Meta. Google. Microsoft. These companies have spent years engineering faster vector-search systems. A basic nearest-neighbor query computes the distance between the query and every stored vector. Its cost grows linearly with the size of the database and becomes impractical