DEV Community

2x lazymac
2x lazymac

Posted on

From Side Project to SaaS: Shipping 30 APIs in 3 Days

#saas #webdev #startup #api

Last month I shipped 30 APIs on Cloudflare Workers in 3 days. Not prototypes — production APIs with documentation, rate limiting, and MCP support. Here is how.

The Stack

  • Runtime: Cloudflare Workers (V8 isolates)
  • Storage: KV (cache), D1 (SQLite), R2 (files)
  • Deployment: Wrangler CLI + GitHub Actions
  • Monitoring: Cloudflare Analytics + custom logging

Total infrastructure cost: $5/month.

Day 1: The Template (Hours 1-4)

I built one base template that every API shares:

// Base API template
export default {
  async fetch(request, env) {
    const url = new URL(request.url);

    // Standard middleware
    if (request.method === 'OPTIONS') return corsResponse();
    const rateLimited = await checkRateLimit(request, env);
    if (rateLimited) return rateLimited;

    // Route handling
    const handler = routes[url.pathname];
    if (!handler) return jsonError('Not found', 404);

    try {
      const result = await handler(request, env, url);
      return Response.json(result, { headers: corsHeaders });
    } catch (err) {
      return jsonError(err.message, 500);
    }
  }
};
Enter fullscreen mode Exit fullscreen mode

Key decisions:

  • Every API returns JSON
  • Every API has CORS enabled
  • Every API has rate limiting
  • Every API has an MCP endpoint

Day 1: First 10 APIs (Hours 4-10)

With the template ready, each new API took 20-30 minutes:

# My workflow for each API
cp -r template/ new-api/
# Edit the route handlers
# Add API-specific logic
wrangler deploy
# Test with curl
# Done
Enter fullscreen mode Exit fullscreen mode

APIs shipped on Day 1:

  1. QR Code Generator
  2. Email Validator
  3. Color Palette Generator
  4. Password Strength Checker
  5. Fake Data Generator
  6. Text Analysis (sentiment, keywords)
  7. Regex Toolkit
  8. IP Geolocation
  9. Markdown to HTML
  10. URL Shortener

Day 2: The Data-Heavy APIs (Hours 11-20)

Day 2 focused on APIs that needed external data or computation:

# Testing each API as I built it
import requests

apis = [
    ("AI Spend Calculator", "/ai-spend/calculate?model=gpt-4&input_tokens=1000&output_tokens=500"),
    ("Tech Stack Detector", "/tech-stack/detect?url=https://github.com"),
    ("Crypto Signals", "/crypto-signal?symbol=BTC"),
    ("SEO Analyzer", "/seo/analyze?url=https://example.com"),
]

for name, path in apis:
    resp = requests.get(f"https://api.lazy-mac.com{path}")
    status = "OK" if resp.status_code == 200 else "FAIL"
    print(f"[{status}] {name}")
Enter fullscreen mode Exit fullscreen mode

Day 3: MCP + Documentation (Hours 21-30)

The final day was about making every API AI-accessible:

// Added to every API
if (url.pathname === '/mcp') {
  return Response.json({
    name: apiName,
    tools: apiTools,
    version: "1.0.0"
  });
}
Enter fullscreen mode Exit fullscreen mode

And auto-generated docs:

# Each API self-documents at /
curl https://api.lazy-mac.com/ai-spend/
Enter fullscreen mode Exit fullscreen mode

Lessons Learned

  1. Template-first development saves 80% of the time
  2. Deploy early, iterate fast — Workers deploy in < 2 seconds
  3. KV is your best friend for caching expensive computations
  4. MCP support doubles your audience — AI users are a huge market

The Numbers

  • 30 APIs deployed
  • Average response time: 45ms globally
  • Infrastructure cost: $5/month
  • Revenue: on track for $500/month from API subscriptions

Try Them Out

All 30 APIs are live at api.lazy-mac.com. The AI Spend Calculator and Tech Stack Detector are the most popular.

curl "https://api.lazy-mac.com/ai-spend/calculate?model=gpt-4&input_tokens=1000&output_tokens=500"
Enter fullscreen mode Exit fullscreen mode

AI FinOps API on Gumroad | Tech Stack API on Gumroad | Full API Hub

Top comments (0)