Proxy-Pointer RAG: Multimodal Answers Without Multimodal Embeddings

# Researchers have found a way to get AI systems to answer questions using both text and images without needing expensive specialized technology that handles both types of information at once. Instead of forcing the AI to process images and text together, the new approach uses a smarter organizational structure that lets the system point to the right image when needed, then provide answers more efficiently. This could make it cheaper and easier for businesses to build AI tools that combine visual and written information.