When Every Employee Manages 200 AI Agents

Igor Rikalo is President and COO at o9 Solutions.

getty

​By the early 2030s, I expect agentic AI to significantly reshape the workplace. Over the next five years, a top product manager within an organization may supervise 180 agents, and a head of supply chain may coordinate 300 specialized digital co-pilots that are tasked with running forecasts, negotiating contracts, testing scenarios and flagging risks in real time.

The dynamics introduced by agentic AI have the potential to boost productivity, compress decision speed and increase revenue per employee. Yet, there may be impediments to this potential as well. The limiting factor won't be data, computational capabilities or capital—it will be how well humans can utilize it.

AI Will Multiply Knowledge Workers, Not Replace Them

When the merits and drawbacks of AI are debated, the possibility of job displacement is often a central focus, but another focus should be how AI can be leveraged by a workforce. AI can move beyond task automation and ultimately multiply the capabilities of each knowledge worker who uses AI tools.

For example, if a product manager with 200 agents delivers five times the previous output, this could structurally change revenue per employee as operating margins expand and decision latency shrinks from weeks to hours. This increased productivity will likely be rewarded at the executive and board levels, but the velocity it creates produces secondary issues that will have to be addressed.

The Cognitive Bottleneck

As AI agents generate more and more alerts and recommendations, the volume of incoming information outpaces human capacity to process it into insights that guide decisions. As AI systems continue to enhance and accelerate planning and decision-making cycles, human oversight becomes increasingly constrained.

As a result, new risks emerge. As employees focus on overseeing agentic AI tasks and automations in real time, they could experience attention fragmentation across hundreds of agent signals, endure decision fatigue caused by constant prioritization and develop an overreliance on automated recommendations that could result in a degradation of decision quality, missing signals that are lower frequency but higher impact, and automation biases.​

The New Role Of The Manager

As I mentioned in a previous article, incorporating agentic AI into the workplace will significantly alter responsibilities across roles. In this dynamic, management responsibilities will move from supervising people to orchestrating agent systems. In the years ahead, managers will be responsible for designing a portfolio of agents, determining which decisions are made by humans versus automated systems and which require escalation based on specific criteria.

More broadly, leaders across an organization will be charged with setting decision rights between humans and AI agents, establishing override thresholds, auditing model bias, measuring return on investment and configuring agent networks to align with business strategies and initiatives. Managers don't become less important. Their focus changes from supervising activity to governing decisions at scale.​

How Middle Management Shifts Its Focus

While middle management may feel it's in the crosshairs, it's highly unlikely that further AI implementations will eliminate management roles. Instead, it will be the catalyst for transforming middle management roles and daily responsibilities.

For example, if today's manager oversees a team of eight people, in three to five years, they could manage eight people in addition to 400 agents. As a result, the span of control becomes multidimensional, potentially causing some manager layers to shrink, while other roles will emerge around governance, cognitive load coordination and decision architecture.

It's necessary to consider the risks that can occur if complexity isn't appropriately managed, because if an organization activates hundreds of agents without clear decision trees, accountability and structured escalation logic, small errors could amplify in scope and affect multiple areas of their organization and value chain. Therefore, it's imperative that companies treat cognitive architecture as seriously as financial architecture.

How Roles Change For Senior Executives

At higher levels of leadership, roles will likely evolve as well, becoming less focused on gathering information and more on actions involving judgment, clarity and sequencing. For example, a senior executive could begin her day reviewing 60 AI-generated briefs that need to be escalated based on risk probability and financial exposure. By mid-morning, she receives approval alerts from three agents requesting reallocation of capital in response to real-time demand. In the afternoon, she's reviewing actions suggested by a supply agent that flagged a projected service risk and needs to sign off on a suggested multimillion-dollar inventory move. The executive has made fewer decisions throughout her day, but each decision carries more impact, and any mistakes can have a broader reach.

How Boards Should Prepare For Agentic AI In The Workplace​

Traditionally, company boards have used revenue per employee as a benchmark to measure productivity. Although this measure is still useful, it needs to be combined with other measures to provide a fuller scope of productivity across an organization. Boards should also consider incorporating metrics like revenue per employee and agent clusters, agent density per critical role, average decision latency, human override rate, escalation load per executive and error containment speed. The above-mentioned metrics can measure both productivity and cognitive stress and indicate whether an organization's AI deployment increases productivity or amplifies fragility.

Overall, deploying AI agents will increase decision velocity, but it also raises the question of how to protect judgment quality as AI usage increases. Before introducing agentic AI programs within your organization, it's important to consider developing clean decision rights before scaling autonomy, simplifying workflows before multiplying inputs and training leaders in agent orchestration, not only technical adoption.

A successful agentic AI deployment will depend on how well your organization structures the relationship between human cognition and machine scale. AI will be capable of multiplying intelligence, but leaders must be able to multiply clarity.

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?