Glossary
What is LLM (Large Language Model)?
Last updated:
An LLM (Large Language Model) is a neural network trained on very large text corpora that can understand, generate and reason over text. Examples are GPT, Claude and Gemini.
Technically, a language model predicts the most probable next word at each step; billions of these steps produce answering behaviour. Two consequences follow directly: models hallucinate (they produce plausible but false text), and they preferentially cite content that is clear, evidenced and machine-readable.
For websites, the training approach decides whether content appears in answers: what is missing from the training and retrieval data cannot be cited. That is why crawler access and citability matter more than positions in results lists.
Why it matters for the mid-market
LLMs are the engine underneath AI applications from phone agents to copy generation. The model choice decides cost, privacy and quality; it belongs in strategy, not in IT alone.