Embeddings

An embedding is a string of numbers, called a vector, that represents a piece of text by its meaning rather than its exact words. Text with similar meaning ends up with similar numbers, so a search can find a relevant passage even when it shares no words with the question. This is what powers semantic search.

Embeddings are strong on natural language and weaker on exact terms. On text full of precise tokens (product codes, ticker symbols, dates, line-item labels) they can blur the distinctions that matter, which is why plain keyword search sometimes beats them. They are stored and compared in a vector store.