How I Built 4 Apps Without Writing Code — A Designer's Journey with AI
I spent ten years making things look good for other people. Photoshop, Illustrator, wireframes, revisions, more revisions, "can you make the logo bigger" revisions. That was my life. A good life, honestly. Predictable. Comfortable. Suffocating.
I knew some HTML and CSS. Enough to break things. Not enough to fix them.
The idea of building actual software — like, real apps that real people use — felt like something that belonged to a different species. The species that understood what async await means without Googling it every single time.
And yet. Here I am. A designer lost in the land of code, with four projects I built from scratch. No CS degree. No team. No idea what I'm doing half the time.
This is that story.
## The itch
You know that feeling when you design something beautiful, hand it off to a developer, and get back something that looks like it was assembled during a turbulence? That was my career.
Every handoff felt like putting a message in a bottle. You throw it into the ocean and hope someone on the other side understands what you meant. They never do. Not really.
One night — I think it was 2 AM, the hour when bad ideas sound brilliant — I thought: What if I just build it myself?
The logical part of my brain laughed. The stubborn part opened a terminal.
## Hũ Vàng: The one that proved it was possible
Hũ Vàng tracks gold prices and stocks in Vietnam. Real-time data. Profit/loss calculations. The kind of app that sounds simple until you actually try to build it.
I didn't know React. I barely knew JavaScript. But I had Claude, an AI that doesn't judge you when you ask "what is a useEffect" for the fourteenth time.
Here's the thing nobody tells you about building with AI: the code is the easy part. The hard part is still the same hard part it always was — figuring out what to build and why. Which data sources to trust. How to organize information so a human brain can actually process it. How to make something that doesn't just work, but feels right.
Ten years of design taught me all of that. I just never had the tools to use it.
Now I did.
## Flashbee: The one that humbled me
Flashbee is an English learning app for Vietnamese kids. Spaced repetition, illustrated vocabulary, text-to-speech. I'm still building it. It's nowhere near done. Some days it feels like it'll never be done.
This project taught me something important: AI is a multiplier, not a magician. You give it a smart idea, you get smart code. You give it a dumb idea, you get dumb code really, really fast.
I spent weeks on the learning algorithm. Not coding it — thinking about it. How do kids actually memorize words? When should a card repeat? What makes a six-year-old want to open an app again?
The AI wrote the code in minutes. The thinking took weeks. Nobody talks about that ratio, but it's the real one.
## VocabVault: The one that scared me
After building things people can see, I went somewhere I couldn't see — the backend. VocabVault is a vocabulary illustration library. An API. Flashbee uses it as its core image source.
API design. Database architecture. Storage optimization. Authentication. These words used to make me physically uncomfortable. Like hearing a dentist describe a root canal in detail.
But here's what I discovered: you don't need to understand everything before you start. You need to understand enough to ask the right questions. And guess what designers do every single day? Ask the right questions. "What does the user need here?" is the same muscle as "What should this API endpoint return?"
Same skill. Different costume.
## SCOUT: The one that got meta
SCOUT is an AI tool that analyzes market opportunities. I built an AI to help me figure out what to build with AI. Yes, I see the irony. No, I don't care. It's useful.
It's a personal tool — not public, not polished. Just something I use to evaluate ideas before spending weeks on them. The kind of tool you build when you realize your biggest risk isn't bad code. It's building the wrong thing.
By this point, my workflow was something like a recipe:
- >Think about the user experience. Really think.
- >Break it into pieces small enough that each one makes sense on its own.
- >Let AI write the code.
- >Read the code. Not to understand every line, but to understand the shape of it.
- >Ship it. Even if it's ugly. Especially if it's ugly.
## What I know now that I didn't know then
Design skills are building skills. Product thinking. User empathy. Information architecture. Visual hierarchy. The industry spent decades calling these "soft skills" and putting them in a separate box from "real engineering." That box was always artificial. AI just made it obvious.
AI doesn't replace thinking. It replaces typing. The distance between "I have an idea" and "I have a working prototype" got shorter. The distance between "I have a bad idea" and "I have a good idea" didn't move an inch.
Not everything needs to be public. Some projects are personal tools. Some are half-finished experiments. That's fine. You're building. You're learning. You don't need an audience for that to count.
Ship ugly. My first versions of everything were embarrassing. They worked, though. And a working ugly thing beats a beautiful thing that only exists in Figma.
The lonely parts are still lonely. I can build now. Marketing, selling, growing — those are still hard. Building the product is the fun part. Everything after is the actual job. AI helps with some of it. But sitting with the discomfort of putting yourself out there? That's still a human problem.
## So what now?
I'm writing about all of it. The tools, the workflow, the mistakes, the small wins that keep you going. If you're a designer wondering whether you could build something real — you can. I'm proof that the bar is lower than you think. The only thing high is the stubbornness required.
No hype here. No "AI will change everything" speeches. Just a designer, lost in the land of code, figuring it out one project at a time.
If that sounds like something worth following, subscribe to The AI Builder. I write weekly about building products with AI — the honest version, not the LinkedIn version.
And if you ever want to talk, I'm around. Preferably over text. I'm much better at writing than talking. Obviously.
Sơn is a designer who accidentally became an indie maker. He builds things with AI from Vietnam and writes about it at gianghaison.me.