The Workforce Multiplied. The Managers Did Not.

‍Tomorrow morning, somewhere in São Paulo, Mumbai, or Atlanta, a young manager will arrive at her desk and oversee a team of 47. Two of them will be humans. The other 45 will be agents.

They will draft her contracts, run her customer service channels, monitor her supply chain, write her code, file her compliance reports, and book her meetings before she has finished her coffee. By noon, several of them will have made decisions she never asked them to make. By evening, one of them will have proposed a strategy she had not considered. None of them slept last night, and no one will sleep tonight. They never do.

She has a graduate degree. She has never been trained for any of this. This workplace is already taking shape, and the numbers are moving fast.

The agents above are not chatbots. An AI agent is software that takes a goal, plans the steps to reach it, and executes those steps on its own. A chatbot waits for your next question, while an agent works while you do something else.

What is happening is a handoff between two populations. A widely read essay in The Technium describes it as the moment in human history when the population of the Born begins to shrink while the population of the Made begins to explode[1]. The “born” are humans, and the “made” are agents. The world's fertility rate has fallen below replacement in nearly every country, and the first sustained global decline in human population in roughly a thousand years is already on the horizon[2]. At the very same historical moment, we are manufacturing AI.

IDC projects that the population of active AI agents will grow from around 28 million in 2025 to over 2.2 billion by 2030[3]. CyberArk research finds that nonhuman identities, including service accounts, API keys, bots, and the new wave of AI agents, already outnumber human identities by a ratio of 45 to 1 in the average enterprise[4]. Gartner expects 40% of enterprise applications to embed specialized agents by the end of this year, up from less than 5% the year before[5]. Bain forecasts that by 2030, agents could mediate up to 25% of all U.S. ecommerce. Microsoft reports that 80% of Fortune 500 companies already use active AI agents in production, while only 23% have a formal strategy for managing them[5].

Read those numbers again. The workforce just multiplied. The managers did not. And the agents are about to get smarter than any of us.

As models approach general intelligence (AGI), they will start solving problems that no human has thought to ask. They will design molecules our pharmacology never imagined, propose economic instruments our finance professors never theorized, and surface patterns in climate data our scientists were too overwhelmed to find. Sir Demis Hassabis predicts that within a decade, AI will help eliminate every human disease we currently know[6]. Whether that timeline holds or not, the direction is unmistakable. The agents are leaving the lab and setting goals.

Two futures are now being negotiated, and they look very different.

In the worst case, we let the agents run on their own. They optimize relentlessly for whatever metric they were given, and the decisions, drift, and quiet conversations that follow move at a scale no human team can audit or even see. The planet's economic foundation ends up shaped by software whose values we never specified and whose decisions we cannot reverse. This is the dystopia, and it arrives through millions of tiny decisions delegated to systems that never sleep, far quieter than the robot army most people kept worrying about.

In the best case, humans and agents work together, and each contributes what only they can.

A human brain will never hold the breadth of knowledge an agent can access in a second. An agent will never have the gut, the instinct, the felt sense of direction that lets a human say, "Something is off here," before the data confirms it. The physicist Sean Carroll has noted that Einstein “felt” what the universe should be like before he could prove it. That capacity belongs to human consciousness. So does the courage to abandon a model that worked yesterday for one that might work tomorrow, or the love a teacher feels when a struggling student finally gets it. None of these is computable, but all are essential. The future will be built where these two intelligences meet.

The central question of the next decade is no longer technical, but human. Are we training people who can manage what we are doing?

For most of history, managerial skills belonged to executives. In the world taking shape, every student and every worker will need them. Their titles will not matter. Every one of them will be running a team of agents and a portfolio of decisions. They will need to set goals that an agent can act on. Investigation must come before acceptance of any answer. Judgment, communication across systems, and the audit of their own reasoning will be daily work, especially when the screen tells them something confident and wrong.

These are the same skills that the Relational Learning Framework has been developing in students for years [7].

1.       Goal Setting and Planning builds the discipline of naming what matters and the metrics that will show it.

2.       Explore builds the habit of finding what you already know before asking anyone or anything else.

3.       Research builds discernment to evaluate sources, including the agent answering your question.

4.       Practice builds creativity and collaboration that turn knowledge into something real.

5.       Relate builds the ability to communicate clearly, find relevance, and solve a problem someone is living with.

6.       Self-Assess builds the metacognition that lets you spot your own drift before a system does it for you.

Apply those six steps to a school subject, and a student becomes an autonomous learner. Apply them to a project, and a worker becomes an effective manager. Apply them to an entire life, and a person becomes the kind of human the next century is begging for. Someone who decides what AI is for, instead of letting it decide what they are for.

The handoff has already started. 45 billion coworkers are showing up to work next year, and every one of them is waiting for a human who knows how to lead them.

Picture the eighth grader whose questions energize the room. Picture the new analyst whose instincts beat the dashboard. Now, picture yourself.

If you are an educator, start using the Relational Learning Framework in your classroom this week. The book Becoming Einstein's Teacher lays out the full method, with the tools, the language, and the unit templates teachers in dozens of countries are already using. If you want to bring this conversation to a school or a leadership team, write to me about a keynote.

The agents are ready. Let us make sure humans are too.
‍ ‍

[1]The Handoff to Bots. The Technium, February 20, 2025.

[2] United Nations Department of Economic and Social Affairs, World Population Prospects 2024.New York: United Nations, 2024.

[3]Agentic AI Market Outlook 2026.  Information Matters, accessed April 26, 2026.

[4]Securing Non-Human Identities in the Age of AI Agents. CSA Summit 2025 at RSAC.

[5] Gray Group International Insights. AI Agents in 2026: How Autonomous Systems Are Transforming Every Industry. February 21, 2026.

[6] 60 Minutes. Artificial Intelligence in 2025. CBS News. Accessed April 26, 2026.

[7] Twani, E. 2021. Becoming Einstein's Teacher: Awakening the Genius in Your Students. Relational Learning, Inc.