A traditional database finds books by title or ISBN. A vector database finds books by meaning — 'show me books similar to this paragraph.' It stores the essence of data as mathematical coordinates and finds neighbors in that space.