Workday’s new Global Workforce Report offers a sharp, if bounded, signal about how employers are describing AI skills. In job requisitions from companies using Workday Recruiting, demand for basic AI skills such as simple prompting fell 25% after peaking in January 2026. At the same time, demand for hands-on skills tied to building AI tools, automating workflows, and AI engineering rose 51% between September 2025 and July 2026.
That does not mean prompting has suddenly become irrelevant, or that every employer is hiring the same way. It does suggest that in one large enterprise recruiting dataset, the market is putting more weight on turning AI into working systems than on basic familiarity with a chatbot. The useful distinction is between an introductory capability and the ability to design, automate, evaluate, and improve a workflow.
What Workday’s data actually covers
The report should not be read as a census of the entire labour market. Workday says its figures combine de-identified workforce data from active HR-software customers with at least 250 employees, a global survey of 6,001 employees and business leaders, an August survey of 5,944 workers, and skills data from roughly 550 enterprise employers using Workday Recruiting. Its recruiting data cover September 2025 through July 2026.
That scope is a strength and a limitation. It can surface changes inside large organisations that use Workday’s systems, including internal movement and job-requisition language that public data often do not capture. But it cannot establish that the same pattern holds for small firms, employers using other systems, or every country and occupation. Workday’s own economic analysis makes the same broader point: some hiring patterns differ materially by industry and company size.
For leaders, the report is more useful as a directional prompt than a hiring forecast. A company that has taught staff to write better prompts may now need to decide which teams can safely automate a process, which people can assess an AI output, and where technical ownership sits. Those are different skills from tool access alone. They also require documentation, governance, and time for people to learn—not simply a new line in a job description.
The AI skills training gap
Workday found that 79% of surveyed workers said they knew the skills they needed to succeed, while 66% said their employer was helping them develop those skills. That 13-point gap is more revealing than a generic claim that workers must “upskill.” If organisations want more people to build with AI, they need to make the path concrete: identify the work that can be redesigned, teach the underlying process and data constraints, and give employees a way to test, review, and challenge the result.
The report also found that 40% of business leaders expected AI to help them get more from current employees, while 28% expected it to reduce headcount. Those are expectations reported by decision-makers, not evidence of future job losses or gains. Still, they help explain why development matters. In a workplace where job tasks are changing before roles disappear, the quality of training and job design may shape who benefits from the transition.
That concern connects with Unhyd’s recent coverage of Data & Society’s Worker Lens programme, which is examining how AI changes work through management choices, workplace design, and workers’ ability to question a system. Workday’s report does not answer those questions. It adds a current view of what large employers using its products appear to be asking for.
Mobility may be the harder problem
The report’s other warning is about the routes people normally use to adapt. Workday said internal moves declined year over year at 57% of employers in its data, while promotion rates were essentially flat. A separate measure showed the median number of applicants per filled role rising from 58 to 69, with particularly large increases in technology and media and in financial services. Workday links some of that application volume to broader AI use by job seekers, but the report does not establish that AI is the sole cause.
That caveat matters. Slower mobility can reflect hiring freezes, pay, location, management support, or a mismatch between applicants and open roles. It should not be casually attributed to AI. But the combination of changing skills, heavier application funnels, and fewer internal openings raises a practical question for employers: are workers being given a credible route into the new work, or only being told that the old work is changing?
For workers, the implication is not to chase every AI label. Basic fluency remains useful, but the more durable evidence of capability may be a demonstrated ability to improve a real process: specify the task, use appropriate tools, check output, understand failure modes, and communicate the result to colleagues. For employers, that means measuring whether training creates those abilities rather than counting course completions or prompt-library downloads.
Workday’s October release is not a final verdict on the AI labour market. It is a timely look inside a defined set of enterprise systems, and it makes a focused point: the demand signal may be moving from using AI to building accountable work around it. The next test is whether organisations create the training and mobility that let employees make that move.
