Ersultan Ermanov launched his first startup back in his student years. Together with his groupmates, he created two travel portals for the Embassy of Kazakhstan in Singapore. That was followed by a major project Kupi.kz, a large marketplace they eventually had to shut down at the peak of its popularity. Then came the search for a much-needed solution in the education space. After several unsuccessful attempts, Ersultan and his team finally hit on a real market pain point. That’s how TrustExam.ai was born — a project that initially helped combat cheating during exams. Today, it has evolved into a comprehensive anti-cheating system designed to ensure fairness and integrity in any kind of knowledge assessment, both online and offline.
As part of the joint project by Digital Business and Astana Hub, 100 Startup Stories of Central Asia, Ersultan shared why the marketplace story didn’t work out — and how he came up with the idea for TrustExam.ai while at a maternity hospital. We also talked about how cheating has evolved, the tools being developed to fight it, and the project’s international expansion into European and African markets.
«We got $600,000 and were offered to kick off a full-on marketplace»
– My mum’s a mathematician, and I was always pretty good at it too. But after school, I enrolled at Al-Farabi Kazakh National University to study international relations. After a year, though, I realized it just wasn’t for me. So I transferred to James Cook University in Singapore and switched my major to Computer Science.
Learning wasn’t easy for me, so I decided to dive into programming through real, hands-on experience in commercial projects. An internship at the Embassy of Kazakhstan helped spark an idea. I noticed that many Singaporeans came in to apply for visas, but often didn’t really understand where they were going or what they needed. Some didn’t even know the minimum visa processing time.
To solve that problem, my friends and I launched two informational travel portals: SeeKazakhstan.com and Discover-Kazakhstan.com. They had all the essential info for tourists — everything about the country, how to apply for a visa properly, and what documents were needed. We built the websites using online tutorials, and for the content, we pulled information from books about Kazakhstan we found at the embassy. We scanned them with an OCR tool and uploaded the content to the site.
Getting traffic was actually pretty easy. We worked out a deal with the embassy to hang a banner on the premises and leave some business cards around. The next step was figuring out how to monetize it.
- We integrated the Booking.com API.
- We signed partnership deals with taxi services and hotels in Kazakhstan.
- We started collecting and processing requests for services like lawyers, tour guides, and translators.
So over time, our websites grew from a student project into a profitable business that helped us pay for our education and enjoy a student life without financial stress. That was my first real experience — the one that got me started on my path in the IT world.
– What happened with the projects after you returned to Kazakhstan?
– I wanted to scale the projects to other countries and reached out to the Ministry of International Affairs for support. But it didn’t work out — the Ministry already had its own plan for developing informational portals.
But the experience with the travel websites taught us how internet traffic and SEO algorithms work. We used that knowledge to launch new e-commerce projects. We built niche websites around different topics, like beauty, sports, and fitness devices, and managed to bring them to the top of search results. It was around that time, in 2015, that I met my current co-founder and CTO, Eduard Zaukarnaev. Since then, he’s been in charge of the technical side of all our projects.
The success of our e-commerce projects caught the attention of investors, who provided us with $600,000 and offered us the chance to launch a full-scale marketplace. That’s how Kupi.kz was born, a platform featuring over 120,000 products and handling more than 500 orders per day at its peak. It was a big success.
However, due to internal issues within the organization that started to impact the whole team’s work, we made the decision to shut the project down. I was 25 at the time, and it hit me hard. We had experienced such dramatic growth and then, almost suddenly, it all came crashing down.
«To this day, I don’t know if that move was brave or just kind of foolish»
– After the tough experience with the marketplace, what made you want to launch another startup, especially in the education space?
– After we shut down Kupi.kz, I spent about six months thinking about what to do next. In 2018, our team from the earlier projects decided to give EdTech a try. It seemed like a fast-growing, promising field. Our first product was a basic LMS (Learning Management System for online teaching — note by Digital Business).
The thinking was simple: existing systems like Moodle (a popular open-source LMS used by thousands of universities worldwide — note by Digital Business) were too complex. They were hard to customize and usually required hiring IT specialists just to get them up and running. We wanted to build a similar product, but make it much simpler and more accessible for any company to use.
For two years, we were in talks with potential customers, but it didn’t lead to much. We only managed to implement a few commercial LMS projects. To keep ourselves afloat, we took on small website jobs here and there, just enough to cover our operational expenses.
– When did the idea for TrustExam.ai come about?
– It all happened unexpectedly. On March 6, 2020, my daughter was born. I was standing at the maternity hospital when I got a sudden call from the vice-rector of one of the country’s largest universities — we had previously shown him our LMS. He said, “Ersultan, there’s a pandemic, and we have an urgent question. Does your system have proctoring?” At that moment, I didn’t even know what the word proctoring meant. But I said, “Yes, we do.” He replied, “Can you show it to us?” I said, “My daughter was just born and I’m at the hospital right now. Can I get back to you in a week?” He said, “Sure, no problem. Talk to you later.”
During that week, my team and I quickly figured out what proctoring actually was. Basically, it’s computer vision technology used to verify that the person in front of the camera is really who they say they are. We put together a demo version and presented it to the client — and they liked it. That’s how the story of TrustExam.ai began. And to this day, I’m still not sure if that move was brave or just kind of foolish.
– How did the project start to grow from there?
– Our first client was one of the largest universities in the country and many other institutions look to them when it comes to adopting new approaches. After we successfully implemented TrustExam.ai there, other local universities started reaching out to us as well.
But when the pandemic ended, universities went back to traditional offline formats. That’s when we started having internal debates about the future of the project. It felt like the industry we were working in might disappear altogether.
In September 2021, I was invited to join the Ministry of Digital Development as Deputy Director of the IT Department. I stepped away from the day-to-day operations of the projects but continued to take part in strategic sessions. During my time at the Ministry, the team made a huge leap forward.
After three years, I returned to the project. By that time, the team had built up enough use cases and confidence to take the company international. So my international experience from the Ministry turned out to be really valuable.
«Kazakh users are incredibly entrepreneurial — they always find a workaround or an alternative solution»
– Can you share how the project evolved while you were away?
– Today, TrustExam.ai is a unified anti-cheating system designed to ensure honesty and integrity in any kind of knowledge assessment, whether online or offline.
Since 2020, when our product focused only on webcam video analytics, cheating has evolved dramatically. Now, for example, there are AI extensions that can answer questions instantly. You just take a screenshot, and the AI generates a solution in real time.
Cheating has now moved to the hardware level. One of the most common methods today involves broadcasting the screen from the exam computer to a third-party device using special equipment. Students also use micro earpieces, hidden cameras, and even remote-controlled keyboards and mice.
It’s a real arms race. No matter how hard you work to secure the system, there’s always someone with big earnings at stake — and they’ll keep looking for loopholes.
– What does your system look like today
– We offer three levels of protection, which can be deployed in different configurations depending on the client’s needs.
The first level is software-based security.
For online exams:
- The basic protection includes webcam image analysis, along with monitoring key biometric indicators of the test-taker and the surrounding environment.
- A background application that tracks all processes running on the computer.
- A comprehensive exam application that completely locks down the computer, basically turning it into a “vegetable” for the 2–3 hours of the exam. The user can do nothing else but take the test.
For offline formats, we developed a proprietary operating system based on Linux. If anyone tries to make changes to it, the exam program immediately stops working. That’s crucial because our testing centers are spread across the entire country, and it’s hard to monitor each one in person. It used to be common for people to break in at night and try to install unauthorized screen-sharing software. Now, those attempts are blocked.
The second level is hardware detection.
We use specialized sensors to detect micro earpieces. For offline exams, it’s a USB device connected to the computer that picks up radio waves. In other words, it can actually detect what’s being transmitted to the test-taker. For online exams, we’re beta testing a Bluetooth scanner that analyzes the user’s environment. By measuring signal distance, the system can identify which devices are physically close to the computer. If anything suspicious shows up, the exam can be paused and the user is asked to disconnect any unnecessary devices.
The third level is AI-based behavior analysis.
We trained our model on data from over 1 million users. By analyzing 30 different signals, it distinguishes typical user behavior from that of a cheater, allowing it to predict the likelihood of fraud.
– How do you stay on top of new cheating methods and detect emerging vulnerabilities?
– We conduct daily analysis of exam results
- We analyze statistics on exam pass rates, track result trends, and study behavioral patterns.
- We flag suspicious users.
- We analyze the collected data and look for anomalies.
- When we identify a new case, we immediately start working on how to close that vulnerability.
Thanks to how entrepreneurial Kazakh users are, no matter what you come up with, they’ll always find a workaround. Because of that, we’ve seen nearly every cheating scheme out there. Honestly, I’d say that’s become one of our competitive advantages on the international market.
– Why do we even need such a complex system for exams? Is cheating really that big of a problem?
– The consequences of fraud can be catastrophic. For example, there was a recent fatal car accident in Florida involving a truck driver who didn’t speak a word of English and couldn’t understand road signs, yet he had a commercial driving license. During the investigation, authorities uncovered a criminal group that was helping people obtain licenses illegally. So this goes far beyond academic dishonesty. When it comes to licenses — whether for drivers or professional certifications — cheating can directly put lives at risk.
«We have a chance to win £300 million in the UK»
– In addition to Kazakhstan, TrustExam.ai is already active in international markets. How do you find clients in other countries?
– First of all, we monitor the media for reported cases of cheating. If there’s been a scandal, we reach out and say, “Hey, you’ve had a problem but we have a solution that can help reduce the risks.”
Participating in the StartX accelerator helped a lot. When you’re part of the Stanford community, many questions during negotiations just disappear on their own. It acts as a kind of validation that you’re a serious business in the eyes of potential international partners.
Another underrated tool I always recommend to startups looking to enter international markets is the Ministry of Foreign Affairs of Kazakhstan. We sent them an official letter describing our project and asked them to forward it to the embassies. When a company receives an email from an address ending in “embassy.gov,” the chances of getting a reply are almost 100%. That’s actually how we entered the Ethiopian market.
– Could you share more details about this case?
– Ethiopia has a population of over 130 million, and the number of school graduates grows every year. The organization responsible for managing national graduation exams was struggling to handle the increasing workload, so they turned to us for help.
During the negotiations, I asked how they’d heard about us. I assumed it was through LinkedIn or a Google search. But it turned out our embassy in Ethiopia had sent them a presentation about TrustExam.ai — and that’s what sparked their interest.
So our current plan is to run a pilot exam across Ethiopia, track key metrics, and then move forward with a full-scale implementation of the anti-cheating system.
– This summer, you announced a partnership with the Driver Instructor Association (DIA) in the UK. How did you manage to secure an agreement with such a major player in a competitive market?
– With DIA, we had already run a pilot project to implement TrustExam.ai in the driver licensing system for taxi and bus drivers. For us, this is first and foremost a status project. Breaking into the conservative UK market was a major milestone — and DIA is the number one driving school association in the UK and Europe, bringing together thousands of organizations.
The partnership with DIA gave us the opportunity to jointly participate in a major tender from the UK Ministry of Transport to implement a nationwide driver testing system. We’re talking about 4 million exams per year. The total value of the tender is £300 million, with a contract duration of 7 to 9 years.
We’re competing against 15 other companies in the tender, all of them local. But we feel confident, largely thanks to our partnership with DIA. What also sets us apart is our focus on the technical detection of cheating, something none of the other participants are emphasizing, even though it’s a real issue in the UK.
– How many clients does TrustExam.ai currently have?
– We currently work with around 40 organizations. About 60% of them are based in Kazakhstan, and the remaining 40% are international. Some clients hold exams just once a year, while others run testing sessions on a daily basis.
We’ve worked with over 20 universities, but over time we realized that combating cheating isn’t always their top priority. That’s why we’ve shifted our focus to industries where exams and certifications have a direct impact on licensing, safety, and reputation — in other words, sectors where cheating is a critical and highly sensitive issue.
– What is your pricing strategy?
– Most companies price their services based on the cost per exam hour. For us, the pricing is built around two key components: headcount and computing power. Over time, we’ve learned how to manage both efficiently.
There are services on the market that charge up to $40 per proctored exam. Our pricing starts at just $1, depending on the volume and complexity of the case.
We don’t follow a single standard pricing model. Each project is priced individually. For social initiatives, like programming olympiads, we often just ask to be mentioned as a sponsor. On the other hand, commercial projects with complex requirements may come at a higher cost, depending on what’s involved.
«My wife likes to joke that she’s our main investor»
– So am I right in understanding that TrustExam.ai has been developed entirely with funding from the founders?
– Right now, we’re financing the startup with the revenue it generates. But in the early stages, we invested money we had earned from previous projects. One of those is finance.kz, which is still profitable today, and we actually have big plans for its future development.
It’s hard to say exactly how much we’ve invested in TrustExam.ai — we’ve always had more of an approximate sense of debit and credit. But one thing was clear: we never spent more than we earned. In the beginning, we rented a tiny office in the Samal district of Almaty. It was a basement with no windows, barely enough space for three people. Back then, server hosting costs were sometimes covered by my mum and my future wife. My wife still jokes that she’s our main investor.
Now, everything we earn goes straight back into development. And as for the servers, those are fully ours now.
– Do you set any specific financial goals for the future?
– We don’t have strict KPIs set for the next 5–10 years. But roughly speaking, if we’re valued at $100 million in a year, that would be great. We jokingly measure our goals by the size of our families. I have three kids, my co-founder is about to have a baby, and we’ve got to feed this whole crowd. So the plan is simple — we need to earn enough to keep everyone well-fed, dressed, and with a little left over for some rest and travel.
But on a serious note, our main goal is to succeed internationally. That’s our top priority. We earn at home, reinvest in ourselves, scale globally, and bring export revenue back to Kazakhstan.
– What are your plans for the future development of the product?
– I’m confident that demand for our solutions will continue to grow. The cheating industry is rapidly evolving across all sectors, and what’s even more concerning is that it’s now attracting venture investment.
For example, there was a widely discussed case where a startup called Cluely, which actually helps people cheat during online interviews, raised $15 million at a $120 million valuation. To counter tools like that, we’re about to launch our own anti-cheating solution specifically for online hiring. It’s actually easier to pitch now. We simply say, “We help companies protect themselves against Cluely,” and that sparks real interest.