Comment by cheesyFishes

This is actually what llama-index does if you don't use a vectordb integration