Embedding

An Embedding is a numerical representation of text, images, or other data as a vector in a high-dimensional space. Semantically similar concepts are placed close together — "dog" and "cat" are closer than "dog" and "stock market." These vectors are the foundation of modern AI applications.

In practice, embeddings are generated by specialized models (e.g., OpenAI text-embedding-3, Cohere Embed, Google Gecko). A typical embedding vector has 768–3,072 dimensions. They are stored in vector databases like Pinecone, Weaviate, or Chroma, which enable efficient similarity searches.

Embeddings are the technical foundation for RAG systems: documents are stored as embeddings, and when a user query arrives, its embedding is computed and the most similar document chunks are found. Semantic search, recommendation systems, duplicate detection, and clustering are all based on embeddings.

Particularly relevant for marketing: embeddings enable semantic content analysis (finding thematically similar articles), audience clustering based on behavior patterns, and intelligent product recommendations. Costs are minimal — OpenAI's embedding model costs just $0.02 per million tokens.