What an AI agent actually is (not the marketing definition)

The exact clicks

1. Take whatever you are calling an agent — a GPT, a chatbot, an automation.
2. Give it a job with more than one step, and do not tell it the steps.
3. Watch whether it plans the steps itself, or waits for you to name each one.
4. Check whether it used a real tool — sent, wrote, fetched, booked — or only described one.
5. Check whether it looked at what came back before continuing.

What changes

• Plan, act, check, on a loop. That loop is the entire difference.
• A chatbot answers. An agent acts — it decides the steps itself.
• Acting means a tool that touches something real, not a description of the tool.
• If it cannot act without you, it is a chatbot with a better name.

What it will not do

• Agents do not fail like chatbots. A chatbot gives you a bad answer; an agent fails confidently, halfway through, having already done part of the work.
• That is why the first one you build should touch something reversible — a draft, not a send.
• Nothing here makes an agent reliable. It makes the word mean something, which is what the rest depends on.
• The loop is the definition, not a product. Two tools can both have it and one can still be useless for your job.