Isomorphic Labs previewed a new Isomorphic Labs Drug Design Engine (IsoDDE) that aims to move past AlphaFold-style structure prediction toward more actionable small-molecule design. The company says IsoDDE doubles AlphaFold 3 accuracy on a protein–ligand structure generalization benchmark and predicts binding affinities at accuracies that it claims exceed physics-based methods. IsoDDE is positioned as a unified computational drug-design system that can generalize to unseen biological targets, including identifying novel binding pockets from amino-acid sequence input. The preview references Isomorphic Labs’ 2024 AlphaFold 3 release with Google DeepMind as the starting point for this next step. For biotech teams, the key development is the explicit shift from in-silico biomolecular modeling toward multi-property predictive performance needed for real-world medicinal chemistry workflows. If validated in broader benchmarks and prospective programs, IsoDDE could shorten lead-optimization cycles. The announcement also highlights scalability and reduced compute/cost claims versus traditional physics-based approaches, targeting one of the bottlenecks that typically limits how far computational screening can go in early discovery.