For most organizations, the AI conversation typically revolves around technology, dominated by questions like: What can we automate? Which tools should we deploy? How can we best support employee adoption?
But once AI agents enter the picture, the conversation must change. As AI agents increasingly execute work, interact directly with customers, make decisions and operate with greater autonomy, they become part of the workforce.
That shift raises new questions strikingly similar to the ones we ask about our people: What work should they perform, and who decides that? Who is accountable for their actions? How does their work change that of the people working alongside them? What new skills will they need?
IT isn’t equipped to answer those questions, or in many cases, even thinking about them. While IT can tell the organization what an agent can do, HR can provide the perspective on what it should do. That’s why HR can’t be brought in as a post-deployment afterthought.
Instead, HR needs to be part of building an agentic talent strategy. Here are four critical roles HR should play in a successful agentic AI rollout.
Define the Agent’s Role Before It’s Deployed
Unlike traditional software tools that perform isolated, predictable functions, AI agents can execute tasks, interact with other systems and operate with unprecedented autonomy.
In many ways, they need something resembling a job description: key responsibilities, scope of authority, and parameters for acceptable outcomes. IT may define technical capabilities, security and integration requirements, but HR brings deep expertise in role clarity, performance expectations and accountability.
That makes HR essential in establishing the “rules of engagement” before an agent ever enters the workflow. If an agent will materially affect how people work, HR should help design its role.
Think ‘Agents for Tasks,’ Not ‘Agents for Jobs’
There’s a lot of fear among the workforce that AI will take their jobs, but smart agentic implementation should focus on making people more effective, not simply replacing them.
Consider a sales development rep whose job includes research, prospecting, outreach, scheduling and relationship-building. AI can handle many of those activities, but you still need the person to build relationships. Instead of replacing them, AI creates an opportunity to redesign the job role around higher-value capabilities where human judgment and relationships matter.
HR is uniquely positioned to see that bigger picture. Rather than allowing automation to drive headcount, HR can identify which tasks move to agents and then determine how employees can be reskilled for higher-value responsibilities. This maximizes human capacity by offloading mundane tasks while retaining high-potential talent who can be redeployed elsewhere.
At Criteria, our AI agents have taken on portions of our tactical customer onboarding work, allowing our customer success team to focus more heavily on strategic customer conversations, renewals and growth opportunities—the relationship building only humans can do. But the transition required more than technology. Employees needed new skills, new performance expectations, and KPIs for what success looks like. That’s the difference between software implementation and workforce transformation.
Own the Human Side of AI Governance
Governance is critical in AI deployment. Companies must set guardrails for policy, procedures, security, integration, and responsible use.
IT is well-equipped to monitor technical compliance and performance, but HR must help establish human accountability, expectations for how humans work with AI and escalation procedures for when something goes wrong. That becomes especially important as agents encounter sensitive data or make consequential decisions. Someone must ultimately be responsible, and building those parameters in from the beginning is a decidedly HR function.
What happens when an agent needs to be decommissioned? HR can play a vital role in the offboarding process, including how tasks will be reallocated and how workload, capacity, and skills will be realigned.
Addressing model bias also falls within this scope. AI systems can reflect bias based on the data they’re trained on, but bias can also emerge from the assumptions and perspectives of those designing them. HR expertise in fairness, equity, employee impact and organizational culture should inform how agents are selected, trained, tested and audited. This perspective can enhance IT’s standards for performance and accuracy.
Prepare Employees for What’s Next
As AI agents take on more work, HR’s most important role may be helping employees navigate the transition. More than instructions on how to use AI, people need clarity on what AI means for their careers. They need to know which parts of their jobs will change, what skills will matter most, how performance measurement will change and where their human judgment will remain essential.
Without those conversations, employees are left to fill the void themselves, and “AI is coming for my job” is an easy conclusion.
HR can replace that fear and uncertainty with confidence and clarity. Transparent communication about what’s being automated, what’s still needed from humans, and how the organization will help them grow into those roles is vital. This candid conversation can preserve years of institutional knowledge and nuance that cannot simply be programmed into an AI agent.
HR can help ensure organizations don’t pursue efficiency at the expense of automating away capabilities they still need.
HR Shouldn’t Wait for an Invitation
As AI agents become more sophisticated, they cannot function in a silo. The scope of collaboration will expand far beyond IT and HR to include legal, compliance, procurement, and other business functions.
The goal for AI should be to create the right combination of human and AI capabilities that allows humans to perform at their maximum potential, rather than just automating the maximum amount of work. Getting there requires IT’s expertise, but it also demands HR’s perspective on job roles, expectations, skills, performance, culture, and people.
Most organizations don’t hire people and then find a role to plug them into. Why would they do that with AI agents? Building an agentic talent strategy (one that defines roles, requirements, expectations, onboarding, and decommissioning) before AI agents start is critical. Waiting until they arrive on the job puts the organization at risk.
Jillian Phelan has over 25 years of experience in senior HR roles and currently serves as Chief People Officer at talent assessment SaaS company Criteria Corp where she spearheads the HR function for 200 employees. She has held several HR leadership roles including at Ernst & Young, McGladrey & Pullen, Avetta and more, making her well-versed in HR best practices across a diverse range of industries and company sizes.