Retrieval augmented generation (RAG) empowers generative language services by integrating extensive context from external data sources (a.k.a. knowledge bases). The current RAG-enhanced generative services are predominantly hosted in cloud environments, relying on static knowledge bases without real-time sensory information which may lead to constrained scalability, responsiveness, and overall service quality. One promising opportunity is to extend the deployment of such services to the network edge, leveraging the anticipated capabilities of 6G networks. In this article, we propose a deployment framework for RAG-enhanced generative services in 6G. We address the key challenges at the convergence of service deployment, 6G networks, and user interactions. Additionally, we explore potential techniques to enhance RAG-based services through data fusion, dynamic knowledge base deployment, service customization, and interactive user experiences. Lastly, we shed light on future paths toward the effective deployment and delivery of RAG-enhanced generative services.