Today's Edition
BATOOL GHAITH (ABU DHABI)
The UAE’s push into Agentic Artificial Intelligence could reshape government services from systems that respond to requests into ones that can identify needs, recommend actions and carry out defined processes proactively, according to Mohammed Alketbi, NEP-AI Expert in Agentic AI.
The UAE’s National Experts Programme (NEP) serves as a launchpad for Emirati experts to play a leading role in transforming future-growth sectors aligned with the country’s national priorities.
Alketbi said the earliest opportunities for agentic AI are likely to emerge in areas where government processes are already highly digitised, data-driven and governed by clear rules.
These could include service delivery, internal government operations, case management, compliance monitoring, data analysis and decision-support functions.
“With integrated government systems already in place and increasingly proactive, data-driven service delivery, Agentic AI can help shift government from systems that respond to requests to systems that can proactively identify needs, recommend actions and execute defined processes,” Alketbi told Aletihad.
He stressed that the objective should not simply be to automate as many government tasks as possible. “It should be to identify high-impact use cases where autonomy can deliver measurable improvements in speed, productivity, service quality and public value,” he noted.
Identifying the Right Use Cases
The UAE’s National Agentic AI Framework sets an ambition to integrate Agentic AI into 50% of government entities within two years, while 80,000 federal employees are expected to receive training in Agentic AI technologies and tools, he said.
“Scaling Agentic AI requires technology, talent and governance to advance together. The UAE has invested significantly in its digital infrastructure and data ecosystem.
The next challenge is ensuring that government organisations have people who can understand what Agentic AI can realistically achieve, identify the right use cases and manage the transformation responsibly,” Alketbi explained.
The objective is not simply to create AI users, but professionals who can evaluate AI solutions, understand their economic and operational implications, and make informed decisions about their deployment, he added.
He also emphasised that Agentic AI differs from systems that simply generate information or respond to prompts because agents can understand objectives, plan steps, make decisions within defined limits and execute tasks with a degree of autonomy.
In government, Alketbi said that could open the door to redesigning both public services and internal operations. The strongest candidates for delegation are high-volume, repeatable and rules-based tasks that can be clearly measured, which include routine administrative workflows, data validation, document processing, monitoring, reporting and certain service transactions.
But he also cautioned against treating complete delegation as the end goal. “There is an important distinction between delegating execution and delegating accountability.”
Clear Governance Frameworks
He stressed that the future should not be viewed simply as replacing government employees with AI agents.
Instead, AI could take on appropriate operational complexity while employees focus more heavily on leadership, judgement, innovation and decisions requiring context and accountability, Alketbi said.
“The AI system itself cannot carry institutional accountability. Accountability must remain with the government entity responsible for the service and the decision,” he added.
Before deploying an agent, he said government leaders need to understand its capabilities, limitations, data dependencies and decision boundaries, and to establish what an agent can do independently, which actions require human approval and how its decisions can be monitored and challenged.
“This is why Agentic AI must be deployed with clear governance frameworks, human oversight, auditability and escalation mechanisms,” he said.
Alketbi also said the potential becomes even greater when multiple specialised AI agents work together rather than a single agent handling one task, as one agent could analyse information, another identify possible actions, another coordinate the workflow and another execute an approved step.
But autonomous interaction between multiple agents also introduces another layer of complexity, he noted, as each agent would need clearly defined roles and permissions, while systems would require monitoring and escalation mechanisms.
“Ultimately, government should be able to understand not only what an individual agent did, but how a chain of AI-driven actions produced a particular outcome,” Alketbi explained.
That requires understanding both the technical architecture behind agentic systems and the governance, risk and decision-making implications surrounding them, according to Alketbi.
He indicated that the most important shift is to stop thinking about Agentic AI simply as another digital tool. Instead, it represents an opportunity to rethink how government works.
The UAE has already built two decades of digital-government capability, with Agentic AI potentially representing the next stage of that journey.
“National expertise is the bridge between technological capability and government transformation,” he said, adding that the aim should be to use AI not merely to automate existing processes, but to create a more proactive, intelligent and effective government that delivers greater value to people and society.
