The rapid growth of Mobility-on- Demand (MoD) services such as Uber, Lyft and car2go, and more recently ridepooling options such as Via and Bridj, are a clear indication of consumer demand for more flexible and convenient transit services. These services allow users to access a fleet of shared vehicles by simply using a smart-phone to request a vehicle. Thus, MoD services provide the convenience of a private car without the hassles of maintenance, financing, insurance etc. While these services have the potential to improve urban transportation and pave the way for more scalable and sustainable transportation systems, it is not clear that this will in fact be the case in their current form of operating as isolated services.In this work, we wish to develop fundamental theory that may enable MoD systems to integrate and collaborate with masstransit systems, and provide a more scalable and sustainable unified urban mobility service. In particular, we will develop theory that characterizes the fundamental limitations of a system that intends to leverage both flexibility and dynamic high capacity ridepooling in real-time. We will model this service as a variation of the set cover problem, and we will study the theoretical performance limits of such a model. We will propose a competitive algorithm that enables the real-time operations of such MoD system. The intent is to enable the operations of an integrated system that is both more efficient to operate and provides a better user experience.