Large Language Model (LLM)

A large language model is an AI trained on enormous amounts of text to predict the next word (technically, token) in a sequence. That simple objective, at scale, produces systems that can summarise, answer questions, and write fluently. LLMs are the engines behind ChatGPT, Claude, and Gemini, and increasingly behind search itself.

For SEO, the practical points are that modern search routes many queries through LLMs to generate direct answers rather than just returning links, and that LLMs can misattribute or hallucinate. So the goal is to be the clearest, most consistent, most authoritative source on your topics, so that when a model assembles an answer it represents you correctly and cites you rather than a competitor. Entity clarity and consistency across the web directly reduce misattribution.

LLMs combine a static training snapshot with live retrieval in search settings, so you influence them through two channels: the corpus they train on (long-term reputation and consistency) and the passages they retrieve at query time (freshness and structure). Senior thinking treats "how the models perceive our entity" as a brand-and-authority problem, recognising that the associations built consistently across the web today shape which brands the models volunteer tomorrow.