Chapter 14: Authentication, Databases & User Management¶
“Projects grow when users trust them.”
This chapter takes us from solo demos to real users, real data, and real structure. If your AI project is going to grow, it will need:
✅ Authentication ✅ A database ✅ User management
This is how you go from fun app to platform.
This chapter covers:¶
- Why auth & user data matter in AI apps
- Supabase vs Firebase (and how to choose)
- Adding login, storing usage history
- Schema design for prediction-based apps
- Builder’s lens: building with identity in mind
Opening Reflection: The Shift from Tools to Ecosystems¶
“When one user logs in… you’ve moved from app to experience.”
An AI model can:
- Classify text
- Draw an image
- Predict stock prices
But without:
- A login button
- A way to store results
- A user dashboard
…it’s just a moment of interaction.
Once you add identity, you gain:
- ✅ Memory
- ✅ Personalization
- ✅ Security
- ✅ Community
This is how apps grow into ecosystems.
14.1 Why You Need Auth & Data¶
Even the simplest AI app benefits from:
| Feature | Why It Matters |
|---|---|
| Login | Track who used what, when |
| History | Store past predictions (e.g., memes, text) |
| Usage Tracking | Show stats (generations, liked results) |
| Billing Tiers | Limit features per user plan |
14.2 Choosing Your Stack: Supabase vs Firebase¶
| Feature | Supabase | Firebase |
|---|---|---|
| Language | SQL/Postgres, JS/Python clients | NoSQL/Firestore, JS-heavy |
| Auth | Email, OAuth, Magic Link | Email, OAuth, Phone, Anonymous |
| Realtime DB | ✅ Yes (Postgres pub/sub) | ✅ Yes (Firestore) |
| Storage | File storage (avatars, images) | Cloud Storage |
| Open Source | ✅ Yes | ❌ Proprietary |
| Great For | SQL lovers, devs who want control | Fast prototyping, Google integration |
🔧 Recommendation for AI Builders: Use Supabase if you want SQL + auth + file storage all in one stack.
14.3 Adding Authentication (Supabase Example)¶
Install the Python client:
pip install supabase
Create a project → Get your URL and anon/public API key
Sign up at https://supabase.com
Python Code¶
from supabase import create_client
url = "https://xyzcompany.supabase.co"
key = "your-public-anon-key"
supabase = create_client(url, key)
# Sign up a new user
auth_response = supabase.auth.sign_up({
"email": "user@example.com",
"password": "strongpassword"
})
✅ Use frontend (React or Gradio) to call this via Supabase API or JS SDK.
14.4 Design Your Schema (Prediction-Based App)¶
Table: users¶
| id | created_at | |
|---|---|---|
| 1 | user@a.com | 2025-04-23 |
Table: predictions¶
| id | user_id | input_text | result_text | model_used | timestamp |
|---|---|---|---|---|---|
| 1 | 1 | "Is this positive?" | "POSITIVE" | bert-base-uncased | 2025-04-23 14:12 |
Benefits:¶
- Track usage per user
- View prediction history
- Filter by model/task
- Export data for analytics
14.5 Frontend Integration (React or Gradio)¶
Gradio Example:¶
gr.Textbox(label="Username")
gr.Textbox(label="Password", type="password")
React + Supabase/Firebase:¶
- Supabase JS SDK
- Firebase Auth
- Custom backend endpoints (
/login,/signup)
Add these for production readiness:
- JWT tokens for session management
- LocalStorage for persistent login
- Role-based routing (e.g.,
user,admin)
14.6 Builder’s Lens: Identity Unlocks Continuity¶
“Without users, you have usage. With users, you have relationships.”
When users log in:
- They expect quality
- They invest in results
- They complete the loop:
input → feedback → return
This is how you build real AI products — not just demos.
Summary Takeaways¶
| Concept | Why It Matters |
|---|---|
| Auth = ownership | Ties data and actions to a real person |
| DB = memory | Enables history, analytics, billing |
| Supabase = full stack | SQL + auth + file store + REST API |
| Identity = UX layer | From toy → tool → trusted experience |
🌟 Closing Reflection¶
“Every user login is a vote of trust. The database is where you honor it.”