AI lets you prove software. That changes who should decide.
A working prototype used to come after the budget. Now it can come before. What that means for a council director, a strata manager or a school business manager.
Twenty years ago software companies built first and looked for customers afterwards. Five years ago the minimum viable product reduced the spend, but it still required a build before a buyer. Today a credible, clickable prototype with a mobile experience, a dashboard and an assistant can be produced in days.
That changes the order of events. Identify the business problem. Prove it with a prototype. Show it to the person who owns the outcome. Get agreement. Then, and only then, build the production version.
The wrong claim and the right one
"AI lets me build software" is the wrong claim. Plenty of tools generate code now, and the code they generate is not what an organisation's IT function will accept into production. Identity, integration, security, governance and support are still engineering work, and they still have to be done properly.
"AI lets me prove software" is the right claim. The prototype is no longer the product. It is the evidence. It proves that the journey matters, that people would use it, and that the sponsor is willing to put a name to it. That is what secures the decision.
What this means for the person deciding
If you own a platform budget or a group of users, the sequence above puts you in a different position. You do not have to approve a six-figure build on the strength of a slide deck. You can see the thing working, in your context, with your journeys, for a fixed fee and a fixed duration, and then decide.
It also means the prototype should be treated as what it is: a decision tool. Ours is retired when the production build starts. The production version is built from a hardened scaffold on platforms you own, so nothing you rely on is metered by a prototyping vendor.
Where to start
Not with a build, and not with a prototype either. Start with one platform and the people who use it, and ask where the cost and the frustration sit. That is what an Opportunity Review does, and it is free. Sometimes the answer is that nothing is worth proving yet. When something is, the next step is an AppProof Sprint.