I don’t have OpenAI access; how do I implement some of these advanced rag features with Gemini Pro (free API version)?