Google DeepMind demonstrated an AI agent that searched thousands of compounds and proposed new molecular structures with routes to make them in about an hour, a computational design workflow Google said would typically take days. The demonstration started from two existing lung cancer medicines, crizotinib and lorlatinib, and examined related compounds to suggest what a chemist might make next. The example examined restricting how freely a molecule moves so it can bind more effectively to its target: crizotinib has five major rotatable bonds, while lorlatinib’s ring structure reduces the effective number to zero, contributing to greater potency. The agent gathered compounds from the literature and connected them to reported data, with an interactive viewer allowing scientists to inspect individual records and explore relationships between molecular properties and potency. Antigravity’s Cheminformatics Skills coordinate existing resources, including RDKit for molecular calculations and the ChEMBL and PubChem databases, while the language model directs the workflow and chemistry tools execute the calculations. The agent compared closely related molecules to examine the effect of structural changes and built predictive models to inform subsequent proposals, and scientists could review the outputs before deciding which designs deserved further work. The demonstration comes as Alphabet applies its models to drug research through Isomorphic Labs; CEO Sundar Pichai said during the company’s latest earnings call that “Isomorphic is using Argon to support drug design research.”