Robotic Process Automation (RPA) recently gained a lot of attention, in both industry and academia. RPA embodies a collection of tools and techniques that allow business owners to automate repetitive manual tasks. The intrinsic value of RPA is beyond dispute, e.g., automation reduces errors and costs and thus allows us to increase the overall business process performance. However, adoption of current-generation RPA tools requires a manual effort w.r.t. identification, elicitation and programming of the to-be-automated tasks. At the same time, several techniques exist that allow us to track the exact behavior of users in the front-end, in great detail. Therefore, in this paper, we present a novel end-to-end approach that allows for completely automated, algorithmic RPA-rule deduction, on the basis of captured user behavior. Furthermore, our proposed approach is accompanied by a publicly available proof-of-concept implementation.