AI startups President Tokayev sees as future unicorns

Surfaice

Surfaice is a tech startup founded by Kazakhstanis in Silicon Valley, USA. The company develops vertically integrated AI employees to automate and manage the construction of standard facilities such as schools, hospitals, energy infrastructure, and retail stores.

Technically, the solution is implemented as a fleet of specialized agents operating through a unified knowledge graph that stores all regulations and activity logs.

The company is led by CEO Alim Uderbekov, an aerospace engineer, former employee of the Mission Control Center for the International Space Station, founder of Central Asia’s largest building materials marketplace, and former consultant at McKinsey and KPMG.

The startup has already raised $1.5 million from leading venture capital firms, including Shadow VC, Sturgeon Capital, iSeed VC, Plug & Play, Most, and others. Surfaice also completed the Astana Hub × Draper University Hero Training accelerator program.

Surfaice’s client portfolio covers projects with a combined construction budget of over $360 million. In the United States, its clients include JD Sports (revenue $13 billion, more than 3,400 locations) and Five Below (revenue $4 billion, 1,800 locations). In Kazakhstan, its AI solutions are already used by Magnum, Jusan Bank, Citix, and other partners.

On July 9, 2025, the team presented its technology to the President, highlighting how Surfaice could help address the shortage of over 280,000 school places (about 500 schools) through AI-driven construction management systems.

Following the meeting, instructions were given to work on a pilot project with government participation. The company is now launching a pilot to build three schools in Almaty with fully AI-automated processes.

CEO Alim Uderbekov told Kazinform News Agency about short-term goals of the project:

“In terms of scaling, our goal over the next two to three years is to become the leader in AI software for retailers that build standardized stores in the United States.”

He explained that their vision is for the AI agent to take on every stage of a client’s construction process.

“Our goal is for our AI agent to handle any task involved in the construction projects of our clients. For example, a McDonald’s manager could say: “Calculate the budget,” “Prepare the project schedule,” “Initiate document signing,” “Launch the tender,” and so on.”

The core idea behind their vertically integrated AI agent, he added, is similar to how tools like Cursor or ChatGPT can be used to write code, with ChatGPT aiming to deliver a complete project, such as a website, from start to finish. In their case, they view a standard construction project as the equivalent of a landing page in the world of software development.

The team consists of 16 members, including PhDs, experienced engineers, international competition winners, and young talents. The main engineering team is based in Almaty, Kazakhstan.

Assessing the competitive landscape, Uderbekov emphasized:

“Competition is quite strong. Whenever you create a disruptive project that aims to transform existing approaches, you inevitably face significant market resistance. No one will simply hand you the market. But we believe we can become the leader because Surfaice is essentially creating a new category. Until now, there has never been software based on artificial intelligence designed specifically for building standardized projects. It simply did not exist. We are now defining both the concept and the term.”

WDSOFT

This AI-powered system is designed to diagnose lung cancer, breast cancer, and COVID-19 viral pneumonia. With a sensitivity rate exceeding 90 percent, it improves early cancer detection by 15 percent. The AI system is already in use in medical institutions across several regions of Kazakhstan, including Atyrau, Almaty, Zhetysu, Zhambyl, Abay, and East Kazakhstan. In the Almaty region, the system’s introduction has already led to an 11 percent increase in pathology detection.

The technology follows a multi-step process. It begins with training on thousands of pre-annotated medical scans containing both normal and pathological cases, enabling the AI to learn the defining characteristics of cancerous tumors. The system then pre-processes new images to improve quality, reduce noise, and enhance contrast, before analyzing tissue structures and density to identify anomalies such as nodules or shadows. The Japanese oncology institute contributed early-stage radiology images and data for algorithm development.

Once potential anomalies are detected, the AI classifies them using contour and texture analysis to distinguish between malignant and benign growths. By assessing shape, boundaries, and density, and drawing on CT and MRI data, it calculates the probability of malignancy. The system then cross-references its findings with millions of previously processed scans, further improving diagnostic accuracy through its growing database.

The AI outputs a probability score, such as “85% likelihood of malignancy”, and highlights suspicious areas, directing the doctor’s attention to critical details. It also continuously improves through feedback from physicians, increasing both its accuracy and efficiency over time.

Among its key advantages are rapid processing, handling large volumes of scans in just minutes, accuracy rates above 90 percent, and the ability to detect subtle changes that may be missed during visual examination.

Yerlan Zhanibekov, commercial director of WDSOFT, told Kazinform News Agency about the project’s uniqueness:

“Our project is unique in its scale and has no direct analogues. While similar AI systems exist worldwide, only in Kazakhstan has such technology been deployed to cover an entire region and its entire population. The centralized use of AI across all medical institutions in the region, including remote districts, enables large-scale coverage of the country’s population.”

CodiPlay

CodiPlay is an education startup that makes learning to code accessible and engaging through a smartphone. Its ecosystem brings together three core products: the CodiPlay mobile app, CodiKit robotics sets, and the CodiTeach web platform. This integrated approach supports both students and teachers, creating a complete learning environment.

The CodiPlay app introduces popular programming languages — Scratch, C++, and Python, through game-based learning. Arcade-style challenges turn coding into an exciting experience, helping children grasp complex concepts more easily. With CodiKit sets, students can design their own Internet of Things (IoT) projects, building not only technical knowledge but also problem-solving and engineering skills.

The CodiTeach web platform is designed for teachers, helping them plan lessons, track progress, and build curricula. It makes the teaching process much simpler and more effective. Together, the CodiPlay products foster digital skills and IT literacy among children.

The startup has already expanded internationally, with its solutions adopted by schools in Kazakhstan, Uzbekistan, South Korea, Qatar, the United Kingdom, Saudi Arabia, Azerbaijan, and Indonesia.

Its impact has also been recognized globally: CodiPlay was awarded Digital Public Good (DPG) status by the Digital Public Goods Alliance (DPGA), an initiative backed by the UN Secretary-General. This acknowledgment made CodiPlay only the third company in Kazakhstan to receive the title.

AI Kitap

Another startup President Tokayev referred to as a potential unicorn is AI Kitap, a platform for digital AI-powered textbooks that provides personalized learning tailored to each student’s knowledge level, performance, and individual preferences.

Earlier, Kazinform News Agency reported on how algorithms shape the new digital landscape.