The File Search API looks very promising and we would love to use it for vector creation, automated (semantic?) chunking etc. It works quite well with the kinds of documents we want to use it with (legal).
The problem is that we need to be able to do pre-filtering for the similarity search part based on customers/users and also document categories.
We currently use a third party vector db for that and we would like to keep that one. Is it possible to get the results of the file search embeddings process and store it manually in another vector store?
If currently impossible, when can one expect this to work?
Thx!