Three Employees, 19 Agents: How AI Is Reassembling the Manager Role

A two-hour daily management meeting at CVS has shrunk to roughly 15 minutes after agents took over report generation and dashboard preparation. The change at the U.S. pharmacy and insurance giant is one of several concrete shifts Fortune's reporting identified as AI rewires the manager role from the inside out.

At Edward Jones, CIO Kevin Adams built an AI system that continuously absorbs documents, decisions, history, open issues and commitments from each of his enterprise programs and committees. A colleague calls it his "chief of staff." Adams sets the boundary himself: "I don't let the machine tell me what to do. I let the machine prompt me about what I should be doing."

Pushpendu Pal, Chief Digital and Technology Officer at CVS Health, told Fortune that the time AI freed from budget administration and project-status reviews has not produced a shorter day. He now spends more hours coaching employees, working with engineers and developing organizational capabilities.

What is actually changing inside the manager role?

Fortune's editors, after interviewing CIOs and senior executives, argue the right frame is no longer automation versus augmentation. AI is unbundling parts of management and adding new parts. Some responsibilities are migrating into infrastructure; others are expanding to cover digital labor.

The CVS meeting illustrates the first shift. Agents now run reports overnight, aggregate results and prepare dashboards before the team arrives. The meeting no longer exists to construct a shared picture of the business. The picture already exists. Attention moves to exceptions, problems and decisions.

Edward Jones shows what persistence looks like at the individual level. Adams keeps adding relevant documents, decisions and open commitments to each environment over time. The accumulated context surfaces connections, preserves continuity and prompts follow-through that once depended on a manager's memory.

Who still needs a middle manager?

Layers whose work centered on collecting, repackaging and transmitting information are now easier to question. The relevant organizational test has shifted. It is no longer "How many people report to this manager?" It is "What distinct management responsibility resides at this layer?"

If the answer is unclear, the layer may reflect an information architecture that AI is dismantling.

What happens to the capacity AI removes?

A real redesign asks which responsibilities should grow as others shrink. If the saved hours simply fill with more projects and meetings, capacity expanded. The manager role did not change.

Pal at CVS described that redistribution. His reclaimed time now flows to coaching, mentoring and direct collaboration with engineers on service improvements. "AI is assisting in creating more time for leaders to perform their primary task more effectively, like coaching and mentoring, being consultative with business stakeholders and working with engineers in sculpting the improved services," Pal said.

What does it mean to manage a digital workforce?

CVS's Alan Rosa offered the clearest data point. He spends 70% of his time training AI systems and 30% using them. That work covers giving agents relevant context, tuning performance and determining what they can safely handle.

Tyler Derr, CTO at financial technology firm Broadridge, posed the question that frames the next decade of organizational design: "How do you manage the change when you've got a leader with three human employees and 19 agents?"

Is span of control still a single number?

The traditional measure breaks into at least four separate metrics:

• Human span: employees under direct people-management responsibility.
• Agent span: digital workers or systems being directed.
• Workflow span: bodies of work someone owns.
• Total productive leverage: output the combined system produces.

Those numbers will diverge. A manager with three employees and 19 agents illustrates why a single span-of-control figure obscures more than it reveals.

Where does accountability land when work is distributed?

Whirlpool drew an early line. CIO Eduardo Salas described agents that create content and run legal, regulatory and brand review steps. A human owner still approves customer-facing material. The work moves through agents without leaving ownership of the consequential decision unclear.

IBM pointed to a related shift. Matt Lyteson, IBM's CIO for Technology Platform Transformation, warned that organizations can easily mistake AI adoption for AI value. Usage is measurable. Outcomes are harder. "There is going to be a bigger focus on outcomes versus output," Lyteson said.

If an outcome is produced by two employees, six agents and an automated workflow, managing each step is no longer enough. Someone must own the overall result. The object being managed expands from the team to the human-agent operating system.

What should CIOs and CHROs do next?

Fortune recommends abandoning the job title as the starting point. Break the manager role into its parts: who gathers information, monitors progress, maintains context, follows up, allocates work, develops employees and owns the outcome.

Then decide where each responsibility belongs. Some stays with formal managers. Some moves to individual contributors. Some becomes persistent AI infrastructure. Some moves to agents. Some disappears. New responsibilities get added.

The future manager, the piece concludes, is unlikely to be today's job description with 30% of the tasks crossed out. It will be a different bundle, reassembled around a mix of human and digital work that most HR systems are not built to capture.

Broadridge is already working through the practical questions: how to identify agents across the company, what work they should perform, what systems they can access, who owns them and how their behavior stays continuously inside human oversight. Those answers will determine whether the next manager's job description reads like a people role, a platform role, or both.