NudgeLearn AI: Smart Revision & Retention Coach

• 

  title page

• 

  project structure

Inspiration

Students often spend hours reading notes, PDFs, and study materials but still forget important concepts within a day or two. We were inspired by the real problem of poor knowledge retention and unstructured revision habits.

Traditional studying is passive—students reread notes but rarely know:

what they are weak at what to revise next when to revisit concepts

What it does

NudgeLearn AI is an AI-powered smart revision platform that helps students retain knowledge through personalized learning loops.

Users can:

📄 upload study notes 🧠 get instant AI-generated summaries 🎯 identify weak concepts 📝 receive concept-based quizzes 🔔 get smart revision nudges 📚 store previous learning history ♻️ revisit older sessions anytime

The platform converts static notes into:

summary → quiz → nudge → history → revision loop

How we built it

We built NudgeLearn as a full-stack AI web application.

🐍 Backend FastAPI (Python) Gemini API + local fallback AI quiz generation service smart nudge engine REST API architecture 🌐 Frontend HTML CSS Bootstrap JavaScript fetch API localStorage-based history 🧠 AI workflow

We integrated:

Gemini API for note analysis local heuristic fallback summarizer concept-aware quiz generation smart nudges based on weak areas

This architecture allowed us to keep:

frontend backend AI services history layer

cleanly separated.

Challenges we ran into

We faced several real engineering challenges:

API authentication issues

Initially, Google Cloud CLI and Vertex authentication caused setup issues on Windows.

quota limits

Gemini free-tier limits quickly caused 429 quota exceeded errors.

frontend-backend fetch issues

CORS, local HTML serving, and API routing mismatches caused “Failed to fetch” errors.

project structure bugs

Backend service imports and folder separation caused module loading errors.

adaptive quiz design

Making quiz questions feel concept-specific instead of generic required redesigning the quiz service logic.

Each challenge improved the final architecture significantly.

Accomplishments that we're proud of

✅ full-stack working MVP ✅ separate frontend + backend architecture ✅ real AI note analysis ✅ concept-aware quizzes ✅ smart revision nudges ✅ study history persistence ✅ Gemini fallback for quota-safe demos ✅ portfolio-ready SaaS-style UI

What we learned

how to integrate AI into real workflows beyond chat the importance of fallback systems for reliability why product stickiness comes from retention loops how frontend/backend separation improves maintainability how adaptive UX dramatically improves perceived intelligence how to design for hackathon speed without sacrificing scalability

What's next for NudgeLearn AI: Smart Revision & Retention Coach

We see strong startup potential for NudgeLearn.

Next steps include:

📄 PDF and handwritten note OCR 🎤 voice tutor mode 📈 progress analytics dashboard 🔥 streak tracking and gamification 👥 collaborative study groups ☁️ cloud-based history sync 📱 mobile responsive PWA 🎯 exam-specific revision modes (GATE/JEE/placements) 🤖 personalized AI mentor agents

Built With

• 5
• bootstrap
• css
• fastapi
• fetch
• googlrgeminiapi
• html
• javascript
• python
• uvicorn
• vanilla