Automated decision-making (ADM) tends to be discussed as a process of delegating and automating decisions from people to machines in the context of automation. This chapter shows instead how ADM can function as a mediating algorithmic logic that manages and generates new relationships between actors, and economies of value that may not pre-exist or are in the process of being automated. We develop this argument drawing on empirical work from an experiment with ADM in Sweden that sought to optimise and steer heat provision in everyday life through ‘smart’ thermostats. Our approach is informed by science and technology studies and perspectives on socio-technical experiments, considering them as generative of new environments and social relationships. We show how experiments with ADM where algorithms take decisions about the steering of thermal provision in everyday life can redefine the understanding of control, and power relationships between energy companies, data-driven economies, and people brought together through a common concern about temperature.