The Utah-based pharmaceutical company has trained a large language model to enable scientists to tap into dozens of machine learning models at once, saving them time during drug development.