NVIDIA and Pure Storage, an enterprise storage company, have partnered to develop retrieval-augmented generation (RAG), which is intended to enhance LLMs.

The collaboration focuses on specific initiatives such as 1) a RAG pipeline for AI inference, using NVIDIA NeMo Retriever microservices and GPUs alongside Pure Storage's enterprise storage solutions to enhance the accuracy, currency, and relevance of inference capabilities for LLMs and 2) vertical-specific RAG solutions for the financial sector with versions apt for the healthcare and public sector set to be released in the future.

The firms claimed that this partnership could improve AI models by decreasing false information while increasing enterprise-specific customization.