Back to Glossary
TechnicalAI Glossary

Embedding

Quick Answer

An embedding is a numerical vector representation of text, images or other data that captures semantic meaning. Items with similar meaning produce similar vectors, which makes embeddings the backbone of semantic search, recommendations and RAG systems.

In Depth

What Embedding really means

An embedding typically has a few hundred to a few thousand dimensions. Cosine similarity or dot product between two embeddings gives a useful measure of how related the underlying items are, independent of their surface wording.

Embeddings are produced by dedicated models, which may be much smaller and cheaper than full LLMs. Choosing the right embedding model and keeping embeddings refreshed as content changes are important operational considerations.

Why It Matters

Business relevance for UK organisations

Embeddings power semantic search across knowledge bases, duplicate detection in CRM data, clustering of support tickets, and 'more like this' product recommendations on UK e-commerce sites.

Real-world example

How this shows up in practice

A Leeds e-commerce brand replaced its keyword search with embedding-based semantic search and saw a 22% uplift in search-to-purchase conversion.