From Zero to 6 Apps in 6 Weeks: The Honest Retrospective
I built 6 educational apps in 6 weeks with AI. Here's what the tool can do — and what it can't do for you.
Six apps. Six weeks. One solo developer. And a question I'm still sitting with: was it the right call?
The honest retrospective — not the euphoric Twitter thread about "the vibe coding revolution."
What happened
Between January and February 2026, I built six educational apps for the French market: citizenship exam prep (CitoyenApp), driving theory (PermisB), HACCP, first aid/SST, electrical safety certification, and TCF/TEF for immigration candidates. Real niches. Verifiable demand. Users who search for these courses every day.
Main tool: Claude Code. Stack: Next.js, Supabase, Vercel, Tailwind. Each app deployed as a PWA — Progressive Web App — to avoid App Store review delays and stay agile.
The speed result: real. What would have taken a small team six months happened in six solo weeks. Core functionality, interfaces, quiz flows, progression systems — AI absorbed 80% of the repetitive technical load.
The impact result: more complicated.
What AI can't do
Building an app is solving a technical problem. Finding users is solving a distribution problem. These are two different problems, and AI only solves the first one.
I assumed product quality would generate organic traffic. Classic mistake — and one that's documented everywhere. The rule that applies: app stores promote apps that already have traction, not apps that just launched. The algorithm distributes toward apps with reviews, installs, momentum. A new app starts with a structural disadvantage, not a blank slate.
This isn't a critique of AI. It's a fundamental reality of digital distribution that I should have integrated into the strategy before building.
The PWA trap
Choosing PWAs was logical for deployment speed. It was a strategic mistake for distribution.
A PWA doesn't appear on the App Store. It doesn't appear on Google Play. These are two channels that concentrate millions of daily searches on terms like "French citizenship exam" or "HACCP certification prep." Without presence in these stores, you're invisible to 70% of natural discovery channels.
The only alternative: organic web SEO — which takes six to twelve months to produce meaningful results on competitive keywords — or paid advertising, which completely transforms the project's economics.
I should have anticipated this. PWAs are a fast deployment tool, not a distribution tool.
The portfolio paradox
Six apps simultaneously was a diversification strategy: if one fails, others compensate. In theory.
In practice, it means 1/6th of the attention on each product. Marketing — social media, SEO, partnerships, feedback loops with early users — requires time and repetition. Six products at once means six simultaneous launch cycles, six audiences to build at the same time.
The data on the "app factory" model is clear: it works for experienced operators who have systems in place, existing audiences, and first products already on autopilot with stable revenue. That wasn't my situation.
The sequential strategy — one product, real traction, then the next — is slower short-term. It's more solid medium-term.
What the numbers actually say
The median for an indie app: under $50/month after a full year. Only 17% of apps ever reach $1,000/month. The top 100 subscription apps capture 81% of all subscription revenue. The concentration is extreme.
This isn't to discourage — it's to calibrate expectations. AI compressed build time by 80%. It didn't change the fundamental economics of mobile app monetization.
The real calculation: how long does it take to reach 1,000 active users in a niche like French citizenship exam prep? With organic SEO: 6 to 18 months. With App Store presence and well-executed ASO from day one: potentially less. Without either: indefinite.
What I'd do differently
One product first. CitoyenApp — French naturalization exam prep — is the most differentiated and the most urgent for its users. It deserved all the attention.
Native app from the start. The PWA stays useful as a prototype, but the App Store version needed to be planned in the initial scope, not added as an afterthought.
ASO before launch. Stores give new apps an algorithmic boost in their first weeks. If the listing isn't fully optimized at publication — title, keywords, screenshots, icon — that window is lost. It's an irreversible decision.
Distribution budget planned upfront. Not after building. The distribution budget needed to be in the initial roadmap, not as a response to "nobody's coming naturally."
What AI actually changed
Despite all of this, something happened that I couldn't have done before.
In 2024, building six educational apps in six weeks solo was impossible — not just in terms of code, but in terms of UX design, quiz systems, structured content, user state management. AI absorbed the technical complexity and let me focus on product decisions.
The real shift: the bottleneck moved. Before, the question was "can I build this?" Now, the question is "can I distribute it?" These are very different problems — and the second one doesn't get solved with a better prompt.
For solo developers looking at AI tools with excitement: the build speed is real. Take it. And plan distribution with the same rigor you plan architecture.
The code is 20% of the work. The remaining 80% has no AI shortcut — not yet.
Felipe Díaz Marín has twenty years of hospitality operations experience across Chile, Malaysia, Spain, and France. He is a lecturer in organizational leadership, marketing, and entrepreneurship at CY Cergy Paris Université, and advises hotel and F&B teams on operational transformation. Based in Paris.