How to use multiple documents in Assistants API to get best result?

I am developing a customer service assistant tool application, specifically for e-commerce sellers who need to use AI customer service assistants to serve their buyers.
To help the AI assistants provide better consultation services about products and services, these sellers will need to upload materials introducing their products, services, and company. Each seller might upload one or more files, and the importance and priority of these files may vary in different scenarios. Therefore, I plan to use OpenAI’s Assistants API to implement the RAG functionality.
The question is: once the sellers have uploaded these documents, in the conversation between buyers and the AI customer service assistant, how does the Assistants API determine the importance and priority of these files for different types of questions, and is additional coding required to achieve better retrieval results?
Thanks for your help.