Governments worldwide are embracing the evolution of digital economies and automation for efficiency gains to overcome scarce resourcing. However, achieving efficiencies without burdening the government or community they serve is not trivial. We explore a case study following the creation of an automated system to recover debts from welfare recipients who supposedly misrepresented their income, colloquially known as ‘Robodebt’. Business process management was extensively used in the project’s design and implementation; however, the owners and agents struggled to achieve their desired outcomes. On the flip side, the recipients of the automated system faced erroneous debts and a faceless system that ignored their existence. The system would ultimately fail; ruled illegal due to its process assumptions and, never came close to the goal of $4.772 billion in savings. Our analysis synthesises the project’s design choices, deriving how the process mindset failed the owners and agents. We posit that public process documentation is required to prevent similar failures in the future.