The text discusses the implementation of Retrieval Augmented Generation (RAG) with multimodal data, which can improve the accuracy of Large Language Models (LLMs). The article covers three key patterns to implement multimodal RAG: grounding all modalities into one primary modality, embedding them into a unified vector space, and employing hybrid retrieval with raw image access. The choice of pattern depends on the specific needs of the AI application. Additionally, the text highlights the importance of scalability in implementing multimodal RAG systems, particularly when dealing with large amounts of data. It also introduces Milvus, a vector database that offers advanced features and easy integration with popular tools for multimodal RAG. The article concludes by emphasizing the significance of using a scalable vector database system like Milvus for AI applications that require efficient and accurate response generation.