When Workday unveiled its agentic workforce platform at its annual Rising conference last autumn, the reaction in the HR technology industry was something between awe and alarm. The platform — which the company says can autonomously handle 80 percent of routine people-operations tasks — represents the most ambitious bet by a legacy enterprise software vendor on the future of AI-driven HR.
We spent three weeks testing the platform in a controlled environment with the cooperation of two enterprise customers. What we found was a system that is genuinely impressive in narrow domains and genuinely unreliable in others — a portrait of a technology that is further along than most people realize, and further from readiness than its creators suggest.
The platform excels at tasks with clear parameters: scheduling, routine compensation adjustments, benefits enrollment nudges, compliance documentation. It struggles with anything requiring nuanced judgment about individual employees — performance disputes, leave requests with extenuating circumstances, the thousands of small decisions that define the actual texture of HR work.
“The 80 percent figure is real if you define the work narrowly,” said one HR technology analyst who reviewed our findings. “But the 20 percent that remains is often the 80 percent of the value.”
