A Dynamically Configurable Autonomous Vehicle (DCAV) is a new class of autonomous vehicle concept using a separable design of lower and upper parts—carriers and modules—to allow more flexible operation. A fleet of DCAVs consists of a set of carriers and a set of compatible modules. Different, possibly crowd-sourced, modules can increase the number of use-cases for DCAVs, possibly leading to disruptive changes in the transport sector. This study investigates the use of DCAV system operating on an Autonomous Mobility-on-Demand (AMoD) scenario, combining passenger and freight transport flows. The novel problem is denoted as the Dynamically Configurable Autonomous Vehicle Pickup and Delivery Problem (DCAVPDP). We propose a mixed-integer linear programming (MILP) model aiming to minimize DCAV-fleet size and distance traveled. We compare the performance of a DCAV fleet to the performance of a typical single-purpose fleet (consisting of dedicated passenger and freight vehicles). The numerical study, with 360 instances for each fleet type, considering four people-and-freight demand distribution scenarios, the inclusion of ridesharing, module-and-carrier (de)coupling locations, and different simulation horizon lengths, shows that the proposed modular DCAV system can fulfill a mixed people-and-freight demand using, on average, 18.77% fewer carriers than a regular AMoD system comprised of single-purpose vehicles while increasing on-duty fleet utilization by 4.82%.