Vector databases use approximate nearest-neighbour indexes to search millions of embeddings in milliseconds, often combined with keyword search and metadata filters. They are a common component of retrieval-augmented generation and recommendation systems.
Many general databases now offer vector search too. See vector databases explained.