LTL (Less than Truckload) shipping has an efficiency problem, due to inflexible schedules and the inability to address short-term changes in demand. We propose a technical solution for improving the efficiency of LTL shipping by automating aspects of operational decision-making using rich, real-time information. In particular, we propose a framework for collecting heterogeneous data from Internet of Things (IoT) sensors and data services in the context of logistics networks to maintain a rich state of the environment as a basis for routing and scheduling decisions. We use this information for optimizing freight streams by applying standard Operations Research methods to the problem. In particular, we introduce a methodology for computing dynamic vehicle schedules based on dynamic demand in a level of detail that allows schedules to be planned and adapted in an autonomous decision process in day-to-day operations. To that end, we formally define Dynamic Transshipment with Time Constraints in a Finite Planning Horizon and Vehicle Allocation, two mathematical optimization problems that build a schedule based on dynamic demand, considering constraints on delivery times, capacities of vehicles, loading terminals and hubs. Finally, we demonstrate the performance of our approach with evaluations based on operational data from an Austrian logistics company. In these evaluations, we are able to reduce the total amount of driven kilometers by 15.2% and improve vehicle utilization by 2.8 percentage points. By introducing and testing a mechanism for incident handling, we also demonstrate the robustness of our framework against unforeseen delays, such as vehicle breakdowns or traffic congestions.