Combinatorial Problems have been a classic problem for the man. The objective of global optimality versus cost that this implies is an unresolved issue by researchers. In this sense, problems such as Travel Salesman Problem (TSP) or Vehicle Routing Problem (VRP) relate to combinatorial optimization problems arising everyday for a group of people would like to have the solution to such problems as close as possible. Thus, the idea that such problems purely operational, would be solved by heuristics in devices for normal use for sellers or drivers of the car's makes sense. This paper proposes an architecture that supports the resolution of problems in complex combinatorial mobile devices across heuristics or metaheuristics embedded in the devices themselves and working from a set of data from a central server.