Contextual AI, a start-up co-founded by Douwe Kiela, has emerged from its stealth mode with $20 million in seed funding. The company aims to create “next generation” large language models (LLMs) for enterprises. LLMs have the potential to revolutionise industries, but they do come with limitations, particularly in reliability, compliance and governance issues. To address these challenges, Contextual AI has developed a new text-generating AI technology based on retrieval augmented generation (RAG), a technique that augments LLMs with external sources to improve their performance. Contextual AI’s co-founder, Kiela, led research into RAG while at Meta.
RAG technology allows LLMs to develop a “context-aware” response to a question. This can be useful for companies that want to utilise generative AI for specific business applications. Contextual AI is looking to explore these potential applications with Fortune 500 companies. The start-up plans to spend the majority of its seed funding on product development, including the construction of a compute cluster to train LLMs.
While several generative AI companies have stated they will pursue the enterprise market, Contextual AI claims it will take a different approach by building a much more integrated solution tailored for enterprise use cases. Contextual AI hopes its focus on accuracy, reliability and traceability will help enterprises unlock the true potential of generative AI.
