Make undruggable targets a thing ofthe past


The current pharmaceutical research and development model is broken and unsustainable.



With the current pharmaceutical R&D model, it takes about 12 to 15 years to bring a new drug to market.



The estimated cost of research and development sits at over $2.5 billion per drug approval, increasing at a rate of 8.5% each year.​

success rate


More than 90% of drug candidates that enter Phase 1 clinical trials fail to reach regulatory approval, often due to the limited predictive value of preclinical models of disease.​​



Drug approvals remain stagnant in the current model. Fewer than 70 new drugs enter the market each year. With such a low yield, scalability is nearly impossible.​​​


We must take more shots on goal.

To decrease failure rates and increase R&D productivity, we must take more shots at a variety of disease-causing targets and therapeutic modalities.

Using machine learning, our multi-module drug discovery platforms generate more novel drug candidates, shorten the timeline, cut operational costs and increase the probability of success of pharmaceutical R&D.

One important distinction: we believe AI and computational technologies are impactful in drug discovery only when biology is our North Star. Our platforms combine domain expertise in biology and pharmacology with AI and machine learning to provide an end-to-end solution.

  • SUEDE™ captures compound desirability with validated pharmacophores. With SUEDE™, we screen billions of molecules a million times faster than traditional methods, enabling rapid hit discovery for diverse targets.

  • BAGEL™ designs novel scaffolds from molecular templates using neural networks. To prioritize novelty and bioavailability, BAGEL™ learns pharmacologic features from datasets vetted by experienced medicinal chemists.

  • We own a state-of-the-art drug property prediction platform built on graph neural networks (GNN). GNNs excel at QSAR tasks, consistently outperforming industry-standard techniques in virtual screening and lead optimization.

  • We provide end-to-end computational drug design and high throughput lab automation capabilities. Our wet lab platform generates billions of proprietary biological activity data points, which further enhance our AI models.

  • Built on a proprietary data layer of twelve million three-dimensional hydropathic interaction (HINT) maps, we leverage ROSALYND™ for de novo Protein Structure Prediction, Therapeutic Protein Design, and Protein Structure Refinement.

  • Three-dimensional hydropathic interaction (HINT) maps reflect the favorable and unfavorable substructures of predicted protein folds. HINT goes beyond traditional methods when evaluating the physical forces acting on proteins.

  • Structure-based drug design is most effective with accurate side chains. We own a proprietary side chain prediction platform built on advanced AI models that drive our protein refinement and de novo design pipelines.