What a product costs now that the code is cheap.
Running a live software product costs less every month than a phone line. That changes the arithmetic on whether to build one. It does not touch the part that decides whether it works.
Gulf Coast Bite is a live software product built and operated by this practice. It scores fishing conditions across six waters on the Gulf Coast, publishes to Instagram and Facebook on its own schedule, and sends a weekly email written separately for each water. It has a trip planner, a charter directory, a community catch board, and about fifty pages of guides.
Most people assume a thing like that carries what it would have carried in 2020: a developer, a designer, and a retainer that renews whether or not anything ships. It does not, and the gap is not a small one.
What the arithmetic used to be
A software product used to require funding, or at minimum a technical cofounder willing to work for equity. Hosting cost money. Scheduled jobs cost money. Sending email at any volume cost money. Every one of those was a monthly line item with a person attached to it.
None of that is true at the scale a small business product actually operates. Hosting, scheduled jobs, image generation, form handling, email to a real list, and the AI doing the daily production work all sit at or near the entry tier for this kind of traffic, and most small business products will never outgrow it. Advertising is a separate decision with a separate budget, and it is the only line that behaves the way it used to.
The cost of building software has collapsed. The cost of knowing what to build has not moved at all.
What did not get cheaper
The real expense was never the infrastructure. It is the judgment about what to build, what order to build it in, and what to leave out.
Gulf Coast Bite has had more removed from it than added. A regulations section came out because the homepage was getting busy. A red tide box came out. The trip planner was rebuilt twice and renamed once. Every one of those decisions took longer than the code did, and every one of them mattered more.
That is the part that does not get cheaper, and it is the part nobody quotes you a price for. An AI will build whatever you describe, quickly and competently, including the wrong thing. It will build the wrong thing beautifully, and it will not mention that it was the wrong thing.
What this means if you are weighing a product
A working product, in market, operating itself, is now within reach of a business that could not have considered one two years ago. That is a real change and it is worth taking seriously.
What has not changed is that most product ideas are not worth building. The cheap part got cheaper. The expensive part is still expensive, and it is still the part people skip.