Can ChatGPT build me an app? AI app builders versus a developer

Yes, up to a point. AI tools, ChatGPT included, can produce a working prototype quickly: screens you can click through, a form that saves data, sometimes a small app you can share with a link. That is a cheap way to test an idea. An app that signs people in, stores customer data, takes payments or goes through app store review still needs a developer to secure, test and maintain it. So the practical way to develop an app now often uses both: an AI tool to explore the idea and show people what you mean, and a developer to build the version real customers depend on. This guide covers what AI app builders do well, where they stop, how to get from a prototype to production, and what each path costs.

Used for the right job, they are excellent. Five jobs suit them.

• Turning an idea into something you can click. Describing an app in a sentence and seeing screens appear is the fastest way to find out what you actually mean.
• Showing rather than telling. A rough prototype explains an idea to a co-founder, an investor or a developer better than a page of notes.
• Testing demand cheaply. Put the prototype in front of five people who would use it, and watch where they get stuck, before spending money on a build.
• Small tools for your own use. A calculator, a checklist or a simple tracker that holds no one else's data and does not take payments.
• Finding the hard questions early. Every point where the tool gets confused, such as who can see what or what happens when a payment fails, is a question the discovery workshop would ask anyway.

A prototype only has to work once, for one person, in a demo. A product has to work every day, for everyone, including the people trying to break it. The gap shows up in the same places.

Security and permissions. A demo usually has one user. A real app has many, and none of them may see anyone else's data. In Reckonly, a product our co-founder Anis built before ELORA DIGITAL, the organisation is applied as a filter on the server and the browser never supplies it, so there is no request a user could alter to read another business's records. That kind of rule has to be designed in, and then tested.

Data that has to last. Customer records need a database designed for the app, backups that can actually be restored, and a safe way to change the structure when the app changes.

Payments. Taking money is the easy part. Failed cards, refunds, plan changes and the messages a payment provider sends back to your app all need handling, and every path needs testing.

App store review. Apple reviews every app against its App Store Review Guidelines, however the app was made. Two of them matter for most first apps: an app "should include features, content, and UI that elevate it beyond a repackaged website", and an app that supports account creation must also offer account deletion within the app. Google Play has its own policies. Meeting them takes testing on real devices and responding to review feedback.

Maintenance. The libraries underneath any app release security fixes, phones get new operating systems, and someone has to apply the updates and check that nothing broke.

Ownership and hosting. Some builders host the app on their own platform. Check whether you can export the code and run it on hosting in your name, before customers depend on it.

Running costs. If the app itself uses AI, every request costs money. A production app caps that spend before it happens; Reckonly runs a billing check before each AI call for exactly this reason.

Keep the prototype, and treat it as the brief.

1. Write the one sentence. Who the app is for and the one thing it lets them do. The prototype shows the screens; the sentence decides what the first version is.
2. Bring the prototype to discovery. A paid discovery workshop ($500 to $1,500, credited to the project if it goes ahead) turns it into a written scope: the workflows, the kinds of user, the integrations and what is deliberately left out. The price is fixed once you sign the scope.
3. Decide what carries over. The screens and flows usually carry over as design input. The code often does not, because the data structure and the security rules work best when they are designed in from the start.
4. Build in short steps. Two-week sprints, each ending with a working demo, so the first real users can try it early.
5. Launch on accounts in your name. Hosting, domain, code repository and app store accounts, with documentation and a 30-day warranty that sits alongside the consumer guarantees under the Australian Consumer Law and does not limit them.

What goes in the first version of an MVP covers step one in more detail, and our process page walks through each stage.

AI features inside the app, such as reading documents or a chat assistant, are an add-on from $3,000, and after launch a care plan from $300 a month covers hosting, updates and fixes. How much it costs to build an app in Australia explains what moves a project within its range.

An AI tool on its own is enough when the idea is untested, the app holds no customer data, nobody pays through it and you are the only person who depends on it.

Bring in a developer when real customers will sign in, the app holds information about people, money changes hands, it has to pass app store review or your business will depend on it working tomorrow.

Most good apps pass through both stages. If you have a prototype, or only an idea, tell us about it. A founder replies with an estimate range within one business day, and if the honest answer is to keep testing with the prototype for a while longer, we will say so.