Skip to main content
AI & Agents

Embedding

An embedding is a list of numbers representing the meaning of a piece of content, produced by a model so that semantically similar items sit close together in the resulting space. Comparing embeddings lets software judge whether two texts are about the same thing without matching any of the same words.

Embeddings are what make semantic search possible. A query about "reducing customer churn" can surface a document about "improving retention" because their embeddings are near each other, despite sharing no vocabulary. This is also their limitation: proximity in meaning is not precision, which is why embedding search alone performs badly on exact codes and identifiers.

Two operational rules matter more than model choice. Embeddings from different models are not comparable, so changing the embedding model means re-embedding the entire corpus. And what you embed matters as much as how — chunking documents along their real structure, rather than every fixed number of tokens, is usually the difference between a retrieval system that works and one that returns fragments.

FAQ

Embedding
FAQ.

Common questions about embedding.

Ask Us Anything

No. Each model produces vectors in its own space, so distances between embeddings from different models are meaningless. Switching embedding models requires re-embedding the whole corpus — worth planning for, because it makes the choice more consequential than it first appears and re-embedding a large corpus takes real time and budget.

Related terms

Vector Database

A vector database stores text, images or other content as numerical embeddings and retrieves items by semantic similarity rather than exact keyword match.

RAG (Retrieval-Augmented Generation)

RAG is a technique where an AI system retrieves relevant passages from your own documents at query time and generates an answer grounded in them, with citations.

Semantic Search

Semantic search finds results by meaning rather than by matching words, so a query about "reducing customer churn" can surface a document about "improving retention" even though they share no vocabulary.

Browse the full glossary →