Increases in market turbulence, a broad spectrum of customer requirements, and the complexity and dynamics of highly saturated markets pose significant challenges to production systems and their production planning and control (PPC). In addition to coping with an increased variety of products, these systems must enable high throughputs and offer a high degree of adaptability against process fluctuations and unforeseen events. To address these challenges, reinforcement learning (RL) has emerged as a promising approach for optimizing PPC. In contrast to other machine learning methods, RL interacts directly with its environment, enabling real-time responses to system changes and direct interventions in the PPC. This work's main subject is the realization of an RL agent focusing on production control and a deep dive into order release. Specifically, an RL agent, which uses Q-learning as a widely used RL algorithm, is trained to automate the task of order release as a central PPC task. From a production logistics point of view, the agent integrates the load- and schedule-oriented perspectives in its decision-making processes. The implementation of the RL agent is anchored in the process chain of the IFA Learning Factory. The findings indicate that the RL agent can reliably automate the task of order release following a discernible learning process. Concurrently, the RL agent can adjust the workload of individual production systems while meeting stipulated deadlines.