McKinsey has 25,000 AI agents. Here's what actually matters.

McKinsey just announced it has 25,000 AI agents working alongside 40,000 humans.

The headlines are treating this like a workforce expansion story. More agents, more capacity, more output. We’re staring at the number and missing what actually matters.

AI agents aren’t employees. They’re ephemeral: spun up for a specific task, dissolved when done, reconfigured for the next problem. Some exist for a single client meeting. Others become templates, reused and remixed across projects. The count is already outdated by the time you read it.

Which means McKinsey isn’t announcing a workforce expansion. They’re announcing the end of a 4,000-year-old bargain: time for expertise.

The Old Deal

For four millennia, expertise came from repetition. You started as an apprentice doing grunt work. You made mistakes. You built pattern recognition. Eventually, after years, you developed judgment.

The junior consultant spent two years building financial models before “earning” the right to do strategy. The young lawyer reviewed 10,000 documents before developing judgment about what mattered. The architect hand-drafted plans until spatial intuition became second nature.

That loop just disappeared. When AI agents handle the modeling, the document review, the drafting, the junior professional never gets those formative repetitions. They skip straight to supervising outputs they lack the experience to evaluate.

Everyone’s calling this the “apprenticeship crisis.” But what if we have it backwards?

What if the traditional path to expertise was never actually that good at teaching the skills that mattered? What if we confused the vehicle with the destination? What if we told ourselves the grunt work was “foundational” because we had no other way to develop people?

Now we do.

What Actually Changes

A junior consultant joins McKinsey. Day one, they’re handed a struggling retail client. Instead of spending weeks building a market model, they work with an agent to generate fifty scenarios in an afternoon: different competitive dynamics, different consumer behaviors, different economic conditions.

Their job isn’t reviewing spreadsheets. It’s recognizing that forty-nine of those models are asking the wrong question entirely.

They spend their first six months testing business hypotheses that would have taken the previous generation five years to even get access to. They develop judgment not through artificial scarcity of opportunity, but through massive iteration on what actually matters: seeing patterns, asking better questions, synthesizing across domains.

Apprenticeship isn’t disappearing. It’s compressing and intensifying.

The Real Shift

We’re not losing the path to expertise. We’re discovering that the old path was an artifact of scarcity, not a requirement for mastery. The constraint was never “not enough dues to pay.” It was “not enough chances to practice what actually matters.”

When a junior lawyer reviewed 10,000 documents, they were developing pattern recognition. But they were also developing tolerance for tedium, deference to hierarchy, and comfort with work that didn’t require judgment.

Two possibilities:

1. These weren’t features of expertise development, they were bugs we rationalized as necessary. The next generation develops judgment faster, sees possibilities wider, builds solutions we couldn’t imagine.

2. They develop something else entirely. Pattern matching without intuition. Synthesis without understanding. Confidence without scar tissue.

We'll know we've succeeded when they can spot what the AI misses. When they look at fifty perfect models and recognize the real problem isn't being modeled at all. When a client asks 'what should we do?' and they can say 'here's what the data shows, and here's why we should do something different' and explain exactly why their judgment matters more than the model's confidence.

The Question

So the question isn’t “How many agents do you have?” It’s: “What becomes possible when your people can test any idea, model any scenario, and iterate 1000x faster than ever before?”

When intelligence becomes abundant, the valuable skill isn’t operating the tools. It’s knowing what’s worth building.

So what becomes worth building when you can build anything?

The first generation that doesn't have to choose between grunt work and judgment is finding out right now,

Ram

— 

Ram Srinivasan

MIT Alum | Author, The Conscious Machine | Global Future of Work and AI Adoption Leader published in Business Insider, Fortune, Harvard Business Review, MIT Executive Viewpoints and more.

—

A Message From Ram:

My mission is to illuminate the path toward humanity's exponential future. If you're a leader, innovator, or changemaker passionate about leveraging breakthrough technologies to create unprecedented positive impact, you're in the right place. If you know others who share this vision, please share these insights. Together, we can accelerate the trajectory of human progress.

Disclaimer:

Ram Srinivasan currently serves as an Innovation Strategist and Transformation Leader, authoring groundbreaking works including "The Conscious Machine" and the upcoming "The Exponential Human."

All views expressed on "Substrate" and across all digital channels and social media platforms are strictly personal opinions and do not represent the official positions of any organizations or entities I am affiliated with, past or present. The content shared is for informational and inspirational purposes only. These perspectives are my own and should not be construed as professional, legal, financial, technical, or strategic advice. Any decisions made based on this information are solely the responsibility of the reader.

While I strive to ensure accuracy and timeliness in all communications, the rapid pace of technological change means that some information may become outdated. I encourage readers to conduct their own due diligence and seek appropriate professional advice for their specific circumstances.