How I
build things.
AI changed how quickly I can turn an idea into software. So my process isn't about spending months building first. It's about deciding what deserves to be built, building small, testing early and learning from what happens.
AI made building easier. It didn't make choosing what to build easier.
That's the interesting part.
When software becomes cheaper and faster to make, the bottleneck moves upstream. The question becomes: is this actually worth building?
That's why I care about small products, quick experiments and putting things in front of real people instead of trying to predict everything from a laptop.
Question
Something feels unnecessarily difficult, someone has a problem, or an idea keeps coming back. Before writing code, I try to understand the actual problem.
Stress-test
Is this a real problem? Who has it? How are they solving it now? Does an app actually make sense? This is where an idea earns the right to become a build.
Build small
I build the smallest useful version I can. Not every feature. Not every edge case. Just enough product to make the idea real and testable.
Test
Then I try to break it. I test the flows, look for confusing moments, check the experience on mobile and pay attention to what doesn't work the way it should.
Ship
If it's useful enough to put into the world, I launch it. A live product creates better feedback than a perfect product sitting in development.
Learn
What people click, ignore, ask for, struggle with or actually use becomes the next piece of information. Then the loop starts again.
AI is part of my building stack.
I use AI to move faster — from thinking through product ideas and generating code to debugging, writing copy, exploring interfaces and working through problems. But AI doesn't decide whether something deserves to exist. That's still the human part.
Build faster. Think harder.Small beats bloated.
I would rather build one useful thing properly than bury a simple problem under twenty unnecessary features.
Test before guessing.
Real usage is more useful than endlessly imagining what people might want.
Shipping starts the learning.
Launching isn't the end of the process. It's when the product finally starts giving you information.
Not every idea
gets built.
And that's a feature, not a failure. Some ideas are better as experiments. Some need more validation. Some turn out to be unnecessary. The point of the process is to find that out before too much time and money disappear into the build.
Curious about the things that made it through the process?
Explore the apps →