Two weeks. $340 in tools. A working product with paying customers.

Here’s exactly how I did it.

The Product

A simple SaaS for small agencies. Proposal templates with e-signatures. Nothing revolutionary. Clear problem. Clear solution.

The goal wasn’t building something amazing. It was validating demand fast.

Day 1-3: Planning and Research

Claude Pro ($20/month)

I dumped my idea into Claude. Asked it to poke holes. Find competitors. Identify risks.

Then I used it to draft user stories and feature requirements. Two hours of back-and-forth gave me a clear scope.

Outcome: 12 user stories. Clear MVP scope. List of 8 competitors to study.

Day 4-7: Building the Product

Cursor Pro ($20/month)

Cursor is VS Code with AI built in. I’m not a developer but I can read code and describe what I want.

I described features in plain English. Cursor wrote the code. I tested. Fixed errors with Cursor’s help. Repeated.

For authentication, payments, and database, I used Supabase (free tier) and Stripe.

Four days of 8-hour sessions. Working product.

Outcome: Functional app. User auth. Stripe payments. Basic proposal builder.

Day 8-10: Design and Copy

v0 by Vercel ($0-20/month)

Described the UI I wanted. v0 generated React components. Better looking than anything I’d design myself.

Claude for copy

All landing page copy, email sequences, and in-app text drafted by Claude. Edited for voice. Maybe 30% rewritten.

Canva ($0 - free tier)

Logo, social images, favicon. Good enough for MVP.

Outcome: Landing page. Email sequence. Decent-looking product.

Day 11-12: Testing and Polish

ChatGPT Plus ($20/month)

Used for QA. Described features, asked it to think of edge cases. Found 6 bugs I’d missed.

Also used for writing test user scenarios and documentation.

Friends as beta testers

AI can’t replace human feedback. Five friends tried the product. Found UX issues AI didn’t catch.

Outcome: Squashed bugs. Smoother UX. Ready for launch.

Day 13-14: Launch

Claude for launch content

Product Hunt description. Twitter thread. LinkedIn post. HackerNews intro. All first-drafted by Claude.

Spent an afternoon making it sound like me, not a robot.

Launch day: Posted everywhere. Got 12 signups. 3 converted to paid ($49/month each).

The Costs

Tool	Monthly Cost
Claude Pro	$20
Cursor Pro	$20
ChatGPT Plus	$20
Supabase	$0
Vercel	$0
Stripe	2.9% + 30c per transaction
Canva	$0

Total tools: $60/month during build.

I also bought a domain ($12) and had existing hosting.

Total MVP cost: ~$72

What I Learned

AI doesn’t replace skill. It amplifies it.

I have enough technical understanding to direct AI. Complete beginners would struggle.

But I couldn’t have built this alone two years ago. AI closed the gap.

Speed kills perfectionism

Two weeks forces simplicity. No scope creep. No polish paralysis. Ship ugly.

Validation > Features

My MVP was missing obvious features. Didn’t matter. Three paying customers proved demand. Features come later.

The Caveats

This approach works for simple B2B SaaS. Clear problem. Clear solution. Limited complexity.

It doesn’t work for:

• Complex technical products
• Anything requiring custom AI/ML
• Consumer products needing beautiful design
• Regulated industries

But for validation? For proving an idea? AI tools let one person move like a small team.

If you’re sitting on an idea, you’re out of excuses.