Purpose In preclinical medical education, students may appear to be coping until summative assessment reveals deeper problems with integration, study strategy, or participation. In HyFlex teaching, however, earlier signals often emerge through attendance analytics, engagement metrics, and participation patterns. This work-in-progress describes the development of a proactive learner-support framework that uses those signals to identify students earlier and connect them with timely support before academic difficulty becomes failure. Methods Drawing on prior HyFlex implementation, the framework is aimed to combine attendance and engagement data, live-session participation indicators, asynchronous access patterns, and performance trends to flag emerging concern. Students meeting predefined thresholds are routed to tiered interventions like brief check-in outreach, coaching on study strategy, AI-generated microcases for clinical reasoning practice, etc. The microcases are being designed as curriculum-aligned, reusable, faculty-reviewed activities so that educators remain responsible for facilitation and feedback while workload is managed through reusable prompts. Results As a work in progress, the current outputs are a signal-to-support map, a triage workflow, and a pilot implementation plan for this term. Evaluation will examine learner engagement with outreach and microcases, median time-to-intervention, student perceptions of support, and performance trends. The intended contribution is a practical model that converts HyFlex participation data into earlier support actions. Conclusion Reframing HyFlex as both a teaching model and an early-support infrastructure aligns learner analytics, feedback, mentorship, and second chances. Paired with carefully governed AI microcases, this approach aims to provide a scalable pathway for recovery and growth in preclinical medical education while helping programmes respond to the realities of an AI-shaped educational environment.