COMPAS aims at developing novel compact models and ultra-compact digital twins. The compact models capture nonlinear, transient, and coupled (i.e., multi-physics) situations. The digital twins can self-sufficiently cast decisions (ultimately in real-time) for prognostic health management. COMPAS will develop them by using the example of thermo-mechanical reliability of high-tech systems, such as motor control units for automated factories, smart infrastructures (streetlights, power grids), or autonomous vehicles. These high-tech systems integrate numerous highly complex components, which require several landmark innovations in model order reduction, artificial intelligence techniques, and software coding. These innovations will have a substantial impact on the simulation software market for high-tech systems since they enable unprecedented design optimization and thus reduce experimental development cycles. Moreover, they provide advanced health management for fail-safe high-tech systems with less hardware redundancy.