In the classical Vehicle Routing Problem (VRP), it is assumed that each worker moves using an individually assigned vehicle. Removing this core hypothesis opens the door for a brand new set of solutions, where workers are seen as transportable resources that can also move without the help of a vehicle. In this con- text, motivated by a major European energy provider, we consider a situation where workers can either walk or drive to reach a job and where carpooling is enabled. In order to quantify the potential benefits offered by this new framework, a dedicated Variable Neighborhood Search is proposed to efficiently tackle the underlying synchronization and precedence constraints that arise in this extension of the VRP. Con- sidering a set of instances in an urban context, extensive computational experiments show that, despite conservative scenarios favoring car mobility, significant savings are achieved when compared to the solu- tions currently obtained by the involved company. This innovative formulation allows managers to reduce the size of the vehicle fleet while keeping the number of workers stable and, surprisingly, decreasing the overall driving distance simultaneously.