I Built a 192K-Line SaaS Platform in 5 Months With AI While Working at a Lumber Mill

I'm not a developer. Let me get that out of the way first.

I don't have a computer science degree. I never worked at a tech company. I didn't go through a bootcamp. The closest thing I have to a technical education is a semester of computer science at a CEGEP in Quebec City that I didn't finish. That was in 2002.

For the last 20 years, I've done everything except software development. Seven years in the Canadian Armed Forces doing signals intelligence. HR management for a group of companies with 90+ employees. Co-founded a food festival. Ran an event management business. And for the past year, I've been working as a logistics clerk at a lumber mill in the Beauce region of Quebec, helping produce 56 million board feet of SPF lumber per year.

In November 2025, I started building a software platform. Five months later, it had 192,000 lines of code, 96 components, 11 AI integrations across three providers, real-time video debates via WebRTC, payment processing through Stripe and PayPal, support for 11 languages, and a gamification system with 99 levels.

I built the entire thing with AI. Specifically, Claude Code. No co-founder. No development team. No funding. Just me, a laptop, and a tool that turns plain language instructions into working software.

This is the story of how that happened, what I learned, and why I think it matters for anyone who has an idea but thinks they can't build it.

The Idea

The idea behind ELBO had been bouncing around in my head for years. Not the technical details. The core insight.

AI is going to replace most screen-based work. Data entry, reports, translations, customer support, basic coding. Machines are already doing these things better, faster, and cheaper. Within a decade, most of what people do on a computer won't require a human anymore.

So what will matter? The skills that AI can't replicate. Critical thinking. Persuasion. Negotiation. Public speaking. The ability to defend an idea under pressure. The ability to tell someone something they don't want to hear and do it with empathy.

The problem is there's nowhere to practice these skills. You can read about negotiation. You can watch a TED talk about critical thinking. But reading about swimming doesn't teach you to swim.

I wanted to build a place where people could actually practice. A training ground for the skills that will matter most in the next decade. Not a course. Not a workshop. A live arena where you practice against AI opponents that challenge you, audiences that judge you, and a scoring system that tracks your improvement.

I had the vision. I had zero ability to build it. Until Claude Code changed that equation.

The First Week

I remember the first night. I sat down, opened Claude Code, and typed something like "I want to build a debate platform where users can argue with an AI that plays devil's advocate." I had no idea what a framework was. I didn't know the difference between frontend and backend. I didn't know what an API was.

Claude Code didn't care. It asked me clarifying questions, suggested a technology stack (Next.js, React, Supabase, Tailwind CSS), and started generating code. Actual, working code. I'd describe what I wanted in plain French or English, and it would build it.

By the end of the first week, I had a working prototype. A page where you could type an argument and an AI would argue back. It was ugly. It barely worked. But it was real. Something I described in words was now a thing on a screen that did stuff.

That moment changed everything. Not because the prototype was impressive. But because I realized the barrier between "I have an idea" and "I have a product" had just collapsed.

The Grind

The next five months were intense. I was working full-time at the lumber mill during the day. Nights, weekends, and every spare moment went into ELBO.

The process was always the same. I'd describe what I wanted in plain language. Claude Code would generate the code. I'd test it, find problems, describe the problems, and Claude Code would fix them. Sometimes I'd describe a feature that seemed simple and it would take three days of back-and-forth to get right. Sometimes I'd describe something complex and it would work on the first try.

I learned as I went. Not how to code, exactly. But how to think like a developer. How to break big features into small steps. How to describe what I wanted precisely enough for the AI to build it correctly. How to debug by describing symptoms instead of reading error logs.

I also learned what AI is bad at. Long-term architectural decisions. Understanding the full context of a large codebase after thousands of changes. Remembering design decisions made three months ago. Consistency across files. These required me to be the project manager, the architect, the quality controller. The AI was the builder, but I was the brain.

What Got Built

Here's what ELBO looks like today.

A live debate arena where users argue on camera via WebRTC (LiveKit), with an audience that votes in real time. An AI Devil's Advocate (we call him Einstein) that you can debate 24/7 using voice, powered by ElevenLabs text-to-speech and Simli animated avatars.

Eleven AI integrations across three providers. Claude handles the core debate analysis and scoring. Gemini powers alternative perspectives. Groq handles speed-critical operations.

A full gamification system. XP, 99 levels, an AURA score that tracks six skill dimensions (persuasion, clarity, logic, empathy, rigor, listening). Crews, alliances, leaderboards.

A three-currency economy designed from day one to comply with Quebec financial regulations (AMF). Pièces for cosmetics, Points ELBO for rewards, XP for progression. The currencies are deliberately separated so they never cross regulatory lines.

Payment processing through both Stripe and PayPal. A cosmetic shop with avatar skins and effects. Anonymous participation so visitors can interact without signing up.

Four interconnected universes. ELBO (the public arena), NOVA (education), APEX (corporate training), and VOIX (civic democracy). All unified by the ECHO profile system, a lifelong record of demonstrated skills that follows you from school to career to civic life.

Support for 11 languages. Server-side rendering for SEO. A daily content system (Tribunal du Jour) that keeps the platform active. And a 50% weekly profit redistribution model where half of all revenue goes back to active users.

192,000 lines of code. 96 components. Built by one person with no coding background in five months.

What I Learned

Building ELBO taught me things no business school or coding bootcamp ever could.

First, the barrier to building software is gone. Not lowered. Gone. If you can describe what you want clearly, you can build it. The skills that matter now aren't technical. They're clarity of thought, persistence, and the ability to manage complexity. These are human skills, not coding skills.

Second, AI is an incredible builder but a terrible architect. It will do exactly what you ask with impressive skill, but it won't tell you whether what you're asking for is a good idea. The vision, the strategy, the "why" behind every feature... that's still 100% human.

Third, solo founding with AI is both empowering and lonely. You can move incredibly fast. You can build things that would have required a team of ten just two years ago. But there's nobody to argue with, nobody to catch your blind spots, nobody to tell you when you're wrong. I've had to find that in other ways, including, ironically, arguing with AI about my own decisions.

Fourth, building the product is the easy part. Seriously. Getting people to use it is ten times harder than building it. I have a platform with 192K lines of code and essentially zero paying users. The marketing, the design, the positioning, the cold start problem... these are the real challenges, and AI can't solve them for you.

Why This Matters Beyond My Story

I'm not writing this to brag. Honestly, ELBO might not succeed. The cold start problem is brutal, the design needs work, and I'm one person trying to compete in a market where well-funded companies are adding AI features to their existing platforms every week.

But here's what I know for sure. If a logistics clerk at a lumber mill in rural Quebec can build a 192K-line SaaS platform in five months with AI, the old rules about who gets to build technology are finished.

You don't need a CS degree. You don't need to live in San Francisco. You don't need venture capital or a technical co-founder. You need an idea, clarity about what you want to build, and the stubbornness to keep going when things break at 11pm on a Tuesday and you have to be at the mill at 6am.

The tools exist. The excuses don't. Not anymore.

If you want to see what one person and AI can build together, check out ELBO.

And if you're sitting on an idea because you think you can't build it, you're wrong. You can. The only question is whether you will.