Angstrom AI

Angstrom AI (YC S24) builds GenAI-based molecular simulations to substitute wet lab experiments in the drug development pipeline. Angstrom AI’s models learn directly from the equations of physics, meaning they do not require training data, bypassing the main limiting factor in machine learning for bio. This ensures quantum mechanical accuracy of the results and avoids the hallucinations seen in other GenAI technologies (e.g., generating an image of a person with six fingers). Simulating physics allows Angstrom’s models to perform experiments on a computer and get the same results as the wet lab. This can be used to evaluate quantities like drug solubility— which is important for predicting bioavailability— and drug-protein binding affinity. Laurence Midgley, Javier Antorán, Jose Miguel Hernandez Lobato, and Gabor Csanyi met at the University of Cambridge, where they researched Machine Learning for Computational Chemistry. Building on a strong academic track record, with over 40k combined citations and over 30 years of combined experience in the field, they've developed the first physically accurate AI-based simulation of multiple molecules interacting. Learn more at https://lnkd.in/gM5rnPG4.

Angstrom AI (YC S24) builds GenAI-based molecular simulations to substitute wet lab experiments in the drug development pipeline. Angstrom AI’s models learn directly from the equations of physics, meaning they do not require training data, bypassing the main limiting factor in machine learning for bio. This ensures quantum mechanical accuracy of the results and avoids the hallucinations seen in other GenAI technologies (e.g., generating an image of a person with six fingers). Simulating physics allows Angstrom’s models to perform experiments on a computer and get the same results as the wet lab. This can be used to evaluate quantities like drug solubility— which is important for predicting bioavailability— and drug-protein binding affinity. Laurence Midgley, Javier Antorán, Jose Miguel Hernandez Lobato, and Gabor Csanyi met at the University of Cambridge, where they researched Machine Learning for Computational Chemistry. Building on a strong academic track record, with over 40k combined citations and over 30 years of combined experience in the field, they've developed the first physically accurate AI-based simulation of multiple molecules interacting. Learn more at https://lnkd.in/gM5rnPG4.