This ANR project CoolMagEvo is focused on the development of an alternative technology for cooling applications: magnetic cooling systems using magnetocaloric alloys. The innovative goal of this project is to explore the relevance of an artificial evolution-based digital approach for the design and the optimization of magnetic cooling systems. These devices are complex systems because they involve a large number of entities in multi-physics interactions which are responsible of emergent behaviors. The novel stake of the project is to associate different fields of research that are usually not combined together with experts of experimental measurements and material sciences towards specialists of multi-physics and multi-scales simulation and finally specialists in artificial intelligence and bio-inspired methods. This project will require first a detailed characterization of the physical properties of several new and more powerful families of alloys with a giant magnetocaloric effect around their 1st order magnetic phase transition. These measurements will be then incorporated into a multi-physics model ensuring the thermodynamic consistency of the various physical parameters in order to calculate the performance of magnetic cooling systems in terms of temperature span, cooling power and energy efficiency. This simulation tool, experimentally validated on prototypes, constitutes the evaluation function that will be implemented into an evolutionary algorithm combined with deep learning artificial intelligence (AI) and associated with massively parallelized computing capabilities. These bio-inspired computation methods will potentially provide optimizations and innovative forms of architecture that the human mind has not considered yet.