Large language model (LLM)

A large language model (LLM) is an AI system trained on massive amounts of text to understand and generate human language by predicting likely sequences of words. LLMs like the models behind ChatGPT, Gemini, and Claude are the engines behind generative search — they synthesize the answers that increasingly sit between a user and your website.

Why it matters for search

LLMs changed search from retrieving links to composing answers. Two properties matter for visibility. First, an LLM’s built-in knowledge is frozen at its training cutoff and can be wrong or vague about specifics, which is why current answer engines pair the model with retrieval (RAG) to ground responses in fresh sources. Second, an LLM works with whatever text it’s given, so clean, well-structured, unambiguous content is easier for it to interpret and reuse correctly. You don’t optimize the model; you optimize the content it retrieves and the entity signals that shape how it represents your brand.

Where it applies

• Answer and generative engines — the core technology behind AI search results.
• Content operations — LLMs used to scale content production and optimization pipelines.
• Brand representation — what a model “knows” and says about your brand when asked directly.

What matters from each seat

• In-house — focus on the retrievable, structured content and entity signals a model draws on, not on gaming the model.
• Brand — monitor how models describe your brand; a wrong default answer is a reputation issue.
• Agency and consulting — use LLMs to scale production, but keep human review on accuracy and originality.