"Find yourself in the pages."
A calm, immersive, AI-powered book discovery experience.
Most platforms make reading feel like:
- ❌ Endless scrolling
- ❌ Algorithm overload
- ❌ No emotional connection
BiblioDrift is different.
It feels like:
📚 Walking into a quiet bookstore
☕ Picking a book based on mood
🌧️ Letting the atmosphere guide you
- 🧘 Zero UI Noise → No clutter, no distractions
- 🎭 Vibe-First Discovery → Search by feeling, not metadata
- 📖 Tactile Interaction → Books behave like real objects
- 🤖 AI as a Bookseller → Not recommendations, but conversations
- 3D books you can pull, flip, and explore
- Shelf-based organization (Want / Reading / Favorites)
- Mood-based recommendations (e.g., “rainy mystery”)
- Dynamic AI-generated blurbs
- Conversational assistant → Elara, the Bookseller
- Glassmorphism interface
- Ambient sounds (rain, fireplace)
- Emotion-based tagging system
- Skeleton loaders (smooth loading)
- LocalStorage persistence
- Seamless interactions
| Layer | Technology |
|---|---|
| Frontend | HTML5, CSS3 (3D), Vanilla JS |
| API | Google Books API |
| Backend | Flask, SQLAlchemy, JWT cookies |
| AI | LLM-powered notes, chat, and mood analysis |
| Storage | LocalStorage |
BiblioDrift follows a strict rule:
- ❌ No hardcoded lists
- ❌ No manual curation
- ✅ 100% AI-generated discovery
- Mood
- Emotional tone
- Intent
- Vibe
- 🤖 AI-powered recommendations (core)
- 🧠 Conversational librarian (Elara)
- 🌧️ Mood-based discovery engine
- 🎧 Ambient environments
- 📊 Emotion analytics (future)
Frontend = Librarian
Backend = Curator
graph TD
A[Frontend UI] -->|Mood Query| B[Flask Backend]
B -->|Prompt Engineering| C[LLM / AI Service]
C -->|Generated Insight| B
B -->|JSON Response| A
A -->|Book Data| D[Google Books API]
A -->|Persistence| E[LocalStorage]
BIBLIODRIFT/
│
├── backend/ # Python backend logic
│ ├── app.py
│ ├── ai_service.py
│ ├── cache_service.py
│ ├── config.py
│ ├── error_responses.py
│ ├── models.py
│ ├── security_utils.py
│ ├── validators.py
│ │
│ ├── mood_analysis/ # mood-based recommendation logic
│ └── purchase_links/ # purchase link generation
| ├── price_tracker/
│
├── frontend/ # UI (client-side)
│ ├── pages/ # HTML files
│ │ ├── index.html
│ │ ├── auth.html
│ │ ├── chat.html
│ │ ├── library.html
│ │ ├── profile.html
│ │ └── 404.html
│ │
│ ├── js/ # JavaScript
│ │ ├── app.js
│ │ ├── chat.js
│ │ ├── config.js
│ │ ├── footer.js
│ │ └── library-3d.js
│ │
│ ├── css/ # Styles
│ │ ├── style.css
│ │ ├── style_main.css
│ │ └── style-responsive.css
│ │
│ ├── assets/ # Images, sounds
│ │ ├── images/
│ │ └── sounds/
│ │
│ └── script/ # extra JS (header scroll etc.)
│
├── config/ # ⚙️ Configuration
│ ├── .env.development
│ ├── .env.example
│ ├── .env.testing
│ ├── requirements.txt
│ └── runtime.txt
│
├── docs/ # 📚 Documentation
│ ├── contributing.md
│ ├── Open-Source-Event-Guidelines.md
│ ├── TUTORIAL.md
│ └── page.png
│
├── tests/ # 🧪 Test files
│ ├── test_api.py
│ ├── test_llm.py
│ └── test_validation.py
│
├── .gitignore
├── README.md
├── LICENSE
├── netlify/ # deployment config
├── script/ (if any left)
├── venv/
└── .vscode/
BiblioDrift follows a strict AI-only recommendation model.
- All recommendations must be generated dynamically using AI/LLMs.
- Manual curation, editor picks, static mood lists, or hardcoded book mappings are not allowed.
- AI outputs should be based on abstract signals such as:
- Vibes
- Mood descriptors
- Emotional tone
- Reader intent
This ensures discovery stays organic, scalable, and aligned with BiblioDrift’s philosophy of vibe-first exploration.
-
Clone the repository: bash git clone https://github.com/devanshi14malhotra/bibliodrift.git
-
Open index.html in your browser.
- That's it! No build steps required for the vanilla frontend.
The Flask backend powers authentication, library sync, AI blurbs, chat, mood analysis, and other API flows.
- Netlify should serve the static frontend from the generated dist/ bundle.
- The Flask backend, database, Redis, and AI services are not hosted by Netlify.
- To make the API work in production, deploy the backend separately and point the frontend MOOD_API_BASE to that host.
To keep the frontend and backend synced, use the following mapping:
| Feature | Frontend Call (app.js) | API Endpoint (app.py) | Logic Provider (ai_service.py) |
|---|---|---|---|
| Book Vibe | POST /api/v1/generate-note | handle_generate_note() | generate_book_note() |
- Endpoint: POST /api/v1/generate-note
- Logic: Processed by ai_service.py
Method: POST URL: /api/v1/generate-note Description: Generates an AI-powered "bookseller note" based on the book's vibe, mood, and metadata.
Headers
json { "Content-Type": "application/json" }
Body
json { "title": "The Night Circus", "author": "Erin Morgenstern", "mood": "mysterious, magical, slow-burn romance" }
Success (200 OK)
json { "status": "success", "note": "A dreamlike duel unfolds in a wandering circus of shadows and light. Perfect for readers who crave atmospheric magic and quiet intensity." }
Error (400 Bad Request)
json { "status": "error", "message": "Missing required fields: title or mood" }
- Frontend sends a POST request from app.js to /api/v1/generate-note.
- The Flask backend (app.py) receives the request via handle_generate_note().
- Input data (title, author, mood) is validated.
- The request is passed to generate_book_note() in ai_service.py.
- The AI model generates a contextual "bookseller note".
- The backend returns the generated note as a JSON response.
- Frontend displays the note in the book popup UI.
We welcome contributions to make BiblioDrift cozier!
- Fork the repo.
- Create a feature branch such as feature/cozy-mode.
- Make your changes and test them locally.
- Push your branch and open a Pull Request.
See CONTRIBUTING.md for the fuller workflow and contribution rules.
MIT License.
Built by Devanshi Malhotra and contributors, with ☕ and code.
If you like this project, please consider giving the repository a ⭐ STAR ⭐.