An AI platform for small-molecule drug design built around two core models: NeuralPLexer, a co-folding model for predicting protein and protein-ligand structures, and Enchant, a multimodal transformer for predicting molecular properties. The models are integrated with automated chemical synthesis and experimental execution so novel designs are turned into biological data each week.

• ·NeuralPLexer co-folding model for protein and protein-ligand structure prediction
• ·Enchant multimodal transformer for molecular property prediction
• ·Physics principles embedded in the AI architecture for data efficiency
• ·Automated chemical synthesis integrated with high-throughput experimentation