AI Agents Will Force HR to Rethink How Work Gets Assigned and Evaluated
HR leaders face a question most organizations have barely begun to answer: How do you manage AI agents once they're doing actual work?
That gap between capability and management is where SAP SuccessFactors is positioning itself. The company is building tools to let enterprises assign permissions, track costs, and evaluate performance for AI agents the same way they do for human employees-because soon, both will sit side by side on org charts.
Maryann Abbajay, Chief Revenue Officer at SAP SuccessFactors, discussed the shift at SAP Sapphire in Orlando. She outlined three immediate pressures: skills management, workforce planning, and the operational reality of mixing human and AI labor.
Skills and Succession Planning Drive Current Demand
HR departments are shifting focus away from generic talent management toward skills-based workforce planning. Companies want to identify what skills exist in their workforce, where gaps appear, and how to build capabilities before people leave.
The problem is acute in specialized fields. Payroll systems, for example, require years of expertise that younger workers don't automatically possess. When SAP restructured in 2024, the company had to quickly upskill senior staff to cover payroll gaps left by departures. That kind of planning is becoming standard.
Abbajay said most large enterprises aren't yet fully executing skills-based workforce planning, but the shift is underway across industries.
AI Agents Need Employee IDs and Performance Reviews
SuccessFactors now allows companies to create employee IDs for AI agents and place them on org charts. This isn't administrative theater-it's functional necessity.
An agent needs permissions to perform certain tasks and restrictions on others, just like a human employee. A manager using the system can define skill requirements for a project and have AI surface internal candidates ranked by match percentage-80%, 95%, and so on.
Performance evaluation is the next step. Managers will need to assess whether an agent is producing correct results and actually completing assigned work. The evaluation framework doesn't exist yet at most companies.
Cost Accountability Is Missing
Organizations assume AI agents cost less than people. Abbajay said that assumption hasn't been tested.
Companies are still deciding what work to assign to agents versus humans, what oversight to provide, and how to measure ROI. Few are tracking the infrastructure, consumption, and operational costs of running an agent. That scrutiny is coming as deployments scale.
The question will shift from "Can an agent do this?" to "Does it make financial sense for an agent to do this?"
Resistance Is Real, But Experience Reduces It
HR leaders report resistance to AI adoption even from their direct reports. One CHRO addressed this by assigning work that required AI tools and setting a tight deadline. Resistance softened once people experienced how much time AI saved-three to four hours compressed into 15 minutes.
Adoption requires more than tool availability. Long-tenured employees especially need structured enablement plans. Abbajay said the shift from basic to advanced AI usage won't happen organically.
Compliance Becomes More Complex With Agents
Regulated industries face infrastructure and citizenship requirements that complicate AI deployment. Government agencies often require systems to run in-country and may restrict who can manage the environment based on security clearance levels.
SAP holds FedRAMP certification in the US and operates data centers in multiple countries including India, Germany, the Netherlands, Australia, and the UK to meet these requirements.
Certification tracking also matters. Disney uses SAP's Learning Management System to certify staff on rides-without verified certification, employees can't work. In healthcare and manufacturing, that becomes a safety and liability issue.
What Managers Need to Know
SuccessFactors has moved from a peripheral HR function to mission-critical infrastructure in the last five years. As agents enter workflows, the system becomes the place where organizations define what humans do and what agents do.
Managers should expect two conversations to dominate the next year: agent performance metrics and agent cost allocation. Both require frameworks that don't yet exist in most organizations.
Start thinking about how your team would evaluate an AI agent's work. The answer will shape how you structure roles, assign responsibilities, and measure outcomes.
Learn more about AI for Human Resources, or explore the AI Learning Path for CHROs to understand how HR leadership is adapting to agent-based workforces.
Your membership also unlocks:
