Contextual AI announced it secured $20 million in funding led by Bain Capital Ventures (BCV) with participation from Lightspeed, Greycroft, SV Angel, and prominent angel investors.
The newly formed company specializes in the development of large language models (LLMs) for enterprise applications that prioritize data privacy and offer enhanced safety and trustworthiness. The LLMs are unique on the market as they can be customized to specific business needs, and include greater efficiency compared to others currently available on the market.
Contextual AI aims to address the current challenges by developing LLMs that are suitable for work environments. The company's objectives include tackling issues such as hallucination, where LLMs generate false information with high confidence. They also aim to improve attribution, allowing for verification and error correction.
Compliance is another issue it plans to focus on, aiming to enable efficient removal or revision of information in LLMs to mitigate risks. The company is also focused on customization that will adapt to new data sources seamlessly. Additionally, with its LLMs, data privacy is prioritized, ensuring companies can retain their valuable private data without the need to share it externally.
“Knowledge workers of the future need LLMs that work accurately, efficiently and effectively over huge private datasets, in a way that companies can trust,” said Douwe Kiela, co-founder and CEO.
Douwe Kiela and Amanpreet Singh, co-founders of Contextual AI, have extensive backgrounds in AI research. They initially met as research leaders at Meta AI in 2016 and continued their collaboration at Hugging Face. Contextual AI's proprietary research builds upon the established Retrieval Augmented Generation (RAG) method, which Kiela played a pioneering role in during his time at Meta.
“With 10-plus years of experience as researchers in AI, natural language processing and machine learning, my cofounder Amanpreet and I are on a mission to take this powerful technology to the next step, where it can be used in real-world enterprise applications,” said Kiela.
Edited by
Greg Tavarez
