The question isn’t whether AI agents will join the workforce—it’s whether an organization is ready to manage them when they do. Most HR leaders are focused on which AI tools to buy, but Scott Schmidt has different concerns: Who’s going to train these digital employees, review their performance and handle it when they mess up?

Schmidt, global deputy talent leader at EY, has watched his firm integrate AI for 400,000 employees worldwide. His conclusion: The biggest AI challenges in the workplace aren’t technical—they’re management challenges. And HR departments aren’t prepared for them.

“We may not be preparing adequately for agents that can manage performance calibration or compensation analysis,” Schmidt warns. He says the realignment from thinking about AI as a helpful tool to AI as a workforce member requires HR leaders to put aside the typical assumptions about how work gets done.

1. Decide who will manage AI agents—and how

The most weighty question isn’t technical—it’s organizational. Schmidt believes HR departments should take ownership of AI agent management, treating them similarly to human employees.

“Questions such as, how we train the agents, review their performance and address underperformance will require HR to take the lead,” he says.

This means developing new processes for AI agent onboarding, setting performance standards and creating feedback loops. Just as a team wouldn’t deploy a new employee without training and oversight, says Schmidt, AI agents require structured management approaches.

He encourages leaders to think about the practical side of AI agent management by reflecting on these questions. These aren’t technical issues, he explains—they’re HR challenges that call for HR-driven solutions.

• Will there be dedicated AI agent managers?
• How will the org measure agent performance?
• What happens when an AI agent underperforms or makes mistakes?

2. Prepare for individualized learning at scale

AI integration creates a learning challenge unlike anything HR has faced before. Schmidt’s observation is stark: “No single person will have the same AI learning experience.”

This reality demands a complete rethinking of training approaches. He says universal programs won’t work when every employee needs different AI capabilities based on their role, experience level and learning style.

At his firm, the response includes options ranging from bite-sized learning modules to intensive programs like the EY Tech MBA and master’s in business analytics, which are free degree programs designed exclusively for EY professionals. According to Schmidt, the organization has logged 4.44 million AI learning hours, with 89% of employees completing foundational AI courses.

The scale of this challenge requires preparation, not reaction. The key is building a flexible learning infrastructure before it is needed, says Schmidt. He says HR leaders should consider how they’ll assess individual learning needs, deliver personalized content and track progress across diverse learning paths.

Read more: As AI agents enter people operations, here’s how HR is adapting

3. Build a strategic ‘translation’ capability

Schmidt predicts that HR teams able to “translate AI-generated insights into strategic advice for business leaders will be the most valuable in the future.” This represents a growing core competency for HR professionals.

AI agents will generate vast amounts of data about workforce performance, engagement patterns and operational efficiency. Schmidt predicts the differentiating skill won’t be generating this data—it will be interpreting what it means for business strategy and organizational success.

This requires developing analytical thinking skills and business acumen within the HR team. Schmidt advises leaders to consider who on their team can bridge technical AI outputs with strategic business recommendations. If that capability doesn’t exist, it needs to be developed or hired.

The goal is positioning HR as the interpreter between AI capabilities and business value, not just the implementer of AI tools, according to Schmidt.

4. Foster the right culture before tech deployment

Culture determines AI success more than technology selection. Schmidt emphasizes that “a culture of curiosity and a willingness to explore AI’s possibilities will help determine future success.”

The biggest risk isn’t technical failure—it’s employee resistance or fear. Schmidt notes the importance of helping people “see AI as a partner in their work” rather than a replacement threat.

This cultural foundation requires deliberate cultivation. Schmidt points to his firm’s “thrive time” program, which provides dedicated time for personalized growth opportunities and learning. The program signals organizational commitment to employee development alongside technological advancement.

Before deploying AI agents, assess the organization’s comfort with change, willingness to experiment and openness to new ways of working. If these cultural elements aren’t present, Schmidt suggests HR leaders focus there first. He emphasizes that “any transformation must be human-centric, aiming to change mindsets so professionals can use AI to enhance their expertise.”

5. Develop speed and comfort with ambiguity

The pace of AI development creates unprecedented uncertainty for HR planning. “We don’t know exactly where technology is headed or how quickly it will evolve,” Schmidt observes, “so HR departments must build the capacity to move quickly and be comfortable with ambiguity in decision-making.”

He says agility isn’t about making faster decisions—it’s about making good decisions with incomplete information and adapting quickly when circumstances change. Schmidt identifies speed and culture as the key differentiators between HR departments that will lead to transformation versus those that struggle to keep up. Building these capabilities requires practice with smaller decisions before facing major AI integration choices.

He presents the following thinking points when HR leaders consider how their team and workforce currently handle uncertainty and change:

• Are there processes for rapid decision-making?
• Are team members comfortable with experimental approaches?
• Can the team pivot quickly when initial plans don’t work? 

Schmidt says these capabilities become essential when managing AI agents that may evolve faster than traditional planning cycles can accommodate. As Schmidt notes, “We are still at the beginning of the AI lifecycle, but the shift from experimental to essential AI is underway.”

HR Executive presents this piece as part of a September 2025 series covering AI agents: Where are they now?