I Built a SaaS Product With AI and Zero Coding Experience
Twelve months ago I couldn't have told you the difference between a React component and a Node.js server. Today I run a live SaaS product with real users, Stripe payments, a Supabase database, and an SEO content engine that drives organic traffic every week.
I didn't hire a developer. I didn't use a no-code builder. I used Claude, a Terminal window, and a lot of patience.
This is the real story — the stack, the workflow, the mistakes, and what I'd do differently.
The Problem I Was Solving
I work in Healthcare IT. I've watched colleagues get laid off, struggle to update their resumes, and spend hours tailoring applications for jobs that were a bad fit. The existing tools — resume builders and ATS checkers — felt generic and disconnected from how hiring actually works.
I wanted something that would take your resume and a job description and tell you precisely how well they matched, then help you close the gaps. AI-powered, specific, actionable.
That became Get Resumatch.
The Stack (And Why I Chose It)
Every decision was made based on one question: can I operate this from a Terminal with Claude's help?
- Frontend: React (Vite) — Claude knows it extremely well, errors are easy to diagnose
- Backend: Node.js + Express on Railway — simple deploys, straightforward logs
- Database: Supabase — Postgres with a dashboard I can actually read, generous free tier
- Auth: Supabase Auth + Google One Tap
- Payments: Stripe — the industry standard, Claude has deep knowledge of the API
- AI: Claude API (claude-sonnet) for all resume and job analysis
- Email: Resend — simple API, dead easy to set up
- Hosting: Vercel (frontend) + Railway (backend)
Total monthly infrastructure cost in early stages: under $30.
How the Build Actually Worked
I didn't sit down and architect the whole system upfront. I built feature by feature, always starting with the simplest version that proved the idea worked.
The workflow was the same every session:
- Upload a handoff doc describing the current state and what I wanted to build next
- Claude would give me commands or files to create
- I'd run them in Terminal and paste the output back
- We'd iterate until the feature was live
Some features took an hour. Some took three sessions. The job search integration — pulling live listings via API and scoring them against a user's resume — took about four sessions to get right.
The Mistakes That Cost Me the Most Time
I'm including these because the tutorials never do, and they're the most valuable part.
- Not setting hard constraints early. I'd let Claude refactor things that were working just fine, and break something in the process. I now maintain a strict list of things that must never change — specific library versions, CSS rules, component patterns — and I include them in every handoff.
- Summarizing errors instead of pasting them. Every time I paraphrased an error message, I added a loop to the debugging cycle. Paste the full output. Always.
- Building features before validating the core. I shipped a job tracker, a LinkedIn rewriter, and a cover letter generator before I had a single paying user. Focus on the core value proposition first.
- Ignoring SEO until month two. Organic search is the highest-leverage channel for a solo SaaS. I should have started building content and landing pages on day one.
Where It Stands Today
The product has real users, a paying subscriber, and growing organic traffic from 35 SEO-optimized blog posts. Google Search Console is showing all-time highs on impressions and queries. Monthly recurring revenue is early-stage, but the trajectory is there.
More importantly: I built something real. Something that works. Something that helps people. Without a development background, without a team, without outside funding.
What This Means for You
If you have a problem worth solving and you're willing to learn the Terminal workflow, you can build it. The AI handles the code. You handle the product thinking, the user empathy, and the persistence.
The gap between "I have an idea" and "I have a live product" has never been smaller. The bottleneck is no longer technical skill. It's clarity of thought and willingness to ship.
Pick your stack. Learn the workflow. Start small. The rest compounds.