Organisations in the UAE are running AI agents in production at a scale few other markets match, according to new research from Dataiku. The study, Global AI Confessions: CIO Edition, was carried out by Harris Poll and compares how companies in eight markets are adopting and managing agentic AI.
Sixty-two per cent of UAE chief information officers estimate that more than 50 AI agents are active in production inside their organisations. Fifteen per cent of UAE organisations said they operate more than 500 agents, the highest share among the markets surveyed and well above the global average of 9 per cent. Hiring is following the deployments: among CIOs creating or planning embedded roles for AI experts or forward engineers, 22 per cent in the UAE expect to recruit more than 10 people in the next year, again the top figure in the research.
Oversight has not kept pace. Only 18 per cent of UAE CIOs said they have a fully unified view of the AI agents working across their business applications, data environments and digital platforms, and just 12 per cent said agent lifecycle management is standardised across their organisation. The research suggests that a company could spot a problem agent in one system while it keeps running through another. Every UAE CIO surveyed also reported more pressure from their board than in 2025 to show measurable returns on AI spending.
Florian Douetteau, chief executive and co-founder of Dataiku, said UAE organisations are deploying agents at a scale that remains uncommon worldwide. He added that the companies appear well equipped to detect harmful agent behaviour, but that detection does not always mean an agent can be stopped quickly across every environment where it runs.
Sid Bhatia, Dataiku’s area vice president and general manager for the Middle East, Turkey and Africa, said the UAE has built a strong base for agentic AI through large deployments, investment in specialist skills and work to limit the impact of AI incidents. The next phase, he said, is linking those safeguards across wider technology estates. That means giving CIOs a single view of their agent population, setting clear authority over what agents can do and making it possible to step in quickly, regardless of where an agent was built or which system it runs in.
