AI Agent

An autonomous system driven by AI that can perceive its environment, make decisions, and take actions to achieve specific goals.

How it works

An AI agent operates in a perceive-reason-act loop. It observes the current state of its environment (via text, tool outputs, sensor data), reasons about what to do (using an LLM or other AI system), and takes an action that changes the environment (calling a tool, writing a file, browsing the web, generating output). The result of each action is fed back as the next observation, continuing the loop until the goal is achieved or a stopping criterion is met.

Why it matters

AI agents represent the shift from AI as a question-answering system to AI as a worker that can complete multi-step tasks autonomously. Rather than answering 'how do I do X?', an agent can actually do X. This dramatically expands the economic value of AI: agentic systems can conduct research, write and run code, manage files, and interact with external services with minimal human oversight. Agents like AutoGPT, Devin, and those built on Claude/GPT-4 are the leading edge of this shift.