# What is LLM (Large Language Model)?

Source: https://mushbloom.uk/resources/glossary/what-is-llm.html

By Mushbloom. Published and updated: 2026-10-12.

## Definition & overview

A large language model is a model trained on large amounts of data to process and generate language. Its generated text is not necessarily a database lookup or a live web search.

## Analogy & mental model

It is a skilled language assistant, not an always-current encyclopaedia.

Give the model reliable context and check important output rather than assuming fluency equals accuracy.

## Why it matters

LLMs are useful for language-heavy tasks, including summarisation, drafting and interpretation. Their fluency can make errors difficult to notice. A business should separate the model’s language capability from the external information and permissions needed to do a task correctly.

## Practical business example

A model rewrites a staff-approved explanation in simpler language, while the actual product facts remain anchored in the supplied source.

Illustrative scenario, not a client case study or promised result.

## How it works

1. Provide instructions and appropriate evidence for a defined language task.

2. The model generates output based on the input and learned patterns.

3. Validate the result and use separate software controls for consequential actions.

## Core components

- Trained model

- Task context

- Output validation

Learned information may be incomplete or outdated, and fluent output can still be false.

## LLM — Large Language Model vs Search engine

A search engine retrieves indexed material. An LLM generates language; an application may combine both but the model alone is not necessarily searching.

## Frequently asked questions

What does LLM mean?
A large language model is a model trained on large amounts of data to process and generate language. Its generated text is not necessarily a database lookup or a live web search.

How is it different from Search engine?
A search engine retrieves indexed material. An LLM generates language; an application may combine both but the model alone is not necessarily searching.

What is a sensible first step for a small business?
Give the model reliable context and check important output rather than assuming fluency equals accuracy. Provide instructions and appropriate evidence for a defined language task.

What limitations should I keep in mind?
Learned information may be incomplete or outdated, and fluent output can still be false.

## Sources

Google Cloud: Retrieval-augmented generation: https://cloud.google.com/use-cases/retrieval-augmented-generation

No rankings, AI citations or commercial outcomes are guaranteed.