SimuExam AI - Adaptive Exam Simulator & Diagnostic Engine

💡 Inspiration

Students spend endless hours summarizing textbooks and highlighting notes, yet they still struggle during real exams. The problem isn't a lack of studying—it's passive studying and an inability to pinpoint why an answer was wrong. We built SimuExam AI to transition students from passive note-reading to active recall and targeted error diagnosis.

🚀 What It Does

SimuExam AI is an intelligent study and assessment ecosystem that turns raw study materials into adaptive, real-time mock exams.

Key features include:

Adaptive Mock Exam Generator: Upload lecture notes, PDFs, or slides to instantly receive structured exams with multiple-choice and conceptual questions.
Smart Error Diagnostics: Instead of a simple percentage score, our AI categorizes mistakes into three core root causes:
1. Concept Gap: Misunderstanding underlying principles.
2. Question Misread: Misinterpreting prompt instructions or conditions.
3. Memory Slip: Forgetting specific details or formulas.
Spaced Repetition Engine: Failed concepts automatically queue into personalized daily review schedules using optimal recall timing intervals.

🛠️ How We Built It

We built SimuExam AI using modern web architecture integrated with high-performance AI engines:

Frontend & UI: React, Tailwind CSS, and Lucide Icons for a clean, dark-mode first, responsive dashboard.
AI Processing: Generative AI models for structured JSON question generation, semantic evaluation, and natural language error classification.
Data & Logic: Base44 app architecture with dynamic exam state management, score tracking, and automated question scheduling.

📐 Mathematical Formulation of Spaced Repetition

To model the forgetting curve and optimize the review interval ( I ), we apply an adapted memory retention function:

$$ R(t) = e^{-\frac{t}{S}} $$

Where:

( R(t) ) is target memory retention at time ( t ).
( S ) is the calculated memory stability based on historical quiz performance:

$$ S_{new} = S_{old} \times (1 + \alpha \cdot \text{Accuracy}) $$

This ensures that hard-to-remember concepts return in shorter time intervals, maximizing study efficiency.

🛑 Challenges We Faced

Prompt Engineering for Structured Outputs: Ensuring the AI reliably outputs strict JSON structures for exam rendering without hallucinating option formats.
Accurate Error Classification: Fine-tuning the reasoning prompt to distinguish between a student genuinely misunderstanding a concept versus simply misreading a question word.

🎓 What We Learned

Active Recall > Re-reading: Building this app reinforced how crucial immediate feedback loops are for retention.
Base44 Full-Stack Speed: Leveraging Base44 allowed us to rapidly iterate from prompt design to a fully functional interactive UI.

🔮 What's Next for SimuExam AI

Voice-based Socratic Tutor: Oral exam preparation using real-time speech-to-text.
Collaborative Study Rooms: Multiplayer test sessions with synchronized timer countdowns.