Definition
Embeddings convert text, images, or other data into vectors. Items with similar meaning sit closer together in vector space, which makes semantic search and recommendations possible.
How it works
An embedding model reads the input and outputs a list of numbers. Search systems compare those vectors to find content that is semantically similar rather than just keyword-matched.
Why it matters at work
Embeddings power RAG, semantic search, clustering, recommendations, deduplication, and personalization. They help AI systems find relevant context before generating an answer.
Workplace example
A support knowledge base uses embeddings so a customer question about billing failure can retrieve articles about payment retries even if the exact words differ.
Frequently Asked Questions
Are embeddings only for text?
No. Embeddings can represent text, images, audio, code, products, users, and many other data types, depending on the model and use case.