When you are promoted into a management role, you learn to lead through other people. Your team takes responsibility for execution, while you gain the decision rights, ownership and control that come with the position.
But that is only true when the people you manage are people.
If, instead, what you manage today is a team of AI agents, something fundamentally changes. As AI agents increasingly redesign knowledge work, they assume more and more of the execution. Yet the decision rights, ownership, authority and responsibility for directing outcomes do not automatically move with that execution.
AI is expanding what we can describe as human agency, yet people are still evaluated based on what they personally produce, rather than on how well they decide, direct and own the work.
Now imagine that same dynamic playing out across an entire organization.
That disconnect has a name: “blocked agency.”
Microsoft’s 2026 Work Trend Index, built on a survey of 20,000 workers across 10 countries together with analysis of trillions of Microsoft 365 productivity signals, provides clear evidence of this. As AI increasingly takes over execution, people should gain what the report defines as agency: the ability to direct work, make decisions and own outcomes. The pressing challenge for organizations is turning that expanded agency into business value.
Most organizations are not doing that, and the problem is not that employees are unprepared.
Here is what the research reveals about why organizations are blocking the agency AI is giving their people, and what leaders need to do about it.
The AI Capability Gap: What Employees Can Do Versus What Organizations Permit
When employees begin working with AI, something changes in what they are capable of accomplishing. They no longer spend most of their time analyzing information, producing deliverables and executing tasks.
Instead, they move toward shaping outcomes. They begin redesigning workflows, coordinating work across AI systems and operating at a level of thinking that previously belonged to more senior roles. Among AI users surveyed in Microsoft’s research, 58% say they are now producing work they could not have completed a year earlier.
Nearly half of all Microsoft 365 Copilot conversations support higher-order cognitive work, including analysis, problem solving, evaluation and creative thinking.
The capability is there — the ability to act on it is not.
The reason is simple: the organizational structure has barely changed. Org charts still determine ownership. Decision-making authority continues to follow hierarchy. Performance measures still reflect an older definition of the job.
Employees are embracing AI tools and dramatically expanding what they can accomplish. But they quickly encounter the limits imposed by role definitions, job descriptions and the authority associated with their level.
They are capable of doing more. The organization has not redefined what they are empowered to do, rewarded them for doing it or created the conditions that allow those capabilities to flourish.
During a conversation on The Future Of Less Work podcast, Jared Spataro, Chief Marketing Officer for AI at Work at Microsoft, captured the issue simply: “When the system itself has a governor on the speed that it can go, it doesn’t matter how fast an individual can run.”
The Microsoft 2026 Work Trend Index puts numbers behind this disconnect. Only about 9% of AI users fall clearly into the blocked agency category, highly capable people whose organizations have not evolved alongside them. But the broader pattern extends much further.
Roughly half of all AI users operate in what the report describes as the “emergent zone,” where both individual capability and organizational readiness are still developing, often at different speeds, with employees progressing ahead of the systems around them.
The Transformation Paradox: Why AI Adoption And Work Redesign Collide
What makes this a structural challenge rather than simply a motivation issue becomes clear in the data describing what employees are actually experiencing.
Sixty-five percent of AI users say they worry about falling behind if they fail to adapt quickly to AI. At the same time, 45% say it feels safer to focus on existing goals instead of redesigning how they work. Only 13% report being rewarded for reinventing their work with AI, even when they continue delivering results.
Taken together, these numbers tell an important story. People recognize that they need to change. Many are eager to do so. Yet the organizational systems surrounding them make it safer not to.
Microsoft refers to this as the “transformation paradox.” The very forces accelerating AI adoption are simultaneously slowing the work redesign required to create meaningful value. Employees are ready to rethink how they work. The metrics, incentives and organizational norms surrounding them continue rewarding the previous way of working.
This is not simply a matter of weak incentives or slow technology adoption. It reflects a deeper conflict between improving existing work and fundamentally redefining what work should become. As Spataro explained, “we’re not trying to make the processes that we have just a little bit more efficient… I think the firm of the future is a totally different place.”
That tension appears clearly in how organizations perceive their own transformation. Only one in four AI users believes leadership is clearly and consistently aligned around AI transformation. Leaders are also considerably more likely than employees to believe experimentation feels safe and that reinvention is rewarded. That difference in perception deserves attention.
Some organizations have already managed to close this gap. Microsoft refers to them as “Frontier Firms,” representing roughly 16% of organizations where individual capability and organizational readiness reinforce one another. Managers openly demonstrate AI use while setting clear expectations for AI-assisted work. Teams regularly share lessons learned as well as mistakes. Experimentation is encouraged rather than punished.
Employees in Frontier Firms are twice as likely to say they are rewarded for reinvention regardless of outcome. According to the research, culture, managerial support and talent practices account for more than twice the impact on AI success compared with individual mindset and behavior, 67% versus 32%.
The single biggest factor determining whether AI creates real business value is not the technology itself. It is the environment organizations build around the people using it.
How To Remove Blocked Agency: The Organizational Redesign Leaders Must Lead
The transformation paradox is fundamentally a systems problem. And systems do not change themselves.
Redesigning the system so it matches the new reality of work is one of the most important responsibilities leaders have today. That means performance metrics, incentives and expectations need to recognize people for changing how they work, not simply for continuing to deliver work the same way they always have. It means building a culture that views AI as a strategic advantage and actively encourages experimentation.
It also requires managers who demonstrate AI use themselves while rewarding it in others, alongside talent practices that develop new capabilities and create real opportunities for people to apply them.
The opportunity for employees to create meaningful impact has never been greater. The research leaves little doubt about that. What remains uncertain inside many organizations is whether the systems surrounding employees are evolving quickly enough to unlock that potential.
Think back to the manager we met at the beginning of this article. Promoted because of results, now leading AI agents that increasingly handle execution, yet still being evaluated on outputs rather than on the quality of judgment, direction and ownership brought to the work. That is a design problem, not a people problem. The role has changed, but the organizational system supporting it has not.
Every leader owns part of that system. Which means every leader should be asking a simple question: What needs to change on my team so the people capable of doing the highest-value work also have the authority and freedom to act on it?
That is where true organizational redesign begins.
