Embedding
A list of numbers standing in for a piece of text, arranged so that two texts about the same thing land close together and can be found by meaning instead of by matching words.
Keyword search fails the moment the reader and the document use different words for the same thing. An embedding model turns both into points in one space, and the distance between the points answers the question of whether they are about the same thing.
The length of the list is fixed per model — a few hundred to a few thousand numbers — and embeddings from two different models cannot be compared at all. Switching models therefore means re-embedding the whole corpus, which is the cost that gets left out of the plan.
It does not replace keyword search. Exact names, product codes, and version numbers are where meaning-based search is weakest, so production systems usually run both and merge the two lists.