Eng. Yaser Alshehhi, NEP-AI expert in the AI strategy development sector, discusses closing the gap between AI experimentation and impact in this exclusive op-ed.
Walk into almost any boardroom today and you will hear the same claim: “We are already using AI.” Dashboards show pilot projects. Slide decks list use cases. Budgets have been approved, vendors engaged, proofs of concept delivered. By every visible measure, adoption is underway.
And yet, when you ask the harder question, what has AI actually changed about how this organisation performs? the room often goes quiet.
This is the gap defining this stage of the AI era. Not a gap in ambition, and increasingly not even a gap in access to technology. It is a gap between experimentation and impact. Across sectors, organisations have become remarkably good at starting AI initiatives and far less consistent at finishing them in a way that moves a financial or operational needle. Pilots multiply. Value does not compound.
Having worked across sectors on AI strategy, I have come to see this less as a technology problem and more as an organisational one. The tools have matured faster than the operating models built to use them. Closing that gap requires a deliberate shift in how organisations think about AI, from a portfolio of experiments to a system for delivering value.
Why pilots stall
Most AI pilots are not failures in a technical sense. Models work. Proofs of concept prove the concept. What they consistently fail to do is survive contact with the organisation around them.
Three patterns show up again and again. First, use cases are chosen for feasibility rather than value, organisations pursue what is easiest to prototype, not what is most consequential to the business. Second, capability is built around the pilot rather than the enterprise, so a successful use case in one team rarely transfers to another. Third, governance is treated as a final checkpoint rather than a design input, which means questions of accountability, data ownership, and risk are addressed only after a model is already in production, often too late to do so well.
The result is an organisation that can point to dozens of AI initiatives and still struggle to name one that changed a P&L line, a customer outcome, or an operating cost in a measurable way.
From access to activation
Deloitte’s 2026 State of AI in the Enterprise report describes a similar shift: most organisations have deployed AI tools, but far fewer have achieved meaningful usage. The gap between availability and adoption is now a primary barrier to value. While the report draws on a global survey of more than 3,200 business and IT leaders, including 130 respondents in the UAE, its findings point to a broader challenge that is highly relevant to organisations operating in the region: moving from access to meaningful activation.
Organisations that achieve stronger results tend to look beyond simply providing AI tools and focus on how employees incorporate them into their daily work. Encouraging teams to explore practical applications, exchange lessons learned and demonstrate tangible improvements can help build wider organisational engagement. Leadership support remains essential, but lasting adoption also depends on employees recognising how AI can improve the way they work.
Deloitte’s research also highlights the importance of preparing the organisational and technical foundations needed to scale AI initiatives. Addressing integration requirements, data access, and operational challenges from the beginning can help organisations move beyond experimentation. These considerations become increasingly important as AI is integrated into physical operations, including robotics, connected devices, and industrial equipment. For organisations in the UAE, designing AI initiatives with long-term implementation in mind can help translate successful pilots into measurable operational improvements and more sustainable returns.
From use cases to value cases
The shift from adoption to value starts with a change in vocabulary: from use cases to value cases. A use case answers “can AI do this?” A value case answers “should we, and what happens to performance if we do?”
Building a portfolio of value cases means prioritising ruthlessly. AI applications that deserve investment are the ones that should be selected against clear criteria, the size of the opportunity, the organisation’s actual readiness to execute, and a credible path from pilot to scale. This is uncomfortable work, because it means saying no to interesting experiments in favour of a smaller number of initiatives with real economic weight behind them. But it is precisely this discipline that separates organisations generating measurable returns from AI and those still funding a growing list of proofs of concept.
Capability, not just capacity
Turning a prioritised value case into delivered value requires capability that most organisations underinvest in: not more data scientists, but the connective tissue between technical teams and the business functions that own outcomes. This means product owners who can translate a business problem into an AI-ready one, operations leaders who can redesign a workflow around a model’s output rather than bolting the model onto an unchanged process, and critically, a layer of leadership fluent enough in AI to make sound investment and risk decisions without needing every detail translated for them.
This last point deserves particular emphasis. The organisations pulling ahead are not simply the ones with the best models. They are the ones whose leaders understand AI well enough to ask the right questions of it: where it should be trusted, where it should be challenged, and where its output should change a decision rather than simply inform it. Developing that kind of leadership fluency, grounded in practical judgment rather than technical trivia, is, in my view, one of the most under-resourced investments in the current AI landscape. It is also a key focus of NEP-AI: developing leaders who can convert AI knowledge into decisions that move performance, not just presentations that describe potential.
Governance as an enabler, not a brake
The final shift is in how organisations approach governance. Leading organisations are increasingly embedding governance into the design of AI initiatives from the outset. They define decision ownership, establish how performance will be monitored, and plan for how systems should respond when outcomes fall short, before the model enters a live process. Done well, governance is an enabler of responsible innovation, building the trust and confidence needed to scale AI successfully.
The measure that matters
We need to change what we measure. The right question is not how many AI initiatives have been launched, but how many have been retired, scaled, or replaced by something better, because that is the signature of an organisation that has moved from experimenting with AI to extracting value from it.
The technology is ready. The opportunity is real. What will determine which organisations capture it is the discipline of their choices and the readiness of their leaders to act on them.
Image Credit: Eng. Yaser Alshehhi
Source: Tahawul Tech

