Two weeks ago I published Why I'm Killing My SaaS Before Launch, the story of spending 2 months building 70,000 lines of enterprise-grade infrastructure and never shipping.

Today I'm launching Huesly. Built in 2 weeks. Live users. Actual revenue.

Here's what I did differently.

What Changed

Ryva (killed):

• 2 months building
• 70,000 lines of Go
• Custom backend, CI/CD, email queues
• Zero users
• Never launched

Huesly (shipped):

• 2 weeks building
• Supabase + Next.js + Vercel
• Live on Product Hunt, Hacker News, X
• First signups in 24 hours
• Actually making money

The difference? I finally listened to my own advice: ship fast, scale later.

The Stack Comparison

Ryva	Huesly
Custom Go backend with service layers	Next.js API routes
Multi-stage CI/CD with load balancing	Vercel deploy button
Custom email queuing system	Resend (one API call)
3 containerized Docker instances	Single Vercel deployment
Custom auth with JWT + refresh tokens	Supabase Auth
2-week billing system integration	Stripe Checkout (30 minutes)

Every "enterprise feature" I over-engineered for Ryva had a boring alternative that let me ship 10x faster.

What Huesly Does

Simple problem: developers spend 3+ hours picking colors, fonts, and spacing for side projects that might never ship.

Huesly generates complete design systems from moods in less than 1 second.

How it works:

1. Enter a mood (Minimal, Bold, Professional, Elegant, etc)
2. Get instant WCAG-compliant colors, font pairings, spacing scales
3. Export to CSS variables, Tailwind v4 config, SCSS, or JSON tokens

That's it. No Figma. No design skills needed. Straight from idea to production-ready code.

The Technical Decisions

Why OKLCH Color Space?

Most color generators use RGB or HSL. They create mathematically correct palettes that look terrible together.

OKLCH is perceptually uniform - colors that are mathematically similar actually look harmonious to humans. When you adjust lightness, the hue stays consistent. With HSL, yellow at 50% lightness looks way brighter than blue at 50%.

This is why Huesly's palettes feel cohesive instead of random.

Why Next.js + Supabase?

After Ryva, I had a choice:

• Reuse my 70k lines of production-grade Go infrastructure
• Start fresh with boring technology

I chose boring. Here's why:

For an MVP, Supabase gives you:

• Auth (GitHub OAuth in 5 minutes)
• Database (PostgreSQL with RLS)
• Real-time subscriptions
• Row Level Security for multi-tenancy

What took me 2 months to build in Go, I got in 2 days with Supabase.

The rule: Use custom infrastructure when you're serving thousands of paying customers and Supabase becomes the bottleneck. Not before.

Early Results

Launch day (24 hours):

• Posted to X, Hacker News, Dev.to, Product Hunt
• 2 signups
• Active discussions on HN
• Email from old Ryva waitlist converting to Huesly users

Not massive numbers. But users are using it. People are exporting design systems to real projects. That's infinitely more than Ryva's zero.

What I learned:

• 2 real users > 0 hypothetical enterprise customers
• Shipping creates momentum
• Feedback only comes from real usage
• Revenue proves value, not architecture

The Pricing Experiment

I'm testing two models:

• Lifetime: $30 (capped at 100 users)
• Monthly: $15/month unlimited

Early theory: developers hate subscriptions for simple tools. They want to own it once and use it across multiple projects.

So far, lifetime is converting 3:1 vs monthly.

This might be a mistake long-term (lower LTV), but for early traction, it's working. Developers are more willing to try something when they know they won't get hit with recurring charges.

I can always introduce v2 features as subscription-only later.

What I'm Doing Differently

1. Building in Public

Ryva was built in silence. I thought code quality would attract users.

Huesly is different. I'm:

• Posting daily updates on X
• Sharing metrics (even small ones)
• Writing about technical decisions
• Documenting failures

Attention is the bottleneck, not engineering skill.

2. Simple Problem, Fast Validation

Ryva tried to solve developer productivity across Slack, GitHub, email, and project management tools. Complex integrations, rate limits, syncing hell.

Huesly solves one thing: getting from mood to production-ready design system in seconds.

Simple problem = fast validation = quick iteration.

3. Content-First Strategy

I'm learning:

• High-level video editing
• Viral hooks for short-form content
• Long-form storytelling
• Platform-specific content (X threads, TikTok demos, blog posts)

Engineering got me to 0 → 1. Marketing will get me to 1 → 100.

The Ryva Code Isn't Wasted

Those 70,000 lines of Go? Still valuable.

When Huesly actually needs scale - when Supabase becomes the bottleneck serving thousands of paying customers - I'll have battle-tested patterns ready:

• Multi-tenant architecture with RLS
• Service layer with dependency injection
• Comprehensive testing infrastructure
• Production-grade error handling
• Multi-stage deployment pipelines

The code isn't wasted. The timing was just wrong.

The lesson: Build enterprise infrastructure when you have enterprise problems, not before.

What's Next

Short-term (next 30 days):

• Reach 100 users
• Get 10 paying customers
• Collect testimonials
• Improve based on feedback
• Learn what features actually matter

Medium-term (3-6 months):

• Hit $1K MRR
• Launch v2 features
• Build viral content engine
• Grow through word-of-mouth

Long-term (before graduation):

• $100K ARR
• Prove a 16-year-old can build real businesses
• Document everything publicly

The goal isn't just revenue. It's proving that:

1. You can ship fast without sacrificing quality
2. Boring technology beats interesting architecture for MVPs
3. Marketing matters more than perfect code
4. Age is irrelevant if you solve real problems

To Everyone Who Over-Engineers

I spent 2 months building infrastructure I didn't need.

You might be doing the same right now:

• Perfect CI/CD before you have users
• Custom auth instead of Supabase/Clerk
• Microservices for a product with 0 customers
• Abstraction layers "for when we scale"

Stop.

Ship something ugly. Get users. Iterate based on real feedback.

You can always refactor later. You can't get time back.

Try Huesly

If you've ever spent hours tweaking hex codes or searching for the perfect font pairing:

👉 huesly.app

Enter a mood. Get a design system. Export to code. Ship your project.

And if you're building your first SaaS: learn from my mistakes. Ship fast. The perfect infrastructure can wait.

Follow the journey:

• X: @egewrk