What's the difference between a chatbot and an LLM?

A chatbot is a product built around a back-and-forth chat interface, while an LLM is the underlying technology that actually generates the words — and a chatbot may or may not be powered by one.

Was every chatbot always powered by an LLM?

No. Before LLMs existed, most chatbots were rule-based: a decision tree of pre-written responses triggered by specific keywords or menu clicks. The customer service bot that made you type "1 for billing, 2 for support" had no real understanding of language — it was matching patterns, not thinking. Modern chatbots like ChatGPT, Claude, and Gemini are different: they're LLMs wrapped in a conversational product.

The LLM is the "brain" doing the actual language generation, while the chatbot layer adds the chat history, turn-taking, safety filters, and memory features that make it feel like talking to something.

How do you tell a chatbot and an LLM apart?

The clean way to keep these straight: chatbot describes the form (a conversational, back-and-forth interface), and LLM describes the technology (the model predicting the next word). The two don't require each other.

Can you have one without the other?

Yes, in both directions. You can build a chatbot without an LLM — old-school rule-based bots did it for years. And you can use an LLM without any chatbot in sight, like an API call that summarizes a document in one shot, with no conversation happening at all.

That's why the distinction is worth holding onto: the interface and the engine are separate choices, and knowing which is which tells you what a given system can actually do — and whether there's any real language understanding behind the chat window.