You’ve probably heard it: these days you type a few sentences and AI makes you a website. It’s partly true. We use AI ourselves every day. But between “something opens in the browser” and “this earns money for my business” there’s still a long way. This article is about where AI really helps, where it stops, and why a good product still needs people who take responsibility for the result.
What AI already does well
Let’s be honest: AI has sped up almost every stage of our work. It helps us draft copy, try several design directions, write routine code quickly and find a bug that used to take half a day to track down.
But the time we save doesn’t disappear. We put it where AI doesn’t help: into conversations with the client, into decisions about how the product should work, and into checking that everything really does work.
A page that opens is not a product yet
Tell an AI tool what kind of page you want, and a few minutes later your browser shows something that looks like a website. For a personal page, an idea you want to test or a one-off campaign, that may be enough. We’ll say that honestly.
The problems start when the page meets real people. The form works on your laptop, but enquiries never arrive. The menu breaks on one particular phone. The page loads slowly on mobile data. Google indexes pages that were never meant to appear in search.
You can hand each problem back to the AI. Sometimes it finds the fix straight away. More often it adds a patch on top of a patch, and after a few rounds nobody understands the project any more, including the person who generated it.
Your business is not a template
AI doesn’t know how your company actually works. It doesn’t know who your customers are or how they book. It doesn’t know what happens when a customer doesn’t show up, or which data must never end up in the wrong hands. Without those answers, all it can generate is an average website for an average business.
Take OWL Barbershop. They needed their own booking system so they could leave expensive third-party platforms and their subscription fees behind. The result was a website and two apps, for iOS and Android. Customers book a visit in a couple of taps at any hour, and the system sends reminders on its own, which significantly reduced no-shows.
Those decisions didn’t come from one sentence in a chat. They came from conversations about how a barbershop runs day to day and what the owner wants to see in the admin panel. Today it holds the whole customer base, the schedules and the visit history, and the owner runs it all without calling us about every small change.
Looking good has become cheap. Good design hasn’t
AI produces a finished-looking design in seconds. So appearance alone no longer tells you much about the quality of the work.
That doesn’t make good design easy. AI offers options, but it can’t tell which one is right. It has to be steered: you need to know what to ask for, what to throw away and what to refine. And you need to recognise the moment the product is truly good, not just finished. That takes design experience and taste. No prompt can replace either.
The important questions come after the first impression. Does the page hold up with real content, long Latvian words and three languages? Does the customer understand where to tap next? What does the screen show when there’s no internet, when a list is empty or when a payment fails?
We use AI to try ideas faster. But the direction is chosen by a person who knows your customers and your goal. A design can look great and still be wrong for your project.
Writing code is easy. Answering for it isn’t
AI writes code very fast. We treat it like a very fast junior developer: in experienced hands it saves hours, but its first version must never be taken as finished.
AI doesn’t know why earlier decisions were made the way they were. When it fixes a bug, it often adds another condition or workaround instead of noticing that the structure itself needs to change. And there are areas where a mistake costs a lot: payments, customer data, security, GDPR, App Store and Google Play requirements.
That’s why our developers’ work has changed. They review what AI proposes, keep the architecture simple, remove what isn’t needed and decide when generated code is safe to keep and when it has to be rewritten.
When AI is enough, and when to come to us
If you need a personal page, a simple one-page site or a prototype of an idea, try AI. It may well be enough, and we’ll tell you so even if you come to us.
A team starts to matter when the website or app starts earning money. Then it has to attract customers, take bookings or payments, connect to your systems and work in three languages. It has to be fast and secure, and someone has to take responsibility for it after launch too.
That’s what we do: websites, mobile apps, mini apps, chatbots and CRM systems. AI helps us work faster, but a person answers for every decision. Tell us what you want to make. We’ll tell you honestly whether an AI tool is enough or you need a team.