The tax profession's traditional way of building expertise has always run on a hierarchy: junior professionals handle routine work — data entry, reconciliation, first-pass classification — and the judgement they build doing it becomes the foundation for more senior roles later. Chapter 14 of Real-Time Tax Transformation (forthcoming) makes a specific argument about what happens to that pipeline once AI absorbs the routine work it depends on: an AI system that handles the routine collapses that pipeline, because the mechanism that used to build judgement no longer operates in the form junior professionals relied on.
The World Economic Forum's 2026 analysis with PwC puts numbers behind the shift. Entry-level roles in the highest AI-exposure quartile show nearly twice the rate of net skills change compared to non-entry-level roles globally — a measure of how fast the required skill mix is turning over, not how much work exists. For the UAE specifically, entry-level roles in the highest-exposure quartile show more than four times the net skills change rate of roles in the lowest-exposure quartile. A static training model that assumed skills would stay roughly stable for a few years while someone worked their way up is no longer a safe assumption to build a development programme around.
What replaces the ladder
The WEF's finding on career progression is the part tax and finance leaders should read most carefully: as AI takes on task execution, progression stops following a linear, time-in-role sequence and becomes what the report describes as a non-linear mosaic — built from project outcomes and demonstrated capability rather than years served. The driver is a growing expectation that professionals at every level move from executing tasks to orchestrating the AI agents performing them. For a junior tax professional, that means the work that used to build judgement — routine classification, first-pass reconciliation — is exactly the work an agent now does, and the judgement still has to be built somewhere.
The WEF report identifies which skills are in the most acute shortage across AI-exposed sectors, and the list is consistent: critical thinking, domain judgement, and the ability to work effectively with AI agents. These are hard to automate, hard to assess in a hiring process, and — the part that matters most for tax functions — hard to build once the routine tasks that used to develop them have been automated away. An organisation that has not deliberately redesigned how junior professionals build judgement will find, within a few years, that it has capable AI systems governed by professionals who never had the chance to develop the judgement the governance role now requires.
This is the specific gap the Fusion Professional™ profile is built to close, as a practical answer to a structural problem. The profile combines domain tax judgement with enough technical fluency to specify, interrogate and correct what an AI agent produces, and it has to be built deliberately, through exposure to exception-handling and system governance work, because the traditional apprenticeship route that used to produce it no longer runs through the same routine tasks. Organisations that redesign their development model now, while their real-time tax implementations are still creating the institutional context to build this capability in, are preparing for a shift that is already underway inside their own systems, not one that might arrive later.