|1","ClassId":1073872969,"Properties":[469775450,"Normal1",201340122,"2",134233614,"true",469778129,"Normal1",335572020,"1",469777841,"Arial",469777842,"Arial",469777843,"Arial",469777844,"Arial",469769226,"Arial",335551500,"0",268442635,"22"]}"--}|1","ClassId":1073872969,"Properties":[469775450,"Normal1",201340122,"2",134233614,"true",469778129,"Normal1",335572020,"1",469777841,"Arial",469777842,"Arial",469777843,"Arial",469777844,"Arial",469769226,"Arial",335551500,"0",268442635,"22"]}">Remote laboratory environments are increasingly central to distance and online STEM education, yet students often lack access to immediate, situated tutor support during live experimental activity. This paper presents OELAssist, an AI-driven adaptive support system designed to provide real-time, context-aware feedback to students working in remote engineering laboratories. OELAssist integrates learning analytics with retrieval-grounded generative artificial intelligence to identify interaction patterns associated with student difficulty and to deliver timely, personalised guidance aligned with course materials and safety protocols. The system was developed and piloted within the Open University’s Open STEM Labs, focusing on the Pressure Vessel experiment in a core undergraduate engineering module. Using large-scale interaction data and practitioner-informed evaluation, the study examines how AI-mediated feedback can support student self-regulation, improve procedural decision-making, and provide scalable alternatives to synchronous human tutoring. Findings indicate that adaptive, AI-generated feedback can enhance student engagement, promote safe and efficient experimental practice, and offer valuable diagnostic insight for tutors and module teams. The paper contributes a practice-oriented model for integrating AI-driven adaptive support into remote laboratory environments and discusses implications for digital pedagogy, equity, and institutional strategies for scalable student support in distance STEM education.