Alibek Polatov is originally from Shymkent. After finishing school, he enrolled at the Kazakhstan branch of Lomonosov Moscow State University, where he studied systems programming. After graduation, he returned to Kazakhstan and quickly found his path in entrepreneurship. He first launched a web studio focused on website development, and since 2020, he has been building his own startups. This year, Alibek became the CEO and co-founder of shai.pro, a platform that enables users to create AI agents and digital assistants. The company has already raised $6 million in investment.
As part of the joint project «100 Startup Stories of Central Asia» by Digital Business and Astana Hub, Alibek shared why his previous products didn’t take off, how he had to make an emergency flight to another city to keep talks going with an investor, and what sets shai.pro apart from its competitors. We also found out how the startup managed to sign a contract with a subsidiary of Alphabet, how the team sped up its deal-closing cycle, and what plans they have for the coming year.
«I realized you can raise serious money even at the idea stage»
– Alibek, how did your journey in IT begin?
– I graduated in 2013 from the Faculty of Computational Mathematics and Cybernetics at MSU. Pretty soon after that, I got into entrepreneurship. I started a web studio doing websites and digital marketing. But deep down, I always wanted more than just providing services. I wanted to create a real product, something of my own that I could sell and actually make money from.
I got into the startup world in 2020, when my former classmate Magzhan Ikram flew in from Singapore. He told me he had raised $1 million in investment for his IT product — and it was still just an idea. That’s when I realized there’s serious money in this space. I suggested we launch a project together.
At first, we thought about building a chatbot constructor, but we couldn’t raise investment for it. Then we shifted to a financial tool in the digital currency space. But after six months of trying, we still couldn’t get a license from the AIFC, so we had to put the project on hold. Instead, we decided to build our own blockchain-based analytics system. It allowed users to access wallet info, track transactions, and analyze user behavior on crypto exchanges without registration. That’s how the Oraclus project was born.
In 2022, we raised $400,000 from a major blockchain data provider. Then in 2023, we were accepted into the Hero Training program led by Tim Draper. At that time, AI solutions were already gaining momentum in Silicon Valley, and we started thinking about how to integrate AI into our own project. One idea was to create something like Perplexity (an AI-powered search engine — note by Digital Business) but focused on the blockchain space.
But to keep growing, we needed funding. In 2024, we announced a $2 million round. At the time, our global competitors were raising much larger amounts, such as $30 million or even $60 million. We knew it would be hard to move forward without investment. Unfortunately, we weren’t able to close the round, so we had to pause work on Oraclus.
– When did the idea to create AI agents first come up?
– Around the same time, another university friend flew to Kazakhstan from the UK. He was working at the pharmaceutical company GSK in London, where he was involved in developing GenAI products and was already using multi-agent systems.
He explained that agents can communicate with each other, make decisions autonomously, assign tasks, verify their completion, and so on. I was really surprised by that and suggested we build an AI solution that could provide investment recommendations for companies.
At the end of 2024, I shared the idea with the owner of several large companies operating in Kazakhstan, Uzbekistan, Hong Kong, and Cyprus, as well as with Ruslan Adilbaev, the main beneficiary of Uzbekistan’s AVO Bank. He saw potential in it, but it later became clear that other players in the market weren’t quite ready for such advanced solutions. Instead, one of the investment companies we spoke with asked us to implement AI on a more basic level — specifically, to create a chatbot based on local data that would operate in a closed environment. We signed a contract and got to work.
«In our previous projects, we focused too much on building the product when we should have been focusing more on selling it»
– What happened next?
– In March, I met with Ruslan Adilbaev again in Astana. He had just an hour before his flight to Almaty. I started explaining in detail how multi-agent systems work and what they’re capable of. I also told him about the solution we were building for one of the investment companies. We ran out of time, so I quickly bought a ticket for the same flight and flew with him. We kept talking during the flight and at the hotel. The next day, we put together a business plan and signed an investment agreement. Ruslan invested $1 million into the project, which went toward product development and supporting the team. That was the beginning of shai.pro.
– What challenges did you face early on?
– At that time, many company leaders didn’t fully understand how exactly AI could help them. So we approached the problem as rationally as possible. First, we studied the business processes of potential clients. To do that, we selected a focus group of employees and ran workshops to explore how AI could simplify their work. Then we created a list of tasks that AI agents could realistically speed up and deliver visible results.
After that, management would decide which agents to launch first, and we’d get to work. This approach helped us build solutions that were actually useful, rather than implementing AI just for the sake of it.
For example, one of the client requests led us to develop a vertical agent called SHAI Compliance AI, which checks documents for compliance with international and industry standards like ISO, GOST, API, and ASME. What used to take a person several hours, the AI agent can now handle in just a couple of minutes.
However, the experience we gained from previous projects helped us avoid many common mistakes while building shai.pro.
– What kind of mistakes are you referring to?
– The first mistake was the lack of promotion. My partners and I are programmers, and we’re pretty modest people. In the past, we built products but never really talked about them publicly. Even when we raised $400,000 for Oraclus in 2022, we didn’t announce it in the media or on social networks. At the time, it felt like the amount wasn’t big enough to mention. But looking back, if we had made more noise, maybe those earlier projects would have had a future.
It’s also important to be based in the same country where most of your clients are, especially if you’re selling a complex product. With Oraclus, most of our potential customers weren’t in Kazakhstan. We often traveled to places like Singapore, the UAE, and other countries where blockchain was growing fast. But we couldn’t close any major deals. When you’re signing a large contract, it’s crucial to be physically close to the other party.
When opening a funding round, it’s important not to think along the lines of «we’re in Kazakhstan, so everything is cheaper here». If you want to compete globally, you need the ability to hire top talent from around the world. This helps solve complex problems faster and move your product forward. That’s why it’s essential to raise a substantial amount of funding.
Another mistake we made was losing focus. In our previous projects, we concentrated too much on building the product when we should have focused more on selling. With shai.pro, we took a different approach. We pitched the idea to our first client, an investment company, secured funding, and used that money to develop the product.
«We have specialists on our team who’ve built GenAI solutions at international corporations»
– What does shai.pro offer now?
– Our main product is a platform that allows companies, including banks, quasi-governmental organizations, and other large corporations, to independently create AI agents and digital assistants.
We also offer several ready-made AI agents. The first one we developed was SHAI Summarization Agent, which creates concise summaries based on conversations in corporate chats. Later, we added SHAI Report Generator Agent, which can process multiple documents and generate a report based on a given template. There’s also SHAI Voice Assistant Agent, a voice-based tool that transcribes audio files. And that’s just part of the product lineup.
In addition, we’re developing Financial Agent Studio for investment companies, a solution that includes around 10 AI agents designed to simplify the most common tasks for this type of client.
– Do you have many competitors?
– In my opinion, about six months ago there were hardly any competitors in Kazakhstan. But now, companies developing similar solutions are starting to appear. Still, we don’t see that as a problem. First, we’re several steps ahead in terms of expertise. Second, the presence of similar projects actually helps with sales — more and more large clients are becoming aware of AI agents, so we don’t have to explain the value of the technology to potential customers as often as we used to.
– What makes shai.pro different from other AI agents?
– Most of the solutions on the market are more or less similar. That’s why the uniqueness of shai.pro lies in its team and the expertise of our people. We have specialists who’ve built GenAI solutions at major international corporations. Thanks to them, we can respond quickly to market changes and develop truly powerful and useful products.
In addition, most of our products operate on-prem (a deployment model where the product or platform is installed within the client company’s own infrastructure, rather than in the cloud — note by Digital Business). This approach is typically preferred by enterprise clients who handle a lot of confidential and sensitive information. We pay special attention to data security, and we have a dedicated team focused on that.
«We’re discussing contracts worth 100 million tenge per year with some clients»
– Who are the clients of shai.pro?
– We have around 30 clients in our portfolio, including government organizations, quasi-governmental bodies, and financial companies. Most of them are based in Kazakhstan. The average deal cycle is about three months, but some clients are ready to sign the paperwork within a week.
At the same time, selling the platform itself is quite challenging, as many clients still don’t fully understand what kind of solutions they actually need. Once we introduced vertical agents, the deal-closing process became much faster.
In addition, the product isn’t cheap, and enterprise clients need to carefully calculate both the tangible and intangible benefits of using our solutions. That’s much easier to do with ready-made AI agents.
There aren’t many clients yet who buy the platform and build their own agents. Most prefer us to handle everything from setting up the platform with AI agents to configuring the LLM. The technology is still relatively new, and many companies don’t yet have in-house specialists with experience in this area.
– What is the startup’s business model?
– We mainly sell annual licenses, but the final cost is customized and depends on the number and complexity of the agents. For example, until the end of December, you can get a one-year subscription to Salyq.ai, an AI-powered tax advisor, for just 5,000 tenge. After that, the price will go up.
Within the AI agent platform, contract amounts are usually much higher. Setting up and maintaining a single AI agent costs around 10 to 20 million tenge per year. With some clients, we’re discussing contracts starting from 100 million tenge annually.
– Have you reached the break-even point?
– Not yet, but we’re making progress. In our field, it’s hard to calculate MRR because we might sign several large annual contracts in a single month, but the revenue has to be spread over 12 months. By the end of the year, we’ll be able to share exact figures.
«The founder’s main task is to keep the team together»
– How much has been invested in shai.pro so far?
– We’ve raised a total of $6 million for the project from Ruslan Adilbaev, with $5 million of that coming in September. Most of the funds go toward supporting the team and developing the product. Currently, around 70 people work at shai.pro, with 90% of them based in Kazakhstan. We’re actively hiring right now, especially looking for strong GenAI specialists.
– How will the project develop over the next year?
– We’re focusing on R&D to stay flexible and quickly adapt to market changes.
We’re also in the process of finalizing agreements with Kazakhtelecom, which would allow them to offer our solutions to Samruk-Kazyna portfolio companies. Previously, deals were often delayed because many companies simply didn’t have the GPUs and graphics cards needed to run our products. Now the situation has changed. Kazakhtelecom has the hardware, and we have the technology.
We’re also looking into entering the European market. One of our co-founders, Dilshat Uteshev, lives in the UK and meets with potential clients. For example, our solution is currently being actively tested by Galvani Bioelectronics Limited, a joint venture between Alphabet (Google’s parent company) and pharmaceutical giant GSK. Our product completed a task in just a few minutes that used to take the corporation three months. If we gain more clients in the region, we’ll open a local office.
– You’ve been launching products in Kazakhstan for over five years. How do you assess the state of the startup ecosystem in our country?
– In Kazakhstan, there are more and more opportunities each year to grow a profitable product within the country.
In the past, everyone was chasing investment. Now we have venture funds, business angels, and soon a fund of funds will be launched. Accelerators and support programs from Astana Hub and other organizations are being launched regularly. We’ll start seeing the results from those who went through them very soon. I’m confident that a new generation of young entrepreneurs will emerge in Kazakhstan — small teams of 3 to 5 people who will be able to build serious products and make real money from them.
– What advice can you give to aspiring startup founders?
– Don’t go through accelerators just to check a box. If something doesn’t work out the first time, you need to keep going and try new solutions. The reason for failure usually isn’t the program itself. It could be the market, the team, wrong expectations, or investments that were too early or not the right fit. The knowledge and experience you gain from every program stay with you. You just have to put them into practice.
Over the years, I’ve realized that the most important thing in a startup is not letting the company die. The founder’s job is to keep the team together, because these are the people who shape the culture, believe in the idea, and follow your lead. You have to give them the sense that they’re doing something meaningful, that they’re part of something bigger.
It was thanks to the team that we made it through tough times. We looked for ways to earn quickly and cover cash flow gaps. Many startups shut down in those situations. We managed to stay afloat because the team stayed together.