Why I Made AI Fluency a Job Requirement at Kentico

You can tell a company is serious about AI by how it hires. 

For most of the past year, I have been asking myself that question about Kentico. We have invested in tools. We have set up internal training. We talk about AI in every leadership conversation. But our hiring process and our promotion criteria did not actually require any of it. 

So, a few weeks ago, I changed that. As of May 1, every new hire at Kentico must demonstrate AI fluency to be hired. As of the July promotion cycle, every internal promotion requires it. By the end of 2026, every person at Kentico will need to meet a fluency bar appropriate to their role. 

It is the most explicit policy I have set as CEO. It is also the one I should have set earlier. 

The Piece I Was Missing 

Last week, I argued in this blog that AI tools producing only 10 percent productivity gains are a cost, not a multiplier. The unspoken assumption in that argument is that your people can tell the difference. They can look at a workflow, identify where AI actually compounds value, and walk away from the cases where it doesn't. 

That capability is not a soft skill anymore. It is the job. If you are running a software company in 2026 and your team cannot make that call, no amount of tool investment will save the math. You will end up paying enterprise prices for novelty. 

The reason I had to formalize the requirement is that the market stopped giving software companies room to drift. The big tech layoffs of the past year were not isolated events. Block let go of roughly 4,000 people, around 40 percent of its workforce. Atlassian let go of 1,600, reportedly including their CTO over AI fluency concerns. These are not panic moves. They are companies acknowledging publicly what most leaders are still saying privately: the same work can now be done with fewer people, and the market expects it. 

When a company my size leaves AI fluency as a vague preference, we will find ourselves on the wrong side of that math, fast. 

Carrot and Stick, On Purpose 

The framework I have been using internally to manage this transition is old and simple. Carrot and stick. I am explicit about it with the team because I think the honesty matters. 

On the carrot side, we have tried to remove every reasonable barrier to learning. Everyone at Kentico has access to frontier tools, including Microsoft Copilot, Claude, ChatGPT and others, name it. We created new roles specifically for the transition: an AI enablement advocate inside R&D, an AI officer at the company level, and 2 dedicated internal AI developers. 

Every Friday is set aside for AI experimentation and learning, with no project pressure and no expectation that you produce something useful. We have published practical guidance on which tools are approved and how to actually use them. 

On the stick side, we have been just as deliberate. We have an explicit hiring freeze, which means we are intentionally limiting headcount even as the business grows. AI usage shows up in every performance review. Team leaders across functions are actively pushing adoption rather than waiting for it to happen on its own. And then the new requirement on top of all of that: fluency for new hires, fluency for promotions, and within this year, fluency for everyone. 

Removing barriers is not enough. People also need to know that the floor moves. 

What "AI fluency" Actually Means at Kentico 

The part I want to be most precise about is what AI fluency actually means at Kentico. It is not a single corporate standard. We do not have one test, one certification, or one definition of what fluent looks like. 

What we have is a structural decision: each team lead and department head defines what fluency means for their function. Fluency for a developer is not fluency for a support engineer. Fluency for someone in sales is different again. The bar is set close to the work, by the people who understand the work, and it is expected to change as the tools and workflows change. 

That decentralization matters more than the policy itself. A centralized AI fluency standard would have been performative. A decentralized one forces leadership across the company, not just at the top. Every manager has to think, seriously, about what their team needs to be able to do with AI, and then defend that definition. If a department head cannot define fluency for their own team, that is itself a signal worth paying attention to. 

Why I Went This Far 

I went this far because AI adoption is not a training problem. It is a leadership problem. In every company I have seen wrestle with this, the people closest to the work are ready to experiment. The blocker is usually higher up. The team lead who is not personally fluent. The executive who treats AI as a tools-team responsibility. The manager who does not know what to ask for. Setting a fluency bar at every level of progression, including promotions, is how you stop that pattern from quietly taking hold. 

It also clarifies what the company is paying for. If we are going to invest in better tools, dedicated AI roles, and protected time for learning, we have to expect the corresponding shift in how work gets done. A vague encouragement does not produce that shift. A requirement does. 

I do not enjoy setting hard policies. I prefer expectations that emerge from culture. But I have stopped pretending that the AI transition will happen at the speed we need it to if we leave it optional. 

So here is the question I would put to other CEOs. You have all said publicly that AI matters. The question is whether you are running your hiring, your promotions, and your performance management as if it matters. 

If not, your AI strategy is a slide. 

Make it a requirement.