This article focuses the policy discourse that Swedish municipal personal social services (PSS) must engage with when implementing automated decision-support systems; how these tools are conceptualized in the context of social work and what outcomes they are expected to yield in the PSS organizations. Applying an adapted version of Bacchi’s WPR framework, results indicate that the three main policy actors directing the Swedish PSS portray a future where the capacity of the welfare state is threatened, thus suggesting digital automation as an objective and politically neutral tool for saving the PSS from this worrisome prospective. This article, however, argues that by uncritically promoting a particular form of digital automation within the PSS, the policy discourse risks overlooking the characteristics of digital technologies, thus both disregarding its consequences and amplifying the neoliberal ideals that award private enterprise the role of the main supplier of public welfare.

Many European countries’ public organisations have turned to digital automation as a strategy to enhance the efficiency of their service delivery. One example of this is the use of Robotic Process Automation (RPA) in the provision of Social Assistance (SA) in Sweden. Swedish SA is a means-tested subsidy intended as a last resort for the economically most vulnerable persons, and caseworkers exercise a high degree of discretion when deciding on clients’ eligibility for the benefit. Based on semi-structured interviews with stakeholders at the national and local levels, this study explores how these actors conceptualise and rationalise the adoption of RPA in the SA context and discusses how this relates to the mutual configuration of RPA and SA. The results show that RPA is framed as a simple technology to alleviate caseworkers’ administrative workload so that they can instead focus on supporting clients towards self-sufficiency. However, local actors describe RPA adoption as a challenging process with modest outcomes. As such, the study suggests that stakeholders’ simplified conceptualisation of RPA allows them to shift responsibility for its practical adoption to local PSS organisations, which ultimately pushes the PSS organisations towards configuring their SA casework processes around the limited capabilities of the technology.

In line with the global trend towards digitalization, digital automation has become a politically endorsed strategy to enhance efficiency and transparency in public service delivery. One such example is the adoption of Robotic Process Automation (RPA) in the context of social assistance (SA) in Sweden. While SA is a municipally organized means-tested subsidy for the most financially vulnerable, RPA functions as highly structured software executing administrative tasks based on predetermined rules. Employing a sociomaterial perspective, this article analyses group interviews conducted in four Swedish municipalities, to explore if and how RPA adoption configures the dynamics of SA casework administration and practice. Findings suggest that RPA adoption does not inherently lead to expected all-encompassing enhancements of faster and fairer eligibility determinations as well as a more client-centric approach in SA casework. Instead, the instrumentalist approach of leveraging digital tools to achieve specific outcomes, combined with the task delineation required by RPA, appears incongruent with the nature of SA casework practice. As caseworkers attest to organizational adjustments and a more instrumental approach towards clients’ situations alongside RPA adoption, this study underscores the inadequacy of the rationalist notion of segmenting SA casework in order to achieve efficiency in casework that, by law, should be based on individual judgements by professional caseworkers.

Many European countries employ Robotic Process Automation (RPA) in the administration of public benefits. However, there is limited understanding of how RPA is applied at the client level. This article investigates the utilization and impact of RPA use on social assistance (SA) distribution in Sweden, drawing on a sample of 800 SA applications in four Swedish municipalities. The results show that RPA use correlates with applicants' country of birth, age and duration of SA receipt. Additionally, RPA implementation coincides with less generous decisions, disproportionately affecting financially vulnerable groups. Rather than a correlation between generosity and the technology itself, the results suggest a conflict between the reorganisation of SA administration during RPA implementation and the principle of individualized judgments inherent in SA casework. Hence, public organisations are encouraged to ensure that their adoption of RPA neither exacerbates unequal access to services nor compromises professional discretion in favour of efficiency-driven measures.