Abu Dhabi’s public-sector AI programme is past the pilot stage. As of October 2026, more than 100 AI use cases are live across more than 40 government entities, with a pipeline of more than 200 in development. The figure that matters most to the people running the programme is not the count of applications but the shared infrastructure underneath them: sovereign data systems and common AI capabilities that entities draw on rather than build separately.
The delivery story traces back to TAMM, a unified platform whose Arabic name translates as “consider it done.” Originally launched under the Emirate’s Ghadan 21 accelerator programme to consolidate dozens of fragmented agency portals into a single digital gateway, the platform has since entered its next phase as TAMM 4.0, anchored by the Abu Dhabi Government Digital Strategy (2025-2027), which targets complete AI integration across public entities by 2027.
Today the platform brings together more than 1,170 services, from transport and health to business and community services, in one system used by more than 4.5 million people. It processes more than 17 million transactions annually. Ninety-six per cent of them are completed autonomously. For users, the operational shift is concrete: instead of physically visiting government centres or navigating 80 different websites and apps to work out which entity is responsible for what, they now access one app.
To understand how that integration was executed on the ground, The Recursive spoke with His Excellency Wesam Lootah, Director General of GovDigital at the Department of Government Enablement Abu Dhabi (DGE), the technology architect guiding the Emirate’s centralized digital architecture. A fuller account of the deployment is available at https://www.therecursive.com/inside-abu-dhabis-journey-to-making-ai-work-across-40-government-entities-tamm/.
H.E. Lootah describes the challenge of making 40-plus government entities function as one coherent experience as primarily organisational before it is technical. Each entity has its own data, its own processes and its own institutional logic. What made integration possible was building shared digital foundations, sovereign infrastructure, common data standards and shared AI capabilities, that entities connect into rather than maintain separately.
The clearest expression of what that integration enables is TAMM AutoGov, an agentic AI tool that proactively completes services for users. When a licence renewal is approaching, AutoGov detects it, checks eligibility, coordinates across the relevant entities and completes the process without the user initiating anything. That cross-entity, event-driven delivery, in H.E. Lootah’s words, gives millions of people back something valuable: their time.
Delivery at this scale also depended on workforce preparation, which the programme treated as a deliberate precondition rather than an afterthought. By the end of 2025, 95 per cent of Abu Dhabi public-sector employees had completed AI training, a baseline that required significant coordination across more than 40 entities. The harder effort, H.E. Lootah notes, sits in the gap between completing a training programme and genuinely changing how people work. To close it, the government built practical tools alongside learning programmes, including Tomouh, a personalised AI-powered training platform that provides tailored learning pathways for employees, particularly in AI and digital literacy.
Meanwhile, the operational results are visible in specific services. In healthcare, more than 10,000 physicians use AI at the point of care to access structured patient information and clinical context in real time, supporting better-informed decisions without replacing clinical judgement. Across the city’s road network, AI continuously analyses feeds from more than 625 cameras for potential incidents and hazards, with a reported detection time of three seconds per camera, giving municipal teams situational awareness that would be difficult to sustain manually.
The change is also reshaping what government employees do with their working day. A few years ago, every request for a government service, whether a licence renewal, permit approval or title deed issuance, required manual processing. Today 96 per cent of TAMM requests are handled by AI, a transformation that has returned millions of hours to both employees and the people they serve. As AI absorbs more of the information-heavy routine, employees’ attention moves toward complex cases and decisions that carry consequence. The pace is uneven across roles and entities, but the underlying pattern is consistent across government.
Partnerships have been structured to protect delivery conditions. The Frontier Employee Programme deploys Microsoft 365 Copilot to 35,000 civil servants across 27 government entities, with data remaining within Abu Dhabi’s sovereign digital infrastructure. The principle applied across every significant partnership, H.E. Lootah explains, is that the Abu Dhabi Government controls the data, the governance framework and the conditions of deployment, while technology partners provide capability within that. DGE works with a range of technology organisations spanning cloud infrastructure, cybersecurity, AI development and workplace productivity, and remains open where alignment on the governance model exists.
For other countries watching whether this model actually works, the central lesson is about sequencing. Governance decisions, H.E. Lootah argues, need to be made before deployment, not after. Every AI application entering the Abu Dhabi Government faces an explicit judgement about what it can do on its own and where a human professional must review or decide. AutoGov completes eligible routine services proactively, but only within conditions and preferences the user has already set. In health services, AI supports physicians with clinical information and risk indicators, but every clinical decision stays with the doctor. When the boundaries are explicit, the system earns trust at scale, from people, from employees and from the institutions responsible for outcomes. Start there, before the technology choices, and the rest follows more reliably than people expect.