I want to make a small, cranky request on behalf of clear thinking everywhere: stop calling everything an agent. The word has been stretched so far that it now means anything from a genuinely autonomous system to a chatbot that calls one API. When a word means everything, it means nothing, and the fuzziness is not an accident. It is marketing.
What an agent actually is
An agent is a system that takes actions in a loop, decides what to do next based on the results, and keeps going toward a goal without a human driving every step. The defining feature is the loop plus the autonomy. It observes, acts, observes the outcome, and adjusts, on its own, until it is done or it gives up. That is the bar.
What keeps getting called an agent and is not
- A chatbot with a nice prompt. That is a chatbot.
- A model that calls a single tool once and returns the result. That is a function call.
- A workflow where a human approves every step. That is automation with a human in charge, which is often exactly what you want, but it is not an agent.
If a human has to press the button at every step, you do not have an agent. You have a very talkative macro.
Why the sloppiness matters
This is not pedantry for its own sake. The vague use of "agent" lets thin products borrow the excitement of genuinely autonomous ones. A company ships a glorified chatbot, calls it an agent, and rides a wave of hype it did not earn. Buyers, told everything is an agent, cannot tell the ambitious systems from the rebranded ones, and they overpay for the difference or get burned by autonomy they did not know they signed up for.
The ask
Use the word precisely. When someone sells you an agent, ask the two questions that cut through it: what does it actually do in a loop without a human, and what happens when it is wrong. If the answers are "not much" and "we are not sure," you are looking at a chatbot in a costume, and you should price it like one. I love real agents. That is exactly why I want to stop pretending everything is one.