CAS has announced the introduction of science-smart agentic AI designed to enhance research workflows and accelerate scientific discovery.
The new capabilities integrate intelligent agents trained on scientific content into CAS BioFinder® and CAS SciFinder®, enabling automation of complex research workflows. These agents allow scientists to ask direct questions and obtain reliable, evidence-backed responses grounded in the curated data of the CAS Content Collection™, distinguishing them from generic AI tools that rely on unverified or incomplete sources.
In the coming months, CAS BioFinder users will gain access to a conversational AI agent that supports complex scientific inquiry and interactive follow-ups. To ensure accessibility for a global audience, users will be able to communicate with the agent and receive detailed responses in their preferred languages. A literature review agent is also planned to summarize relevant publications and facilitate broader exploration.
CAS SciFinder will integrate similar agentic AI features before the end of 2025 to expand support for chemical research workflows.
CAS has emphasized that the new agentic AI experience upholds its long-standing standards of scientific rigor and transparency. Responses are linked to CAS entities and external sources to enable validation and further exploration. Judgment agents identify potential inaccuracies, and users can resume or download their session histories for review. These capabilities have been developed in accordance with the CAS Ethical Approach to AI, which underscores the importance of pairing technology with scientific expertise to promote responsible research aligned with the values of the global scientific community.
These advancements demonstrate the broader commitment of CAS to prepare researchers for the next era of scientific discovery. Agentic AI, trained on scientific language, workflows, and the CAS Content Collection™, is being integrated across CAS solutions to enhance scientific workflows, within internal operations to improve efficiency and accuracy, and through CAS Custom Services℠ to help organizations strengthen their own AI strategies.
Traditional application interfaces provide structured exploration and reliable outcomes. The new agentic AI experience builds on this foundation by allowing researchers to use natural language to pose questions and receive accurate, evidence-based answers. Designed to augment human expertise rather than replace it, this approach offers a more intuitive way to navigate scientific information while preserving the flexibility of existing workflows.
By leveraging the CAS knowledge graph, agentic AI connects concepts, relationships, and data points that may not be immediately apparent through conventional search methods. This deeper contextual understanding helps researchers identify insights more efficiently and make informed decisions.
Agentic AI in CAS BioFinder has begun assisting researchers in exploring complex disease areas. For example, users investigating ovarian cancer have engaged with a conversational agent that guided their inquiry and uncovered new research directions. While these insights could also be achieved through existing CAS BioFinder functions, the agentic AI experience introduces a new way to interact with scientific content—enabling natural-language queries and generating structured, report-style responses that streamline exploration.
A complex query on ovarian cancer has been interpreted by the conversational agent, producing a curated response based on verified scientific data.
Manuel Guzman, President of CAS, has highlighted that agentic AI helps scientists save time and focus on higher-value work. He has added that these intelligent agents enhance innovation by streamlining data exploration, surfacing relevant insights, and supporting faster, evidence-based decisions.
CAS has reaffirmed its commitment to advancing agentic AI capabilities to promote reliable research acceleration in a wide range of environments. Future development will focus on expanding agent functionality to address complex challenges in predictive modeling and decision support.
These evolving capabilities are expected to help organizations extract deeper insights from the human-curated data in the CAS Content Collection™, refine research strategies, and accelerate innovation across multiple scientific disciplines.
Click here to read the original press release.
More News in this Theme
