Full speed ahead with agentic AI
The Middle East is all-in on artificial intelligence (AI), and with good reason. Analysis from PwC reveals AI could boost the region’s GDP 8.3% by 2035 – generating $232 billion in additional value – as long as adoption is widespread, responsible, and focused on productivity gains. What’s more, embracing AI-powered technologies is a key part of national transformation agendas like the Qatar National Vision, Saudi Vision, Abu Dhabi Economic Vision, Dubai Vision 2030, and Oman’s Vision.
Agentic AI deployment in particular is progressing rapidly in the Middle East. Three-quarters (74%) of organizations in the GCC plan to adopt AI agents, and almost a fifth (19%) are already progressing from pilot projects to full-scale implementation.
To support these AI ambitions, businesses in the region are having to modernize their tech stacks, with a particular focus on sovereign infrastructure, including sovereign cloud. So how will these upgrades help advance agentic AI, and what else will Middle East businesses need to succeed?
Legacy infrastructure slows down agentic AI
Organizations in the Gulf Cooperation Council (GCC) are dedicated to making AI systems work, with 72% saying senior leaders support their AI strategy and are committed to a clear, well funded roadmap of use cases. Yet their technology and data infrastructure doesn’t necessarily support this commitment. Just 37% say their technology foundations are well established and they have the strong data fundamentals needed to support any AI capability.
Continued reliance on legacy systems – especially among the government-adjacent organizations, banks, utilities, healthcare providers, and energy companies in the region – makes it difficult to scale agentic AI for multiple reasons:
- Disconnected systems and manual workarounds make it difficult to see how an enterprise runs, so it’s hard to identify and act on high value use cases for agentic AI.
- AI agents need rapid, governed access to the right business data to work effectively, but siloed legacy systems hinder real-time access and orchestration.
- Any advanced AI model requires scalability in processing power and data storage that can rarely be delivered by legacy setups.
- Proving ROI from agentic AI (RoAI) is almost impossible when legacy systems prevent businesses linking AI agent activity to business outcomes.
To make agentic AI work, businesses across the Middle East need to modernize their Enterprise IT. This is likely to include transitioning from legacy on-premises servers to scalable cloud computing environments, improving systems integration, implementing robust security and governance controls, and upgrading to a unified data architecture.
- Find out how Celonis helps your business transform strategically, disrupt minimally, and modernize on your own terms.
Sovereignty increases trust in Enterprise AI
Digital sovereignty has become a core theme as Middle East organizations modernize their tech stacks for agentic AI. While digital sovereignty used to be about data sovereignty or data residency – where data is stored, how it is processed and who can access it – the emergence of AI has expanded its scope to include AI models, training data, inference environments, and AI governance. Being able to clarify who controls these elements, what rules they apply, and what the alternatives are, helps improve the resilience of the AI stack, meet regulatory requirements, and build stakeholder trust.
IDC’s March 2026 Tech Buyer Survey reveals businesses are redesigning their enterprise architectures for greater sovereignty and resilience in four key ways:
- Multi-regional and distributed architectures are the new baseline.
- Energy resilience has moved to the core of cloud decisions.
- Workload portability and provider diversification are non-negotiable.
- Sovereign cloud has become a boardroom priority.
The fourth point – sovereign cloud – is becoming critically important. In response to both regulatory pressure and geopolitical risk, every major cloud provider – including Oracle, Google, Microsoft, and AWS – now offers sovereign cloud environments designed to ensure data and infrastructure operations remain strictly within a defined geographical and legal jurisdiction.
Although sovereign cloud infrastructure-as-a-service (IaaS) spending in the Middle East is currently relatively low compared to other global regions, it is set to double in the next 12 months. As, Gartner’s Sr Director Analyst, Rene Buest, explains, “As geopolitical tensions rise, organizations outside the U.S. and China are investing more in sovereign cloud IaaS to gain digital and technological independence.”
Beyond sovereignty controls in the cloud, many countries in the Middle East are looking to establish entire sovereign AI ecosystems that leverage regional infrastructure, models, workforces, and networks to reduce their dependence on other global regions.
Saudi Arabia’s HUMAIN is a perfect example of this AI ecosystem trend. Designed to position the kingdom as a global AI hub, the initiative will provide a range of AI services, products, and tools, including next-generation data centers, AI infrastructure, and sovereign AI cloud capabilities. It will also offer advanced AI models and solutions including one of the world’s most powerful multimodal Arabic large language models (LLMs). But the goal is not isolation or total self-sufficiency, with HUMAIN collaborating with a range of global partners like NVIDIA, AWS, and Cisco.
Sovereign AI infrastructure is becoming a global trend, with IDC predicting that by 2028, 60% of multinational firms will split AI stacks across sovereign zones. However, it’s not the only upgrade Middle Eastern enterprises need to get value from AI agents.
Eliminate AI blind spots: Operational clarity for AI agents
Even when organizations have a modern, sovereign, scalable infrastructure in place to support agentic AI, it can still fail to deliver the ROI they’re hoping for. This is due to a lack of operational context.
Without an in-depth understanding of how a specific business operates – combined with knowledge of things like organizational hierarchies, vendor relationships, and strategic objectives – AI agents are still generalists, not specialists.
The Celonis Context Model (CCM) provides operational context through a dynamic, real-time, system-agnostic digital twin of operations. In doing so, the (CCM) gives AI agents the context they need to deliver meaningful value. It transforms them from generalists to specialists by providing a deterministic foundation of:
- Hindsight: What has happened across historical workflows.
- Insight: What is happening right now in active processes, and why.
- Foresight: What should happen next to achieve the optimal business outcome.
When they have the right context, both humans and AI agents can reason correctly, decide sensibly, and act reliably, producing meaningful business impact. By providing this context, Celonis helps companies in the Middle East bridge the gap between AI ambition and operational execution, enabling them to:
- Understand how processes run across systems and departments.
- Identify high-impact agentic AI use cases.
- Provide the operational context AI agents need to make the right decisions.
- Support agent orchestration across teams, workflows, and systems.
- Measure the business impact of AI over time.
The Middle East AI race requires an ecosystem
No single vendor can provide everything Middle East businesses need to make agentic AI work and meet the AI ambitions of national transformation agendas. A composable ecosystem of partners, combining best-in-class vendor solutions with open-source software, reduces risks related to vendor lock-in, cloud concentration, and single points of failure. It allows businesses to maximize resilience, innovate freely, and integrate emerging AI technologies without overhauling their entire stack.
To make the most of agentic AI, businesses need multiple partners working together, including:
- Cloud and sovereign cloud infrastructure for scale, security, and data residency, from hyper-scalers like Oracle, AWS, Google, and Microsoft, or regional cloud providers.
- Data platforms like Databricks or Snowflake to organize, govern, and activate enterprise data.
- AI platforms and agents, often from providers like Anthropic, IBM, Nvidia, or OpenAI, to reason, recommend, and act.
- A platform like Celonis, that connects the ecosystem and provides the right operational context for AI decision making.
- Business teams and governance models to guide, monitor, and improve AI-enabled work.
- Explore the Celonis partner ecosystem which includes global professionals in services, technology, and academia.
Moving from IT modernization to AI-ready operations
Here are six practical steps businesses in the Middle East can take to move ahead with agentic AI:
Step one: Understand your existing operations. Create a real-time digital twin of your operations that shows how your systems interact and how work happens across departments.
Step two: Assess legacy constraints. Use your digital twin to identify where your infrastructure, systems, data, and processes limit readiness for agentic AI.
Step three: Modernize your infrastructure. Define your cloud and data strategy, including which workloads require sovereign infrastructure.
Step four: Identify AI use cases. Prioritize high-impact agentic AI use cases where you can quickly realize and prove significant value.
Step five: Orchestrate across the ecosystem. Analyze, design, and operate AI-driven processes, giving agents the right context and orchestrating them alongside your people and systems.
Step six: Measure value continuously. Track how agentic AI improves speed, cost, compliance, service quality, resilience, and business outcomes.
Agentic AI needs more than infrastructure
Modernizing enterprise IT – including adopting sovereign infrastructure – is vital for Middle East businesses to achieve their AI ambitions. But agentic AI cannot succeed on infrastructure alone.
An ecosystem of trusted providers, and the operational context AI needs to act reliably are both prerequisites to turning the region’s agentic AI ambitions into measurable business outcomes.