The leadership gap holding back enterprise AI

Executive Summary

Most organizations are treating AI like a technology upgrade. Raf Postepski, Senior Director at Alvarez & Marsal, explains why the real transformation starts when leaders stop layering AI onto existing structures and start redesigning how their business operates, decides, and scales across markets.

Drawing on nearly two decades of experience advising Fortune 500 and mid-market firms, Raf outlines a familiar pattern. AI initiatives are multiplying and experimentation is accelerating but for global organizations, the gap between experimentation and advantage is widening. Without a unifying strategy, governance model, and operating redesign, value remains fragmented.

His core message is clear, AI changes how decisions are made. Organizations that succeed will redesign around data, decision intelligence, and customer outcomes. Those that don’t will accumulate complexity faster than advantage.

From banking cubicles to enterprise transformation

Raf Postepski’s career did not begin in AI. It began in early-2000s banking in Toronto, inside rigid operating structures that left little room for curiosity. A move to Dun & Bradstreet shifted his thinking toward data-driven outcomes, and that thread pulled him to London, into e-commerce, and eventually into large-scale operational transformation across Asia. 

“That was kind of it,” he says. “I was hooked.”

From there, digital transformation consulting followed, eventually evolving into the AI strategy work he leads today at Alvarez & Marsal.

Why AI feels familiar and why that’s dangerous

One of Raf’s most important observations is historical. He sees AI today in the same place enterprise technology sat roughly 15 years ago.

Back then, technology lived inside operations. It solved functional problems like email, printers and infrastructure. Over time, leaders realized technology was not just operational, and CIO roles expanded as digital became a board-level concern.

“I think we are now at a point where we are seeing that with AI,” Raf explains.

The danger is many organizations are making the same mistake again. Treating AI as something to layer onto existing structures rather than something that reshapes them.

AI is following the same path as enterprise technology

Today, AI often sits inside IT or under the CIO by default and that placement limits its impact.

“If you really want to truly exploit what AI is capable of,” Raf argues, “you need to carve it out of technology and have its own pillar.”

Just as enterprise technology broke free from pure operations, AI must break free from being governed solely as a technical function. Its impact extends across strategy, operating models, decision-making, and organizational design.

Why layering AI on top doesn’t work

A recurring misconception Raf encounters is the belief that AI can simply be added to existing processes without rethinking what surrounds them.

“Too many leaders think you can just layer AI on top of the existing operating model, and you don’t need to change much.”

In reality, AI reshapes how work is done across the organization. Processes change, roles evolve and decision rights shift in ways that demand redesign rather than incremental adjustment. 

AI strategy, in Raf’s view, must inform business strategy, rather than sit alongside it.

Strategy before experimentation

Raf is not anti-experimentation, but experimentation without direction creates noise.

Organizations need a clear AI strategy that cascades beyond the boardroom. Leaders at every level of the organization, from direct reports to frontline managers, must understand what AI means for their day-to-day work and how it changes the way they contribute.

He describes a workshop where leaders believed an AI vision existed, but could not translate it into anything actionable.

“They said, we have one. But we have no idea what that means.”

Without clarity, enthusiasm turns into scattered pilots and duplicated effort. Strategy provides the coherence that experimentation alone cannot..

AI as augmentation, not replacement

Raf is explicit about AI’s role today.

“We see AI very much as an augmentation, not pure automation.”

The immediate risk is displacement, where those who do not understand how to work with AI will struggle. Those who do will move faster, think broader, and make better decisions.

The competitive divide will emerge through capability, not headcount.

Mid-market speed versus enterprise inertia

Raf sees a sharp contrast between mid-market firms and large enterprises.

Mid-market organizations benefit from minimal legacy, allowing them to move quickly, adopt new architectures, and experiment without constraints that will slow them down. Their weakness is often long-term planning and governance, the strategic scaffolding that keeps momentum from becoming chaos.

Enterprises face the opposite challenge. Legacy systems, data silos, and organizational fatigue slow progress before it begins.

“They see AI as another layer of complexity,” Raf notes. “That becomes a point of exhaustion.”

Why data matters more than systems

A key shift Raf emphasizes is perspective. Business systems are no longer the source of advantage. Data is.

“I see them as just boxes holding data. What I care about is the data that it holds.”

From an AI perspective, systems matter only insofar as they enable data access, fluidity, and reuse. Organizations that focus on extracting and orchestrating data gain speed and flexibility that system-centric thinking cannot deliver.

Decision intelligence, not just efficiency

Most AI initiatives today focus on efficiency. Doing things faster, cheaper, or at greater scale.

Raf argues that this is only half the value.

AI also enables better decisions. Scenario modelling, forecasting, and causal analysis allow leaders to explore outcomes before committing resources.

“It’s the decision science,” he explains. “That’s the part people haven’t gone all in on yet.”

He cites pharmaceutical clients using AI to improve clinical trial feasibility, accelerating time to market and improving go-to-market decisions in an industry where first-mover advantage is enormous.

Speed with thought, not speed alone

Speed without judgment creates risk and Raf frames AI as enabling what he calls “speed with thought.” This is where machines expand the decision space leaders can explore, testing assumptions without committing capital prematurely.

“The machines are going to give me the opportunity to better scenario model,” he says.

Governing AI across global complexity

Global organizations cannot adopt AI uniformly  and this is where the challenge of scaling across markets becomes most acute.Regional regulation, data sovereignty requirements, and cultural context  all shape what is possible and how quickly teams can move.

Middle East data sovereignty requirements, European AI Act risk classifications, and GDPR obligations each introduce constraints that demand localized governance rather than centralized mandates.

“You can’t have a blanket approach,” Raf warns.

AI governance must be adaptive, regionally informed, and embedded into operating processes rather than enforced after the fact.

For companies expanding internationally, the implication is significant. AI governance must be adaptive and regionally informed, embedded into operating processes rather than enforced after the fact. The same principle applies to every layer of how a business communicates across markets, from the data it uses to the language it speaks to customers.

Build for advantage, buy for parity

One of Raf’s clearest principles concerns differentiation.

Models themselves are becoming commoditized and the value no longer  sits in the algorithm.

“The model is just a dumb parrot,” he says. “It’s not thinking.”

Differentiation comes from what surrounds the model. Proprietary data, orchestration layers, business logic, and process integration are where competitive advantage actually lives.Organizations should build where advantage matters and buy where parity is sufficient.

Where differentiation really lives

As models and agents become interchangeable, architecture and data strategy become the deciding factors.

“You can buy the model,” Raf explains. “But you still need to have it looking at the right processes and the right data.”

Competitive advantage shifts upstream into data organization and downstream into experience, and the experience layer becomes decisive.

AI’s impact ultimately surfaces in how customers interact with businesses,regardless of where those customers are or what language they speak.

Raf sees the middle of the value chain becoming less visible. What matters are the edges and how data is translated into decisions, and how experiences feel to customers.

“It’s how you bring people closer to it in an organic, seamless way.”

For organizations operating across markets, this means the experience layer must work consistently in every language and cultural context. The seamlessness Raf describes requires more than good technology; it requires infrastructure that adapts to how customers actually engage, wherever they are.

This elevates ethical considerations, transparency, and trust as AI-generated interactions become more human-like.

Quality of interaction over quantity

Personalization, Raf argues, must mature beyond volume..

“It’s going to be quality of interaction versus quantity of interaction.”

AI enables restraint as much as reach, and knowing when not to communicate becomes as valuable as knowing when to engage. Human-centric design, informed by behavioural models and context, determines whether AI enhances trust or accelerates fatigue.

The mindset leaders must adopt now

Asked for a single mindset shift, Raf does not offer a technical answer.

“Understand what you actually want and are asking for.”

Leaders must accept uncertainty, admit gaps in understanding, and remain curious enough to keep learning as AI evolves faster than any individual can master. Progress depends on humility, experimentation with intent, and openness to learning from unexpected sources, whether that means leaning on junior team members who understand the technology or engaging peers who are further along the journey.

“Have the humility to say, no, I actually don’t understand this. And that’s ok.”

Reframing transformation

Raf Postepski’s message reframes what AI transformation actually requires. What matters is how organizations think, decide, and adapt, particularly as they scale across borders, languages, and regulatory environments. AI rewards those who move deliberately, redesign intelligently, and remain curious long after the first experiments launch. 

The future belongs to organizations that treat AI as a catalyst for rethinking how they create value in every market they serve.