Automation has sparked dreams of overcoming human boundaries since at least the 1950s, and recent decades have seen an upswing in imagining a bright technological future in which artificial intelligence (AI) will solve social, economic, and political challenges and improve public administration and welfare provision. However, the introduction of algorithmic automation can cause friction. On the one hand, enthusiasm for algorithmic automation in the industrial and public sectors is increasing, and widespread digitalization and automation projects are revolutionizing public administration and welfare. On the other hand, public discourse is increasingly painting a dystopian picture of the digitally automated welfare state, critiquing biases and raising issues of accountability and autonomy. In this article, I take the frictions, conflicts, and contradictions that emerge from technological change as a starting point for considering the temporalities of algorithmic automation, particularly focusing on the Swedish context. Considering three temporal layers of welfare automation-timing, belatedness, and perpetual emergence-I outline the contours of the contemporary Swedish digital welfare state and discuss the critical implications of delegating decisions to algorithmic systems and their mediating features. Drawing on a diverse set of empirical data drawn from in-depth interviews with civil servants and software developers and observations of Swedish public agencies, I engage with the unfulfilled promises of the fully automated welfare state.