AI Study Game Generator

• 

  Home Page

• 

  Upload Page

• 

  Game Page

📚 About the Project — AI Study Game Generator

💡 Inspiration

As students, we constantly face the same problem: we spend hours reading notes, textbooks, and past exam papers, but still struggle to truly understand and retain the material.

Traditional revision is:

• passive
• repetitive
• often ineffective

What’s missing is active engagement — the ability to test understanding in a structured and meaningful way.

At the same time, creating good practice questions manually is time-consuming and requires skill.

This inspired us to ask:

What if any study material could instantly become an interactive learning experience?

---

🛠️ What We Built

We built an AI-powered Study Game Generator that transforms raw study materials into a progressive learning game.

Users can upload:

• Text files
• PDFs
• Images (future extensions)

The system then:

1. Extracts and processes the content
2. Uses AI to generate exam-level questions
3. Organizes them into difficulty-based levels
4. Creates an interactive game where users:

• answer questions
• receive feedback
• progress through levels

---

⚙️ How We Built It

We used MeDo as a full-stack AI co-developer, iterating through multiple structured prompts to build the system.

🧩 System Architecture

User Upload
   ↓
Text Extraction (PDF / OCR / TXT)
   ↓
AI Question Generation
   ↓
Validation & Structuring
   ↓
Game Engine (Levels + Feedback)

🔧 Key Components

• Frontend

  • Upload interface
  • Game UI with levels, scoring, and feedback

• Backend

  • File processing pipeline
  • AI integration for question generation
  • Database storage

• AI Integration

  • Gemini API for generating:
  • concept-based questions
  • multiple-choice options
  • detailed explanations

• Database & Storage

  • Supabase for:
  • storing uploaded files
  • managing generated questions
  • tracking processing status

---

🧠 How We Used MeDo

MeDo was not just a tool — it acted as a co-developer.

We used multi-turn conversations to:

1. Generate initial full-stack structure
2. Identify weaknesses (e.g. shallow questions)
3. Refine prompts to enforce:

• concept-based questions
• real exam-level difficulty
• explanation-driven learning

We iteratively improved:

• backend logic
• AI prompts
• UI/UX

This process turned a basic quiz generator into a smart learning system.

---

🚧 Challenges We Faced

1. Low-Quality AI Questions

Initially, the system generated shallow questions like:

• “What is the main topic?”

👉 This provided no real learning value.

Solution: We redesigned prompts to enforce:

• concept-based reasoning
• application-level questions
• strict grounding in the material

---

2. Ensuring Reliability

AI outputs can be inconsistent or malformed.

Solution:

• Added validation layer:

  • ensures 4 options
  • checks correct answer
  • verifies structure

• Fallback strategies were redesigned to avoid low-quality output

---

4. Making It Feel Like a Game

A plain quiz is not engaging.

Solution:

• Introduced levels (easy → medium → hard)
• Added scoring and progression
• Provided instant feedback with explanations

---

📈 What We Learned

• Prompt engineering is critical — small changes drastically affect output quality
• AI is powerful, but requires strict constraints and validation
• Building with AI is not one-shot — it’s an iterative process
• User experience matters as much as technical functionality

We also learned how to:

• design AI-driven systems
• integrate APIs effectively
• structure full-stack applications quickly using MeDo

---

🎯 Impact

This project transforms learning from passive reading to active, game-based understanding

Instead of memorizing, users:

• engage
• think
• apply knowledge

---

🏆 Final Thought

This project shows how AI can move beyond content generation into:

creating personalized, interactive learning experiences

---

🔥 One-Liner

Turn any study material into an interactive learning game in seconds.

Built With

• googleaistudio
• medo