Applications

AI Agent

An AI system that can perceive its environment, make decisions, use tools, and take autonomous actions to achieve a goal.

An AI agent is a system built on top of a language model that can take actions in the world, not just generate text. Rather than answering a single question, an agent receives a goal, breaks it into steps, uses tools (web search, code execution, file access, APIs), observes the results, and iterates until the task is complete.

The key difference from a standard LLM: agents have a loop. They think → act → observe → think again. This enables them to handle multi-step tasks that require dynamic decision-making — researching a topic, writing and running code, booking a flight, or managing a workflow.

Core components: A language model (the "brain") + tools (search, code exec, APIs) + memory (context, vector store) + a loop (plan → act → observe → iterate). The model decides which tool to use and when.

Agent Types

ReAct agents — alternate between reasoning and acting
Plan-and-execute — make a full plan, then execute each step
Multi-agent systems — multiple specialized agents collaborate
Browser agents — control a web browser to complete web tasks
Coding agents — write, run, debug code autonomously

Popular agent frameworks include LangChain, LlamaIndex, AutoGen, and CrewAI. Function calling APIs from OpenAI and Anthropic make it easy to give models access to structured tools. The main challenges are reliability (agents can get stuck in loops), safety (preventing unintended actions), and cost (multi-step tasks consume many tokens).

← Back to Glossary