← Knowledge Base

AI Agent

An autonomous software system that uses large language models to reason, plan, and execute multi-step tasks by interacting with tools and data sources.

Definition

What Is AI Agent?

An AI agent is a software system that uses a large language model (LLM) as its reasoning engine to autonomously plan and execute multi-step tasks. Unlike a simple chatbot that responds to individual prompts, an agent can decompose a complex goal into subtasks, select and invoke tools (APIs, databases, code execution), evaluate intermediate results, and iterate until the goal is achieved. Agents operate with varying degrees of autonomy — from fully supervised (human approval at each step) to fully autonomous (executing complete workflows without intervention).

Origin

Where It Came From

The concept of software agents predates LLMs by decades, originating in AI research from the 1980s and 1990s with systems like the Belief-Desire-Intention (BDI) model. The modern LLM-powered agent emerged in 2023 with systems like AutoGPT, BabyAGI, and LangChain's agent framework, which demonstrated that LLMs could serve as general-purpose reasoning engines for tool use. The agent paradigm accelerated with the introduction of function calling in GPT-4 and tool use in Claude, which provided structured interfaces for LLMs to invoke external tools. Current agent architectures range from simple ReAct-style loops to complex multi-agent hierarchies with orchestrators, subagents, and specialized skills.

Applications

Use Cases

  • AI coding assistants that can read codebases, write code, run tests, and iterate on failures autonomously
  • Research agents that search multiple data sources, synthesize findings, and produce structured reports
  • DevOps agents that monitor infrastructure, diagnose issues, and execute remediation steps
  • Customer support agents that troubleshoot issues, access knowledge bases, and escalate to humans when needed

Ecosystem

Related Technologies

  • Claude Code, Cursor, GitHub Copilot (coding agents)
  • LangChain, LangGraph, CrewAI (agent frameworks)
  • Model Context Protocol (MCP) for tool integration
  • OpenAI Function Calling / Claude Tool Use
  • SQS, Kafka, Redis (agent coordination)