The UAE wants half of its government sectors, services and operations running on autonomous AI agents within two years. Sheikh Mohammed bin Rashid Al Maktoum, the country's Vice President and Prime Minister, announced the plan Thursday under directives from President Sheikh Mohamed bin Zayed Al Nahyan.
What's actually being announced
In a cabinet announcement, officials described "agentic AI" as models that monitor operations, produce recommendations, manage processes, and execute actions without constant human input. Sheikh Mohammed, writing on X, framed it less like infrastructure and more like a hire: "AI is no longer a tool. It analyses, decides, executes, and improves in real time." The claim that an AI will become a government "executive partner" is the kind of phrasing that sounds bolder than it tends to be in practice.
Performance reviews have teeth. Ministers, directors-general and federal entities will be scored over two years on how quickly they implement the new standards, according to Gulf News. Every federal employee is slated for specialised AI training.
The "first in the world" problem
Officials are pitching this as a world first. That claim deserves a squint. Estonia has been quietly automating public services for years. Singapore runs agentic workflows inside specific agencies. What the UAE is actually claiming is scale and timeline, not the underlying idea, and the 50% figure is doing a lot of work without a clear definition of what counts as a "service" running on AI versus one that merely uses it.
There's also the question of what agentic means inside a bureaucracy. The technology, as of 2026, is still young enough that major commercial deployments regularly hit reliability walls. Rolling it across ministries in 24 months is ambitious before you factor in procurement cycles, legal review, and the usual friction of public-sector change.
Training, oversight, Mansour
Sheikh Mansour bin Zayed Al Nahyan will oversee the transition, with a taskforce chaired by Cabinet Affairs Minister Mohammad Al Gergawi, according to The National. Implementation is being phased across ministries, with continuous performance and impact assessment built in.
Thursday's plan builds on the UAE's National AI Strategy, announced in October 2017, and two decades of e-government work. The country appointed the world's first Minister of State for AI that same year, a piece of symbolism that aged better than most.
What it doesn't answer
Nothing in Thursday's announcement specifies which vendors, models, or infrastructure will underpin the shift. The UAE has deepened ties with US AI firms, including an agreement to build a large Abu Dhabi AI campus, and state-linked G42 has spent the last year stockpiling compute and chips. Whether citizens will know when an AI agent rather than a civil servant is adjudicating their visa, licence, or benefits claim is a question nobody addressed Thursday.
Civil liberties advocacy in the region is quieter than elsewhere, but the surveillance and accountability concerns are familiar. A system that "executes and improves in real time" inside a government is, by definition, one making decisions few humans will review case by case.
Work begins immediately. The two-year clock, going by Sheikh Mohammed's framing, started Thursday.
