An introduction to hyper‑personalised blended learning (HPBL)

Delivered at Bett Asia 2026, this session was presented by Dr David NV, Associate Professor at Universiti Teknologi MARA (UiTM) in Malaysia. In his address, Dr David introduced Hyper-Personalised Blended Learning (HPBL)—an innovative pedagogical framework designed to re-engineer traditional assessment feedback loops and align Higher Education course outcomes directly with real-world industry skills.

Key Takeaways
What is Hyper-Personalised Blended Learning (HPBL)? HPBL is a pedagogical framework that integrates domain-tuned artificial intelligence into blended learning environments. It shifts generative AI from a passive answer generator into an active, iterative practice scaffold and reflective thinking partner.

Re-Engineering the Feedback Loop in Higher Education
In traditional Higher Education, technical and skills-oriented instruction follows a rigid, linear trajectory: accredited curriculum delivery, periodic summative assessment, formal grading, and eventual feedback. However, this model suffers from a structural flaw—delayed feedback. By the time students receive commentary on their submitted work, the academic timetable has already moved on to subsequent topics or semesters. Consequently, feedback rarely contributes productively to ongoing concept mastery.

Real learning occurs in the formative space between formal assessments, where students engage in self-directed study and practical application. Yet, this formative space is frequently constrained by intrinsic psychological barriers. Many learners experience significant inhibition—the fear that asking basic or conceptual questions will expose weakness or lack of preparation in front of peers and academic staff. Learners frequently encounter the dilemma of not knowing what they do not know.

Presented at Bett Asia 2026, Hyper-Personalised Blended Learning (HPBL) restructures this dynamic. By introducing a fine-tuned, subject-grounded AI companion, HPBL creates a safe learning environment where students can ask foundational questions without judgment, receive real-time formative feedback, reflect, and immediately re-attempt concepts until confidence and competence are established.

The OmniSkills.ai Platform Architecture and Ecosystem
Rolled out at Universiti Teknologi MARA (UiTM), the HPBL ecosystem is operationalised through the OmniSkills.ai platform. Designed with pedagogical intentionality, the platform moves generative AI away from simple content retrieval toward continuous micro-assessment and skill development across five core layers:

Onboarding and Anxiety Reduction: Learners access the platform via Single Sign-On (SSO), undergo automated diagnostic profiling, and interact with structured preset prompts that eliminate initial generative AI hesitation and establish immediate baseline readiness.

The Living Skills Model: The platform maps academic course outcomes alongside Washington Accord Program Outcomes directly to real-world job roles and industry competencies tailored to Malaysia's education and economic sectors.

Domain-Tuned AI Companion (AMA): Acting as a tailored academic guide, AMA adapts its interaction style based on discipline demands. In quantitative technical subjects such as mechanical vibrations, it functions as a step-checker for mathematical derivations. In qualitative subjects such as engineering ethics, it serves as a reflective thinking partner to cultivate critical dialogue.

Automated Formative Scaffolding: Through integrated Self-Check Quizzes (SCQ) and routine classroom exercises, AMA automatically evaluates draft derivations, provides real-time rubric-aligned justifications, and generates personalised mock examination papers for targeted self-remediation.

Learning Analytics and Educator Dashboards: Moving beyond surface clickstream data, analytics track cognitive progression using the Engagement Quality Index (EQI) and real-time persona shifts, offering educators a bird's-eye view that connects individual mastery to employer skill demands.

Tracking Cognitive Growth and Prompt Maturity
To evaluate the impact of HPBL, UiTM conducted longitudinal tracking across eight-week modules in two contrasting undergraduate mechanical engineering courses: Vibrations (a quantitative, highly mathematical subject involving 75 students) and Engineers in Society (a qualitative, reading-intensive subject involving 53 students).

Rather than relying on basic usage metrics, backend analytics evaluated deeper cognitive transformation through specialised course-level and student-level indicators:

Prompt Complexity Classification (PCC): Evaluates the sophistication and structure of student inputs over time.

Cognitive Demand Index (CDI): Measures the depth of thinking required by analysing prompt intent.
Course Competency Alignment Index (CCAI): Tracks how closely prompt interactions align with curriculum goals and industry skills.

Engagement Quality Index (EQI): Combines multiple metrics into a composite score of meaningful platform interaction.

Persona Classification and Progression: Tracks individual learner evolution from basic clarification seeking to self-regulated analysis.

Empirical Shift in Student Interactions
The backend data revealed a clear cognitive evolution across both disciplines over the eight-week period:
Quantitative Case Study ( Vibrations ): In week one, students asked elementary recall questions, such as asking for the definition of vibration transmissibility. By week eight, students formulated complex, multi-variable prompts requesting parameter variation analysis, transmissibility curve shifts, and physical explanations near resonance.

Qualitative Case Study ( Engineers in Society ): In week one, prompts focused on basic definitions, such as distinguishing engineering ethics from professional responsibility. By week eight, students submitted detailed ethical scenarios involving team reporting delays, asking the AI companion to evaluate relative professional responsibility based on specific professional codes of conduct.

Across both academic subjects, students consistently progressed from lower-order recall toward application, analysis, and evaluation on Bloom's taxonomy, becoming more independent and self-regulated learners.

Quantifiable Student Outcomes Across 500+ Undergraduates
The integration of HPBL provided measurable proof of improved academic mastery across a broader cohort of more than 500 engineering undergraduates at UiTM:

Cohort 1 Performance: HPBL-assisted students achieved a 13.7 per cent higher median score in complex evaluation tasks compared to non-HPBL cohorts.

Cohort 2 Performance: HPBL participants outperformed their peers by a massive 19.1 percentage points in Course Outcome 3 (CO3) and secured the top three academic positions across the entire engineering faculty.

Failure Rate Reduction: In historically challenging technical modules, institutional implementation of HPBL contributed to reducing overall course failure rates from a historical 78 per cent down to 24 per cent.
Strategic Implications for Higher Education and TVET

The findings presented at Bett Asia 2026 demonstrate that generative AI in Higher Education does not replace teaching staff; rather, it expands educator capacity. By automating routine formative feedback and providing scalable, non-judgmental practice environments, institutions can offer individualised support that was previously unachievable at scale. Furthermore, by integrating Educator Dashboards that map real-time student competency progression against open-market job requirements, HPBL transforms standard course grades into transparent, future-ready career pathways for Higher Education and TVET institutions alike.

Audio and written summaries powered by Gemini Notebook, a free generative AI tool grounded in your provided sources. Try Gemini Notebook for free. You can listen to the audio overview of the session here.