Thought Leadership
IT departments are driving AI transformation across the rest of the organization — while struggling to evolve their own. The consequence is already visible: employees are simply solving the problem themselves.
There is an old saying: the shoemaker’s children go barefoot. In other words, those who spend all day building solutions for others rarely find the time to improve their own. That is precisely what many IT teams are experiencing today.
IT departments are building, securing, and managing AI initiatives across business units, yet often lack the capacity to modernize their own way of working. On the surface, this looks like a resource issue — and it is. But underneath lies something more structural: for years, IT has been trained to respond to the needs of others. The CRO demands faster sales processes, the COO pushes for operational efficiency, HR wants better candidate experiences — and IT is expected to make it happen. In that constant delivery role, there is little room left to ask a fundamental question: what does IT itself actually need?
The IT That Has No Time for Itself
This is not about a lack of intent. IT organizations are historically shaped functions that have repeatedly absorbed major shifts: cloud migration, evolving security requirements, new operating models, and an ever-growing number of systems and integrations. The result is an IT function that is, in many cases, simply overloaded.
That is exactly what makes the situation so difficult. IT is not only externally stretched — it is also internally cautious. Historically, IT has been the organization’s natural skeptic. Too often burned by overhyped technologies or left to deal with the fallout of rushed decisions made elsewhere, IT has learned to measure twice and cut once. That discipline built modern enterprise IT. But in the context of rapid AI transformation, it can also slow everything down.
There is also a cultural dimension. IT leaders tend to be more structured and risk-aware than other executive functions. That is usually an advantage. But in the context of AI adoption, that same mindset can result in organizations moving too slowly.
Bring Your Own AI: Employees Are Moving Ahead
So what happens when IT cannot keep up? Employees start bringing their own AI tools into the workplace.
And this is no longer a fringe phenomenon. A recent UK study by NTT DATA Business Solutions, surveying 1,657 employees, found that 53% already use personal AI tools at work. Among employees aged 25 to 34, 75% have bypassed official systems to use their own AI workarounds.
This is no longer traditional shadow IT. In the past, entire departments like marketing or sales would adopt tools without IT approval. With Bring Your Own AI, the dynamic is different: it is personal, decentralized, and far harder to control. It is no longer departments making decisions — it is individual employees independently choosing the AI tools that make them more productive.
The signal is clear: the workforce is often ahead of the official enterprise infrastructure. Many employees now see AI as a personal productivity upgrade. And if companies are unwilling to invest even small monthly costs per user for AI tools, they risk losing control over their AI adoption strategy entirely.
The Real Risk: More Than Just Loss of Control
Bring Your Own AI is not driven by malicious intent. It reflects frustration and organizational inertia. But it introduces real risks: unvetted tools, private accounts, and uncontrolled data flows. Security, compliance, and governance quickly become weak points — not because employees are careless, but because the organization has not kept pace.
The most dangerous outcome, however, is a credibility gap. If other business functions begin to be seen as the real AI innovators while IT lags behind, IT risks losing its role as a strategic enabler. Instead of shaping transformation, it becomes a reactive service layer in a shift led by others.
Three Steps That Matter Now
The way forward is not complicated—but it requires action rather than perfection.
First: start simple. AI agents can already be integrated into existing IT processes without waiting for a perfect architecture. One immediate use case is ticket deflection in IT service management. Many support requests are repetitive. Preconfigured agents can already resolve a significant share of them today, freeing up human capacity immediately.
Second: improve data quality. AI effectiveness depends directly on the quality of underlying data. In many organizations — especially outside finance — data quality remains poor, with outdated records and unreliable datasets. Any AI rollout must address this in parallel.
Third: think toward a more autonomous IT model. Not as a sudden overhaul, but as a direction. Technology is already ahead of many processes and cultural structures. Starting now lays the foundation to lead tomorrow, not just support.
Agentic AI will eventually impact every part of the enterprise. That is no longer in question. The real question is who sets the direction. IT should be in that role — but leadership cannot come from behind.
