Clarification on Qdrant Vector Store Behavior When Adding New Documents With Langchain

[crman]

Hello,

I have a question regarding the behavior of the Qdrant vector store in Langchain when adding new documents to an existing vector store.

Scenario:

• I initially created a Qdrant vector store using 3 documents, and their embeddings were stored in the Qdrant cloud.
• Later, I added 2 more documents to the same directory.
• When I create the vector store again, I want to understand how it handles the existing and new documents.

Question:
When I recreate the vector store after adding the new documents, does it:

• Recreate the vector store by reprocessing the original 3 documents along with the embeddings of the 2 newly added documents, and then update the cloud?
• Only create embeddings for the 2 new documents and add them to the existing vector store in the cloud without reprocessing the original 3 documents?

Understanding this behavior is crucial for optimizing our workflow and avoiding redundant processing.

I appreciate any guidance or documentation you can provide on how the vector store handles such scenarios.

Thank you!