A vector is an ordered list of numbers that represents a point in a multi-dimensional space. In the context of AI databases, vectors are how meaning is encoded: a piece of text, an image, or any other data is converted by a model into a vector — for example a sequence of several hundred numbers — that captures its semantic content in a form machines can compare.
The crucial property is that similar content produces similar vectors. Two sentences with the same meaning are represented by vectors that sit close together in the space, even if they use entirely different words, while unrelated content lands far apart. This turns the abstract idea of similarity into geometry, so that finding related items becomes a matter of measuring distances between vectors.
Vectors are the fundamental unit that vector databases store, index, and search. Modern embeddings typically have between a few hundred and a few thousand dimensions, each capturing some learned aspect of the data, though the individual dimensions are not directly interpretable. Everything a vector database does — similarity search, filtering, retrieval — operates on these arrays of numbers.