The Manager's New Role

How management changes

The manager's new role is not to become a better prompt writer. It is to decide how work should be decomposed, what agents should own, when humans must intervene, and how learning should feed back into the system.

From coordination node to system designer

In the old knowledge-work firm, the manager is often the node through which work flows: gathering updates, routing information, resolving handoffs, and coordinating people. In the AI organization, the manager designs the system through which work flows.

What managers govern

Managers govern objectives, permissions, quality standards, escalation paths, feedback loops, and the boundary between agent execution and human judgment. They decide which work should be explicit, which work can be automated, and which work requires accountable human participation.

Why this is not less management

AI does not remove the need for management. It changes the object of management. Instead of managing every motion of human coordination, managers manage the operating environment in which agents and humans each do the work they are suited to do.

The skills that matter

The new role requires systems thinking, process clarity, data literacy, judgment about risk, and the ability to translate business goals into governable workflows. It also requires humility: the manager must learn from interventions and failures rather than treating them as exceptions to hide.

Why the role becomes more architectural

Management becomes more structural. The manager is responsible for the conditions under which agent work is safe, useful, measurable, and continuously improving. That makes management more important, not less.

The old managerial bottleneck

In many knowledge-work firms, managers are the human middleware of the organization. They gather status, interpret priorities, route decisions, remind people of commitments, resolve unclear ownership, and translate between teams. Much of that work exists because the operating system of the firm is incomplete. AI makes it possible to redesign some of that coordination rather than asking managers to carry it personally.

The new managerial questions

The manager now needs to ask different questions. What work should be executed by agents? What information must the agent see? What decisions can be made automatically? Which exceptions require human judgment? What should be logged? What should be reviewed? How will the system improve after each failure? These are design questions, not supervision questions.

Why authority must change

A manager cannot govern agentic work without authority to change the workflow. If they can observe failures but not change data access, policy, permissions, staffing, or escalation rules, they become spectators. The AI organization needs managers with responsibility for systems, not merely responsibility for people operating inside broken systems.

What good management looks like

Good management in the AI organization looks like clear workflow ownership, explicit decision rights, high-quality escalation design, disciplined review of interventions, and continuous improvement of the system. The manager is judged not only by team output, but by whether the system becomes more legible, reliable, and capable over time.