If one has more than one vector store object that requires some of the same files, isn’t it messy to have the same file uploaded again and again through the Platform if you want them attached to the related vector store?
Does anyone else have that issue or are we doing something wrong?
Thanks.
I have the exact same issue and I haven’t found a solution yet. I believe the whole thing is super messy, and very non-user friendly. It seems to overcomplicate things. Shouting this out there to someone who wants to build an “API Files / Assistant / VS Manager” with a nice user interface: I’d pay for that.
Welcome to the community!
Isn’t this what every vector db service provider is doing for you? 
Everyone other than OpenAI’s API Platform you mean?
Well, you can do it natively with openai too: you can use the embedding models to build your own vector store and keep it in memory. A simple vector search is just a simple for loop + dot product, and works fairly well up until around 1000 entries.
You’d just have to manage your own object storage.
That’s a good idea. Thank you.