From NVIDIA: 2024-08-29 12:00:27

Douwe Kiela, CEO of Contextual AI, recognized the limitations of large language models like ChatGPT. In 2020, he introduced RAG, a method for continuously updating foundation models with real-time data. Contextual AI recently closed an $80 million Series A round and plans to double its workforce by year-end.

Contextual AI’s platform, RAG 2.0, offers advanced retrieval-augmented generation. The integration of retriever and language model architectures leads to 10x better accuracy and performance. This optimized solution allows models to run on smaller infrastructure but maintain high accuracy, ideal for edge use cases and LLM-powered chatbots.

The startup’s approach to synchronizing and adjusting neural networks through back propagation leads to significant gains in precision and response quality. RAG 2.0 is LLM-agnostic and can work with different open-source models. Contextual AI overcomes the challenge of identifying relevant data stored in various formats through a “mixture of retrievers” approach, leading to improved performance and customer satisfaction.



Read more at NVIDIA: Startup Contextual AI Uplevels Retrieval-Augmented Generation for Enterprises