Why do companies struggle to move AI into production and core business logic?

At the December AI roundtable, we reviewed perspectives from MIT, McKinsey, and Gartner—and it quickly became clear they all point to the same pattern:

🗝️ AI usually enters organizations through individual functions (marketing, support, analytics, code writing).
🗝️ It’s often driven by individuals or enthusiastic teams.
🗝️ But if AI is meant to change the business model or core processes, it needs a clear mandate from the top—and must be treated as organizational transformation, not just a technology initiative.

A strong debate followed about the reality inside companies:
– The CEO wants AI “at any cost,” and the budget exists,
– but it’s unclear where the initiative should sit,
– who owns it end-to-end,
– how it cuts across departments,
– what ROI looks like and who defines it,
– and what needs to change in processes and IT.

Peter Chrenko summarized it by framing AI as a spectrum of capabilities—from prediction and better decision-making, through cognitive and creative support, all the way to agent-based systems. Without understanding which layer you’re operating in, it’s impossible to identify meaningful use cases or measure their impact.

Examples from manufacturing resonated strongly as well (thanks, Petr Bergl), showing that without clearly defined processes, even digitalization fails—let alone AI optimization. Once you touch one cog in the machine, the entire system moves: quality often improves, but performance can drop in the short term.

AI is a stress test of organizational maturity.

Thanks to everyone for the open and honest discussion. Special thanks to Patrik Horny, Jan Mlynar, and Barbora Liegertová for the organization and preparation.

And if you’re curious where your company truly stands on this journey, try an AI maturity check:
https://lnkd.in/evA8hHXD