This article presents a multi-agent based algorithm for personnel scheduling and rescheduling in the dynamic environment of a paced multi-product assembly center. Our purpose is, on the one hand, to elaborate daily assignment of employees to workstations to minimize the operational costs as well as the personnel dissatisfactions and, on the other hand, to generate an alternative planning when the first solution has to be rescheduled due to disturbances related to operators’ absenteeism. The proposed approach considers the individual competencies, mobility and preferences of each employee, as well as the personnel and competency requirements associated with each assembly activity given both the current master assembly schedule and the line balancing for each product. To benchmark the performance of the multi-agent approach, we use solutions obtained through a simulated annealing algorithm. Experimental results show that our multi-agent approach can produce high-quality and efficient solutions in a short computational time.