A chatbot answers a question and waits for the next one. An agent is given a goal, a set of tools, and the ability to take multiple steps toward completing it (checking a database, calling an API, drafting a document) before reporting back. The distinction sounds academic until you're deciding which one your business actually needs.
When a chatbot is the right tool
If the job is answering repeated questions (pricing, hours, how a feature works), a chatbot is the correct, cost-effective choice. It's a retrieval problem: match the question to the right answer. Building an agent for this is over-engineering; it adds complexity without adding capability.
When you actually need an agent
Agents earn their complexity when the task requires multiple steps and real system access: pulling a customer's order history, checking inventory, and drafting a personalized resolution, not just answering "where's my order" with a canned response. The moment a task requires looking something up, deciding what to do based on what was found, and taking an action, you've left chatbot territory.
The practical advice: start by mapping the actual task, not the technology. Most businesses that ask for "an AI chatbot" are describing a chatbot need. The ones that describe a workflow (several steps, some judgment, a system to check) are describing an agent, whether they use that word or not.