CAS, a division of the American Chemical Society, has introduced CAS Newton, an agentic AI platform designed specifically for scientific discovery. Grounded in the CAS Content Collection, the tool leverages over 150 years of curated scientific literature to provide researchers with reliable, verified insights across chemistry, biology, and materials science.
The platform utilizes a conversational interface that maintains context through complex inquiries, allowing it to synthesize data and refine questions across multiple steps. According to their announcement, this science-smart workflow aims to accelerate decision-making by summarizing large datasets into concise, accurate answers. They share that early user feedback suggests a high level of trust, with three out of four respondents preferring its accuracy over general AI tools.
To support enterprise security, CAS Newton can be deployed within private environments and integrated with proprietary data via APIs. They claim that CAS maintains a strict governance policy, ensuring that user inputs are never shared externally or used for cross-user model training. The technology is now available through CAS SciFinder, CAS BioFinder, and as a standalone interface.
