In the context of variant production, the increasing volatility and customer requirements challenge the profitability of manufacturers. A promising approach to mitigate these challenges could be a dynamic pricing. An intelligent design of a continuous delivery-time-price function allows customers to choose based on their preferences and demand may be shifted to level any peaks. This way, profit, service level, and capacity usage could be improved. This work develops a dynamic pricing model based on reinforcement learning applied to a use case of the automation industry. The results show that the dynamic pricing model performs better than current methods in practice.