Enterprise AI for HR, finance & IT · PUBLIC RESEARCH BRIEF
WorkdayWhich employee request should AI answer, route, or return to a person?
Workday's current AI page describes Sana automating work across HR, finance and IT, using Workday data and business applications to build, orchestrate and manage agents. This brief studies where employee-service assistance should provide an answer, route a request or preserve human ownership; it excludes hiring, performance, pay and other consequential employment decisions.
Sources checked 2026-09-25 · Simulation results not yet generatedCHANGE ONE THING. LEARN WHAT MATTERS.
Three questions for the GTM team.
When a request crosses HR, finance and IT, would automatic routing or a proposed owner and category that the employee confirms create a cleaner handoff? Measure misroutes, repeated explanations, missing fields and appropriate escalation rather than speed alone.
Set up this study →Before a Workday agent prepares an action, would showing the evidence and affected records before the draft or presenting a concise draft with expandable evidence lead to better reviewer corrections? Keep all changes in a sandbox and require an authorized human to approve them.
Set up this study →PROPOSED AUDIENCE
Who should weigh in?
North American organizations evaluating or using Workday for employee service across HR, finance or IT. Include adult employees seeking routine help, service-desk staff, process owners, administrators and privacy or compliance reviewers across different access scopes. Proposed audience; no resolution, productivity or cost outcome is implied.
TWO TIME HORIZONS
Trial today. A habit tomorrow?
Near term · 0–90 days
Over 0–90 days, run bounded answer, routing and draft-review tasks with synthetic or approved de-identified records, current policies and fixed permission roles. Measure factual accuracy, source checking, correct ownership, handoff completeness, unnecessary escalation and reviewer corrections. Do not change payroll, benefits, personnel records or employment status.
Longer term · 3–12 months
Over 3–12 months, follow approved service workflows as policies, roles and integrations change. Examine repeat contacts, unresolved handoffs, policy drift, access errors, employee effort, staff workload and whether reviewers continue to challenge agent output. Productivity or savings claims require observed operating outcomes and comparable baselines.
What would make the result actionable?
Use versioned policy sources, current tenant permissions, synthetic or de-identified requests and independently reviewed expected outcomes. Exclude sensitive personnel and payroll data where unnecessary, log every draft and require human approval for actions. Do not infer employment quality, fairness or compliance from simulated preference or task speed.
A Gather simulation returns hypothetical customer reactions. Quantifying revenue, traffic or retention needs actual business inputs and validation against observed behavior.
Public sources
Workday AI and Sana overview for HR, finance and IT ↗Current product page; checked 2026-09-25