Retrieval Augmented Generation (RAG) on individual users documents for persoanlized document search

Our users store personal notes within our system. We’d like to implement Retrieval Augmented Generation (RAG) so users can search and receive answers based solely on their own notes, without accessing other users’ documents. How can we achieve this personalized search experience while maintaining data privacy i.e it searches in their notes only?

You can use a Custom Information Retriever to connect to any data source to retrieve data with custom queries.

You don’t mention what your storage looks like, but it would be straightforward to query a relational database, and select only the relevant records in the WHERE clause of your query.

But for text based querying we need to do similairty search and normal RDBMS querying will not work. I am using Mongo DB for storgae as of now.

Is this featue only avaiable in 3.9. I am using 3.7