Kvalty Turns One. The Idea Is Almost a Decade Old
Kvalty turns one as a company. The idea sat in my head for years, AI helped me finally build it, and the project keeps growing.
Today Kvalty turns one. Not one year since the idea appeared, because that idea has been sitting in my head for almost a decade, but one year since I founded a company around it. My first company, and, judging by how addictive building things can get, probably not my last one.
It was also recently one year since the first click from Google. Someone I did not know searched for a driving school, Google showed them Kvalty.cz, and they clicked. Since then, the site has kept growing: fewer than 7,000 people visited Kvalty in all of 2025, while almost 30,000 visited in the first half of 2026 alone. That still feels strange to me. Not because Kvalty is finished, because it is nowhere near finished, but because an idea I carried around for years has become a thing people actually use when choosing a driving school.
The Idea That Waited for the Right Moment
Kvalty started from a very ordinary thought. Choosing a driving school should not mean clicking the first result, hoping the price does not ruin you, and praying you do not end up in a car with an instructor who spends the whole lesson shouting. I wanted a clear place where people could compare driving schools by price, data, and real experiences from other students, not just by who has the nicest website, the best ad, or the loudest recommendation from someone nearby.
That is not meant as an attack on the whole industry. I have met good driving schools, and several of them gave me useful feedback over the last year. I just think that when someone is choosing a service that costs tens of thousands of crowns, they should have more to work with than advertising, a first impression from a website, and a few random recommendations. A driving school is not a small decision. You spend dozens of hours there, often under pressure, learning something you will use for the rest of your life. A bad choice can cost money, time, and a lot of nerves.
But an idea is not enough. I have always been mostly an Android developer: Kotlin, mobile apps, Jetpack Compose. That was my world. I did not build websites, and building a serious web product from scratch is not just learning a different syntax. It is backend, frontend, databases, SEO, infrastructure, good UI, usable UX, performance, security, and all the boring details nobody sees until they break. One thing is to picture the product clearly in your head. Another thing is to sit down and actually build it.
That part hurts.
AI gave me a way across that gap. Not in the “write one magic prompt and get a finished product” sense, because that is not how larger projects work and anyone who has tried will find that out quickly. It was more like having a team of virtual developers next to me. My job was to act as the architect: describe what I wanted, review the output, test it, fix the nonsense, throw away dead ends, and iterate. I still had to make decisions, I still had to check the work, and I still had to know when something looked fine on the surface but made no sense underneath. The difference is that without AI, I would probably have postponed the idea for another few years.
At the end of 2024, I finally started working on Kvalty for real.
From the First Data to the First Company
The beginning was mostly about data. I needed to collect driving schools, addresses, contacts, courses, prices, and figure out what could actually be compared. What was public, what could be found automatically, what needed manual verification, and where the reality of the Czech internet began. It became clear very quickly that the Czech driving school market is not exactly a clean data warehouse. Every school has a different website, a different pricing structure, different names for the same services, and a different level of willingness to show prices publicly.
At the same time, it also became clear that once the data was in one place, it started helping people. The first click from Google was nothing in absolute terms. One person, one search, one click. But for me it mattered, because it was not an ad, a friend, or a link I had sent someone. Somebody found Kvalty.cz because they were trying to choose a driving school, and Google offered the site as an answer.
Then the first reviews arrived. The first questions. The first driving schools asking to correct information. And gradually, I realized that Kvalty.cz was no longer just my testing playground. If I was helping people choose a driving school, I had to treat the project more seriously. So a year ago I founded a company for Kvalty. It was not a moment where everything suddenly changed, more a formal admission that this was no longer a toy.
What Changed Over the First Year
In the year since the company was founded, Kvalty grew beyond what I expected at the start. Alongside the comparator, I added Kvalty Academy, a place to prepare for driving test questions. I did not want to build another list of questions with the correct answer highlighted, because that already exists. What made sense to me was adding explanations: why the answer is correct, and smart learning that brings people back to questions they struggle with.
I also added Ukrainian. I do not speak Ukrainian, so I will not pretend this is risk-free or perfect. But it felt important that Kvalty.cz could help people for whom Czech is not easy. A driving license is practical. If someone lives in the Czech Republic, works here, or is raising a family here, language should not be another unnecessary barrier.
Most of the work still disappears into the data, because driving school profiles cannot be a one-time scrape that nobody touches for years. They are continuously checked by an AI system I call the Ralph loop. It goes through driving school websites, checks prices, addresses, courses, descriptions, and page changes, and when something does not match, it prepares a proposal for human review. I wrote about the approach in How I Used 20 AI Agents to Extract Data From 1,700 Driving School Websites. Without that kind of verification, Kvalty.cz would slowly turn into another stale directory with outdated information. There are already enough of those on the internet.
The Reality Behind the Website
From the outside, Kvalty can look like a website with driving schools. From the inside, it gets messy. The servers have been attacked, sometimes hundreds of thousands of attempts per week, and there have been attempts to scrape the data at scale. That part bothers me the most, because the data gives the site value, but it is also supposed to be public, searchable, and useful. Protecting something that is meant to help people find information is a strange discipline.
Then there is communication with driving schools. Some reacted very well, sent corrections, asked how the data works, gave feedback, and honestly surprised me in a good way. Others started with legal threats. I guess that is part of it, but I would like the industry to move toward a place where data, prices, and student experiences are a normal part of decision-making, not something to hide.
Seznam.cz is a separate chapter. Other platforms index more than 10,000 pages from Kvalty.cz, but Seznam.cz has been telling me for months that it struggles to read websites running behind Cloudflare. The result is simple: Kvalty practically does not exist in the largest Czech search engine. I expected more from the largest Czech search engine, especially for a project focused on the Czech market and Czech users.
And I still have a full-time job. Kvalty is my company, but it does not pay my full-time salary. I build it after work, in the evenings, on weekends, and between everything else. This year I also started working on Ferda, a project focused on protecting children online, so some weeks are a lot. But if I did not enjoy the work, continuing would not make much sense.
What AI Actually Taught Me
I learned more over the last year than I expected: databases, backend, CI/CD, UI, SEO, security, working with large amounts of data, product decisions, and communication with companies, authorities, and people who often have a very different idea of what should be visible online. The biggest technical milestone was probably the three weeks in January when I rewrote the whole backend from Directus to a custom system. I wrote about that separately in The 1.4 Million Line Delete.
AI wrote a lot of code during that process. I will not pretend otherwise. But someone had to direct it, hold the line, and decide what belongs in the project and what goes into the trash. The project currently has 596,776 non-blank source lines, which still catches me off guard when I see the number. Not because of the size itself, but because of the responsibility. When you build something like this with AI, you cannot just generate code and pretend the work is done. You need to know what you want, read the output, test it, and spot the things that would fail in production or make the system painful to maintain six months later.
That might be the biggest lesson of the whole year. AI is not a shortcut to a product without thinking. It is leverage. It can give you speed, capacity, and the courage to start things you would not have dared to start alone. But direction, judgment, and responsibility still have to come from a person.
What Comes Next
Kvalty has not earned a single crown yet. It costs money instead, because AI, servers, tools, data, and time all cost something. It has also eaten much more of my time than I would like to admit out loud. But when I see that people really use Kvalty.cz to choose a driving school, write reviews, send questions, or point out mistakes, it makes sense to keep going. An idea in my head became a real thing that helps people avoid a bad choice.
The next big topic is pass-rate data for individual driving schools. I have been negotiating with the Ministry of Transport for the last quarter, because this is information you do not usually see in the Czech Republic, and I think it could change how people choose a driving school. Price matters, reviews matter, and location matters too. But knowing how successful a particular school’s students are at the exams could be a major piece of context.
If it works out, I do not want to turn the data into a weapon against driving schools. I want it to become another fair piece of context for the person choosing one. Clear, understandable, and useful enough to help people see where the price and promises are backed by good teaching.
When I started, I mostly saw Kvalty as an AI experiment. I wanted to find out whether someone who was not a web developer could use AI to build something large and useful. After a year, I see it differently. Kvalty started as an experiment, but now it feels more like a commitment. As long as people are choosing driving schools half-blind, it makes sense to keep working on it.
Happy birthday, Kvalty. Back to work. Kvalty.sk will not build itself.
Thanks to everyone who used Kvalty.cz, sent feedback, wrote a review, reported a mistake, or simply told me the project makes sense.
