Workday used its Rising conference on October 13 to announce agents that act on HR and finance work in the background instead of waiting for a prompt. The part HR leaders will spend time on is the control panel: customers choose which actions an agent takes alone, which need review, and when a person makes the final call.
What Workday announced
In its release, Workday described new agents that “work as autopilots across hiring, payroll, accounting, and compliance.” Joel Hellermark, chief AI officer at Workday, framed the change as a break with the assistant model. He said in the release that “Until now, AI has worked like a copilot” and that the new agents “get the work done on their own.”
Four agents carry the autopilot label. The Talent Acquisition Agent watches open requisitions and finds qualified candidates once a job is posted. The Payroll Agent is meant to spot missing pay data before payroll runs. The Accounting Agent tests financial information continuously. The Deployment Agent flags new capabilities in each Workday release that fit how a customer has configured the system. Workday says nearly 5,000 customers have activated the Deployment Agent since it was announced last September.
The hiring agent is the one that ships first
According to Workday, the Talent Acquisition Agent keeps watch across open roles, flags those at risk of missing their fill date, and starts assisting with requisitions before a recruiter logs in. When the candidate pipeline is thin, it can notify recruiters in tools such as Slack and draft outreach. The release states that “outreach to candidates still requires recruiter approval before anything is sent.”
The agent can also coordinate interview scheduling, give interviewers context and draft feedback for review. Recruiter preferences, such as interview timing or format, can be saved to the agent’s memory. Workday lists Workday Recruiting, HiredScore, Paradox, native Workday skills and connected sourcing providers among the tools it works across. HRTech examined another recruiting agent with saved memory in our piece on recruiting agents that add memory and verification signals.
Workday claims time to fill “moves from weeks to days.” The release gives no customer data or measurement period for that line, so it reads as a vendor expectation until early adopters report their own numbers.
Payroll and accounting come later
Availability splits by agent. The hiring autopilot is available now to early adopter customers, with general availability planned for early 2027. The Payroll, Accounting and Deployment autopilots are planned for early adopters in early calendar year 2027.
For payroll, Workday gives one example: after a series of organizational changes, the agent can audit retroactive adjustments, flag anything outside defined parameters and draft communications to employees for the payroll team to review. Gerrit Kazmaier, president of product and technology at Workday, explained why the company ties autonomy to existing rules. “Probabilistic AI is great at guessing, but in payroll and finance, a good guess isn’t good enough,” he said.
Who sets the limits
Workday says customers “decide which actions an agent can take on its own, which need review, and when a person makes the final call.” Every action is logged, so customers can see what the agent did, what information it used and who approved it. Customers can also set how often agents run, which Workday ties to managing both activity and cost as agent use grows.
A second release the same day extended those rules to agents that customers and partners build. Workday said its Trusted Agent Platform is now generally available, and that every agent, whether built by Workday, a customer or a partner, works under the same Workday permissions, approvals and audit trails. Its Agent-Ready Tools use the Model Context Protocol, so any agent that supports MCP can read Workday data and act through them. BambooHR has launched a connector built on the same idea of following existing permissions, covered in our brief on its Claude connector.
Workday also cites its own research on why the guardrails exist. The release says large language models “can misapply HR policies even when the correct policy is in front of them.” In early 2027 Workday plans to add Grounded Reasoning, which would check a request against company policies, business processes and tenant data before an agent acts. Its example is a manager asking an agent to move someone into a role the manager has no authority to fill: the agent would stop, explain why and point to the next step.
What it means for the HR leader
In our reading, the announcement moves a decision from the vendor to the customer. Workday ships the dial. Each customer decides where to set it for each task, and that setting is a statement about risk that HR, finance and legal will all have a view on.
That decision sits close to the work SAP described on October 7, when it tied HR agents to a redesign of roles in its Autonomous HCM plan. Both vendors are pitching agents for HR work, and both leave the question of how far to let them act with the buyer.
Practical steps for a team evaluating these agents:
- List the actions each agent could take in your hiring and payroll processes, and mark each as automatic, review first or human decision before any pilot starts.
- Ask who on your side owns the log. The release says every action is recorded, which only helps if someone reads it on a schedule.
- Treat the early-2027 payroll dates as a planning window. Payroll errors reach employees directly, so agree the review rules with your payroll lead before early access begins.
- Ask Workday for measured time-to-fill results from early adopters before using the weeks-to-days line in a business case.
Source: Workday, “Workday Launches Autopilot Agents That Get Real Work Done Across HR, Finance, and IT”
