AI Systems Research guide

AI agent or chatbot: three tests before you mix the words

If it cannot refuse a tool call, it is not an agent you can run. It is a demo that can spend money.

Sealed cream envelope on the left of a butcher-block counter, unrolled leather tool roll with keys, flashlight, and stamp on the right
A message that cannot act. Tools that can.

People search “AI agent or chatbot” because a vendor used both words on the same slide. The products are not the same. Mixing them is how a chatbot inherits an API key.

AI agent or chatbot? An agent takes a goal, uses tools, and acts across more than one step without a human prompting each one. A chatbot answers a turn. You type. It replies. The context window is the memory. Close the tab and the work is gone unless you saved it. Useful for drafting. Dangerous the moment you paste its output into a tool it cannot see.

Three tests

Ask the system in the room these three questions.

  1. Does it hold a goal overnight, or only the current prompt?
  2. Does it use tools, or only emit text?
  3. Does it continue across steps without a human typing each one?

If the answer is no, no, no, you have a chatbot. That is a respectable product. It is not an agent. Stop using the word until the tests pass.

A third box exists and people skip it. Automation runs a fixed script: when a PDF lands in this folder, extract the total into the sheet. No goal. No tool choice. A path you named. An agentic workflow is not an automation is that comparison. This page is agent versus chatbot.

Walk one object through the boxes

“Summarise this PDF in the chat” is a chatbot.

“When a PDF lands in this folder, extract the invoice total into the sheet” is automation.

“Chase unpaid invoices this week, using the sheet and the mail tool, and stop if a customer disputes” is an agent, but only if someone wrote the stop.

The stop is the charter. Inputs, outputs, an authority ceiling, success criteria, review, escalation. Without those seven fields you have a demo that can spend money. What is an AI agent? owns the full definition. This page owns the mix-up.

The failure mode is vocabulary

Loose vocabulary is the control failure. A chatbot with a system prompt is still a chatbot. A wrapper around a chat model is still a wrapper. Calling it an agent does not add a ceiling. It adds permission to connect a tool.

The human in the loop is the named person who can fail the next call. If that person is “the team” and the fail line is “looks fine,” you already decided the chatbot may act.

Write the three tests on a card. Run them before the demo becomes a credential. AI Agents for Startup Strategy is live. You do not need the hardcover to refuse the word.

Terminology

Sources

  1. AI agent (glossary)
  2. What is an AI agent?
  3. An agentic workflow is not an automation
  4. Human in the loop meaning
  5. AI Agents for Startup Strategy

Related essays

Further reading

Markdown for LLMs