The success of crowdsourcing systems such as Wikipedia relies on people participating in these systems. However, in this research we reveal to what extent human and machine intelligence is combined to carry out semi-automatic workflows of complex tasks. In Wikipedia, bots are used to realize such combination of human-machine intelligence. We provide an extensive overview on various edit types bots carry out in this regard through the analysis of 1,639 approved task requests. We classify existing tasks by an action-object-pair structure and reveal existing differences in their probability of occurrence depending on the investigated work context. In the context of community services, bots mainly create reports, whereas in the area of guidelines or policies bots are mostly responsible for adding templates to pages. Moreover, the analysis of existing bot tasks revealed insights that suggest general reasons, why Wikipedia’s editor community uses bots as well as approaches, how they organize machine tasks to provide a sustainable service. We conclude by discussing how these insights can prepare the foundation for further research.