Large Language Model (LLM)

What is a Large Language Model (LLM)?

A large language model (LLM) is a neural network trained on very large text collections to predict the next piece of text. That simple objective lets it answer questions, summarize, write and reason over language. LLMs such as GPT, Gemini and Claude are the engines inside AI assistants and AI search products.

How it works

An LLM goes through two broad phases:

• Training. The model learns statistical patterns from public web pages, books, code and licensed data up to a knowledge cutoff. It is then tuned to follow instructions and be helpful.
• Inference. When a user sends a prompt, the model generates a response one token (a word piece) at a time.

On its own, an LLM only knows what was in its training data. AI search products add retrieval-augmented generation: they search the web, pass relevant passages to the model, and ask it to answer from those passages with citations. That is why both your long-term reputation (training data) and your current pages (retrieval) affect how AI engines describe you.

Why it matters for SEO and AI search

• Brand descriptions in training data persist until the next model version, so inaccurate information can linger.
• Retrieval favors pages that are crawlable, relevant and easy to quote.
• LLMs can hallucinate, so clear and consistent facts on your site reduce the risk of errors.

Examples

• A user asks ChatGPT “what does a CNAPP do?” and the model answers from training data, with no sources shown.
• The same question in ChatGPT search or Perplexity triggers retrieval, and the answer cites current vendor and analyst pages.
• A DevTools company finds an assistant describing a deprecated feature because the model’s training data predates the change.

How to apply it

1. Keep core brand facts (category, features, pricing model, integrations) consistent everywhere.
2. Publish clear, crawlable pages for each fact you want models to repeat.
3. Earn coverage on sites models commonly learn from and cite.
4. Test how major LLM products describe you and log errors.

Related terms

• Retrieval-Augmented Generation (RAG)
• Knowledge Cutoff
• AI Hallucination
• LLM Optimization (LLMO)
• BERT