can the Enhanced RAG service class be used with vapi, like using the sdk, to take in the text, transcribe/stt it, run it through the enhanced rag, get the response and preform tts and than send the audio back to the front-end