Workday Finds Basic AI Skill Demand Fell 25% After January Peak

If your 2027 AI training plan is another company-wide prompting course, you are funding a skill employers have asked for less since January. Workday's job-requisition data from roughly 550 employers shows demand for basic AI skills, such as simple prompting, peaked in January 2026 and then fell 25%, while demand for hands-on skills like building AI tools, automating workflows and AI engineering rose 51% between September 2025 and July 2026. The money should follow: fewer seats in generic literacy courses, more depth for the people who own a workflow and can rebuild it.

Workday published the October 2026 Global Workforce Report on October 5. It combines four datasets, and the skills finding is only one of them. The others matter as much for a budget decision, because they show where the internal ladder for those new skills has stalled.

What Workday Actually Measured

The skills numbers come from job postings, not from a survey of what managers say they want. Workday counted skill mentions in requisitions from about 550 enterprise employers using Workday Recruiting, September 2025 through July 2026. Over the same window, mentions of management and leadership skills fell 7% and mentions of training skills fell 13%.

The rest of the report draws on separate sources, and it helps to keep them apart:

1. De-identified HR data from Workday customers with at least 250 employees, matched year over year, for internal moves, promotions and turnover.
2. A global survey of 6,001 employees and business leaders, including about 1,780 decision-makers.
3. The AI@Work Pulse survey of 5,944 workers, fielded in August 2026.

Two caveats before you act on the headline figures. The 25% and the 51% use different starting points: one is measured from a January peak, the other from September 2025, so the gap between them is wider than a like-for-like comparison would show. And the release does not say whether the figures are raw counts or a share of postings. Treat them as direction, not magnitude.

There is also the obvious question of who is talking. Workday sells the HR, recruiting and skills software these employers run on, and its own Skills Cloud is part of that pitch. The underlying data is real customer data, but the framing is a vendor's.

Why Basic AI Skills Dropped Out of Job Ads

The likeliest reading is that basic AI use has become assumed. Workday's own pulse survey found 81% of workers say they have received training on using AI at work, and 76% have adopted AI. When most of the applicant pool already has a skill, it stops being something a hiring manager writes into a requisition.

Prompting still matters on that reading; it just no longer separates one candidate from another. What still separates them is the ability to change how the work gets done: wiring a model into a process, automating a handoff, building the internal tool. A Workday blog post on the report describes the shift as moving from "general AI literacy" to "practical AI execution."

The industry has watched this happen once already. In 2023, "prompt engineer" was pitched as the hot new job title. By spring 2025, Microsoft's Jared Spataro was telling the Wall Street Journal, as quoted by Lead with AI, "Two years ago, everybody said, 'Oh, I think prompt engineer is going to be the hot job.'" and that it was not turning out that way. Microsoft's own 2025 Work Trend Index put AI Trainer, tied with AI Data Specialist, at the top of the AI roles leaders were considering, at 32%. The title faded and the skill got folded into everyone's job. Workday's January peak looks like the same curve, one level down.

Broader labour-market data points the same way. PwC's 2026 AI Jobs Barometer, built on Lightcast postings, found job ads requiring AI skills grew 69% in 2025 against 9% for the market overall, and salaries in what PwC calls "professionalised" roles, where AI automates the routine tasks, rose 42% faster than in "democratised" ones. The demand is real. It has moved up the skill ladder.

---

The Counter-Argument Worth Taking Seriously

Job postings are a weak instrument for workforce strategy, and a careful reader should discount them. Postings are not hires. A requisition tells you what a hiring manager typed, not what the person in the seat needs on day 90.

There is a second objection. Your existing staff are not new hires. A 25% drop in requisitions asking for basic AI skills says nothing about the accountant who has never opened your approved assistant. Workday's survey suggests that group is real: individual contributors are the lowest AI-usage group, at 61%, and 53% of them say their manager has not discussed how AI is changing their role.

Both objections are fair, and neither saves the generic course. The same survey found 57% of workers say they cannot rely on AI output without close checking, and 33% say at least half the time AI saves them goes into correcting low-quality output. Another prompting module does not fix that. Changing the workflow so the output arrives checked, structured and in the right system does. The remaining literacy gap is real but narrow, and the people in it need a manager conversation more than another course.

The Internal Ladder Has Stalled

The skills shift would matter less if companies were moving people into new roles. They are not. Internal moves fell year over year at 57% of employers, and promotion rates stayed essentially flat worldwide. Employees who tried to move most often cited a hiring freeze on the role, a manager who would not support the move, or a selection process that did not feel fair.

Meanwhile 79% of employees say they know which skills they need, but only 66% say their employer helps them build them, a 13-point gap. Workday's Phil Willburn put it plainly in the report's release: "Employees may not be changing jobs, but their jobs are changing around them."

Put the two datasets side by side: employers are posting for AI builders while freezing the internal moves that would let a trained employee become one. The external market will not fill the gap cheaply either: the median filled job drew 69 applicants, up from 58 a year earlier, yet time to fill held at about 60 days. Employers told Workday roles stay open mostly because candidates lack the right skills, or because pay, location or flexibility do not match.

We looked at one industry's version of this in bank job postings for agent orchestration, where a large percentage jump started from a small base of mentions. Workday's dataset spans industries, and the direction is the same.

The Budget Is Shrinking While the Need Moves

The training budget you are reallocating is smaller than it was. SHRM's 2026 L&D benchmarking, covering more than 4,600 organizations, found median L&D spend per full-time employee fell 28% since 2025, training hours held at eight per year, and only 32% of organizations offered AI upskilling beyond basic compliance. External training fell from 25% to 15% of L&D budgets.

Eight hours a year per employee is the whole envelope. If two of those hours go to a prompting refresher most of the workforce has already taken, there is little left for the depth that building skills require. A vendor-commissioned DataCamp and YouGov survey of 500-plus leaders claims organizations with mature upskilling programs are twice as likely to report significant AI ROI. That is a training vendor's claim about training, so weight it accordingly, but it argues for depth over breadth too.

The skill Workday's data says is declining in postings is also the one you need most for this shift. Training skills fell 13% and leadership skills 7% in requisitions. Workday's own blog argues those are the capabilities organizations most need to make AI adoption work. If you stop hiring people who can teach, you will need the builders you do have to teach as part of their job.

---

What to Do With the 2027 Training Budget

Hands-on AI building skills are the ability to change a workflow with AI: connect a model to a business system, automate a multi-step handoff, or build and maintain an internal agent or tool. That is what the 51% rise is measuring, and it is what your money should buy.

This Week:

1. Pull last year's AI training spend and split it into two lines: company-wide literacy content and role-specific build or automation work. Put both numbers in front of whoever owns the 2027 L&D plan.
2. Ask your HR systems lead for your own internal-move rate for the last 12 months against the prior year. If you are among the 57% where it fell, you have the same stalled ladder Workday describes.

This Month:

1. Name three workflows where an owner already uses AI daily and where rework eats the time saved. Fund those owners for a build track on the automation platform you already license, whether that is n8n, Zapier or UiPath, with a working automation as the exit criterion rather than a completion certificate. Before you scale it, read how workflow tools change once they run agents.
2. Replace the next scheduled all-staff prompting session with a 30-minute manager conversation per team on how AI is changing each role. Workday's data says 37% of workers have not had that conversation in three months.
3. Register what employees build. A citizen-built agent nobody tracks is a security problem, as the 44% visibility gap in business-built agents showed.

Before 2027 Budgets Lock:

1. Write internal-move targets for AI-adjacent roles into the workforce plan, and remove the manager veto that employees cite as a blocker. A trained builder who cannot move is a future external applicant.
2. Keep a small literacy line for the individual contributors who have not adopted AI at all, and target it by usage data, not by headcount.

The Bottom Line

"Prompt engineer" took about two years to go from hot job title to a skill folded into every role. Workday's data shows the basic version of that skill following the same path in job ads, with a January peak, a 25% fall within months, and a 51% rise for building skills across the same reporting window. Hiring managers have already moved their requisitions toward builders, while internal moves fell at 57% of employers and the median L&D budget per head shrank by 28%. A 2027 training plan that still spends most of its eight hours on literacy is aimed at last year's requisitions.

Before you renew the enterprise prompting course, check how many of its seats went to people who were already using AI every day.

Continue Reading

• Banks' 1,721% Agent Orchestration Jump Starts From 108 Mentions
• SAP Buys TechWolf, the Skills Engine Feeding Workday's Skills Cloud
• AI Created a 62% Wage Gap Between Two Types of Workers
• Zapier and n8n Alternatives Once Workflows Become Agents
• The 93/7 AI Budget Split: Why Your Next Bottleneck Is Human
• BMW Bets 100 Senior Jobs on AI Agents That Still Need a Signer