What is a large language model (LLM)?

How an LLM works, in plain language

During training the model has “read” an enormous amount of text and learned how words and ideas connect. When you ask a question, it builds the most likely answer step by step. That’s why it writes and understands text brilliantly, but it has no guaranteed knowledge of your company or the latest events.

What an LLM can and cannot do

Can: understand free text, write, translate, summarise, classify, extract data from documents, plan steps (in agents).

Cannot on its own: know your customers, prices and internal rules, guarantee factual accuracy, access your systems.

How to make an LLM useful for a business

• Instructions — how you work and the style you write in.
• Company data — with RAG the model finds information in your documents before answering.
• Tools — access to systems, so the model becomes an AI agent.
• Checks — so that mistakes are caught before they reach a customer.