Custom web apps & portals
-
Updated

Should I use a 14-day AI build shop or a prototype-first developer for my web app?

Use a 14-day AI build shop when the app is simple, standalone and roughly right is good enough. Use a prototype-first developer when the app talks to your accounting, job or CRM systems, when customers will use it, or when getting it wrong costs you money. Applicable's route is a two-week NZ$5,000 Prototype Sprint, then a fixed price from NZ$15,000.

Should I use a 14-day AI build shop or a prototype-first developer for my web app?

Two ways to buy a web app in New Zealand in 2026

The first is new. A handful of Auckland studios now sell custom software "in 14 days from $7,000": AI generates most of the code, engineers finish it, you own the result and pick a monthly plan afterwards. It is a clear offer and, for the right project, a good one.

The second is the one Applicable runs. A two-week Prototype Sprint for NZ$5,000 plus GST, which ends with a clickable prototype, a written scope and a fixed price. Then a build measured in weeks, from NZ$15,000 for a single-workflow web app and NZ$30,000 for a customer or staff portal.

Both put a number on the table before you commit, which is more than most of the market does. The difference is what the number buys.

When the 14-day build is the right answer

  • The app is internal, standalone and simple: a form that feeds a list, a basic CRM, a dashboard over a spreadsheet.
  • It does not need to talk to your accounting package, your job system or your customers' logins.
  • If it is roughly right on day 14, that is good enough, and you can iterate on a monthly plan.
  • You are a founder testing whether anyone wants the thing at all.

In those cases a fortnight and a low five-figure price is hard to argue with. Applicable will tell you so on the first call.

When prototype-first is the right answer

  • Customers or staff will use it every day, and getting a flow wrong costs you money or trust.
  • It has to integrate: Xero or MYOB, a job management system, a payment gateway, a CRM, an existing database.
  • There are rules in it: approvals over a value, trade pricing for some customers, a job that cannot close until a checklist is complete.
  • It replaces something that is already running, so the migration is part of the work.
  • You need to know the total cost before you sign, and hold someone to it.

A sprint costs two weeks and NZ$5,000 to find out what the system actually has to do. That is cheaper than finding out from a live app that customers cannot use.

What each costs, side by side

14-day AI buildApplicable, prototype first
Before you commitA phone call and a priceTwo-week sprint, NZ$5,000, credited against the build
Simple web appFrom $7,000, 14 daysFrom NZ$15,000, 6 to 10 weeks
Customer or staff portalQuoted; the 14 days usually stretchFrom NZ$30,000, 10 to 16 weeks
Integrations"Yes"Checked against your systems during the sprint, priced in the scope
After launchFree self-host to $2,000 a monthFree self-host to NZ$5,999 a month
Who you deal withA delivery managerThe founder and the people who build it
What you ownThe codeThe code, the hosting accounts, the data, the documentation

Prices are NZ$ plus GST. Applicable's are on the pricing page; the sprint is described on the Prototype Sprint page.

How to decide in one conversation

Tell either vendor three things: who will use it, what it has to connect to, and what happens when it goes wrong. If the honest answers are "just us", "nothing" and "we fix it next week", buy the fast build. If any answer involves customers, another system or money, run the sprint. Ask Applicable and you will get a straight answer, including "the other option is fine for this".

Common questions

For a simple, standalone tool, yes: a form, a list, a dashboard, a basic CRM. The price is real and the speed is real. What it is not is a portal that customers log into, integrated with the systems you already run, tested by your staff and supported for years. Those take weeks, and anyone who says otherwise is describing the invoice, not the software.

Because the sprint is where the price becomes fixed. Two weeks of workshops, wireframes and a clickable prototype turn a brief into a scope both sides can read. The NZ$5,000 is credited against the build, and you keep the prototype and scope if you walk away.

It should mean the same thing: the repository, the hosting accounts and the data are in your name. Ask any vendor to put it in the agreement, and ask who can rebuild it if they disappear. Code produced mainly by AI with a thin review is harder for the next developer to pick up; ask to see a handover document.

Sometimes. If the cheap build proves the idea, the rebuild is a normal project. If the cheap build is in front of customers and breaking, the rebuild is a rescue, and rescues cost more than starting properly. Applicable will look at what you have and say which it is.