10 Things I Wish I Knew Before Building My First App with AI
If I could go back to the beginning — before I built my first app with AI, before I set up Kai as my co-founder, before any of this — here's what I'd tell myself.
These aren't theoretical tips. They're lessons I paid for with wasted time, broken builds, and a lot of "why didn't anyone tell me this?"
## 1. Start with a problem, not a feature list
My first instinct was to plan every feature before writing a single line. Authentication, dashboards, analytics, dark mode, animations — the works. I spent days on Figma mockups for an app that didn't exist yet.
What I should have done: pick ONE problem, build the simplest solution, and ship it. I learned this eventually with Hũ Vàng — my wife needed to track gold prices. That's it. One problem. One solution. Everything else came later.
## 2. AI doesn't know your users
AI can write code, generate copy, and suggest features. What it can't do is sit in your living room and watch your wife get frustrated trying to calculate gold profits in Excel. That observation — that real human frustration — is worth more than any AI-generated market analysis.
Design skills matter more than you think in this process. Understanding people isn't something you can delegate.
## 3. Don't let AI write your voice
Early on, I let AI write blog posts without enough direction. They were fine. Professional. Polished. And completely soulless. Nobody wants to read another AI-generated blog post that sounds like every other AI-generated blog post.
Now I brief Kai with my angle, my stories, my opinions. Then I edit everything to sound like me. The AI does the heavy lifting. The human does the soul.
## 4. Your design skills are your superpower
I spent weeks feeling insecure about not being a "real developer." Then I realized: most developers can't design. They can't look at a screen and feel what's wrong. They can't create an experience that connects emotionally.
The gap between designers and developers is closing — but designers have the advantage. We bring taste. That's the one thing AI can't generate.
## 5. Ship ugly, improve later
My designer brain wanted everything pixel-perfect before launch. Beautiful typography. Perfect spacing. Consistent color system. Meanwhile, the app didn't even work.
Now I follow one rule: make it work, make it useful, then make it beautiful. In that order. Always.
## 6. Free tier everything until it hurts
I'm running four apps and a blog on essentially zero hosting costs. Firebase free tier. Vercel free tier. Cloudflare R2 free tier. You don't need to spend money to ship.
The moment you start paying for things before you have users, you're playing pretend. Build with constraints. They force creativity.
## 7. Document everything — it becomes content
Every problem I solved, every mistake I made, every "aha moment" — I started writing about all of it. That documentation became this blog. The blog became my marketing. The marketing brings people to my products.
The work you're already doing is content. You just have to write it down.
## 8. AI amplifies your weaknesses too
AI made me faster. It also made me lazier. When Kai can write a blog post in 3 minutes, it's tempting to just hit publish without reading it. When AI generates code that compiles, it's tempting to ship without understanding it.
Speed without judgment is just faster failure. Learning with AI means reading everything, questioning everything, and deciding everything yourself.
## 9. Nobody cares about your tech stack
I spent way too long choosing between Firebase and Supabase. Between Vercel and Netlify. Between this framework and that framework. Users don't care. They care if the thing works and solves their problem.
Pick something. Commit. Build. You can always migrate later when you actually have users who would notice.
## 10. The hardest skill isn't building — it's marketing
I can build an app in a weekend now. What I can't do easily is get people to use it. Marketing is my weakest point, and it's the most important point.
If I could restart, I'd spend 50% of my time building and 50% telling people about it. Not the other way around. Build in public. Write about your journey. Share your failures. The building is the easy part now — attention is the hard part.
## Bonus: Start before you're ready
I wasn't ready to launch a blog. I wasn't ready to ship apps. I wasn't ready to sell services on Fiverr. I did all of it anyway, and every single one taught me more than another month of preparation would have.
You don't need permission to build. You just need to start. The rest figures itself out along the way.
Currently at 4 apps shipped, 0 revenue, and more lessons than I can count. Building in public one mistake at a time.