CAS Newton brings conversational AI to scientific research
CAS, a division of the American Chemical Society, launched CAS Newton on April 9, an agentic AI system designed to help researchers navigate scientific literature and data. The tool draws on 150 years of curated chemistry, biology, and materials science knowledge from the CAS Content Collection.
The system answers scientific questions by grounding responses in published research rather than generating answers from general training data. In early feedback, three of four users rated CAS Newton answers as more trustworthy than responses from other AI tools.
How it works
CAS Newton carries context forward across multiple interactions, refining questions and synthesizing results as inquiry deepens. Researchers can ask follow-up questions without starting over, reducing the time spent on literature review and data synthesis.
The tool connects concepts across disciplines-chemistry, biology, materials science, and patent data-within a single conversation. This cross-disciplinary linking helps researchers spot connections they might miss using traditional search methods.
Access and deployment
Users can access CAS Newton through CAS SciFinder, CAS BioFinder, or a standalone interface. Organizations can also deploy it in secure environments and integrate it with proprietary data through APIs and third-party AI platforms.
CAS Newton does not share user input outside its application boundary, and queries are never used for cross-user model training. This approach allows R&D teams to use the tool alongside their own data while maintaining data governance standards.
What researchers gain
- Faster access to relevant published science without specialized search expertise
- Concise summaries of large reference sets to support faster decision-making
- Ability to cross-reference findings across chemistry, biology, materials science, and intellectual property data
John Yates, a professor at Scripps Research Institute, said that grounding AI agents in authoritative, curated chemistry data "will significantly expand access and improve efficiency" for researchers moving from question to insight.
For professionals managing research workflows, AI for Science & Research training can help teams understand how to integrate agentic AI tools into existing research operations.
Your membership also unlocks:
