A large language model (LLM) is a neural network trained on enormous amounts of text to predict the next word, and, from that simple objective, to generate fluent language, answer questions, and summarize. It's the engine inside ChatGPT, Gemini, Claude, and the AI features now built into search.
For brands, the important part is how an LLM "knows" things. Some knowledge comes from training: text the model learned, frozen at a knowledge cutoff. The rest comes from live retrieval at answer time, where the model searches the web and grounds its answer in what it finds (see RAG and grounding).
An LLM doesn't store facts like a database; it represents patterns. That's why it can sound confident yet be wrong. An AI hallucination. Influencing what an LLM says about your brand means being well-represented in the text it learns from and being retrievable and citable when it searches live.