ASSISTANT PROFESSOR
For the past few years, most of us have experienced Artificial Intelligence as a helpful assistant—something that drafts an email, summarises a report, or answers a question the moment we ask. Useful, certainly, but always waiting for our next instruction. That relationship is now changing fundamentally. A new generation of AI, described as “agentic”, does not merely respond—it acts. Given a goal, such a system can plan the steps, make decisions, use digital tools, and carry a task through to completion with little human intervention. At JIMS Kalkaji, within the PGDM programme, where the emphasis lies on preparing managers for a fast-evolving business landscape, this shift deserves close attention—because it changes not just what technology can do, but what a manager’s job will actually involve.
From Answering to Acting
The difference between an assistant and an agent is easy to overlook, yet important to grasp. An AI assistant is essentially reactive—it depends on human input and handles one request at a time. An AI agent, by contrast, is goal-directed. Tell it what outcome you want, and it works out how to get there: breaking the goal into steps, choosing which tools to use, and adjusting course when something does not go as planned. In practice, this might mean an agent that handles an entire customer-service resolution, compiles and files a routine compliance report, or coordinates a stretch of the sales cycle from start to finish. The research firm Gartner captures the scale of this change bluntly—fewer than five percent of enterprise applications carried such task-specific agents in 2025, a figure it expects to approach forty percent by the end of 2026. Whatever one makes of the exact numbers, the direction of travel is unmistakable.
The Manager as Delegator, Not Doer
If agents can execute tasks on their own, the manager’s role inevitably moves up a level. For decades, management thinking has drawn a line between doing the work and directing it; agentic AI sharpens that line considerably. The manager of the near future will spend less time issuing step-by-step instructions and more time doing what capable leaders have always done with capable teams—setting clear objectives, defining boundaries, and reviewing outcomes. In truth, delegating to an AI agent is not so different from delegating to a junior colleague. One has to be specific about the goal, honest about the constraints, and clear about which decisions may be taken independently and which must be brought back for approval. The classic managerial questions—what does success look like, where are the limits, who is accountable—do not disappear. If anything, they grow more important, because this particular “employee” works faster, tirelessly, and without the human instinct to pause and ask when something quietly feels wrong.
Why Enthusiasm Alone Is Not a Strategy
It would be a mistake, however, to treat agentic AI as a finished solution waiting to be switched on. Alongside the genuine excitement sits a sobering caution: Gartner estimates that more than four in ten agentic AI projects may be abandoned by the end of 2027, undone by rising costs, unclear business value, and weak controls. A good deal of what is marketed today as an “agent” is really an ordinary assistant dressed in ambitious language—a practice the industry has begun to call “agent-washing”. For managers, the lesson is a familiar one in new clothing. Technology tends to succeed where it is aimed at a real, well-defined problem with a measurable payoff, and to fail where it is adopted merely because it is fashionable. Choosing the right task, setting honest success metrics, and being willing to shut down what does not work are managerial disciplines far more than they are technical ones.
Accountability When the Machine Acts
Autonomy also raises a question that management cannot delegate away: who answers for the agent’s decisions? When a system acts on its own—approving a request, sending out a communication, moving stock between locations—the responsibility for that action still rests with the organisation and the people who run it. This is why oversight, auditability, and clear guardrails are shifting from optional extras to basic requirements. A well-designed agent should operate within defined limits, keep a traceable record of what it did and why, and hand control back to a human whenever a decision carries real weight. Some estimates suggest that within a few years, a meaningful share of routine workplace decisions could be taken autonomously; the organisations that truly benefit will be those that grant such autonomy deliberately, keeping human judgement firmly in the loop for anything consequential.
The Skills That Will Define the Next Manager
All of this points to a quiet but significant change in what makes a manager effective. The premium is moving away from being the person who performs the task and towards being the person who can supervise a system that performs it—framing the problem clearly, judging the quality of an agent’s output, and knowing when to trust it and when to step in. These are not coding skills; they are judgement, communication, and ethical clarity, applied to a new kind of workforce. For management education, the implication is direct. Preparing students for this world means teaching them to treat AI not as a gadget that answers questions, but as a capable yet fallible collaborator that must be directed, checked, and held to account. The managers who thrive in the coming years will be those who learn to lead alongside intelligent systems—delegating boldly, supervising wisely, and never surrendering the human responsibility that no algorithm can carry. Agentic AI may be able to act on its own; deciding what it should be trusted to do will remain, firmly, a human task.
