LLM

Large Language Model

A neural network trained on huge volumes of text that can read, write and reason in natural language — the engine behind ChatGPT, Claude, Gemini and friends.

LLMs are the general-purpose language engines behind modern AI products. They're trained on a snapshot of the web and licensed content, which means they're surprisingly good at general knowledge and surprisingly bad at anything specific, recent or proprietary — like your product range.

Three patterns close that gap: prompt engineering (tell it more), RAG (give it source documents to ground in), and tool use via MCP (let it query your live systems). For product use cases, RAG plus MCP beats relying on what the model 'knows' every time.