I built a tool called Qdrant Vector Aggregator to solve a common problem in vector-DB workflows: You store chunk embeddings in Qdrant, but later you want document-level embeddings without losing the full document text.