Computer-aided molecular and process design (CAMPD) tries to find the best molecules together with their optimal process. If the optimization problem considers two or more components as degrees of freedom, the resulting mixture design is challenging for optimization. The quality of the solution strongly depends on the accuracy of the thermodynamic model used to predict the thermophysical properties required to determine the objective function and process constraints. Today, most molecular design methods employ thermodynamic models based on group counts, resulting in a loss of structural information of the molecule during the optimization. Here, we unlock CAMPD based on property prediction methods beyond first-order group-contribution methods by using molecule superstructures, a graph-based molecular representation of chemical families that preserves the full adjacency graph. Disjunctive programming is applied to optimize molecules from different chemical families simultaneously. The description of mixtures is enhanced with a recent parametrization of binary group/group interaction parameters. The design method is applied to determine the optimal working fluid mixture for an Organic Rankine cycle.