Now That We Can Do Everything With AI, This Skill Is The Most Important
It’s not prompting…
My Medium friends can read this story over there as well.
We’ve reached the strange part of the AI curve.
The part where the hard things got relatively easy.
Building apps. Designing logos. Writing code. Creating landing pages. Writing blog posts. Heck, writing books. Even drafting legal docs or pitch decks.
All suddenly doable in a couple of days, an afternoon, sometimes even in an hour, often for free.
That’s not the future anymore. That’s any given Tuesday now.
But because of that, something flipped. Completely. The most important skill is no longer knowing how to do things.
It’s knowing what is actually worth doing.
And as it turns out, that’s way harder…
The Death of “How”
For decades, leverage came from execution skills.
If we knew how to code, we had power.
If we knew design, we had leverage.
If we knew marketing funnels, SEO, copywriting, we were valuable.
AI just commoditized much of that for many people in most situations.
Today, we can:
Scaffold an app with a prompt
Generate a design system in minutes
Spin up a SaaS backend without touching a server
Ship something that looks “professional” before lunch
Execution didn’t disappear. But it got cheaper. Abundant. Faster.
And when something becomes cheap, the bottleneck moves upstream.
Taste, Judgment, and Direction
What’s scarce now is not ability. It’s judgment.
AI can write hundreds of blog posts in minutes. But are they actually good? That’s judgment. And AI isn’t great at that
The most important skill now, is knowing:
Which problems are real
Which niches are underserved
Which ideas look obvious in hindsight but invisible beforehand
Which things AI makes possible that weren’t viable before
What to build that people actually want
AI is a multiplier. But it multiplies whatever we feed it. Garbage idea in, fast garbage out.
Everyone Can Build Now
But few people know what to build.
It’s kind of like dressing well. Anybody can buy a bunch of clothes. Doesn’t them stylish.
And it’s so obvious in the digital space.
Thousands of products. Polished. Well-designed. Dead on arrival.
Not because the founders are lazy. Or dumb. Or bad at execution. But because they answered the wrong question.
They asked: “Can we build this?”
Instead of: “Should this exist at all?”
AI made it easy to answer the first question. It did nothing for the second. Unless you assist with the research.
Finding Ideas That AI Didn’t Kill (Yet)
Ideas that:
Sit close to real human pain
Are annoying, boring, or unsexy
Live in small niches big companies ignore
Are too specific to be obvious
Are too messy for generic tools
Have been only done badly so far
Have only been done for one market, not more
It’s not easy to find these things.
Niches Over Markets, Always
Big markets are loud and crowded. Small niches are not. They’re often underserved.
AI shines when we bring it into:
Admin hell
Life logistics
Paperwork
Compliance
Transitions
We don’t need AI to build yet another notes app or habit tracker. We have too many of those.
We need AI and people to solve problems small niche companies, workers, services have. And build solutions for them.
“What For” Beats “How To” Every Time
Ask ten people what they’re building and most will explain how. Tech stack. Frameworks. Models. Tools. APIs. That’s not important anymore.
The better question is:
What job does this do?
What pain does it remove?
What friction disappears?
What becomes faster, calmer, or less stupid?
Nobody cares that we used Flutter, Firebase, Anti-Gravity, or a fancy model.
They care that their problem went away.
AI makes it tempting to overbuild and underthink.
Why Original Ideas Look Boring at First
Every good idea starts slightly underwhelming.
If it sounds exciting immediately, it’s probably obvious. If it sounds boring, specific, or oddly narrow, that’s a good sign.
“An app that helps freelancers manage VAT deadlines in Germany” Not sexy. Very useful.
AI lets us build this quietly, cheaply, and test fast.
Builders vs. Choosers
On one side:
Builders who can make anything
But don’t know what to make
Shipping endlessly, learning little
On the other:
Choosers who understand people
See friction
Pick problems carefully
And then use AI to execute
This Skill Compounds
This skill compounds harder than coding ever did.
If we get good at:
Spotting broken workflows
Understanding incentives
Identifying neglected niches
Asking better questions
AI keeps getting better at execution underneath us. Which means our leverage keeps increasing. The “what” skill ages well. The “how” skill expires fast.
The Bottom Line
AI didn’t make creativity obsolete. Not at all. It made choice the core skill.
We can all build now. That’s not special anymore. What matters is deciding what deserves to exist. Finding the right problem. The right niche. The right reason. That’s the real work.
And for now, it’s the one thing AI still can’t do for us entirely. It can help us with research, but if the problem isn’t obvious enough, AI won’t find it just yet.
We do.
