After eight years in Dubai, UI Path’s test shifts from robots to agent governance
UI Path’s presence in Dubai began in 2018, when automation discussions centred on repetitive tasks. The recently published interview with Karl Kroeder, vice-president for the Middle East, Africa, India and Southeast Asia at the company, sketches a different shift: the issue is no longer adding an AI agent to a single role, but managing an entire process that brings together agents, robots, systems and employees, and defining what happens in an exception or when human judgement is required.
The test is no longer about launching a new agentThe interview says companies in the UAE have moved from looking for a task that can be automated to coordinating a workflow from start to finish. It cites a KPMG report that 97 percent of organisations in the UAE embed AI agents in workflows, products or services, versus 87 percent globally. It also notes that the state-owned oil company in the UAE has deployed more than 115 agents across finance, HR, procurement and audit. These figures appear in the interview; they do not alone prove implementation quality, but they explain why managing exceptions, approvals and outcomes has become part of the same discussion.
The interview links this shift to a more sensitive question for enterprises and government bodies: where is the data processed, who knows what the agent was allowed to do, and when does a human intervene. It mentions that UI Path launched its Automation Cloud service in the UAE in October 2025 with an option for local data residency. According to Kroeder, an organisation also needs choices between cloud, on-premises and self-hosted options, and the ability to run agents, applications, APIs and legacy systems within a single managed process.
Localisation is not a translation of a global offeringThat is the lesson the company offers for the region. The interview says copying a global plan and pasting it in the Middle East is insufficient, because public-sector buyers are measured on service quality and transformation, not on owning the technology itself. It also points out that government bodies, regulators, cloud providers and local integration partners all influence whether a deployment can actually happen. Therefore, choosing a deployment model, data security and the system’s ability to connect to existing infrastructure become practical prerequisites before discussing AI’s impact.
In Saudi Arabia, the interview attributes 2026 research indicating that 46 percent of organisations run AI at scale in production, versus 21 percent globally. It says the strongest opportunities appear in the public sector, financial services and healthcare, and that the company’s Riyadh office and its work with the Saudi Digital Academy are part of its local investment. However, the value described does not come from a separate experiment: the cited example of a service provider in the Kingdom starts with an automated summary of a customer’s history, interactions and suggested steps before the call-center call, and the impact is measured by the employee’s ability to handle cases that require human judgement within the broader service.
What should actually be measuredThe most realistic message in this piece is that speed and governance are not competing options. Rapid transition without clear processes or accountability may accumulate pilots without operational transformation. An organisation that builds a process defining the agent’s role, the human’s role, policy and outcome can test improvements in efficiency, customer service, risk or processing time. In the Gulf, where national strategies and corporate spending drive expansion, the practical difference will be between those who prove value in production and those who settle for announcing a new experiment.