The development of predictive methods in structural bioinformatics enabling the control of sequence-structure-function relationships has driven the field of computational protein design into an era of unprecedented development that is playing an increasingly prominent role in medical and industrial biotechnology. In the last years, we have developed advanced computational protein design technologies based on a unique combination of automated reasoning, machine/deep learning and robotics-inspired algorithms, together with molecular modeling approaches. The original automated reasoning capacities combine accuracy, computational efficiency while offering the capacity to integrate together physics-based models, design requirements and information extracted from data (on protein sequence, structure and function) by machine learning. This enables the resolution of challenging design problems, with higher efficiency and an increased capacity in targeting hard to formalize design objectives that can be indirectly machine learned. These new computational protein design methods have already been successfully applied in industrial and academic projects, for designing enzymes with improved or new properties/activities for bioprocesses, self-assembling symmetrical proteins of nanotechnological interest and novel nanobody scaffolds satisfying specific constraints for diagnosis. They offer solutions to accelerate and reduce costs of developing new and optimized proteins with applications in health and bio(nano)technology.