These past six months, plenty of friends have known Aubzhang was messing around with AI. Then yesterday he pushed out the "Wanci Zilaiye" app, and the WeChat messages really started.
Some asked: "You built an app with AI — you quietly making money off it?" Others asked: "I want a side hustle with AI too. Which programming language should I learn first?"
Let Aubzhang come clean first: so far, Aubzhang has not made a single cent from AI.
Aubzhang has always thought of AI as a powerful production tool. A lever. What actually makes you money was never AI itself — it's the scarce value you create with AI, and that value happening to be recognized by a market. Without scarce value, even the best AI just mass-produces industrial garbage nobody wants.
And yet short videos online talk AI up like magic. As if typing "build me a Taobao" into a screen means financial freedom by tomorrow. Faced with all that money-anxiety on the feed, Aubzhang finds it a bit absurd — but he gets it. Everyone's afraid of being left behind by the times.
Still, Aubzhang wants to throw one bucket of cold water: don't rush to make money. If you don't fix your mindset first, if you don't understand what real demand looks like, AI will only drive you up the wall.
Two years of poking at AI, plus months of grinding out actual products — the biggest takeaway for Aubzhang is this: for ordinary people trying to do something with AI, the first task really isn't learning to code. It's mental preparation. Aubzhang sorted through the pits he fell into and pulled out three key points for anyone about to get on board.
1. Drop the Wishing-Well Mindset. Be the Foreman.
Most people use AI badly because they treat it like a wishing well in a temple. Toss a coin, make a wish — "build me a useful tool" — then sit and wait for pie from the sky.
And what happens? AI does spit out a pile of stuff. You look at it and it's nothing like what you pictured.
Why? Because you treated AI as a god. In reality, AI is a brilliantly smart intern with zero opinions of its own.
When Aubzhang first used AI for development, he made the same mistake. Tossed over one line: "optimize this interface." AI froze, then produced a mess of random color blocks.
Later Aubzhang learned. You have to be the foreman. You don't lay bricks yourself (write code), but you must hand the intern a clear drawing:
- Want a bookkeeping app? You can't just say "make an expense app."
- You have to say: "Page one, an input box, bold font. Page two, group by month, meals separate from rides. Overall palette: light green."
- You even have to say: "No ads. Absolutely no ads. Only the most basic features."
You figure out the details. AI produces them. The better you know the specific business, the more accurate AI's work.
2. Your "Not Knowing Tech" Is Actually Your Edge
Plenty of beginners want to sign up for a Python course on day one, convinced that without code they can't use AI. Aubzhang thinks that's digging your own pit.
Aubzhang self-studied Python for a while too. Enough to know a few basics. But when it came time to actually write, he had no idea where to start. Two lines in and his head spun. One error log and his head hurt. Honest truth: ask Aubzhang what language the product he shipped is written in, and he genuinely doesn't know. AI handled all of it backstage. All Aubzhang did was feed it business requirements.
And yet his primary-school daughter looked up a few words and said: "Dad, reading it by roots, it's easier to say out loud."
Why? Because what decides whether a product is good was never how pretty the code is. It's whether you understand the user's pain.
Aubzhang knew the thousands of roots on the market were watery, so he cut to 171. Aubzhang knew stiff robot voices are unbearable, so he stuffed in 20,000 good audio clips even at the cost of a bigger installer. Aubzhang hates shady software himself, so this app has zero permissions — it doesn't even connect to the network.
Is any of that a technical problem? No. That's business instinct accumulated over a dozen years as a product manager.
In the AI era, technology got democratized. AI can cover your coding gap. Nobody can cover your "understanding the business" gap. You know the user, you know the pain — that's your hardest moat.
3. Let Yourself Fall in Pits. Some Dumb Work Can't Be Skipped.
People today chase "fast" too hard. One-click generation, overnight riches. The moment an error pops up, the moment AI doesn't get plain language, they curse and quit.
Aubzhang's first week on his first product, he stayed up past 2 a.m. every night. Half the time debugging code. The other half, arguing with AI.
To package that app, he had to copy 20,000 audio files. Started with the laziest copy command. Twenty-one minutes later, the write count was zero. The machine froze. Aubzhang nearly put his fist through the keyboard.
Then he dug for half the night and switched methods. To this day he doesn't know the principle behind it — only that AI calls it a "streaming pipe." AI said it would work, he did it, 13 seconds, done. Black cat or white cat, the one that catches the mouse is a good cat.
And those nights when a wrong parameter or a missing symbol crashed the page — those were all real pits, actually stepped in.
Some dumb work cannot be skipped. Pay tuition early. Fall in a few more times and you'll actually learn AI's temper. If you can't even handle the basics, what makes you think AI will carry your side hustle?
Final Thoughts
Aubzhang isn't writing this to talk anyone out of it. He wants you to settle down.
AI is not Aladdin's lamp. It's an extremely powerful lever. But how much a lever can lift depends on how solid your fulcrum is — and that fulcrum is called "business understanding."
No rush. Take it slow. Get the mental prep done, then we move.
And since we're no longer spectators tossing coins into a wishing well — what weapon do we pick up? Next episode, Aubzhang talks about the tool he finds easiest to use with the lowest barrier to entry.