In a recent interview with SafetyDetectives, Uakhat Bastimiyev, the CEO and co-founder of Verigram, shared insights into his journey and the challenges faced in building the innovative company. Bastimiyev’s background as a software developer and project manager led him to explore large corporate systems, eventually transitioning into an IT consultant role. Verigram, established in 2017, emerged from a deep interest in new technologies, particularly AI. The initial challenges included acquiring large datasets for model training and building a committed team amid fierce competition for top tech talent. Bastimiyev also highlighted his personal transformation from a reserved project manager to a public-facing leader. The interview delved into the evolution of document recognition technology, emphasizing the significant advancements driven by neural nets and OCR. Bastimiyev discussed the transformative impact of AI on fraud detection, enabling online user verification while acknowledging the rise of deep fakes as a growing concern. Looking ahead, he identified data privacy as a crucial trend in Electronic Know Your Customer (EKYC), envisioning a future where individuals have greater control over their digital identities, citing initiatives like Worldcoin as potential drivers of change in this space.
Can you tell me a little bit about yourself and how you became the CEO of Verigram?
My name is Uakhat Bastimiyev. I am the CEO and co-founder of Verigram. My background is that I started my freshman year as a software developer, moved on to working as a project manager, and then managed an innovative fund.
At some point, I realized that I was more interested in large corporate systems and interested in implementing them. So, I converted myself into an IT consultant, which is how I started on the path toward establishing Verigram.
We started the company in 2017. We started it because my partners and I were really interested in the new technologies at the time. In the beginning, we really had no idea what we’d go through, what kind of services or projects we could do. All we saw was a huge buzz around AI. Right now the buzz is all about generative AI, but back in 2014 and 2015, it was all about neural nets. It was the planning AI, mostly about computer vision, data, about that type of stuff. We were very interested and excited, so we quit our jobs and just started from nothing.
We did it the hard way because no one from our management team, our founding team, had experience with running a tech startup. We just took some contracts to get started, but the idea was to concentrate on AI.
So we were declining contracts to develop some mobile application or website, but if there was an interesting task and we are a hundred percent sure that it was about AI, we took it on. We understood what we can do with our hard skills and soft skills, and how to sell that kind of service.
At some point, we understood that document recognition or facial recognition is the place where we can start boosting ourselves and concentrating on some kind of services or products.
What are some of the challenges you found while building Verigram?
The challenges differ from stage to stage. Initially, when we just started, it was very hard to gain large datasets of faces or something where you train your models. This was really a hard case because you can’t find anything, and you don’t understand how to take it with all the legal restrictions. However, you need those datasets because without it, you can’t have good biometric services, not only for faces but also iris, palm, or fingerprints.
A more technical challenge is our tech leaders. All of these technologies are largely based on deep learning. You need to have a very committed, very talented team of engineers. In a world where Facebook, Apple, and other huge companies are recruiting the top talent, it’s hard to convince someone to come join your team instead of going to the large tech companies.
One of the biggest challenges for me personally was to convert from a shy project manager to the guy doing the talks, presentations, and understanding the needs of the customers. I really needed to become a different type of a person, which was really hard.
Looking at the document recognition technology, how has that changed in the past years, and what are the best applications for it?
It’s changed drastically in the past 10 to 15 years. Before neural nets, all of the computer vision stuff, or document recognition was done exceptionally by people with PhDs or algorithmic researchers. One of our founders has a PhD in computer vision; he’s a mathematician and was involved in creating the systems before neural nets, and it was really hard.
Now, it’s changed drastically with different technologies. All of the document recognition uses optical character recognition (OCR) technology and can be done on your mobile device without any internet connection. The models are highly accurate, and it’s working very fast.
How is AI reshaping the landscape of fraud detection and prevention?
AI has taken a lot of processes that were previously only available offline and made them possible online. For example, you can open a bank account or insurance account online, when that wasn’t an option a few years back. The reason is that the user verification or ID checks are done with AI.
However, at the same time, we’re seeing that there is a big problem with the fraudsters using deep fakes to scam people. So a big concern is that fraudsters are adopting these technologies as well, and it’s becoming harder to understand and detect hard issues or hard cases.
What do you foresee as the future trends in EKYC?
The biggest trend that we see right now is data privacy. Data that belongs to a real person, like you or I, should belong only to us. However, when it goes online, a lot of people are trying to use it somehow or monetize it. The trends that we’re seeing is that EKYC is a bit changing. There’s an initiative from Sam Altman (founder of ChatGPT) called Worldcoin. The idea they want to create your digital identity and only you can give permission to the companies who are trying to use your identity for one reason or another, giving you control over your identity.
For right now, this is just an idea. They already invested some money, but right now it’s kind of hard to do, but perhaps in the future, there will be some big database of all the identities from the world where you can easily verify that I am I and you are you.