The fusion of Retrieval-Augmented Generation(RAG) in Conversational AI has profoundly improved its capacity to deliver accurate and context-aware responses by combining the retrieval of relevant information from the knowledge base with human-like language generation. However, Conventional RAG models are often restricted to static knowledge bases which become outdated over time and hinder the capacity of these systems to adapt to proprietary domains in healthcare, education, finance and investments etc. This research proposes a new continuous learning and dynamic knowledge update framework of RAG to address these limitations. The proposed framework is designed to not only enhance the response accuracy and context relevance but also to improve user satisfaction by incorporating real-time data update and user feedback loops. Specifically, our method uses an implicit and explicit feedback-driven system to gather user feedback after each interaction to guide continuous learning, allowing the model to refine the knowledge base and improve its response. This not only reduces the problem of information staleness, but also enhances customer satisfaction. The proposed model has shown significant gains in performance parameters such as, 15% improvement in response accuracy and 20% improvement in user satisfaction over the conventional RAG system. The proposed framework is also a contribution to the growing field of intelligent conversational agents as it provides insights on building an adaptive RAG model to address the challenges of proprietary and dynamic information environment.