Cube provides data scientists with tools to monitor machine-learning models and predict model failures before they occur. The company is positioning itself as a pioneer in AI safety as a service at a time when AI safety is beginning to gain increased attention in the research community. Cube’s platform packages a range of AI-safety algorithms into a single line of code, which can be implemented into any ML model to ensure it works as expected. These safety algorithms currently focus on guaranteeing the reliability of ML models, but the company is also working towards including algorithms that ensure transparency and security of ML models.