A Sociotechnical Perspective on Digital Twins and Edge AI : A Case Study of the NexTArc Project Through the Technology-Organisation-Environment Framework

Sammanfattning

Digital twins and Edge AI are increasingly discussed as technologies that can support data-driven decision-making in complex environments. However, the implementation is not only a technical issue for the development but also depending on organisational and environmental factors. This thesis examines how digital twins and Edge AI are developed and implemented in a complex sociotechnical setting through an in-depth case study of the EU-funded NexTArc project, with a specific focus on Use Case 1 and the development of a digital city twin for the Marieberg district in Stockholm. The purpose of the study is to identify technological and organisational success factors, as well as challenges connected to the implementation and integration of digital twins and Edge AI. The analysis is based on the Technology-Organisation-Environment (TOE) framework. It is conducted through a qualitative case study, using semi-structured interviews with ten participants involved in the project. The empirical material was analysed through a thematic analysis. The analysis resulted in six themes, consisting of the digital twin as a sociotechnical system, data as the core enabling resource, Edge AI as an enabler of local data processing, integration as the main implementation challenge, collaboration between partners, and NexTArc as a platform for learning and capability-building. Furthermore, the findings show that successful implementation depends on alignment between technological, organisational, and environmental factors. Technological components such as sensors, data pipelines, APIs,metadata, edge devices, and user interfaces are necessary, but they only create value when they are coordinated across organisational boundaries. The study further shows that key enabling factors include access to relevant data, clear data structures, metadata, standardised interfaces, shared definitions, role clarity, physical meetings, and knowledge exchange. The main challenges contain missing or inaccessible data, unclear definitions of digital twins, technical integration, unclear partner contributions, and coordination complexity in a large multi-partner project. The thesis contributes by conceptualising implementation as an alignment process rather than a linear technical process, and by providing practical implications for future digital twin and Edge AI initiatives. The implications for future research include following NexTArc as the project advances from a proof-of-context phase towards more integrated and operational solutions. It also includes the possibility to investigate how additional stakeholders such as urban planners, policymakers, citizens, and future users understand and use digital twins. Further, it would be interesting to compare NexTArc with similar projects to examine which findings are specific to this project and which are relevant to other digital twins and Edge AI initiatives.