AI success begins with leadership and organisational design

London Business School professors Nitish Jain and Alex S. Yang examine how leadership, experimentation and trust shape the success of AI adoption

Across the GCC, artificial intelligence has moved rapidly from experimentation to execution. Governments are investing billions in AI infrastructure, sovereign capabilities and digital transformation, while businesses across financial services, healthcare, logistics, manufacturing and energy face growing pressure to demonstrate measurable returns from AI. 

This regional momentum provides a useful context for case studies we examined at London Business School on how organisations can design their approach to AI experimentation, impact and trust. 

The UAE illustrates this momentum. According to PwC’s 2026 UAE CEO Survey, AI-related investments in the country exceeded AED 543 billion across 2024 and 2025, while 72% of UAE companies now identify innovation as central to business strategy. More than half of CEOs are also rapidly testing new ideas and working with external partners to accelerate innovation. 

The ambition is clear, but many organisations are still approaching AI with the wrong question. Instead of asking, “How should we redesign our organisation to succeed with AI?”, they ask, “Which model should we buy?” or “Which vendor should we choose?” Technology matters. But it is unlikely to be the source of lasting competitive advantage when competitors can access much of the same technology. 

When we co-authored two London Business School case studies, on UltraTech Cement, one of the world’s largest cement manufacturers, and Crisil, a global financial analytics company, we came to a different conclusion. Despite operating in very different industries, both cases showed us that successful AI adoption depends less on the technology itself than on how leaders redesign decision-making, enable experimentation, establish effective governance, and shape collaboration between humans and AI. 

Stop asking AI to prove its ROI too early
One of the biggest mistakes organisations make is demanding immediate financial returns from every AI initiative. AI experimentation should not always be evaluated in the same way as conventional investment, particularly when organisations are still learning where the value lies. When leaders insist that every experiment deliver immediate financial returns, teams naturally become more cautious. They pursue incremental efficiencies rather than testing applications that could create greater long-term value. 

UltraTech took a broader approach, evaluating AI initiatives through improvements in decision quality, operational effectiveness, and the potential to scale solutions across multiple plants. Financial value remained important, but it was not the only measure of success. 

Crisil adopted a similar philosophy. Rather than using AI to replace analysts, it focused on augmenting them. AI accelerated research, data compilation and first drafts, allowing experts to spend more time on judgement, interpretation and client engagement. 

The lesson is particularly relevant for GCC organisations moving quickly to deploy AI at scale. Early value may appear as faster learning, better workflows and stronger organisational capability before it becomes visible in financial statements. Leaders who demand immediate ROI risk cutting short the learning cycle that creates those returns. 

Leadership determines where AI should move fast
The strategic question is not simply where to adopt AI, but where to move quickly and where to apply greater control. 

UltraTech partnered extensively with AI startups and external innovators, testing solutions directly in manufacturing environments. Operational errors could generally be identified and contained quickly. The greater risk was moving too slowly. 

Crisil faced a different environment. Its AI outputs influence financial analysis and client decisions, making trust, regulatory expectations and reputational risk far more significant. It therefore prioritised internal development, iterative testing and rigorous human oversight before deploying AI into client-facing workflows. Neither organisation was inherently more innovative. They simply recognised that the cost of failure differs by context. 

That distinction matters across the GCC, where AI is increasingly being deployed in healthcare, financial services, education, energy and public services. Customer-service automation may justify rapid experimentation. AI supporting medical decisions, financial supervision or critical infrastructure requires significantly stronger governance. 

Effective AI leadership is therefore not about moving as fast as possible. It is about knowing where speed creates value and where control protects it. 

The real AI challenge is trust
Better models do not automatically produce better outcomes. The greatest barrier to AI adoption is often human trust. Employees need confidence that AI systems are reliable, transparent and genuinely improve their work. Both UltraTech and Crisil involved end users throughout the development process. Plant operators tested solutions in real operating environments, while financial analysts helped shape workflows, refine outputs and determine where human judgement remained essential. Neither organisation attempted to force adoption. Trust developed through experience and evidence that AI could enhance, rather than replace, professional judgement.  This is increasingly important as the GCC’s AI ambitions reshape the workforce. Regional data provides further context: PwC’s latest AI Jobs Barometer shows the UAE among the world’s fastest-growing AI talent markets, with AI-related job postings more than tripling since 2021. As AI becomes embedded across organisations, workforce confidence, governance and capability development will be just as important as infrastructure investment. 

Three questions boards should be asking
As AI investment accelerates, board oversight needs to move beyond approving technology budgets. The UltraTech and Crisil cases suggest three questions for leaders:  

• Where should we accelerate AI experimentation, and where do tighter controls need to apply? 

• Are we measuring AI only through short-term financial returns, or also through organisational learning and better decision-making? 

• How are we building trust in AI among the people who will use it, particularly where human oversight remains essential?  

These questions shift the conversation from procurement to organisational design. That is where sustainable advantage increasingly lies. The GCC has already demonstrated extraordinary ambition in AI, building enabling policy environments, investing heavily in infrastructure and encouraging public-private collaboration.  The next phase will be determined by something harder to replicate. Not who acquires the most advanced models, but what organisational capabilities they build around them. If competitors can buy the same technology tomorrow, the real source of advantage will lie in how organisations make decisions, experiment, build trust and develop people around it. In the AI era, competitive advantage will belong not to those who adopt artificial intelligence first, but to those who design their organisations to use it best. 

This opinion piece is authored by Nitish Jain, Associate Professor of Management Science and Operations, and Alex S. Yang, Professor of Management Science and Operations at London Business School.