Enterprises in Central Asia and the Caucasus are bringing artificial intelligence into production at double the pace of their counterparts in the rest of the world, according to a KPMG report. 53% of companies surveyed in the region said they had achieved industrial-scale AI implementation with measurable returns, compared to 24% globally.
KPMG interviewed more than 2,500 technology executives from 27 countries to produce the report, including 40 respondents from the Caucasus and 32 from Central Asia. Konstantin Aushev, partner heading KPMG Caucasus and Central Asia's Technology practice, highlighted an important caveat: some respondents may have interpreted the concept of "production" broadly and included fragmented cases such as text and voice bots.
Despite this clarification, there are examples of large-scale projects in the region. Kazakhtelecom, Kazakhstan's national telecoms operator, has incorporated AI since mid-2025 in customer service, procurement, human resources, finance, document management and workplace safety. Yerzhan Assanov, head of AI at Kazakhtelecom, explained that the implementation model is hybrid, with a central team handling complex platforms and an agent creation platform for minor tasks.
Business & Technology Services (BTS), the IT company of the Eurasian Resources Group, uses computer vision to verify the operational status of equipment in mining operations and data analysis to process geological reports. The KPMG report concluded that 44% of companies in the region opt for proactive experimentation, compared to 21% globally.
Rustem Uvaliyev, head of BTS's Digitalisation department, said his company maintains AI pilots on a continuous basis, with around 10% of generative AI pilots reaching production and 80% in the case of "classical AI" such as machine learning and computer vision. Kazakhtelecom launches pilots of new AI solutions almost every month, although it considered it premature to draw conclusions about the pace of industrial implementation.
Assanov noted that at this stage they evaluate the reliability of solutions with real data, how employees use them, and what changes are necessary before scaling. BTS developed a self-service platform called FAIR that allows employees to create digital agents using generative AI without requiring advanced technical training.
Both companies are experimenting with "vibecoding" platforms where employees describe the needed solution using natural language and assemble it quickly without configuring servers or complex technical environments. Aushev said that a couple of years ago the regional approach to AI was more chaotic, but now there is greater understanding of where artificial intelligence is needed.
43% of companies in the region are betting on "fast but selective adaptation", compared to 36% globally, according to KPMG. Assanov said that over the past year Kazakhtelecom's approach became bolder, launching experiments sooner and testing them faster in real tasks, with prior process analysis and introduction of stricter quality requirements.
Uvaliyev warned that "hype" and pressure to implement AI persists, noting that it is important to recognise that it is not a universal solution. He also warned of the risk of scaling AI without assessing costs, a problem that many large international companies already face.
Aushev attributed part of the rapid AI adoption in the region to its relatively high level of digital penetration. He noted that many companies already have the infrastructure and experience necessary to connect systems with data and external sources, facilitating the integration of AI into existing services.
