The Research Group on Knowledge Acquisition and Representation through Expert Systems and Simulation “ARCOSES” has developed a methodology for task scheduling in the service sector, specifically targeting laboratories that perform technical testing in highly variable market environments. This methodology is grounded in identifying the most critical factors influencing service demand, as well as the processes involved in conducting the tests. In addition to employing artificial intelligence techniques, it integrates the analyst’s domain expertise, enabling a deeper understanding and validation of the classification outcomes generated by machine learning. The scheduling rules are evaluated under three distinct scenarios: trend-based, pessimistic, and optimistic. In all cases, the proposed scheduling approach outperforms the current FIFO rule, improving compliance and customer satisfaction levels by an average of 4.2%, which translates to approximately 25 additional clients served on time. This performance also surpasses the theoretical SPT rule, which only achieves an average improvement of 1.9%.