Embeddings

Embeddings are lists of numbers that represent the meaning of a piece of text or an image, so a computer can measure how similar two things are.

In practice: Text with similar meaning gets similar numbers, even when the words differ: "refund policy" and "how do I get my money back" land close together. This is what powers semantic search, recommendations and the retrieval step in RAG.

Why it matters when picking a tool

You rarely see embeddings directly, but they decide how well a tool finds the right document or answer. If a knowledge base or support bot keeps missing obvious matches, weak search based on embeddings is often the reason. Developer tools may let you choose the embedding model.