deploy production RAG

[rapidoo]

Hi,

I succeed to build a multi modal RAG following, this cookbook : https://github.com/langchain-ai/langchain/blob/master/cookbook/Multi_modal_RAG.ipynb that's great,

But I want to deploy this RAG as a langgraph agent, and push it to langchain Cloud,

Everything working wells, I use Qdrand as a vectorStore as Chroma have no saas yet.

And my vision is to load RAG from a service (currently my machine), and allow Cloud to use it for the agent.

The VectorDatabase work fine, but I have difficulty with docStore, ...

As it need to be external too, I use MongoDBByteStrore https://python.langchain.com/v0.2/api_reference/community/storage/langchain_community.storage.mongodb.MongoDBByteStore.html

But as we store String object in the store, the MultiVectorRetriever is buggy, because waiting for Document,

The point is that everything work well on the cookbook with Qdrant an Mongo, but in error with langgraph.

Is everyone use a saas docStore and can help ?