[cranberryturkey]

Doesn't RAG approach (and LangChain) both require you send the context data (ie: your book data) in the prompt query api call? How would you fit 20,000 books in that call?

[abdullin]

It is impossible to fit all that information into the call.

The whole point of RAG - we (somehow) retrieve only the relevant information and put it into the context to generate the answer.