Isomorphic Labs detailed new capabilities for its Isomorphic Labs Drug Design Engine (IsoDDE), positioning the system as a step beyond AlphaFold 3 for real drug discovery tasks. In technical material, the company says IsoDDE more than doubles AlphaFold 3’s accuracy on a protein–ligand structure prediction benchmark, while also improving predictions of small-molecule binding affinities and identifying novel binding pockets using only amino-acid sequence input. The work is framed around generalization beyond training sets—one of the core gaps between protein structure modeling and downstream medicinal chemistry performance. IsoDDE is described as a scalable foundation intended to coordinate predictions across multiple biochemical properties at reduced compute and cost compared with physics-based methods. For the biotech industry, the practical headline is that companies are moving from structure prediction demonstrations to multi-task property prediction and target-pocket discovery workflows that could shorten lead optimization cycles.