FROM A TASHKENT BUS ROUTE TO 3,500 STORES: HOW TASS VISION BUILT AN AI BUSINESS ACROSS MARKETS

FROM A TASHKENT BUS ROUTE TO 3,500 STORES: HOW TASS VISION BUILT AN AI BUSINESS ACROSS MARKETS

Some founder stories begin with an idea. Shakhzod Umirzakov’s began with a problem he could see every day. While working as a ticket inspector on a bus route in Tashkent, he saw firsthand how much revenue was being lost through an inefficient system. That experience eventually led him to build his first passenger-counting device by hand an idea that initially failed, but later evolved into something much larger. Today, Shakhzod is the founder of TASS Vision, an Uzbek technology company that has grown from two founders to a team of around 50 and connected thousands of stores across several markets. Along the way came difficult pivots, technical challenges, major clients, international recognition, and the 2024 President Tech Award in AI Technologies. Georgia has also become an important part of that story. Through local partnerships, TASS Vision has entered the Georgian retail market without building a traditional local office, proving how much the right relationships can matter when expanding internationally. In this conversation, Shakhzod speaks openly about failure, product pivots, trust, entering Georgia, the differences between the two markets, and what he wishes he had understood when he first started building.

1. The Tashkent bus depot and the ticket inspector -how accurate is that, and what do you remember most vividly from that period?

The story is essentially true. I worked as a ticket inspector on a bus route, and I saw with my own eyes how much money simply never reached the till conductors and drivers together were skimming cash every day, and on some routes this “leakage” reached as much as 40% of revenue. This wasn’t an abstract “market problem” from a textbook it was a concrete number I saw every single day. What I remember most vividly is the feeling that this problem was solvable nobody had simply bothered to solve it.

2. Your first passenger-counting device went nowhere how did you take that at the time, and what had changed in you by 2021?

I built the first device literally with my own hands soldering, testing, trying to sell it. It didn’t take off because the market for it was simply too small at the time: transport companies weren’t ready to pay for counting accuracy. Back then it felt like a failure, but in essence it was an early, too-narrow version of the right idea. By 2021, the key thing that had changed was that I stopped thinking about a single vertical (transport) and saw that the same people-counting technology was needed by retail, at a far larger scale. Plus, I already understood how to build hardware and how to talk to a large client experience I simply didn’t have the first time around.

3. Texnomart your first major client, the same year your first cameras went live. Tell me about that meeting.

Texnomart is Uzbekistan’s largest electronics chain, and the fact that they became our first major client mattered enormously: if it works for them, it works, period. But at that point our AI technology was only about 50% accurate not good enough to risk losing the contract. So for a while, every single night we manually recounted visitors from the footage of the cameras we’d already installed, and every morning we handed the client a result accurate up to 99%. From the outside it looked like a working AI system; behind the scenes, a few people were literally finishing by hand what the model couldn’t yet do on its own. That’s the moment I realized they believed us not in a pitch, but in numbers we delivered flawlessly every single morning.

4. Five years, 2021-2026: from two people to a team, from zero to thousands of stores. At which moment were you closest to walking away?

We grew from two founders to a team of 50, and from zero to 3,500 connected stores across several countries-that’s true. But I came closest to walking away not at the beginning, but right after things had seemingly worked out. We pushed accuracy to 90% and started scaling and that’s exactly when it turned out the unit economics didn’t hold: clients were willing to pay no more than $400 a month per store, while our cost for cloud video processing ran to roughly $350 per store. A year earlier we had already made a painful pivot, moving away from sensors toward cloud-based video processing. And now, having barely started scaling on that model, we had to admit it didn’t work either and pivot again, this time to processing directly inside the camera, offline, without constantly streaming video to the cloud. It didn’t help that internet connectivity in the regions left a lot to be desired, so pushing video traffic to the cloud also meant losing data. That was the point a second pivot in a row, right after finally showing growth, and having to rebuild everything again where I came closest to giving up.

5. President Tech Award in 2024, international coverage in 2025 you now represent not just a product but Uzbekistan’s tech sector. How heavy is that weight?

It’s true, after the President Tech Award in the AI Technologies category and the coverage that followed, every conversation I have in a new market also becomes a conversation about what a tech company from Uzbekistan is even capable of. But honestly, I try not to focus on the award itself the focus stays on further expansion into new countries and on giving retailers the ability to make decisions based on the data we provide, as well as on developing advanced AI models that run directly inside our hardware rather than in the cloud. That said, we do feel a real sense of responsibility for being among the first in this space we’re effectively setting the benchmark for the whole Uzbek ecosystem in AI, and specifically in computer vision.

6. You publish your accuracy table openly, including the weak numbers. Whose decision was that, and what did it cost you in sales?

The decision to show not just the strong numbers but the weak ones too was the team’s, and it was deliberate. Any technology operates with a margin of error, and that needs to be acknowledged openly: if your marketing gives the client the impression that everything is perfect, then once they actually get the product their expectations will be inflated and reality won’t match them and the disappointment will be worse than if we’d simply been honest from the start. We deliberately show the real picture, both to demonstrate expertise to the client and to keep our R&D team sharp they have a clear incentive, quarter after quarter, to improve exactly the metrics we’ve publicly admitted are weak.

7. Georgia is among your markets, but you have no office there. When and how did Georgia appear on your map, who installs the cameras in Tbilisi, and how many stores run Vitrac in Georgia today?

Georgia appeared on our map thanks to one of the major retail industry events in Tbilisi, where we met Lado Kandelaki, co-founder of the E-commerce Association, who became our partner in the country. It’s through him that we find contractors across Georgia who handle the on-the-ground camera installations we don’t have our own office or installer staff there. Through this partnership we’ve already connected one of the largest electronics and pharmacy chains in Georgia, and four more pilots with major clothing retailers are running in parallel. By our estimates, by the end of 2026 around 300–500 stores will be connected on Vitrac in the country.

8. Your first visit to Georgia what differed most from what you expected, in business and beyond it?

My first visit to Georgia was in 2023, when we joined the 500 Eurasia program. I was struck by how much Georgian businesses love innovation and are always looking to improve their processes it makes for a very open conversation about technology. But at the same time, I learned that decisions there aren’t always made as quickly as, say, in Central Asia it can take noticeably longer, and you need to be ready for that when setting expectations around deal timelines.

9. Caucasus Ventures is among your investors. How did that relationship begin, and what role did the Georgian connection play in your regional growth beyond the investment itself?

In 2023, after 500 Eurasia invested in us, we started actively looking for our next round and reached out to nearly every venture fund across the Caucasus and Central Asia. One Azerbaijani fund, Caucasus Ventures, believed in our vision and agreed to invest. Beyond the money itself, that Caucasus connection gave us access to a regional network of contacts that directly helped our regional growth, including our entry into Georgia.

10. If an Uzbek tech company were planning to enter Georgia today, what would you tell them at the very first meeting?

I’d say: don’t show up with a solution copy-pasted from Central Asia show up ready to rebuild it. Georgia is a smaller market, but the bar for the quality of negotiations and personal trust is higher deals there are built more slowly, and through relationships, not just numbers in a deck. Second: find a local person who understands the context before you start selling, not after your first failed deal. And separately pay close attention to the legal side and make sure you comply with local law: the regulator in Georgia demands more than in many neighboring countries, and that shouldn’t be underestimated.

11. For a 20-year-old founder with no reputation, network, or money what should they focus on in year one, and what should they deliberately ignore?

Focus on one narrow, real, paying client not on building the perfect product for everyone. Year one is about proving that someone is willing to pay you for solving a specific problem, not about scale. Ignore the polished pitch deck, the perfect brand, and any attempt to look bigger than you actually are; at this stage, neither investors nor clients are buying appearances they’re buying proof that the product actually works.

12. One mistake founders make when they assume success in Uzbekistan will automatically translate to Georgia, or vice versa?

The most common mistake is assuming that if a sales and partnership model worked in Central Asia, it can be copied one-to-one. From the outside these regions can look like a single “post-Soviet” market, but internally the pace of decision-making, the role of personal relationships in business, price sensitivity, and expectations around contracts all differ. Success isn’t about copying the model it’s about adapting the same technology and the same product to a different sales logic. And separately, there’s the legal side: every country has its own requirements and restrictions what’s legal and standard in one jurisdiction can be a direct violation in another, and that needs to be checked carefully rather than assumed by analogy.

13. What did you believe about success back then, building that first device, that you no longer believe today? And what do you know now that you wish someone had told you at the start?

Back then I believed success was primarily about the technology: make the product good enough, and the market will find it on its own. I don’t believe that anymore the technology was necessary but far from sufficient; what turned out to be decisive was the ability to sell, to negotiate, and to hold onto a client’s trust even when something goes wrong. If I could say something to that earlier version of myself, it would be this: don’t expect the first version of the product, or the first market, to be the right one. What matters more than getting it right on the first try is being quick to recognize what isn’t working and rebuilding it that’s exactly how, not on the first attempt, we got to where we are now.

Shakhzod’s story is a reminder that cross-border growth rarely follows a straight line.

The first product may fail. The first business model may not work. A strategy that succeeds in one market may need to be completely rethought in another. What matters is the ability to listen, adapt, rebuild, and earn trust along the way themes that run through almost every stage of his journey.
His experience also reflects something central to the mission of the Georgia-Uzbekistan Business Hub: successful expansion is not only about technology, capital, or market size. It is also about finding the right local relationships and understanding how business is actually done on the ground. We’re grateful to Shakhzod for sharing his story so openly and look forward to following the next chapter of TASS Vision’s growth in Georgia, Uzbekistan, and beyond.