Plant breeding has contributed to the sustained increase crop production and stability in the last century. High population growth challenges achieving food security with current rates of genetic improvement and agronomic practices. Thus the effectiveness of breeding strategies to increase yields in crops must be improved.  Such improvement may come about if some of the complexities in gene-to-phenotype (G→P) relations could be captured in realistic and applicable G→P models. This presentation considers G→P models to propose a synthesis structured within the E(NK) model of trait genetic architecture and focused on breeding applications. A framework that integrates biophysical, physiological and statistical genetics components is demonstrated within an operational breeding program. The behavior of the system is presented through theoretical examples. Then, the application of the framework is demonstrated at a particular stage of a maize breeding program with the objective to improve drought tolerance of maize hybrids for the US Western Corn-Belt. The application of G→P modeling within a breeding program suggests that the improvement of breeding strategies and G→P models is feasible but it will require an iterative learning as we go process, in which all physiology, genetics and phenomics of adaptive traits should be advanced.