Traditionally, “the team” meant a group of people you could turn to and talk with. Today, that same team might include an AI agent drafting reports, another summarizing customer calls, and a third flagging supply-chain problems before a human ever reviews the data.
The challenge is that many leaders are still managing this blended workforce with a playbook designed for all-human teams. Roles may be unclear, managers may struggle to determine when to trust or question AI output, and when something goes wrong, accountability may be difficult to determine.
This shift is redefining AI team management. Leaders who once focused primarily on motivating, coaching, and evaluating people must now manage hybrid teams where human employees and AI systems share responsibilities.
As Harvard Business School Professor Joseph Fuller explains in the online course Management Essentials, “Managers who build adaptive organizations pay attention to those signals and adjust accordingly.”
AI is one of those signals. Responding effectively requires managers to rethink how they structure work, assign responsibilities, measure performance, and lead through change. For organizations using AI, it also means learning how to oversee systems taking on a growing share of work.
Here’s what’s changing about team leadership—and how managers can adapt.
Featured Certificate Course
Management Essentials
Explore a practical, process-based approach for everyday management in a changing world of work.
Self-Paced | 4 Modules | 5-6 Hours per Module | $1,949
Coordination Doesn't Happen Automatically
When everyone works in the same place, coordination can happen informally. A quick question across the room, a hallway conversation, or observing how a colleague handles a difficult client call can help prevent confusion and facilitate learning.
When people worked side by side, coordination, learning, and accountability were supported largely by proximity. As Fuller puts it in Management Essentials: "Place didn't just host work—it quietly structured it."
Hybrid and distributed work changed that. Informal moments no longer happen automatically when people work from different locations. Unless leaders intentionally create opportunities for those interactions to happen, small misunderstandings can linger, decisions can stall, and new employees may struggle to learn how work really gets done.
The same challenge can arise on AI-supported teams. When something goes wrong, leaders may assume that an employee—or an AI system—is underperforming. But the deeper problem may be that the team’s management system was never redesigned for its new way of working.
Fuller offers a concrete example in Management Essentials: A distributed product-launch team falls two weeks behind schedule, several members miss deadlines, and leadership’s first instinct is to blame poor performance and consider replacing people to “reset expectations.”
A better first step is to examine the system:
Was it clear how team members should communicate?
Did everyone have access to the tools and information they needed?
Did people know who was responsible for making each decision?
A project that looks like a people problem may actually be a design issue.
Organizations Are Changing Shape
AI is accelerating another major organizational shift. As generative and agentic tools take on routine work once performed by entry-level employees, organizations may need fewer junior roles while enabling experienced employees to accomplish more.
As Fuller explains in Management Essentials, organizations traditionally relied on a pyramid-shaped structure:
A wide base of entry-level roles
A narrower group of experienced employees who filtered information upward
A smaller senior leadership team
AI is beginning to disrupt that model from the bottom up. The new structure Fuller illustrates in Management Essentials takes the form of a diamond:
A narrower base of entry-level roles
A broader middle of highly capable contributors
A smaller senior leadership team
But this shift creates an important question: What happens to the systems that once developed the people who eventually became senior leaders?
Many talent-development programs were designed for the pyramid. If the organizational structure changes, but those systems don't, companies may see short-term productivity gains while quietly weakening their future leadership pipeline. Fewer entry-level roles mean fewer opportunities to learn through on-the-job experience, while broader middle layers require employees to exercise judgment earlier in their careers.
The solution is not simply to add AI to existing structures. Leaders must redesign how people learn, gain experience, and prepare for greater responsibility.
AI Teams Rarely Consist of Only Employees
Changing organizational structures also changes who does the work.
Management Essentials defines "capability markets" as labor markets in which work is matched to people based on their skills, experience, and credentials. These markets reflect a defining feature of today’s workforce: organizations increasingly rely on a combination of full-time employees, outside specialists, and AI systems.
Consider a company introducing an AI-powered customer platform that flags accounts at risk of leaving. Employees with long-term knowledge of those customers are better positioned to manage the relationships, interpret context, and coach others on how to respond. Shorter-term tasks—such as configuring the platform or testing the AI model for bias—may be better suited to specialists hired for those specific purposes.
Pat Petitti, CEO of consulting firm Catalant, featured in Management Essentials, argues that this sorting process works only when organizations treat outside contributors as integral members of the team rather than an afterthought. Outside talent, he insists, must be regarded “as important to your organization as a full-time employee,” with clear goals and the support needed to succeed.
Hayden Brown, CEO of Upwork—an online platform that connects businesses with independent professionals highlighted in Management Essentials—challenges two common assumptions about freelance talent:
Freelancers are less skilled than full-time employees
Hiring freelancers is inherently riskier
In Management Essentials, she notes that enterprise customers often begin with pilot programs and are frequently surprised by how quickly freelancers contribute and how much value they add.
The goal is not to favor employees, contractors, or AI systems. It's to match each contributor to the work they're best equipped to perform.
Explore how automation and AI can help organizations work smarter in the video below, and be sure to subscribe to our YouTube channel for more explainer content.
Expertise Is Shifting From Doing the Work to Orchestrating It
If AI can draft an analysis or generate a first version of almost anything, what work remains for people?
Human oversight is still essential when work involves nuance and discretion, including:
Identifying the real problem
Setting meaningful constraints
Interpreting results
Recognizing when an answer sounds convincing but is wrong
That distinction is vital in Management Essentials. Featured expert HBS Professor Raffaella Sadun describes a key managerial responsibility: knowing which tasks can safely be delegated to AI—and which still require human experience, context, and judgment.
Within Management Essentials, Sadun compares AI to “a very eager and overconfident trainee”—capable in some areas, but not reliably competent across all of them.
AI can produce work quickly, but speed can't replace expertise. In many cases, it makes sound human judgment even more important.
This shift makes orchestration a central leadership skill. In Management Essentials, Fuller describes the “orchestration model,” which explains how organizations coordinate work across people, teams, tools, and external partners.
The model depends on four capabilities:
Work definition: Breaking work into components that can be sourced, combined, and managed across boundaries
Interface design: Clarifying how people, teams, tools, and partners interact
Coordination and integration: Ensuring that contributions align in timing, quality, and intent
Learning and feedback: Capturing insights across the system and applying them to future work
When AI becomes part of the team, leaders must make responsibility visible. They need to define:
Who reviews AI-assisted decisions?
When is human approval required?
Who can override the system?
How are errors documented and corrected?
How are employees encouraged to report problems?
The more autonomy leaders give AI, the more responsibility they retain for its outcomes. Effective orchestration allows AI to contribute to the work without allowing accountability to disappear into the system.
An AI Agent Is a Team Member, Not a Tool
Leaders often make the mistake of treating an AI agent like ordinary software. But traditional software waits for instructions. An AI agent can take initiative, make decisions, and reshape a workflow when no one is paying close attention.
In Management Essentials, Fuller frames the shift this way: an AI agent behaves less like a tool and more like another team member—one whose decisions can affect everyone else's work.
That means leaders must give AI agents more than a technical assignment. Each agent needs:
A clearly defined role
Specific responsibilities and limits
Decision rights and escalation rules
Meaningful performance measures
A named human owner
A trial period for testing and adjustment
Fuller also recommends giving the agent a name—not to make the system seem human, but to make its decisions easier to discuss. Saying "the system did it" is vague; naming the agent gives the team something concrete to point to when reviewing a decision, he explains in Management Essentials.
Human Motivation Doesn't Vanish
As AI absorbs routine tasks, the work left for humans may become more demanding. Employees and project-based contributors may spend more time exercising judgment, managing relationships, and navigating ambiguity.
Management Essentials offers a useful checklist based on the questions that attract skilled independent professionals to a project—questions managers can also use when designing AI-augmented work:
Will this work build my capabilities?
Will I learn tools or skills that matter to me?
Will I have meaningful autonomy?
Could this opportunity lead to future growth?
Managers who answer those questions clearly are more likely to attract and retain talented contributors. Even as technology reshapes how work gets done, effective leadership still rests on building trust, providing support, and helping people envision a meaningful future with the organization.
Free E-Book
How to Become a More Effective Leader
Explore the difference between leadership and management, the qualities and skills that characterize an effective leader, and how you can unleash the potential in yourself and others.
Build Your AI Leadership Expertise with HBS Online
AI isn't eliminating the need for leadership; it's making leadership more intentional.
As teams become more integrated with AI, managers must determine which responsibilities belong to employees, specialists, or AI systems. They must clarify roles, build trust in AI-assisted work, support people whose career paths are changing, and ensure accountability remains clear.
Successful managers will do more than adopt new tools. They'll design better ways of working—combining AI’s speed and scale with human insight, experience, and strong professional relationships.
Ready to build the skills to lead hybrid human-AI teams? Explore Management Essentials—one of our online leadership and management courses—and download our free e-book to begin preparing to expand your leadership abilities.