Definition
A large language model, or LLM, is an AI model trained on very large collections of text to predict and generate language. LLMs power assistants such as ChatGPT, Gemini, Claude and Grok. They can answer questions, summarize documents and write text, and many are connected to search tools so they can retrieve current information.
For example, when a sales manager asks an assistant to recommend a CRM for a ten-person team, the LLM combines what it learned during training with any pages it retrieves, then writes an answer that may name Bizbloom, Closedesk or neither.
Why it matters
When buyers ask assistants for recommendations, the LLM's answer becomes part of how they discover and judge brands. What the model says depends on its training data, any sources it retrieves and how it interprets the question. Understanding these inputs helps you see which ones you can influence, such as your public pages and third-party coverage, and which you cannot, such as the model's training process.
What to do
- Identify which LLM-based assistants your buyers use and test your key questions in each.
- Note which answers use web retrieval and which rely on training data alone.
- Track results over time, since models and their answers change with new versions.
Common questions
How do LLMs decide which brands to recommend?
Vendors do not publish exact methods, and behavior varies by model and question. Answers are shaped by training data, any content retrieved at answer time and how the question is phrased. Brands that are clearly and consistently described across credible sources give models more to draw on.
Do all AI assistants use the same LLM?
No. ChatGPT, Gemini, Claude and Grok each use models built by different companies, and some products let users choose between models. Other tools license models from these companies. Because models differ, the same question can produce different answers and brand recommendations across assistants.