Researchers unveiled AdaptiveFlow, an AI-informed open-source platform designed to virtually screen billions of drug-like molecules while cutting computational costs by about 1,000-fold versus conventional ultra-large virtual screening methods. The platform is validated by scientists from St. Jude Children’s Research Hospital, the University of Pavia, Dana-Farber Cancer Institute, and Harvard Medical School, with publication in Nature Biotechnology. AdaptiveFlow includes an optimized screening-ready version of the Enamine REAL Space library (69 billion compounds), integrates thousands of docking workflows, and uses property-based prioritization with optional active learning to focus compute where hits are more likely. The team reported identifying nanomolar inhibitors for FSP1 and PARP1 and provided co-crystal structures to support binding insights. The release is aimed at making ultralarge screens routine by leveraging GPU-accelerated methods and near-linear cloud scaling, potentially reducing time-to-hit for early discovery teams.