Comment by paddw

Preprocess the documents to extract certain key features like "Cited Risk Factors", and have the AI model use that instead of each whole document as context?