Retrieval Augmented Generation (RAG) allows enterprises to ground responses from Large Language Models in their specific organization’s data. This helps ensure that AI-powered applications provide responses that are not only accurate, relevant, and consistent, but also aligned with business needs. At Snowflake, we make it simple for our customers to implement RAG, while also enabling the strict governance and privacy controls that businesses require. 

RAG can be implemented in a variety of ways, and Snowflake provides tools and features that allow customers to choose the approach that best meets their needs. To expand these choices, Snowflake announced today that it has invested in Contextual AI, a simple, powerful solution for end-to-end RAG that delivers highly accurate responses.

A typical RAG system combines a vector database, embedding models and other components that must be engineered and integrated together. In contrast, Contextual AI's groundbreaking RAG 2.0 technology seamlessly integrates retrieval and generation to deliver unmatched accuracy for RAG workloads without any additional, error-prone system tuning. 

Following our investment, Snowflake customers will be able to deploy Contextual AI as a Native Application directly within their Snowflake account, simplifying and accelerating RAG deployment for a wider set of users. 

Contextual AI supports a variety of use cases for RAG at scale, including contextual search, chatbots and contextual parsing. For example, its customers include major financial services organizations that depend on Contextual AI to build applications that parse and analyze large volumes of SEC filings to surface insights and guide decision-making.

This investment complements the capabilities of Snowflake Cortex AI, our suite of tools and features for building AI-powered applications. Cortex AI already includes technologies for RAG development, and Contextual AI further expands the options for customers who want a packaged and comprehensive solution that can be deployed quickly and easily.

Contextual AI was co-founded by CEO Douwe Kiela, who previously led research at Meta’s Fundamental AI Research (FAIR) team, where he pioneered Retrieval Augmented Generation (RAG) and other significant AI advancements. Douwe is also former head of research at Hugging Face and an Adjunct Professor in Symbolic Systems at Stanford University. His co-founder is CTO Amanpreet Singh, who has a background in research engineering at Hugging Face and Meta’s FAIR team. The company recently launched RAG 2.0, an update to its Contextual Language Model with benchmarks that outperform alternatives for retrieval augmented generation tasks. 

We look forward to working closely with Douwe and his team to ensure that Snowflake customers can benefit from further enhancements to their product in the future.