What goes wrong.
Each of these cost me a day or a week. They cost you four minutes.
I run a live business built this way. Payments, members, systems running around the clock, close to half a million lines, none of it typed by hand. Most of this I learned by getting it wrong first.
1. Starting too big
Your first build is not a project, it is a proof. Its only job is to exist by the end of the evening.
Instead: no login, no saved data, no other company's account, no money. It will sound too small. That is the point.
2. Not knowing where the money goes
A page that looks right is cheap. The moment it has to remember something, let people sign in, or take a payment, you can spend more than on the whole rest of the project.
Instead: expect the version that saves data to cost more than everything before it. That is the shape of the work, not a mistake.
3. Asking it to fix the bug
"Fix the layout" gets a guess. "On my phone the two buttons sit on top of each other" gets a fix. It cannot see your screen. You are its eyes.
4. Changing five things at once
When it breaks after a five part instruction, you do not know which part did it, and neither does it. One change, look, next change. Feels slower, is faster.
5. Arguing with it
Sometimes it loops, fixing one thing and breaking another. Three rounds is the signal. Ask it to throw that part away and build it again, described better. Starting over costs minutes. Arguing costs an evening.
6. Building something that holds other people's money
One category to leave alone at the start, and it is not about difficulty. It is about who pays when it goes wrong. Other people's money. Logins for strangers. Health details, card numbers.
Instead: keep paying for those. The small stuff is what is now yours.
7. Trusting it because it looks finished
AI output looks confident whether or not it is right. A calculator with the wrong formula looks exactly like one with the right formula. Test it on a case where you know the answer.
8. Trying to finish it
Scope kills more first projects than any technical problem. You start with a price list and by Thursday it has logins, a newsletter and a members area, none of which work. Ship small, use it a week, add one thing.
9. Stopping after one
The first build teaches you almost nothing except that it is possible. The second is where you get good. Most people who quit quit after one, usually because of number one on this list.
Start smaller than feels worth doing. Describe what you see, not what you want. One change at a time. Nothing that holds somebody else's money. Check the maths yourself. Then do it again.
Where to start
If you have built nothing yet: a price list. An hour, prints to a PDF, and nothing about it can go wrong.
A price list or menu 1 hour$12 a monthOr pick the one that matches your week.
Back to the start