Vision 2030’s Real Bottleneck – the Right Job Design

The UAE announced in April 2026 that half of all their government services will run on agentic AI within two years. The message given here was clear – speed. The underlying details of this announcement meant that the main effort would depend on building an “enabling layer” – people whose job is to direct, supervise, and be accountable for a fleet of autonomous agents.  

While most of Vision 2030 conversation frames AI as a skills problem – certification, retraining, competency mapping; it assumes the jobs remain the same and are performed by better-skilled people. Agentic AI breaks this assumption. An AI agent can carry out a multi-step task from start to finish using different methods at a different pace. This forces the unit of work to change and hence the shape of the job, the organisation chart and indirectly the career ladder someone has to climb. No matter how you train your employees, it doesn’t matter if the job won’t exist.

Let's explore a couple of examples. A finance officer who would handle transactions now supervises the agent and looks only at exceptions. A specialist who looked only at one part of the workflow now becomes a generalist who oversees agents across the entire workflow. The middle layer, where the checking happens, will thin out, because routing and checking is the first thing agents would learn. PWC captures this pattern as a move from the pyramid structure to one of an hourglass one, a lean strategic top, a lean middle focused on decisions & exceptions, and a wide early career base because AI-literate newcomers ramp up faster.

For the Middle East region, this pattern is quite important. It collides with the region's most important workforce commitment. Emiratisation and Saudization depend on national graduates entering through entry-level roles and building upward - precisely the roles agentic AI absorbs first. Design the transition carelessly, and you get the opposite of the hourglass: a diamond, with a hollowed-out base that starves the national talent pipeline at its source. No amount of reskilling resolves that tension. Only job design does.

This level of creative talent planning is best handled by the HR team. It is about resketching the architecture of work itself. Three steps will enable this:

1. Reclassify the tasks role by role, before you reclassify the people: What is AI-only, what is human-plus agent and what is human-only.

2. Redesign the entry point: If agents keep learning and taking on junior roles where how to make decisions was taught, build new roles. Agent supervision, exception-handling, quality oversight - where national talent learns it instead.

3. Rewrite the job catalog: Update your job titles and promotion rules to reward how well people direct AI and exercise judgment, rather than the volume of work they get done.

The biggest advantage that the Middle East has: sovereign infrastructure, fast regulators, governments willing to mandate at a pace few can match. But mandates move faster than the organisation charts, and the constraint that decides who delivers is neither the technology or the training. The one piece that sets the leaders apart is quieter: designing the jobs where people and agents will work side by side. That is very much with the HR’s power to shape. The organizations that take it on now won't just keep up with Vision 2030; they'll give their people a clear and confident place to stand as the agents arrive and a real path to grow alongside them.