The UAE’s next AI advantage will come from institutional intelligence

For the past few years, much of the AI conversation has focused on capability.

How powerful are the models? How many people are using them? How many agents can an organisation deploy?

Those questions still matter. But I believe the next phase will be defined by something more fundamental: whether AI makes the institution itself more capable.

A recent conversation with His Excellency Omar Sultan Al Olama brought this into sharper focus for me. The UAE has already moved beyond asking whether AI should become part of government and business. Its ambition now is to embed Agentic AI across services, operations and decision-making at scale.

The more interesting question is what kind of institutions we build when AI agents become part of how work actually gets done.

The answer cannot simply be more automation.

The opportunity is to build organisations that combine human judgement, institutional knowledge and AI capability in ways that allow them to learn faster, make better decisions and improve continuously.

That is what I think of as institutional intelligence.

From deploying AI to redesigning the institution

One of the most important ideas emerging around AI is the concept of the Frontier Firm.

At Microsoft, we describe these as organisations that are human-led and increasingly AI-operated. AI does not sit at the edge of the organisation as another productivity tool. It becomes embedded into workflows, decisions and processes.

The deeper transformation begins when leaders redesign the operating model itself around what people and AI can each do best.

There are already signs that the UAE workforce is moving in this direction. Microsoft’s 2026 Work Trend Index finds that 26% of UAE workers surveyed are Frontier Professionals, compared with 16% globally. These are people who are not simply using AI more often, but applying agents to complex work, redesigning workflows and building more repeatable ways of working with AI.

That is an encouraging signal. But it also raises a bigger institutional question: are the organisations around those people changing just as quickly?

In Abu Dhabi Government, Microsoft 365 Copilot is already being used across tens of thousands of civil servants, while more than 1,000 custom AI agents have been developed inside government. At the federal level, the UAE is targeting the transformation of at least 50% of government services and operations to Agentic AI models within two years. The government has also launched a programme to train 80,000 federal employees in Agentic AI.

The significance is not the number of agents.

It is what those agents make possible when they are connected to institutional knowledge, embedded into workflows and governed within a clear operating model.

AI should compound institutional knowledge

Every organisation has knowledge that gives it an advantage. Some of it lives in databases and documents. Much of it does not.

It lives in years of experience, relationships, decisions, processes, lessons learned and the judgement of people who understand how the organisation really works.

The next generation of AI should help organisations make better use of that intelligence.

Agents need to understand the organisation’s data, policies, history and processes. They need to operate within permissions and guardrails. And over time, the institution should become better at capturing what works, learning from outcomes and improving how work is done.

This is where the distinction between simply using AI and becoming an AI-native institution starts to matter.

A frontier organisation should not simply complete work faster. It should learn from the work it completes.

Better data should enable better agents. Better agents should produce better insights. Better insights should improve decisions and processes. And those improved processes should create knowledge that strengthens the institution again.

The goal is not simply more intelligence inside the organisation. It is an organisation that gets better at using its own intelligence.

That is where AI begins to create a compounding advantage.

As AI gains capability, people need greater agency

This also changes the human-capital question.

Much of the debate around Agentic AI asks how much work machines will eventually be able to perform. I think leaders should ask a different question:

What becomes possible for people when AI takes on more of the execution?

The answer is not that human judgement becomes less important. It becomes more important.

In the UAE, 88% of AI users surveyed say they treat AI output as a starting point rather than a final answer and remain responsible for the thinking.

That distinction matters.

Agency in an AI-powered organisation is not about doing every task yourself. It is about knowing what outcome you are trying to achieve, deciding what should be delegated, setting the quality bar, challenging outputs and remaining accountable for the result.

An effective manager in an agentic organisation therefore needs to know how to redesign workflows, determine which work should be delegated to agents, establish clear decision rights and judge the quality of AI-generated outcomes.

Employees need to know not only how to use AI, but how to direct and challenge it.

And leaders need to create cultures in which experimentation is encouraged while accountability remains clear.

That is why the UAE Government’s programme to train 80,000 federal employees in Agentic AI matters. It recognises that institutional transformation cannot happen through technology alone. People need the skills and confidence to lead that transformation.

The institutional layer needs infrastructure, resilience and trust

None of this can happen without the right foundations. Institutional intelligence depends on trusted cloud and AI infrastructure, secure data, resilient networks, strong cybersecurity and clear governance.

These are the foundations Microsoft continues to invest in across the UAE and the wider Middle East. In the UAE, our planned $15.2 billion investment through 2029 spans infrastructure, operations, talent and ecosystem development, while our broader regional commitments extend that focus to digital resilience, connectivity and skills.

But infrastructure is not an end in itself. Its value lies in what it enables governments, businesses and communities to build on top of it.

An institution cannot become more intelligent if the systems underneath it are fragile. It cannot delegate more work to agents without knowing how those agents are governed. And it cannot become more adaptive unless people retain the judgement and authority to intervene when circumstances change.

This is why resilience in the agentic era is not simply the ability to keep systems running.

It is the ability to keep learning, deciding and adapting as circumstances change, without compromising security, accountability or trust.

The next measure of AI leadership

The UAE has spent years building many of the foundations required for this next phase: infrastructure, policy, institutional capability, skills and partnerships.

The opportunity now is to connect them.

We should continue measuring adoption. But we should also ask more ambitious questions.

• Is AI helping institutions learn faster?
• Are organisations turning what people and agents learn into better decisions, services and outcomes?
• Are people gaining more capacity to exercise judgement and act?
• Can institutions adapt more quickly without sacrificing accountability or trust?

These questions tell us much more about the maturity of an AI economy than the number of models, copilots or agents deployed.

The next phase of AI leadership will not be defined simply by who deploys the most technology. It will be defined by who can turn that technology into enduring institutional capability.

For the UAE, that is the opportunity ahead: to build institutions that become more capable because people and AI are learning, deciding and improving together.

That is what institutional intelligence can make possible.