What is an LLM (Large Language Model)? A guide for businesses

An LLM (Large Language Model) is the AI model behind assistants such as ChatGPT, Gemini, Perplexity and Claude. It is trained on huge amounts of text, which lets it understand questions and write answers in plain language. That matters for businesses, because more and more customers ask an LLM for advice before they choose.

How does an LLM work, in plain English?

During training, an LLM has read billions of sentences: websites, books, articles and forums. From that it has learned how language works and which words and concepts belong together. When you ask a question, the model writes the most likely answer, word by word.

So an LLM does not "know" things the way a person does. It gives the answer that best fits everything it has read. That is also why it sometimes makes mistakes or gives outdated information.

Training and live search: two sources of knowledge

For your business, this is the most important distinction:

• Knowledge from training. What the model read during training. That knowledge stops at a certain date and only changes with a new version of the model.
• Live search. Many AI assistants also search the web when you ask a question and use the pages they find as sources. That information is current.

If someone asks "What is a good plumber in Zwolle?", the assistant usually combines both. That is why what has been online about you for years counts, as well as what you put on your website and in reviews today.

Which LLMs are there?

The best-known AI assistants each run on their own language models:

• ChatGPT by OpenAI.
• Gemini by Google.
• Claude by Anthropic.
• Perplexity, a search assistant that uses several language models and always shows sources.

Each model gives slightly different answers, even to exactly the same question. That is why it makes sense to look at your visibility per model. Read more in why AI gives a different answer every time.

How does an LLM choose which businesses to mention?

An LLM mentions a business when it finds enough reliable information that fits the question. That depends on things like:

• a website that clearly says what you do and where;
• reviews and mentions on sites the model trusts;
• the same details everywhere your business is mentioned;
• pages that answer customers' questions directly.

Our article on how ChatGPT chooses which businesses to recommend goes into this in more detail.

LLM SEO, LLM optimisation and GEO: what is the difference?

There is none, really. They are different names for the same thing: making sure language models mention your organisation when someone asks a relevant question. GEO (Generative Engine Optimization) is the most common term, but you will also see LLM SEO, LLM optimisation and AI search optimisation. Also read what GEO is exactly.

What does this mean for your business?

Three practical points:

1. Make sure an LLM can read you. Do not block AI crawlers and make your key information easy to find. See allowing AI crawlers in robots.txt.
2. Be consistent everywhere. The same name, address and description on your website, in your business profiles and on review sites.
3. Measure what the models say. An LLM changes with every new version. What goes well today may be different next month.

See what LLMs say about you

Curious how ChatGPT, Gemini, Perplexity and Claude talk about your business right now? With the free AI check we ask five customer questions to all four models and you get a report with your opportunities. Want to follow it every week? Use GEO tracking.