Petr Zelenka: Get Rid of AI FOMO

May 28, 2026

According to Gartner data, AI agents will be making 15 percent of everyday business decisions within two years. Fresh research also shows that companies investing in data preparation and analytics for AI tool deployment achieve up to 65 percent better results, ranging from revenue growth to cost savings.

“It is clear to everyone that artificial intelligence will fundamentally change the way business works,” says Petr Zelenka, CEE Lead of the AI division at technology company Adastra, which specialises in developing advanced solutions, connects AI teams around the world, and drives Adastra’s AI business development at a global level.

AI is reshaping the business world, and companies, under pressure from reports about the miraculous capabilities of modern tools, easily fall into a panic that they are falling behind. Zelenka warns of a trap that many companies fall into as a result: they succumb to a sense of FOMO and rush headlong into many projects promising quick results but ultimately delivering minimal value. “The result is a lot of uncontrolled and expensive projects that never deliver real impact,” warns Zelenka. “Successful companies take an almost cold-bloodedly systematic approach. That way they don’t waste money on something they’re not capable of completing,” he adds.

“It is time to stop in that race, take a breath, and assess what is needed to move AI from being a mere assistant in individual processes to a new operational layer that streamlines the functioning of the entire company,” Zelenka explains.

According to Gartner, only four out of ten companies pay sufficient attention to these fundamentals, and last year’s survey also showed that only just under half of ongoing digital initiatives go on to meet or exceed expected business goals.

Google together with Ipsos add further figures from the US in their February report The Path to AI Fluency: 40 percent of employees use AI at work, but only five percent rank among “advanced” users who can truly leverage AI tools to increase efficiency and productivity.

The Ipsos and Google research also reveals that 27 percent of companies provide their people with AI tools but give them no instructions, guides, or rules to go with them, while 37 percent of companies provide guidelines but do not offer adequate tools. Only 22 percent of companies give their people both.

Build the Foundations

Petr Zelenka adds that it is essential for senior company leadership to be involved in the AI strategy, not just the IT department. This is especially important because today it is closely tied to overall business strategy.

He cites the Slovenian-origin NLB Group as an example. It is a leading banking and financial group headquartered in and with exclusive strategic interests in Southeast Europe, with profits exceeding the equivalent of 12 billion Czech crowns, and for which Adastra served as one of the partners in driving strategic change.

The group employs over 8,000 people who already had experience with tools such as Microsoft Copilot, but the goal was a shift toward more autonomous and scalable AI agents. “We identified approximately 500 use cases where AI agents could speed up work and save time for employees,” says NLB Group CIO Dejan Pust.

“At the same time, we were observing promising pilot projects in the market that had to be rolled back due to regulations and banking security requirements, with their costly development starting over from scratch. It was therefore clear to us that we first needed to develop a foundational system that would meet all requirements and subsequently reduce costs and accelerate the development of individual projects,” explains Pust.

So they started by developing a platform – a kind of backbone – through which the company now manages all its AI initiatives. It makes it possible to keep everything under control, from access and model management through cost monitoring to auditing every interaction.

“They first worked through the hard, tedious part that doesn’t appear to bring much value at first glance,” remarks Petr Zelenka, noting that such an approach is not common and companies often start from the other end – with individual projects.

“Those projects do generate initial enthusiasm, but disillusionment soon follows. Projects without a foundational platform frequently run into problems such as the availability of properly grounded data or security, legislative, and regulatory requirements specific to their industry.”

Control Costs and Boundaries

A well-built foundational platform also makes it possible to bring so-called shadow AI under control, where employees connect their own agents to the company’s system. According to both Pust and Zelenka, banning employees’ own tools is not the ideal approach. “However, it is necessary to have control over every point through which tools are connected to the system,” stresses Ales Gorisek, IT and AI architect at NLB.

NLB Group is therefore building, alongside the system for its own generative and agentic AI tools, an AI Citizen platform where employees can connect tools that make their work easier but do not use internal company data. “There always needs to be proper governance in place for the platform – a kind of AI constitution setting out what tools and employees must comply with when connecting to the system,” adds Zelenka.

Building similarly robust foundations naturally costs a considerable amount of money. “But it needs to be seen as a strategic investment and not viewed solely through the lens of short-term returns,” notes Ales Gorisek.

“In our industry, investment in building an AI system is comparable to investment in a digital banking platform. And in the end, it will bring a similar improvement in efficiency,” believes Dejan Pust. “The real return, however, will be calculated once we launch and scale individual projects, which we are planning over a two-to-five-year horizon,” he adds.

NLB Group managed to build the platform in just five months, and the rollout of follow-on projects and innovations should now be faster across all companies in the financial group – tenfold faster, according to Gorisek.

Zelenka points out that investment in a strategic approach to AI also brings a competitive advantage. According to Gartner, by 2029 AI agents will automatically handle 80 percent of routine customer queries and reduce companies’ costs by 30 percent. NLB’s platform will enable faster responses to market changes, better evaluation of customer experience, and the use of data for business decisions.

“We also plan to incorporate multi-agent tools for even greater efficiency. Within one to two years, we would like to reach the point where we can allow agents to also change data based on their decisions, and thereby further increase the productivity of our people,” outlines Dejan Pust.

The interview was published by Forbes Czech Republic.