AI Can Build You a Product in Hours. Selling It Is a Different Question

What shows up in my feed has changed.

It used to be vacation photo grids. Now someone regularly posts a small tool or site they just launched, captioned "built by AI in one night."

You click in. Clean page, working features.

The comments are usually a few likes. Nobody asks how to use it, and certainly nobody asks the price.

I have done this myself, and got stopped cold once.

Last month I built a small tool with AI and sent it to a friend.

My first sentence was about what stack it used and how much work it saved.

He replied: what is this for?

I froze. Everything I said was about what I accomplished. Not one word about what it does for you.

This is not my personal defect.

Engineers are trained to lead with what they have: which framework, how many milliseconds shaved, how clean the code is.

That script works on peers. It does nothing for someone about to pay.

They do not care how you did it. They care whether it solves their specific annoyance, and whether it is worth the price.

I have seen two kinds of introductions. One opens with architecture and response times. The other opens with what this saves you and how soon you can use it. The first reads professional. The second reads like talking to a person. People who know the field care how it was done. People who pay only care whether it was done.

The barrier to building something usable is far lower than a few years ago.

People who cannot code can assemble a site or tool in hours with AI. That part is true.

But "sells" was never the same question as "can be built," and AI never intended to answer it for you.

betterlaunch.co, which makes site-building tools, ran a community survey of indie developers — not an authoritative institution, self-collected community data, so treat it as roughly right: about half of indie developers still sit in the 0 to 1,000 US dollars per month band, many of them with products launched recently.

Jiemian News made a similar point: AI makes products easier to build, and also makes indie developers easier to drown out.

A lot of things used to die at "cannot build it." Now a lot of things die at "built it, nobody saw it."

Lowering the barrier solved who can sit at the table. It did not solve who can play the hand.

Back to the moment I got stopped. What I eventually understood is that my friend was not failing to grasp the tech. I was standing in the wrong position.

I was grading myself on "did I do this job elegantly" instead of "is this job useful to him."

Those two standards are far apart: the answer to the first always lives on my side, the answer to the second always lives on theirs.

The deeper your technical knowledge, the easier it is to sink into the first standard, because it is the only place you know how to score.

Same for freelance work and indie products. A quote sheet packed with your stack is usually a signal that this question has not been thought through — not that the engineering is weak.

Two quote sheets, one leading with the technical plan and one leading with what it solves and how fast it ships, get replies at different speeds. That is my judgment from the last few years, not a controlled experiment, and I am not dressing it up as data.

The idea is not hard. The hard part is remembering to use it before you start.

So I set myself a crude rule: when something is finished, before sending it out, cross out every technical term in the description and see what still convinces.

If nothing is left, the problem is not in the code. It is that I have not worked out who it helps.

If one plain sentence survives, then the tech can come on stage.

The "built by AI in one night" posts will keep coming.

AI can compress building to a few hours. Working out who it is for and what it solves cannot be compressed — you still have to go stand on the other person's side and think it through.