Agentic AI in government just moved from bold ambition to bold execution, and the UAE is leading the charge with real conviction. Building on nearly a decade of consistent investment in artificial intelligence, the country has now begun the hard, essential work of deciding exactly which government tasks a machine may complete on its own, an achievement that deserves genuine recognition as one of the most significant public sector milestones of the year.
A Head Start Nearly a Decade in the Making
The UAE’s AI journey did not begin overnight. The nation published its national AI strategy back in 2017 and, within days, appointed the world’s first minister of state for artificial intelligence. In the years since, it has quietly built the digital identity systems, sovereign cloud infrastructure, and secure data-sharing frameworks that most governments are still working to establish. This patient, foundational groundwork is precisely why the country is now positioned to lead on agentic AI in government rather than simply follow.
That long-term investment came into sharp focus this month when more than 100 federal officials gathered for a workshop marking the strategic launch of the UAE’s agentic AI project. The programme’s central goal is remarkable in scope: converting roughly half of federal government operations, services, and tasks to agentic AI models within just two years. Organised by the National Committee for the Agentic AI Project, the session set implementation priorities, established timelines, and began shaping the frameworks that will determine which government tasks can be entrusted to an autonomous system.
Solving the Hardest Part First
What makes this initiative particularly praiseworthy is that the UAE is tackling the most difficult question head-on, rather than deferring it. The country already has the compute power and infrastructure in place, including a proactive performance system capable of tracking more than 150 million data points every month. The real challenge lies elsewhere: building a defensible method for classifying which decisions a government is comfortable letting an agent complete independently, and which ones still require human judgement.
This task classification effort is the true foundation of responsible public sector transformation. It is a thoughtful, deliberate approach that reflects genuine care about outcomes, not just speed. Huda Al Hashimi, Deputy Minister of Cabinet Affairs for Strategic Affairs, described the workshop as the operational starting point for deploying AI models across government activity, spanning seven pillars including strategy, foresight, policy, governance structures, performance, competitiveness, and innovation.
A Guiding Principle Worth Admiring
One of the most encouraging aspects of this rollout is its guiding philosophy: human leads, AI enables. This principle strikes an important balance between ambition and responsibility. While the founding vision speaks of autonomous systems capable of managing operations and executing independent chains of action, the implementation language firmly emphasises human primacy at every stage.
This balance matters enormously for AI governance. Across a portfolio of thousands of government tasks, both framings can coexist meaningfully, and the classification frameworks being developed now are precisely what will define that boundary in practice. It is encouraging to see a government treating this not as a technical afterthought, but as a genuine governance priority worth solving properly before scaling further.
The Scale of Commitment Is Extraordinary
The scope of this initiative extends well beyond a single workshop. Oversight sits with senior leadership, and earlier this year, the Cabinet approved the first package of government services to apply agentic AI tools. This was paired with the largest training programme in the history of the UAE government, reaching an impressive 80,000 federal employees, from ministers to newly onboarded staff. A national AI healthcare policy was approved in the same session, alongside frameworks establishing digital records as the official source of core government data, all part of a wider push behind the country’s expanding federal AI training agenda.
The momentum has continued steadily since. A Dubai training event in June brought together more than 300 participants from 50 federal institutions, with a clear goal of identifying which services and procedures are best suited for agentic AI within 90 days. A separate initiative that same month, led by the Presidential Court and the Ministry of Cabinet Affairs, extended the same training to 600 additional staff. This kind of sustained, government-wide commitment to federal AI training is genuinely inspiring and sets a high bar for institutional readiness.
Connecting the Dots to Dubai’s D33 Vision
This ambitious rollout of agentic AI fits naturally within Dubai’s broader D33 vision, which aims to double the size of the emirate’s economy and secure its place among the world’s top three global cities within the coming decade. A government capable of moving with speed, precision, and accountability is essential to realising that ambition, and agentic AI in government is a direct enabler of exactly that kind of agility.
By investing so heavily in training, governance frameworks, and thoughtful task classification, the UAE is not simply modernising its public services. It is building the operational backbone needed to support the D33 vision’s broader economic goals, proving once again that public sector innovation and national economic strategy can move forward together with real purpose.
Why the World Is Watching
Governments across Asia and Europe are approaching similar questions from a much earlier stage, typically with a national AI strategy and a coordinating body, but without a firm number or timeline attached. The UAE has done something different: it has committed publicly to a percentage and a deadline. Within two years, there will be tangible, public evidence of what agentic AI in government genuinely delivers in terms of service quality, efficiency, and reliability.
This willingness to move first, transparently and at scale, is precisely what makes the UAE’s approach so valuable to the rest of the world. Other nations do not need to run this experiment themselves; they simply need to decide how quickly they are willing to act once the UAE’s results become public. That is a position of real leadership, and it deserves to be acknowledged as such.
The Document Everyone Should Watch For
The upcoming task classification frameworks will represent something genuinely historic: the first serious, structured attempt by any government to clearly define which decisions it is willing to delegate to a machine. Everything achieved so far- the training programmes, the governance structures, the pillars of implementation- has been building steadily toward this single, defining document. When it is published, it will likely become a reference point for public sector AI governance well beyond the UAE’s borders.
This is an exciting moment to be watching agentic AI in government take shape in real time, and it stands as a genuine milestone for public sector transformation globally. The UAE’s blend of ambition, discipline, and careful governance offers a genuinely encouraging model for what responsible, large-scale AI adoption in the public sector can look like.
For businesses and institutions watching this transformation unfold, the message is clear: intelligent automation succeeds when it is paired with strong governance, clear accountability, and a genuine commitment to getting the fundamentals right. If your organisation is thinking seriously about how agentic AI could reshape your own operations responsibly and effectively, it may be worth exploring what that journey could look like with the right partner. Horizontal8 works with forward-thinking organisations ready to translate this kind of AI ambition into practical, well-governed results.