Building a One-Person SaaS with AI Tools (How I Built RAXXO Studio)
One Person. One SaaS. Zero Employees.
RAXXO Studio is a live SaaS product at studio.raxxo.shop. AI social media copywriter. Generates titles, captions, hashtags, and music suggestions. It has paying users, a 4-tier pricing model, and a full production deployment pipeline.
I built the entire thing by myself. No co-founder, no contractors, no agency. Just me and a stack of AI tools that turned what would traditionally take months into a few weeks of evening sessions.
Solo-founded SaaS companies have been growing faster than venture-backed startups in recent years. The reason is simple: AI tools collapsed the skill gap between "idea person" and "full-stack shipping machine."
Here is exactly how it happened.
The Stack (What Actually Ships)
| Layer | Tool | Monthly Cost | Why This One |
|---|---|---|---|
| Framework | Next.js 16 | 0 EUR | App Router, Server Components, streaming |
| AI Backend | Claude API (Anthropic) | ~15 EUR | Best structured output, reliable JSON |
| Auth | Clerk | 0 EUR (free tier) | Allowlist support, fast integration |
| Database | Neon (Postgres) | 0 EUR (free tier) | Serverless, branches, generous limits |
| Hosting | Vercel | 20 EUR | Preview deploys, edge functions |
| Payments | Shopify | ~30 EUR | Already running store, unified billing |
| Development | Claude Code (CLI) | ~20 EUR (API usage) | The actual builder |
Total monthly running cost: roughly 85 EUR. That is less than one hour of freelance design work at Berlin rates.
The Timeline (Idea to Production)
I built RAXXO Studio in evening sessions after my day job, over the course of a few weeks. There was no structured phase plan. Some nights I worked on auth, the next on UI, then back to fixing API routes. Claude Code handled the heavy lifting while I focused on architecture and UX decisions. The total calendar time was roughly a month, but actual working hours were a fraction of what a full-time build would require.
What AI Handles vs What You Still Do Manually
This is where most "build with AI" articles lie to you. They imply AI does everything. It does not.
AI handles well:
- Boilerplate code (API routes, CRUD operations, form validation)
- Database migrations and schema changes
- CSS layouts and responsive breakpoints
- Writing tests (unit and integration)
- Refactoring and code cleanup
- Documentation generation
You still do manually:
- Architecture decisions (what to build, what to skip)
- UX judgment calls (flow, timing, hierarchy)
- Visual design taste (color, spacing, typography)
- Business logic edge cases
- Debugging production issues at 2 AM
- Deciding when to ship vs when to polish
In practice, AI coding assistants increase output significantly for routine tasks but only modestly for novel problem-solving. That matches my experience exactly. Claude Code is phenomenal at "build me a settings page with these 6 toggles." It struggles with "figure out why this webhook fires twice on Shopify plan changes."
The CLAUDE.md Strategy
The single biggest productivity hack was not a tool. It was a file. CLAUDE.md sits in the project root and acts as persistent memory for Claude Code. Every coding convention, every CSS variable name, every API pattern lives in that file.
Without it, every new conversation starts from zero. With it, Claude Code understands your project like a teammate who has been there since day one. A detailed CLAUDE.md makes a massive difference. It covers theme variables, component patterns, tier naming conventions, deployment procedures, and known gotchas.
If you are building a SaaS with AI tools and you skip this step, you will waste 30-40% of your time re-explaining context. I track this with OhNine, a small tray app I built to monitor Claude API usage. The difference between sessions with and without good context is dramatic.
Cost Breakdown (Real Numbers)
| Category | One-Time Cost | Monthly Cost |
|---|---|---|
| Domain (raxxo.shop) | 12 EUR/year | - |
| Claude Pro subscription | - | ~107 EUR |
| Vercel Pro | - | 20 EUR |
| Claude API (production) | - | ~15 EUR |
| Shopify Basic | - | ~30 EUR |
| All other services | - | 0 EUR (free tiers) |
| Total to launch | ~12 EUR | ~172 EUR |
Compare that to hiring a small dev team. Even a single mid-level developer in Berlin costs 5,000-7,000 EUR monthly. A SaaS MVP from an agency runs 20,000-50,000 EUR. The AI-solo path is roughly 98% cheaper at the build stage.
5 Lessons from the Trenches
1. Ship the ugly version first. My first deploy looked rough. Users did not care. They cared about whether the captions were good. Polish comes after validation.
2. Free tiers are your best friend. Neon, Clerk, and Vercel all have generous free tiers. You can run a real SaaS for under 100 EUR monthly until you hit meaningful scale.
3. AI tools have a learning curve too. Getting good at prompting Claude Code took me about 3 weeks. The investment pays back 10x, but do not expect magic on day one.
4. Version control is non-negotiable. When AI generates code, you need to review every commit. I use Git Dojo (a 5 EUR skill I built for Claude Code) to keep my git fundamentals sharp. AI makes it easy to get sloppy.
5. Automate billing early. I spent more time on Shopify webhook integration than on any other single feature. Subscription billing is never simple. Do it early so you are not retrofitting payment logic into a codebase that assumed everything was free.
Revenue Transparency
RAXXO Studio launched with a free tier and three paid plans. See current pricing at studio.raxxo.shop. The product is still in early growth.
For a solo founder, the first milestone is covering costs. Not hockey-stick growth. Not viral loops. Just: does this thing pay for itself? At under 175 EUR monthly overhead, the answer comes fast.
FAQ
Can you really build a SaaS alone with AI tools?
Yes, but "alone" does not mean "without skills." You need to understand architecture, UX, and business logic. AI handles the typing. You handle the thinking. If you have never built a web app before, AI tools will not magically make you a SaaS founder. They will make an experienced builder 3-5x faster.
How long does it take to build a SaaS with Claude Code?
For a focused product with clear scope, 6-10 weeks is realistic. RAXXO Studio took a few weeks from first commit to production. The key is ruthless scope control. Every feature you add costs time regardless of how fast your AI assistant types.
What is the minimum budget to launch a solo SaaS?
Under 200 EUR monthly including AI subscriptions, hosting, and services. If you already have a Claude subscription, you can get a basic SaaS running on free tiers for roughly 50 EUR monthly (domain + minimal API costs).
Do you need to know how to code?
You need to read code confidently and understand software architecture. You do not need to type every line yourself anymore. Think of it like being a film director: you do not need to operate every camera, but you need to know what good footage looks like.
What would you do differently?
Start with billing integration on day one. Build the CLAUDE.md file before writing any code. And ship to one real user in week 2 instead of week 6. Every week without user feedback is a week of guessing.
Related Reading
- The Real Cost of Running a One-Person SaaS
- How One Person Runs a Full Creative Studio with AI
- AI Caption Generator from Video | How RAXXO Studio Works
Dieser Artikel enthält Affiliate-Links. Wenn du dich darüber anmeldest, erhalte ich eine kleine Provision - für dich entstehen keine Mehrkosten. Ich empfehle nur Tools, die ich selbst nutze. (Werbung)