We develop a macroeconomic theory of the division of household tasks between servants and own work and how it is affected by automation in households and firms. We calibrate the model for the U.S. and apply it to explain the historical development of household time use and the distribution of household tasks from 1900 to 2020. The economy is populated by high-skilled and low-skilled households and household tasks are performed by own work, machines, or servants. For the period 1900-1960, innovations in household automation motivate the decline of the servant economy and the creation of new household tasks motivates an almost constant division of household time between wage work and domestic work. For the period 1960-2020, innovations in firm automation and the implied increase of the skill premium explain the return of the servant economy. We show the robustness of results to the introduction of time trends in skilled-labor supply and the consideration of endogenous demand for leisure. With counterfactual experiments we address the effects of task-dependent disutility of work and of innovations in servant efficiency (the Gig economy). We provide supporting evidence for inequality as a driver of the return of the servant economy in a regional panel of U.S. metropolitan statistical areas for the period 2005-2020.