FRESH TAKES ON BIG IDEAS

For centuries, new technologies—from farming tools to assembly lines to personal computers—have stoked excitement about enhancing productivity and value for businesses. At the same time, they have raised persistent questions about whether certain human skills and roles will need to evolve—or could even disappear.

Today, at the dawn of the agentic AI age, organizations face a similar mix of optimism and uncertainty, paired with new urgency. Leaders are under pressure to move quickly, yet how they act now could shape not just their ability to gain competitive advantage but the future of work itself.

This moment calls for leaders to think holistically about how their organizations create value. Agentic AI is not simply a new tool to deploy; it introduces a new paradigm in which humans and AI agents work together. Value creation will depend less on technical sophistication alone and more on whether people trust, adopt, and effectively collaborate with these systems.

Early experience already shows the difference this makes. One global services company piloting AI agents in customer operations began by redesigning workflows instead of rolling out tools. The organization shifted frontline employees into exception handling and relationship-heavy roles, while training managers to supervise agents rather than tasks. Adoption accelerated, and attrition declined. In contrast, another organization layered agents on top of its current systems to boost productivity but didn’t redefine roles or decision rights. Employees distrusted the outputs, managers worked around the system, and the initiative stalled in pilot mode.

These examples highlight a central truth of the agentic era: Technology alone does not transform work—people do.

That reality places new demands on leadership. As agents take on more analytical, administrative, and coordination tasks, leaders must focus on identifying and cultivating the uniquely human contributions that will matter most. McKinsey research suggests that while technical skills will grow in importance, enduring advantage will increasingly come from socioemotional and cognitive capabilities such as judgment, empathy, decision-making, and systems thinking—all areas where humans complement, rather than compete with, machines.

“Technology alone does not transform work—people do.”

New talent profiles are already emerging in response. Agent orchestrators design and supervise agent workflows. AI coaches help employees integrate agents into daily work. AI-empowered frontline workers focus on high-trust, human-to-human interactions. Managers’ roles are shifting as well: Instead of optimizing individual performance, they are increasingly responsible for designing systems—setting guardrails for agents, coaching people, and intervening only where human judgment truly matters.

Among early adopters, one of the clearest lessons is that workflows must be intentionally redesigned so humans and agents can complement each other. Even sophisticated agentic programs can fail if human input is bolted on as an afterthought. Organizations need to decide where humans stay “in the loop,” where they move “above the loop,” and how people will experience working with agents day to day. User experience matters. Trust and confidence grow when systems are transparent, explainable, and easy to interact with.

At the same time, leaders must confront an unavoidable reality: Much about the agentic future remains uncertain. Most organizations experimenting with agents today are doing so in only one or two functions. The technology is advancing rapidly, but norms, roles, and skills are still taking shape.

That is why cultivating a learning mindset across the organization is so critical. In learning-oriented organizations, early agent stumbles are treated as feedback, not as proof of failure. Teams iterate workflows weekly, refine human–agent handoffs, and update skill profiles in real time. In more rigid cultures, similar setbacks are taken as evidence that the technology is “not ready,” reinforcing resistance rather than progress.

Blending work and learning will become a defining capability of successful agentic organizations. Leaders can accelerate this by investing in upskilling at all levels, encouraging experimentation, and empowering employees to act as change agents. Adoption becomes far more likely when people don’t have systems imposed on them but instead become involved in shaping how agents are used.

As leaders seek to harness agentic AI’s productivity gains and innovation potential, they cannot lose sight of what ultimately determines success. AI agents may become pillars of the next-generation workforce, but the future will be built by organizations that redesign work, leadership, and learning around people. Keeping humans at the heart of the agentic transition is not just a moral imperative—it is a strategic one.

—Edited by Eric Quiñones, senior editor, New Jersey

ABOUT THIS AUTHOR

Sandra Durth is a partner in McKinsey’s Cologne office.