This study examines how a large-scale distance learning institution in the UK designed, implemented, and evaluated an artificial intelligence digital assistant (AIDA) using Sharples’ (2025) embedded systems approach. Through six iterative Design-Based Research studies over 18 months, involving 498 students and 20 staff, we explored three questions: (1) how perceptions of AIDA change with experience; (2) its impact on engagement and learning; and (3) how GenAI can be responsibly deployed at a system level. Early Studies 1–3 gathered student preferences for AIDA’s design and identified benefits of 24/7, context-specific support alongside concerns over ethics, academic integrity, and data privacy. Later Studies 4–6 explored actual hands-on use by student and staff. In a first explorative randomised controlled trial (Study 6), 115 students were able to use AIDA in an OpenLearn Create course, whereby they spent two times longer and visited more pages relative to the 48 students in the control group. Acceptance increased post-use, especially among initially sceptical participants, with 96% expressing interest in AIDA for formal studies. Mapping the six studies against Sharples’ embedded systems approach highlighted enabling factors such as senior leadership sponsorship, cross-unit collaboration, iterative refinement, and data-informed decision-making, while also revealing gaps in institutionalising systems thinking. The findings indicate that purpose-built GenAI tools embedded within the learning environment can enhance engagement and learning outcomes in distance education, provided development is participatory, governance is robust, and integration aligns with institutional strategy. This research offers practical guidance for higher education institutions seeking to adopt GenAI ethically, transparently, and at scale.
Rienties et al. (Fri,) studied this question.