An AI agent is a model plus a harness: the model decides what to do next, and the harness gives it tools and a loop so it can actually go and do it. A chatbot answers and stops. An agent keeps going until the job is done or it gives up.
Most of the quality lives in the harness rather than the model, which is why every AI company now ships developer tooling. Worth knowing that "agent" tells you nothing about how much rope it has: some draft a pull request for you to review, some run unattended against production.
