Considering AI, how much does it cost to build a product?

Every introduction call we run for a new project arrives at the same question: what will this cost? After 7 years of building mobile apps and web platforms, we can share the same version of the answer we give in the call.

The 3 phases of building a product

An app budget makes the most sense when you read it as a sequence of phases. Every project we take moves through the same four, and each one buys you something specific.

Week 1: Discovery

Every project starts with a one-week discovery sprint. We sit down with you and turn the idea into a concrete scope: who the single buyer persona is, which core journey the MVP must nail, which features move to phase two, and what success looks like ninety days after launch. It is the cheapest week of the whole project, and it decides more of the final invoice than any other — every feature cut here saves its full build cost later.

Week 2,3 & 4: Product design

Design runs for a few weeks: user flows first, then wireframes, then high-fidelity screens you can click through and put in front of real people. This is where the product gets its feel, and where iteration is still cheap — moving a screen in Figma costs minutes; moving it in a shipped app costs days.

Week 5-12: Development and launch

The longest phase, and where the feature list you froze in discovery turns into cost. What keeps it down is boring by design: standard auth, an off-the-shelf payment provider, one platform, and features whose patterns are well established. What drives it up is anything bespoke or stateful — custom flows, real-time behaviour, and edge cases where money or security live. Then there are the parts users never see: the admin panel you will want the day your first real user misbehaves, integrations, store submission, QA on real devices, and launch.

What AI changes, and by how much

"Human crafted. Accelerated by AI." is how we describe our work, and the acceleration is specific and measurable. Building the calculator forced us to rate every type of work by how much of it AI can genuinely carry, and the spread turned out to be wide:

Admin panels are the single most automatable slice of a build — roughly 85% of the work is CRUD scaffolding that AI handles superbly. A full dashboard now costs closer to €3,000 than the €10,000 it commanded a few years ago.
File uploads, multi-language support and analytics wiring sit in the same tier: well-trodden patterns AI reproduces almost end to end.
Authentication, payments and scheduling land near 50% — AI writes the plumbing quickly, while a human designs the flows and hunts the edge cases where money and security live.
Onboarding sits at the bottom, around 20%. Great first-run experiences come from taste, user empathy and iteration, and those hours stay human.

Gamifying your product and Keeping your users Active

Most of our portfolio is gamified apps, and building them taught us something budget-relevant: acquisition gets all the attention, while retention quietly decides whether an app survives. The mechanics that bring people back are ordinary features with extraordinary leverage, and they are at their cheapest when designed in from day one:

Daily streaks — a visible run of consecutive days that gives users a reason to open the app tomorrow, and a reason to care about keeping the run alive.
Goal progress — progress bars and milestones that show people how far they have come, turning everyday usage into a story of improvement.
Achievements — badges and unlocks that celebrate the behaviors your product needs most, from a first booking to a tenth session.
Notifications — well-timed nudges tied to streaks and goals, which is exactly where that €1,500 notifications line item earns its keep.

A streak counter or an achievement system is a modest line item during the build and a compounding asset after launch. If your product lives or dies on habit — fitness, learning, finance — budget for at least two of these from day one.

Taking it all in, how much does it cost?

Putting it all together, discovery, design, development and launch, this is the answer. A real, production-ready, well designed MVP, costs between €15,000 and €30,000 to build. Where you land inside that range comes down to the features you choose and how much of the work AI can genuinely carry.

How to keep your cost small

The lower end of the range rewards discipline more than luck. The €15,000 builds we have shipped shared a few habits:

Build for one buyer persona and one core journey. Every extra user type multiplies screens, permissions and test cases.
Start on the web if your audience allows it. A responsive web app skips the two-platform overhead and still lives on a phone's home screen; go native once retention proves the demand.
Let the admin panel be scaffolding. Spend your design budget on the screens customers see — behind the curtain, generated CRUD is perfectly dignified.
Make safe choices. Stripe for payments, standard social logins, off-the-shelf analytics. Every bespoke replacement burns thousands for invisible gains.
Think long term, not everything at launch. Chat, referral programs and personalization make wonderful month-six additions and expensive month-one gambles.

Run your own numbers

Every project is its own equation, which is exactly why we turned ours into a public tool. Spend two minutes with the AI Cost Saving Calculator — pick your platform, tick your features — and it will show where AI can compress your particular build. If you would rather talk it through with a human, book a call or write to hello@untklabs.com; walking founders through this math is genuinely our favorite part of the week.