A consulting firm is building a gen AI assistant that must provide accurate, up-to-date responses using the firm’s proprietary research reports and documentation. The goal is to reduce hallucinations by grounding responses in trusted enterprise content stored in a private database. Which approach should they use to best meet this requirement?