AI Agent

An AI agent is a system that pursues a goal by deciding for itself which steps to take, calling tools, observing results, and continuing until the task is done or it gives up. A plain chatbot answers a prompt in one pass; an agent runs a loop of reason, act, and observe, with a language model as the controller. Tools are what give it reach: web search, code execution, file access, database queries, and calls to internal APIs. Memory carries state across steps, and a planner breaks a large request into ordered subtasks. Coding agents such as Claude Code and GitHub Copilot's agent mode edit files and run tests, computer-use agents drive a browser, and support agents look up an order and issue a refund. Frameworks like LangGraph and the Model Context Protocol standardise how tools are attached. Reliability is the limitation. Small per-step error rates compound across long chains, agents loop or take costly wrong paths, and giving a model real permissions turns prompt injection into a security problem. Production deployments narrow the scope, log every action, and require human approval for anything irreversible.