CramPilot

Inspiration Every student knows the dread: exams are approaching, the syllabus feels infinite, and panic sets in. Traditional study planners fail because they:

Treat all topics equally (they're not) Ignore human psychology (we procrastinate, we stress) Don't adapt when life happens We built CramPilot because we've been there — staring at textbooks at 2 AM, wondering what to prioritize. We wanted an AI that doesn't just schedule, but strategically rescues your exam prep.

What it does CramPilot is an AI-powered exam preparation system that:

🎯 Prioritizes Ruthlessly — Identifies high-yield topics that maximize marks per hour

✂️ Tells You What to Skip — Yes, some things aren't worth your time. We tell you which.

🧠 Manages Cognitive Load — Alternates difficulty levels to prevent burnout

📊 Adapts to Stress — High stress? The AI automatically reduces your workload

🔄 Re-Plans Daily — Missed tasks? It doesn't just push them to tomorrow — it decides if they should be dropped or condensed

How we built it Tech Stack:

Frontend: React 19 + TypeScript + Vite AI Engine: Google Gemini AI (gemini-3-flash-preview) Visualization: Recharts for progress tracking UI Components: Lucide React icons Deployment: Vercel AI Architecture: We engineered a specialized system prompt that makes Gemini think like a crisis manager, not a scheduler:

Time Scarcity Analysis — Calculates remaining days, identifies impossible coverage zones Cognitive Load Management — Avoids consecutive high-effort tasks Weakness-Priority Logic — Focuses on low-confidence + high-impact topics Stress-Adaptive Planning — Automatically adjusts based on reported stress Human Behavior Assumptions — Accounts for procrastination and anxiety Challenges we ran into Making AI "human-aware" — Getting Gemini to understand that students procrastinate, get anxious, and don't follow perfect schedules required extensive prompt engineering

Balancing structure vs. flexibility — Too rigid plans fail when life happens; too flexible plans provide no direction. Finding the middle ground was crucial

Stress detection without being intrusive — We wanted stress-aware adaptation without making users feel monitored

Schema design for structured output — Ensuring Gemini returns perfectly structured JSON for the schedule while maintaining creative reasoning

Accomplishments that we're proud of

✅ The "Skip These Topics" feature — No other study app tells you what NOT to study. This paradigm shift alone can save hours of wasted effort

✅ Stress-responsive AI — When you report high stress, the AI genuinely reduces load instead of adding pressure

✅ Daily adaptation that actually works — Missed tasks don't just pile up; they get triaged

✅ Clean, calming UI — Built with a dark theme that reduces eye strain during late-night study sessions

What we learned

AI prompting is an art — The difference between a generic scheduler and a strategic rescue agent was entirely in the system prompt Students need permission to skip — Psychologically, being told "it's okay to skip this" reduces anxiety significantly Adaptation > optimization — A plan that adjusts to reality beats a "perfect" plan that breaks on day 2

What's next for CramPilot

🚀 Mobile App — PWA or native app for on-the-go check-ins

📚 Syllabus Parsing — Upload PDF/image of syllabus for automatic topic extraction

🤝 Study Groups — Sync plans with friends for accountability

📈 Analytics Dashboard — Long-term insights into study patterns and stress trends

🧪 Practice Test Integration — Auto-generate practice questions for weak topics

Built With

• ai
• gemini
• google
• lucide
• react
• recharts
• typescript
• vercel
• vite