I’ve worked in AI for decades. Agentic AI will irreversibly change our workforce whether enterprises like it or not

The data already tells a remarkable story. KPMG’s 2025 CEO Outlook survey of over 1,300 global CEOs reveals that nearly three-quarters plan to invest 20 percent of their entire budget on AI in the 12 months alone. It is their top investment priority for the second year running. But what matters most is that these leaders aren’t preparing for mass layoffs, they’re actively hiring for AI skills and investing in upskilling. They’re investing in the emergence of what I call the “superhuman” employee — i.e., professionals augmented by AI to achieve what was once impossible.

AI agents are the catalyst for this transformation. Instead of following set steps, they focus on achieving specific goals. With the right context, they independently marshal whatever tools, knowledge, and resources are needed to achieve it, with optimal human agency. They don’t ask “how has this always been done?” Instead, they give you the leverage and power to ask, “what’s the best way to solve this?” — and even, ‘What’s stopping me from doing that?’ It sounds straightforward, but this simple shift has the power to unlock a “superhuman” effect.

This breaks down the artificial boundaries that have constrained how we work for too long — the silos between functions, the limits of individual expertise, the old assumptions about what’s possible in a workday. When a procurement analyst can seamlessly work with AI agents that understand finance, third-party risk, analyze vendor relationships and pathways through complex graphs, and vendor activity investigation and optimization, they’re no longer just doing procurement.

This augmentation boosts economic activity and creates new opportunities. Yes, there’s short-term disruption, but the reward is transformative. We’re rewiring how enterprises operate, creating entirely new ways to solve problems and deliver value that weren’t possible with yesterday’s constraints.

Three Roles, One Revolution

This will fundamentally reshape the workforce itself, and we will see a clear split emerge. Those who build, manage, upkeep, and govern AI agents – the Agent Bosses. Those who evaluate and operate them — the Agent Evaluators. And finally, those who collaborate with these agents as everyday teammates — the Superhumans. But the implications go deeper than job titles.

Traditional organizational structures — built around functional silos and rigid hierarchies — won’t support this new reality. We’ll need to rethink organization and team composition, newer responsibilities (who do you call when there’s a problem with an agent?), decision rights, and how work flows through a business.

A team might consist of three humans and a dozen AI Collaborator agents. New operating models will be essential. How do you onboard an AI agent and give it an identity? How do you measure productivity and business value when a single employee is orchestrating multiple agents across different functions? How do you maintain accountability when decision-making becomes a human-AI collaboration? Who is responsible for the upkeep of said agents?

These aren’t theoretical questions. Forward-thinking organizations are already experimenting with hybrid team structures, creating new roles like Agent Bosses and redefining performance metrics to account for human-AI collaboration. The practicalities matter as much as the vision.

Building the Foundation

But none of this works without the right foundation. The biggest challenge isn’t the technology, it’s context. Much of the expertise within organizations currently resides in people’s heads, and as such is tribal knowledge that’s never been formalized.

For Superhumans and their AI teammates to thrive, enterprises must build robust data foundations and memory systems.  This means capturing and protecting an organization’s collective intelligence — not just data, but the decisions, judgment, intuition, and know-how that usually live in people’s heads. It’s a bit like creating a Pensieve for the enterprise – a living memory any AI agent can draw from. We call these context cartridges and knowledge capsules. Without them, we’re building superhumans on quicksand.

Equally critical is the Agent Control System — a single hub to register, govern, operate, and continuously improve a federated workforce of enterprise agents. Think of it as the operating system for your AI workforce. Just as you have HR systems for human employees, you need infrastructure to manage agent lifecycles, give them identities, monitor performance, ensure compliance, and enable continuous upkeep. You also need to hold their Agent Bosses accountable. Organizations that build this infrastructure now will have an enormous advantage.

The interface revolution is already under way. In the next 18 months, expect to see radically new ways of working with AI agents — natural interfaces across voice, text, visuals, or gesture. Ambient systems that sense and anticipate what you need.

The superhuman workforce isn’t a future vision – it’s already emerging in pockets across enterprises globally. The question isn’t whether this transformation will happen, but whether businesses will shape it or be shaped by it. After more than quarter of a century in AI, I can tell you that the leaders who act now and build the right foundations, won’t just survive this shift, they’ll define it.

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.