---
title: "AI Agent"
description: "An AI agent is a software system that uses a Large Language Model (LLM) to pursue a goal through multiple decisions and actions. The model can interpret the current state, plan or select a next step, use external capabilities, evaluate the result, and continue until it reaches a completion condition."
canonical_url: "https://promptmetheus.com/resources/llm-knowledge-base/ai-agent"
---

> Index of every page on this site: [llms.txt](https://promptmetheus.com/llms.txt)

An AI agent is a software system that uses a [Large Language Model (LLM)](https://promptmetheus.com/resources/llm-knowledge-base/large-language-model-llm) to pursue a goal through multiple decisions and actions. The model can interpret the current state, plan or select a next step, use external capabilities, evaluate the result, and continue until it reaches a completion condition.

Unlike a [chatbot](https://promptmetheus.com/resources/llm-knowledge-base/chatbot) that primarily generates conversational responses, an agent can change external state. It may search a knowledge base, call an API, edit a file, operate a user interface, or delegate work to another agent.

Core components commonly include:

- a goal and instructions supplied through the model's [context](https://promptmetheus.com/resources/llm-knowledge-base/context);
- [tool calling](https://promptmetheus.com/resources/llm-knowledge-base/tool-calling-function-calling) for accessing data and taking actions;
- an [agent harness](https://promptmetheus.com/resources/llm-knowledge-base/agent-harness) that runs the observe-decide-act loop and integrates tools;
- [agent memory](https://promptmetheus.com/resources/llm-knowledge-base/agent-memory) or external state for continuity;
- [agent orchestration](https://promptmetheus.com/resources/llm-knowledge-base/agent-orchestration) for routing, delegation, and control flow; and
- [agent evaluation](https://promptmetheus.com/resources/llm-knowledge-base/agent-evaluation) for measuring task success, cost, robustness, and safety.

Not every multi-step LLM application is fully autonomous. In an [agentic workflow](https://promptmetheus.com/resources/llm-knowledge-base/agentic-workflow), application code may define a fixed sequence of steps. In a more autonomous agent, the model dynamically decides which steps and tools to use.

Reusable [agent skills](https://promptmetheus.com/resources/llm-knowledge-base/agent-skill) and [agent scaffolding](https://promptmetheus.com/resources/llm-knowledge-base/agent-scaffolding) can provide specialized procedures and resources, while [dynamic tool discovery](https://promptmetheus.com/resources/llm-knowledge-base/dynamic-tool-discovery-tool-search) keeps large tool catalogs out of the model's context until needed.

More complex systems may split work across a primary agent and a [sub-agent](https://promptmetheus.com/resources/llm-knowledge-base/sub-agent), or transfer control through an [agent handoff](https://promptmetheus.com/resources/llm-knowledge-base/agent-handoff) when another specialist is better suited to continue the task.

Some agents are reactive, responding only after a [user message](https://promptmetheus.com/resources/llm-knowledge-base/user-message), while [proactive AI](https://promptmetheus.com/resources/llm-knowledge-base/proactive-ai) initiates work from schedules, signals, or events. [Event-driven AI](https://promptmetheus.com/resources/llm-knowledge-base/event-driven-ai) is one architecture for triggering those workflows.

Greater autonomy increases flexibility but also expands the failure surface. Production agents require authorization boundaries, an explicit [agent policy](https://promptmetheus.com/resources/llm-knowledge-base/agent-policy), step and cost limits, [agent guardrails](https://promptmetheus.com/resources/llm-knowledge-base/agent-guardrail), human approval for consequential actions, protection against [prompt injection attacks](https://promptmetheus.com/resources/llm-knowledge-base/prompt-injection-attack), and [traces](https://promptmetheus.com/resources/llm-knowledge-base/agent-trace-observability) that record actions taken.

[Long-horizon tasks](https://promptmetheus.com/resources/llm-knowledge-base/long-horizon-task) usually need additional execution controls, such as an [agent sandbox](https://promptmetheus.com/resources/llm-knowledge-base/agent-sandbox) for isolation, an [agent session](https://promptmetheus.com/resources/llm-knowledge-base/agent-session) for continuity across turns, and a [resumable agent run](https://promptmetheus.com/resources/llm-knowledge-base/resumable-agent-run) for recovering from interruptions without losing state.

## Sitemap

See the full [sitemap](https://promptmetheus.com/sitemap.md) for all pages.