Where a chatbot responds once, an agent loops: it plans, calls a tool (search, code, a database, a purchase), observes the result, and decides the next step until the goal is met. Tool access is what turns a model into an agent — commonly via MCP or function calling.
Agents now shop (agentic commerce), write code, and operate software. Their reliability depends on the quality of the tools and data they can reach — which is why tool descriptions, structured data and server health matter so much.
A chatbot answers a prompt. An agent pursues a goal across multiple steps, calling tools and acting on results autonomously until it's done.