This thesis investigates the dynamic interplay between humans and automation in Intelligent Process Automation (IPA) systems. The research should be focused on defining the respective roles of humans and machines across different stages—ranging from direct action and decision-making to supervision and performance monitoring. The aim is to investigate the optimal balance between human intervention and automated processes in enhancing efficiency and accuracy. Particular attention should be given to activities classification based on their level of human involvement, as well as to frameworks for evaluating overall system performance. A key dimension of the study concerns error management: analyzing and comparing the cost of human versus machine mistakes and their impact on process outcomes. The work aims to generate insights for designing more effective, balanced collaboration models that optimize both efficiency and reliability within automated business environments.