What is vector search?

Vector search works by converting a question and a collection of documents into embeddings, then finding which documents' numbers sit closest to the question's numbers. The result is a search that can surface a relevant passage even when it doesn't contain the words the question used, something ordinary keyword search struggles with.

This is a core piece of how many AI answer tools decide what to read before responding: it's part of what makes retrieval-augmented-generation and grounding work in practice. For a site owner, it explains why a page can get pulled into an AI answer for a question phrased completely differently from the page's own wording, and why loading a page with one exact keyword phrase isn't the lever it would be for classic search, since the comparison here runs on meaning. Most real systems still combine this with ordinary keyword indexation rather than relying on meaning-matching alone.