses tools to actually do it, repeating that until a goal is reached.
What's the difference between an AI agent and a chatbot?
A chatbot replies once per message; an agent takes a goal and keeps acting — calling tools and reacting to results — across multiple steps on its own.
Do AI agents need human approval to act?
It depends on the system — many include human-in-the-loop checkpoints for risky actions, while others run more autonomously; see autonomy-level for the range.

Last checked: 2026-08-28

Related terms

Agentic AI

Agentic AI describes AI systems built to plan and take multi-step action toward a goal on their own, rather than just generating a single response to a prompt.

Agent Loop

The agent loop is the repeating perceive-plan-act-observe cycle by which an AI agent turns a goal into a sequence of actions, checking each result before deciding what to do next.

Multi-Agent System

A multi-agent system is a setup where multiple AI agents, each with a narrower role, work on parts of a larger task and coordinate through some form of orchestration.

Subagent

A subagent is a secondary AI agent that a main agent spawns to handle a specific sub-task, usually with its own focused instructions and tools, and reports back to the agent that spawned it.