Techniques & Methods
Embeddings
How AI represents meaning as lists of numbers — the math behind search and RAG.
Also known as: vector embeddings,word embeddings
An embedding is a numerical representation of meaning. Take a word, sentence, image, or document, and the embedding model turns it into a list of numbers (a vector, typically 768 to 3,072 dimensions). Crucially, things with similar meaning get similar vectors. "dog" and "puppy" end up close together; "dog" and "automobile" end up far apart. Embeddings power semantic search (search by meaning, not just keywords), RAG, recommendation engines, clustering, and classification. Most modern AI products use embeddings under the hood. OpenAI, Cohere, Google, and Voyage all offer embedding APIs; sentence-transformers is the popular open-source option.