Generative AI is eroding critical thinking in education. Students increasingly turn to ChatGPT and similar tools to complete assignments verbatim, bypassing the learning process entirely. The result is a generation that can prompt an AI but struggles to reason through problems independently. Institutions are scrambling to ban these tools, but the genie is out of the bottle — AI isn't going away.
Instead of fighting AI, Socrato flips the script: it uses large language models and gamification as tools for studying, not instead of it. Paste your notes, and Socrato transforms them into structured summaries, interactive quizzes, and flashcards — then rewards you with XP and badges for actually engaging with the material. The AI does the formatting; you do the thinking.
The core insight is simple: AI is great at generating practice material, but the learning still has to happen in the student's head. Socrato sits in that gap.
- Paste your notes into the app.
- Gemini 2.5 Flash generates summaries, multiple-choice quizzes (with topic tagging and correction explanations), and flashcards — all from a single input.
- Take the quiz, get instant grading and per-question AI explanations. Ask follow-up questions if you don't understand why an answer is wrong.
- Review flashcards with an Anki-style spaced repetition system (Again / Hard / Good / Easy).
- Earn XP and badges for completing quizzes and flashcard sessions. A gamification engine tracks streaks, awards bonuses for perfect scores, and evaluates badge triggers after every activity.
Everything is persisted — quizzes, attempts, scores, flashcard reviews, XP — so progress carries across sessions.
| Layer | Technology | Role |
|---|---|---|
| Frontend | Next.js 16, React 19, TypeScript, Tailwind CSS | App UI, auth state, API client |
| Backend | Python, FastAPI | REST API, prompt orchestration, response validation |
| AI | Google Gemini 2.5 Flash | Study pack, quiz, flashcard, and explanation generation |
| Database | Supabase (PostgreSQL) | User data, quiz storage, attempt tracking, XP, badges |
| Auth | Supabase Auth + JWT | Sign up/in, token-based route protection |
┌──────────────┐ Bearer token ┌──────────────────┐ async ┌─────────────┐
│ Next.js │ ──────────────────────► │ FastAPI │ ────────────► │ Gemini │
│ Frontend │ ◄────────────────────── │ Backend │ ◄──────────── │ 2.5 Flash │
└──────────────┘ JSON response └────────┬─────────┘ structured └─────────────┘
│ JSON
│
┌────▼────┐
│ Supabase │
│ Postgres │
└─────────┘
The backend uses a versioned prompt template system (prompts/study_gen_v1.py, quiz_gen_v1.py, flashcard_gen_v1.py) with few-shot examples and strict JSON output constraints. Every LLM response passes through schema validation (field presence, type checking, option-answer consistency) before reaching the client or database.
Study-One/
├── frontend/ → Next.js web app (React, TypeScript, Tailwind)
├── backend/ → FastAPI server (Python)
│ ├── middleware/ → JWT authentication
│ ├── prompts/ → Versioned prompt templates for Gemini
│ ├── services/ → Gemini AI client, Supabase client, gamification engine
│ └── tests/ → Unit and integration tests
├── shared/ → Shared type definitions (API contract)
└── .env → Environment variables (git-ignored)
- Node.js 18+
- Python 3.11+
- A Supabase project
- A Gemini API key
git clone https://github.com/utmgdsc/Study-One.git
cd Study-OneBackend — create Study-One/.env:
GEMINI_API_KEY="your-gemini-key"# Supabase — Dashboard → Settings → APISUPABASE_URL="https://<project-ref>.supabase.co"SUPABASE_SERVICE_ROLE_KEY="eyJ..."SUPABASE_JWT_SECRET="your-jwt-secret"Frontend — copy frontend/.env.example to frontend/.env.local:
NEXT_PUBLIC_SUPABASE_URL="https://<project-ref>.supabase.co"NEXT_PUBLIC_SUPABASE_ANON_KEY="eyJ..."
.envfiles are git-ignored and must never be committed.
cd backend
python -m venv venv
source venv/bin/activate # macOS/Linux
pip install -r requirements.txt
uvicorn main:app --reload # → http://localhost:8000cd frontend
npm install
npm run dev # → http://localhost:3000Run both for full functionality.
All endpoints (except /health) require Authorization: Bearer <token>.
| Method | Endpoint | Description |
|---|---|---|
GET | /health | Health check (public) |
GET | /api/v1/me | Authenticated user identity |
POST | /api/v1/generate | Generate summary + quiz from notes |
POST | /generate-study-pack | Generate study pack (stricter validation) |
POST | /api/v1/quiz | Generate and store a graded quiz |
POST | /api/v1/quiz/attempt | Submit quiz answers, get grading + XP |
POST | /api/v1/quiz/explain | AI explanation for a quiz question |
POST | /api/v1/quiz/result | Record quiz completion for XP/badges |
POST | /api/v1/flashcards | Generate and store flashcards |
POST | /api/v1/flashcards/review | Submit Anki-style card rating |
POST | /api/v1/flashcards/session-complete | Record session for XP/badges |
GET | /api/v1/flashcards/:id/history | Card review history |
cd backend
python -m pytest tests/ -vTests use real JWT tokens signed with your SUPABASE_JWT_SECRET. All Gemini calls are mocked — no API quota consumed.
Built by Arhum, Alex, Jessie, Sean and Usha under Professor Mahmoud.