I need to tell you something that might sound familiar.
In the last few months, I've built 10 pieces of software. A baby gender predictor. A sparkling tea e-commerce site. An AI school platform. Landing pages. Dashboards. Tools.
I'm not a developer. Not an engineer. I don't have a CS degree. I work in business — specifically crypto exchange listings. I'm just a regular business guy who discovered that AI could build things for me.
And yet, every single one of these products works. They're live. Real URLs. Real Stripe checkouts. Real users can sign up, pay, and use them.
And not a single one is making meaningful money.
When I saw Mo (@atmoio) post "I was a 10x engineer. Now I'm useless" — a video that racked up 10,700 likes and 1,100+ replies — I felt something I didn't expect: recognition. Not from the engineering side. From the other end of the same problem.
Because here's the uncomfortable truth that nobody is talking about:
If a 10x engineer feels useless because AI can do what they do — what happens when millions of business people like me discover that building software was never the hard part?
Most people don't appreciate how vertical this curve has gone. Let me walk you through it:
The World Economic Forum projects 92 million jobs displaced but 170 million new jobs created by 2030 — a net gain of 78 million roles. But here's the thing those numbers don't tell you: the new jobs look nothing like the old ones.
Bloomberg ran a piece in February 2026 headlined "AI Coding Agents Are Fueling a Productivity Panic in Tech." A research paper called "Vibe Coding Kills Open Source" argues the entire open-source ecosystem is being undermined. The New York Times declared "The AI Disruption Has Arrived, and It Sure Is Fun."
Fun for whom, exactly?
Here's what I learned the hard way from building 10 products that nobody's buying:
Building was never the competitive advantage. Distribution was. And it still is.
CodeConductor nailed this in their 2026 predictions:
"As AI and no-code remove friction from building, they add friction everywhere else. By 2026, launching a product is trivial."
Read that again. Launching a product is trivial.
That means my 10 apps aren't impressive. They're expected. Anyone with a laptop and ChatGPT can do exactly what I did. The question was never "can you build it?" The questions that actually matter are:
I used to think non-coders like me had cracked the code (pun intended). We could finally compete with engineers! The playing field was level!
But here's what actually happened: the playing field didn't level. It shifted. The game is no longer "who can build the best product." It's "who can solve the best problem for people who will actually pay."
And that requires something AI can't give you: understanding of real human beings in real situations.
I'm not going to sugarcoat this. "AI will take your job" is too simple. "AI won't take your job" is too naive. The truth is somewhere more uncomfortable.
The common thread? Every safe job requires judgment in ambiguous, messy, real-world situations with real consequences. AI is brilliant at pattern matching. It's terrible at navigating the politics, emotions, and chaos of actual human business.
Spending 6 months learning Python when AI writes better Python than 90% of juniors is like learning calligraphy to compete with a printer.
Instead, learn AI fluency: how to direct AI tools, evaluate their output, and combine them into workflows. The new literacy isn't code — it's knowing how to get AI to do what you want.
My 10 apps taught me this the hard way. The skills that matter now:
These are the skills AI cannot replicate, because they're fundamentally about human trust.
McKinsey found 88% of organizations already use AI. The interview question of 2027 won't be "do you know Excel?" It'll be "show me how you use AI agents in your daily work."
Making things is now free. Everyone can make things. The scarce resource is understanding what people actually need badly enough to pay for, and being able to reach those people and convince them you're the one to solve it.
Building software will be like making a PowerPoint — everyone can do it, nobody is impressed.
The value chain shifts from technical execution → problem identification → distribution → trust building.
New monopolies form around proprietary data, customer relationships, and brand trust — not around who has the best engineers.
The biggest winners won't be the best builders or coders. They'll be the people best at understanding human problems and building trust at scale.
After all of this — 10 apps, zero meaningful revenue, and a crash course in what AI can and can't do — I made a decision that might seem counterintuitive.
I started a physical product business.
FLUX Sparkling Tea — a premium jasmine sparkling tea, cold-brewed in small batches right here in Hong Kong.
People ask me: "You just spent months learning to build software with AI. Why on earth are you making tea?"
And the answer is everything I just wrote above.
Because AI can't brew tea. AI can't hand someone a cold bottle at a popup market and watch their face when they taste it. AI can't build the kind of brand trust that comes from showing up in person, from letting people see you, touch your product, hear your story. AI can't navigate Hong Kong food licensing. AI can't do a tasting event at a bar where everyone's choosing between alcohol and nothing.
Software is infinitely replicable. Someone sees your app, they can rebuild it in a weekend. I know — I did it 10 times.
But a physical product has natural moats that digital products don't:
I'm not saying physical products are easy. They're not. They're messy, expensive, and slow compared to shipping code.
But that's exactly the point.
In a world where everyone can build software, the competitive advantage might be building something they can't. Something that requires you to show up. Something that exists in the real world. Something people can hold in their hands.
The digital gold rush is so crowded that millions of AI-empowered builders are all competing for the same eyeballs, the same app store rankings, the same Google search results. Meanwhile, the physical world — local markets, food & beverage, tangible products — is still a place where hustle, taste, and human connection actually win.
Maybe I'm early to this realisation. Maybe I'm wrong. But after building 10 apps that nobody bought, I'd rather bet on something AI can't copy than keep competing in a space where everyone has the same superpowers.
The question isn't whether AI will take your job.
The question is: what will you build that AI can't?
For me, the answer is a bottle of sparkling jasmine tea, brewed in Hong Kong, handed to you with a smile.
Sources: World Economic Forum "Four Futures for Jobs 2030", McKinsey Global Institute, Bloomberg "AI Coding Agents Fueling Productivity Panic" (Feb 2026), MIT Technology Review "Generative Coding" (Jan 2026), NYT "The AI Disruption Has Arrived" (Feb 2026), CodeConductor "AI & No-Code Predictions 2026", @atmoio viral post (10.7K likes, Mar 2026), Wikipedia "Vibe Coding" (Jan 2026 paper)
Are you building in the digital gold rush, or looking for less crowded ground? I'd love to hear your take.