The UAE’s National Agentic AI Committee has released a design guide for building government services around agentic AI, a framework intended to move the burden of navigating bureaucracy from residents to government systems.
The guide was launched during a specialist workshop that brought together officials and AI and service teams from federal ministries and entities. Its stated aim is to help government bodies redesign their services in line with the broader Agentic AI Project’s targets.
Mohamed Rashed bin Tulaie, Assistant Minister for Cabinet Affairs for Government Knowledge Exchange and head of the government services track within the Agentic AI Project, outlined what the shift means in practice. Under the current model, he explained, residents are responsible for identifying the right authority, understanding requirements, submitting documents, and tracking their case through to completion. Under the new model, a resident states what they need and the AI assistant takes over: understanding the request, mapping the steps, executing them, and coordinating with relevant entities.
“The user doesn’t want a chain of procedures — they want an outcome,” bin Tulaie said. “We don’t measure the success of agentic AI by the number of messages exchanged with the user, but by its ability to complete the service and reach what they want.” He described the real shift as one of responsibility: service execution moves from the user to the government.
The guide sets out a set of principles for designing these experiences. Among them: users should be able to express their needs simply; the AI should draw on data already held by government so residents are never asked to resubmit documents they have provided before; and the system should seek the user’s approval before taking any action on their behalf.
The framework also covers edge cases and technical failures, requiring a smooth handover to a human employee when needed, with full context transferred so the user does not have to repeat themselves. It addresses unified and secure payment processing, clear communication of outcomes, and mechanisms for users to question, correct, or contest results. Service feedback is built into the process as a continuous improvement loop.
