AI, LLM, Agent — What's the Difference?

Here's something you've probably noticed. People talk about AI the way they talk about the weather — constantly, confidently, and with a lot of words that seem to mean the same thing. AI. LLM. Model. Agent. Generative AI. They get sprinkled into the same sentence as if they're interchangeable.

They're not. And once you can tell them apart, a surprising amount of the noise out there suddenly clicks into place.

Let's untangle it. No technical background needed. We're just going to sort out which word means what.

Start with the big umbrella: AI

Artificial Intelligence is the broadest term of the bunch. Think of it like the word "vehicle." A vehicle could be a bicycle, a car, a cargo ship, or a fighter jet — wildly different things that all share one idea: something that moves you around.

"AI" is that broad. It's the umbrella word for any computer system doing something that normally needs human smarts: recognizing a face, recommending a movie, flagging a suspicious bank transaction, beating you at chess. Some of that technology is decades old.

So when someone says "this app uses AI," they've told you almost nothing specific — about as much as saying "I arrived in a vehicle." Okay… but which kind?

Zoom in: the LLM

Most of the time these days, when people say "AI," the thing they're actually picturing is an LLM, a Large Language Model. This is the technology behind ChatGPT, Claude, Gemini, and their cousins.

Don't let the name scare you. Break it down:

• Language: it works with words. Reading them, writing them.
• Large: it learned from a staggering amount of text. Think a big slice of the internet, books, articles — more than any human could read in a thousand lifetimes.
• Model: this is just the industry's word for "the trained system." When you hear model, mentally swap in "the AI brain that does the work."

So what is an LLM, in one sentence? It's a system that can explain, summarize, draft, translate, and answer questions in plain conversation — like a remarkably well-read assistant who's brilliant with language.

Give it hands: the agent

Now for the word causing the most confusion lately: agent.

An LLM on its own is a talker. You ask, it answers. Brilliant. But it's stuck inside the chat box. It can tell you the ten steps to book a flight, but it can't actually book it.

An agent is an LLM that's been given hands. It's connected to tools: the ability to search the web, open apps, send an email, run a calculation, click through a website. And it's allowed to take those actions on its own to finish a goal.

The difference, in human terms:

• An LLM is the brilliant friend who explains exactly how to plan your trip.
• An agent is the assistant who goes off and actually books the flights, the hotel, and adds it all to your calendar.

Same language brain underneath. The difference is whether it can only talk, or whether it can act.

How it all fits together

The clearest way to hold all this in your head is as nested circles — each idea sitting inside the bigger one before it.

You'll notice one new term snuck in there: generative AI. It just means AI that creates new things, like text, images, music, or video, rather than only sorting or recognizing them. An LLM is the language-shaped slice of it.

And the word "chatbot"?

One more, because you'll hear it constantly. A chatbot is simply the chat window you talk through. The old "press 1 for billing" bots were chatbots too — clunky ones. Today's chatbots happen to have a powerful LLM behind them, which is why they suddenly got so good. The chatbot is the doorway; the LLM is who's actually home.

The takeaway

That's the vocabulary sorted. But knowing the words raises a better question — one almost nobody stops to ask before diving in: when you say you want to "learn AI," what do you actually mean by that?

That's exactly where we're headed next.