How I Built 8 SaaS Tools in One Night with AI

People ask how I built 8 tools so fast. Here's the workflow.

The Setup

• Editor: Claude Code CLI (terminal-based)
• Stack: Python + FastAPI (simple, fast)
• Hosting: Single $6/month VPS
• Deploy: systemd (no Docker complexity)

The Process

1. Start with a Clear Spec

Don't say "build me a URL shortener."

Say:

Build a URL shortener with:
- POST /api/links - create short link
- GET /{code} - redirect
- GET /api/links/{code}/stats - click stats
- JSON file storage in data/
- FastAPI + uvicorn

The clearer your spec, the better the output.

2. Small Iterations

Don't try to build everything at once.

1. Basic redirect working? ✓
2. Add stats tracking
3. Add custom codes
4. Add expiration

Test each piece before moving on.

3. Let AI Handle Boilerplate

AI excels at: - Project structure - Error handling - API documentation - Standard patterns

You focus on: - Business logic decisions - Edge cases - Integration points

4. JSON for MVPs

PostgreSQL is overkill for most MVPs.

# This is enough for thousands of users
with open("data/links.json", "r") as f:
    links = json.load(f)

Migrate to a database when you actually need it.

5. Deploy Simply

# /etc/systemd/system/myapp.service
[Service]
ExecStart=/usr/bin/uvicorn src.main:app --port 8080
Restart=always

That's it. No Kubernetes. No Docker compose files.

What I Learned

AI is Good At:

• Generating code from specs
• Following patterns
• Writing tests
• Documentation

AI Needs Help With:

• Architecture decisions
• Understanding your specific context
• Edge cases you haven't mentioned
• Integration with existing systems

The Multiplier Effect

A single developer with AI can: - Prototype in hours, not days - Ship more features - Maintain larger codebases - Focus on what matters

But you still need to: - Know what to build - Review the output - Handle deployment - Talk to users

My Stack for Fast Shipping

Try It Yourself

1. Pick a small problem
2. Write a clear spec
3. Use Claude Code or similar
4. Ship in a day

The barrier to building is lower than ever. Go make something.

See what I built: indiekit.ai

GitHub: indiekitai