<p>Automated guided vehicles (AGVs) play a central role in modern manufacturing and logistics systems, where they must execute pickup and delivery tasks efficiently and without conflicts. We study the online dispatching and routing problem for AGVs in loop-based graphs, which integrates the assignment of jobs to vehicles with the determination of conflict-free routes. We propose a loop-based algorithm that handles arbitrary vehicle capacities and jobs with precedence constraints while guaranteeing safe and efficient operation. To evaluate performance, we compare our approach against an exact method, a greedy heuristic, and a metaheuristic. Experiments on both synthetic benchmark instances and a model derived from a real manufacturing plant demonstrate that our algorithm consistently matches or outperforms existing approaches while requiring substantially less computing time.</p>