When Sheikh Mohammed bin Rashid Al Maktoum, Vice President and Prime Minister of the UAE and Ruler of Dubai, set a target for half of all UAE government services to be powered by AI agents within two years, the ambition was clear. The execution, as ever, is the harder part.
Few are closer to that challenge than Dr Adel Alsharji, COO of Presight, the Abu Dhabi-listed AI and big data company that has become one of the UAE’s primary delivery engines for national-scale artificial intelligence. Already reporting a 22.2 per cent revenue increase in Q1 2026 alongside double-digit growth in EBITDA and profit after tax, and with systems that process over two million government decisions every day, Presight is no longer talking about what AI could do for public service; it is the infrastructure quietly doing it.
Recent months have seen the company deepen its footprint considerably, from a strategic partnership with Kazakhstan as the country accelerates its AI-driven transformation, to a commercially significant contract with Khazna Data Centers for an AI-powered unified command and control platform, and a partnership with the Federal Competitiveness and Statistics Centre to build a unified national AI-powered data and statistics platform. In March,
Presight also unveiled the first six companies selected for investment through its AI Innovation Ecosystem, backing the next generation of intelligent systems technologies at national and enterprise scale.
Yet for all the momentum, Alsharji is measured in his assessment of where the hard work lies. Data quality, workforce integration, edge-case reliability at scale — these, he states, are the unglamorous fundamentals that determine whether AI pilots become mission-critical infrastructure or remain permanently at proof-of-concept stage.
In this interview with Gulf Business, he sets out where agentic AI is already delivering in the UAE government, where it is still falling short, and why Abu Dhabi has become the model other governments are now travelling to replicate.
With the UAE leadership’s mandate to have 50 per cent of government services powered by AI agents within two years, where are the biggest execution risks right now?
Execution at this level comes down to getting two fundamentals right. The first is data readiness. Many agentic AI use cases are conceptually strong but struggle to move beyond early deployment because the underlying data is fragmented, unstructured, or difficult to access.
Without well-organised, high-quality data, agents cannot operate reliably or deliver consistent outcomes.
The second is workforce integration. The priority is not just deploying AI agents but embedding them into day-to-day government operations in a way that complements human roles. If these workflows are not carefully integrated, they can introduce duplication or friction rather than improving efficiency and service delivery.
Everyone talks about AI pilots, but far fewer reach scale. What’s the hardest part of turning agentic AI into embedded, mission-critical infrastructure inside government?
The real challenge is making these systems perform reliably within complex, real-world environments. In government settings, agentic systems must operate across legacy infrastructure, fragmented data environments, and multiple agencies, all while maintaining consistency, speed, and accuracy. These are dynamic, high-stakes conditions where systems need to work every time. The pressure increases as deployment expands. Performance that holds in a pilot can break down when exposed to thousands of scenarios, edge cases, and continuous user interaction. This is where robustness, testing, and system design become critical.
Ultimately, this is a systems engineering problem. Embedding agentic AI into infrastructure requires the same standards of reliability, resilience, and governance expected of any mission-critical system.
Across deployments like TAMM and healthcare use cases, where AI agents have clearly delivered better outcomes, where are they still falling short?
We are already seeing measurable impact in real-world environments. Across Presight’s deployments with UAE government entities, our systems are processing over two million decisions per day, handling more than 100 petabytes of data annually, with response times under three seconds, less than 0.01% downtime, and zero security breaches to date. In practical terms, this translates into tangible outcomes. On the TAMM platform, services are becoming more integrated, responsive, and outcome-driven.
Within the Department of Health in Abu Dhabi, AI agents are being used to dynamically coordinate patient flows during emergencies, optimising hospital allocation, ambulance routing, and system-wide readiness. In other areas, such as civil defence and urban planning, agents are enabling more proactive decision-making through simulation and predictive modelling. As these systems scale, the focus now shifts to handling edge cases and variability at scale.
Systems can still struggle with highly unstructured data, unexpected inputs, or scenarios that fall outside their training distribution. Closing this gap is a key focus as deployments continue to mature.
With G42’s AI Agent Factory, how scalable and repeatable is agent development today, and where does Presight fit in that stack?
Agent development is rapidly becoming more scalable and repeatable. G42’s AI Agent Factory is designed to industrialise how agents are built, trained, and deployed, moving away from bespoke development towards standardised, production-ready pipelines. This significantly accelerates the ability to deploy agents across multiple use cases and sectors.
Within that ecosystem, Presight serves as the primary delivery engine, responsible for translating these capabilities into operational systems for government and enterprise clients. We are already deploying agents across our government engagements, demonstrating that this model can scale in practice, not just in theory.
You emphasise augmentation, but in practical terms, which roles are most exposed to automation, and how quickly will government workforce structures need to evolve?
The impact is best understood at the level of tasks rather than entire roles. Most government roles include a significant proportion of administrative and process-driven work, such as handling applications, coordinating workflows, and managing data. These are the areas where AI agents can have the most immediate impact, by reducing manual effort and increasing speed and consistency. This is a shift from automation for efficiency to augmentation for impact.
As agents take on routine tasks, human roles move toward judgment, oversight, and citizen-facing service delivery. The objective is to elevate human contribution by removing administrative burden and enabling more effective decision-making. In terms of timing, this transition is already underway and is likely to accelerate over the next few years, requiring governments to actively rethink workforce structures, skills, and training.
As these systems move into critical areas like healthcare and civil defence, how are you managing risk, accountability, and failure scenarios at scale?
As AI systems become mission-critical, the approach to risk must evolve accordingly. Our systems are designed to be sovereign, secure, and resilient by design, ensuring that data governance and regulatory requirements are fully met.
Equally important is the integration of clear governance frameworks, with transparency, accountability, and human oversight embedded at every layer of the system. We also draw on extensive experience operating in highly regulated, mission-critical sectors, including healthcare, energy and financial services. This informs how we design for reliability, implement continuous monitoring, and plan for failure scenarios.
Ultimately, these systems must meet the same standards as any critical national infrastructure, with robust safeguards, clear accountability, and the ability to operate reliably under pressure.
Presight has been expanding beyond the UAE. What does your current international pipeline look like in terms of signed deals versus active bids, and which markets are gaining traction?
Our international expansion is being shaped by governments prioritising sovereign AI capabilities and national-scale system transformation. More and more, they are moving beyond isolated deployments and looking to embed intelligence into core infrastructure. We are seeing that translate into real momentum across multiple regions.
Recent collaborations in markets such as Kazakhstan and Albania reflect this, with governments looking to replicate elements of the Abu Dhabi model, integrating infrastructure, data platforms, and AI-driven applications into unified systems. What is changing now is the shift from interest to implementation, and that is what is driving sustained international growth.
In international tenders, you’re up against players like Palantir Technologies and hyperscalers. Where does Presight have a clear edge?
Our differentiation lies in both our model and our track record. We take a flexible approach to sovereign AI adoption, working with governments to design systems that align with their national requirements. This can include building local infrastructure within a country or leveraging models such as G42’s Digital Embassies.
At the same time, we bring the advantage of proven deployment at a national scale. Abu Dhabi is not a pilot environment; it is an operating example of how intelligence can be embedded across government systems, delivering measurable outcomes across sectors. What sets us apart is how these capabilities come together.
We focus on integrated, outcome-driven systems, combining data, AI agents, and operational workflows into a single, cohesive architecture. This goes beyond providing tools or analytics and enables governments to fundamentally redesign how they operate.
