AI-powered personalized storybook platform for children, combining Gemini generative AI with ElevenLabs narration and real-time safety moderation.

Every story is unique, illustrated, narrated, and safe.

DataFables lets children design their own character, choose a story topic and age group, and instantly receive a personalized multi-page storybook — complete with AI-generated illustrations and professional narration. Every story is automatically screened for child safety before being shown.

• Personalized stories: Gemini generates unique multi-page stories based on topic, age group, and custom character
• Pixel art character designer: Customize hair, skin tone, outfit, shoes, and accessories
• AI narration: ElevenLabs reads every page aloud in a natural voice
• Child safety: Every story is screened by SafetyKit before display
• Similar stories: Actian VectorAI surfaces related story recommendations
• Achievement system: Kids earn XP, coins, and badges as they read
• Shop: Spend coins to unlock new character items
• Multilingual: Full UI support for 10 languages
• Snowflake integration: Story data and facts stored and retrieved via Snowflake Cortex

git clone https://github.com/jaimeb18/datafables
cd datafables
npm install
npm run dev

Then visit http://localhost:3000

1. Design: User customizes their character (hair, skin, outfit, accessories)
2. Configure: Choose a topic, age group, and language
3. Generate: Gemini writes a multi-page branching story + illustrations in parallel
4. Safety check: SafetyKit screens all content before display
5. Narrate: ElevenLabs voices each page on demand
6. Store: Story metadata saved to Snowflake + Actian for future recommendations

Every story goes through SafetyKit's content moderation API before being shown to the user. If no SafetyKit key is available, a Gemini-based fallback safety check is used. A safety badge is displayed on every story showing the safety score.

Story titles and topics are embedded using sentence-transformers and stored in Actian VectorAI DB. When a story ends, the top 3 most similar stories are surfaced via vector similarity search, with match percentages shown.

datafables/
├── app/
│   ├── api/
│   │   ├── generate/        # Story + image generation endpoint
│   │   ├── narrate/         # ElevenLabs TTS endpoint
│   │   └── recommend/       # Actian similar stories endpoint
│   └── page.tsx             # Main app page
├── components/
│   ├── StoryForm.tsx         # Character designer + story config
│   ├── StoryDisplay.tsx      # Multi-page story viewer
│   ├── AchievementPanel.tsx  # Achievement tracker
│   ├── ShopPanel.tsx         # Coin shop
│   └── SimilarStories.tsx    # Actian-powered recommendations
├── lib/
│   ├── achievements.ts       # XP, coins, level system
│   ├── actian.ts             # Actian VectorAI client
│   ├── safetykit.ts          # SafetyKit + Gemini fallback
│   └── shop.ts               # Shop items + purchase logic
├── python/
│   ├── actian_server.py      # FastAPI proxy for Actian gRPC
│   └── docker-compose.yml    # Actian VectorAI DB container
└── public/
    ├── achievements/         # Achievement badge images
    └── magic-ball.png        # Loading animation

• Node.js 18+
• API keys (see below)

Create a .env.local file:

GEMINI_API_KEY=your_gemini_key
ELEVENLABS_API_KEY=your_elevenlabs_key
SNOWFLAKE_ACCOUNT=your_account
SNOWFLAKE_USERNAME=your_username
SNOWFLAKE_PASSWORD=your_password
SNOWFLAKE_DATABASE=DATAFABLES
SNOWFLAKE_WAREHOUSE=COMPUTE_WH
SAFETYKIT_API_KEY=your_safetykit_key
ACTIAN_PROXY_URL=http://localhost:8001

cd python
docker compose up -d
pip install -r requirements.txt
uvicorn actian_server:app --port 8001

Hacklytics 2026 — Georgia Tech

• @luanamaldaner
• @NicolasMurg
• @jaimeb18
• @viv2612

MIT License — Built for Hacklytics 2026