The other day, I had the honour of moderating a panel discussion at Forbes Tech 2025, dedicated to one of the most pressing issues of our time: AI implementation strategies. Together with leading industry experts – Oleksii Molchanovskyi (UCU) and Oleksandr Krakovetskyi (DevRain, Microsoft AI MVP) – we sought an answer to the question: is the accelerated adoption of AI a sprint or a marathon?
The discussion proved to be deep, and instead of dry forecasts we got several fundamental insights that every leader should understand.
AI is not about "yesterday", it’s about evolution. Before talking about the future, it’s worth looking back. The very term "artificial intelligence" first appeared back in the 1950s. We’ve gone from simple if-else algorithms to the neural networks of the 2010s and today’s boom in generative AI. In other words, it’s not sudden magic, but technological evolution. So the question of whether to use AI is no longer on the table. The question now is – how exactly?
The end of the Prompt Engineering era. One of the most interesting conclusions of the discussion: The profession of "prompt engineer", which has been so widely discussed over the past two years, is virtually disappearing. The models are becoming smarter; they’ve learned to "reflect" and understand even imperfect queries.
Context Engineering is taking its place. Oleksandr Krakovetskyi aptly noted: "The issue isn’t how to ask, but what the model knows about your business. Without your internal data, even the smartest model is just an encyclopedia that doesn’t understand your company’s context. Businesses need to focus on building data infrastructure to "feed" AI the right context.”
Clean data is the foundation. We agreed with Oleksii Molchanovskyi: "If your data is "in Excel on Dropbox" in a chaotic state, AI won’t help you. Garbage in – garbage out. However, there’s a life hack: you can use AI itself to audit that data. Even if you’re not yet ready for full-scale agent deployment, use technology to bring order to your company’s digital ecosystem.”
Who is responsible for mistakes? A painful question for the corporate sector: Who’s to blame if AI messes up? Here, experts are unanimous: Responsibility always lies with the human (Human in the loop). You cannot delegate responsibility to an algorithm. AI is a tool, like an excavator or a calculator. If you operate this tool, the outcome – both profit and loss – is on you. This means businesses must learn to manage risks, rather than shun technology out of fear of mistakes.
Sprint or marathon? So what should you do – rush headlong or wait until the technology ‘matures’? The answer is: do both. It’s both a sprint (you need to test hypotheses quickly) and a marathon (you need to build long-term infrastructure). You shouldn’t wait for the "perfect AI" or an "AI winter". Technology is evolving exponentially. While some wait for stability, leaders are already integrating multi-agent systems, where different AI models debate among themselves to find optimal solutions.
In summary: AI has already become part of our lives. The recipe for success for Ukrainian businesses today:
1. Not be afraid to experiment, but manage the risks.
2. Invest in the cleanliness of your own data (it’s your main competitive advantage).
3. Foster a culture of responsibility, where humans steer AI rather than hide behind it.
Move forward with the right partners, hire experts and remember: The winner isn’t the one with the best model, but the one who has integrated it best into their processes.
Maksym Balaniuk, Director of Innovation & Growth, Metinvest Digital
