# Embeddings

> Embeddings are lists of numbers (vectors) that represent the meaning of a piece of text, an image or other data, produced by an embedding model. Items with similar meaning end up with vectors that are close together, even when they share no keywords. Embeddings power semantic search, retrieval for RAG systems, recommendations, clustering and duplicate detection.

## Why it matters for your business

Embeddings are what let an AI assistant find the policy about remote work when an employee asks about working from home. They are created once when content is indexed and again for each query, and they are inexpensive compared with generating text. The choice of embedding model affects retrieval quality, especially for specialized vocabulary, and switching models later means re-embedding every document. Embeddings are derived from your data, so they should follow the same access and retention rules as the source documents.

## Example

A manufacturer embeds several thousand past quality incident reports. When an engineer logs a new defect described as hairline cracking near the weld on bracket assemblies, the system finds earlier reports with similar meaning, including ones worded differently, such as fractures at the joint, and shows how each was resolved.

Related terms: https://www.lumeor.studio/glossary/vector-database/, https://www.lumeor.studio/glossary/rag-retrieval-augmented-generation/, https://www.lumeor.studio/glossary/chunking/, https://www.lumeor.studio/glossary/knowledge-base/

---

Source: https://www.lumeor.studio/glossary/embeddings/