12 Years, One Firm, and the AI Chasm

I spent 12 years at McKinsey. It’s an experience that feels both like a lifetime and a heartbeat.

I joined as a fresh-eyed engineer who wanted to solve big problems but had no idea how. I left understanding that the “big problem” is rarely the data or the strategy — it’s the human architecture behind it. That shift in perspective took a decade to fully earn. And now, as I watch AI fundamentally reshape the nature of strategy work, I find myself at a strange crossroads: deeply grateful for the toolkit I was given, and increasingly convinced that the model that produced it is facing something it hasn’t faced before.

Here’s what twelve years at McKinsey taught me — and why I believe the next decade of strategy and execution looks fundamentally different from the last.

1. Imposing Structure on Chaos

Clients don’t call McKinsey when things are clear. They call when things are politically charged, emotionally loaded, and genuinely hard to frame. The most valuable skill I developed wasn’t financial modeling or slide-making — it was walking into a room where no one knew where to start and drawing the first box. Taking a paralyzing, undefined situation and converting it into a hypothesis-driven workplan. Creating direction from ambiguity.

This is still the hardest thing to do well. AI can generate remarkable content, analysis, and code — but it cannot yet generate intent from a vacuum. Someone has to define the problem before the machine can help solve it. The human role isn’t disappearing; it’s concentrating at the front end of every hard question.

2. Leadership Has No Single Answer

McKinsey’s feedback culture is rigorous — sometimes painfully so. But what it gave me wasn’t a playbook for leadership. It gave me something more durable: the conviction that there is no single playbook.

I worked with partners who led through sheer intellectual force, and others who led through deep listening. I watched brilliant senior colleagues build followership through humor, through precision, through radical candor, through quiet consistency. What they shared wasn’t a style — it was intentionality. They had each figured out what kind of leader they were, and deployed it deliberately.

What I came to understand is that leadership is situational by nature. The approach that unlocks a demoralized client team is not the same one that drives a high-performing internal team through a brutal deadline. The posture that works with a skeptical CEO doesn’t work with a first-year analyst who needs to believe the work matters. Effective leaders aren’t the ones who’ve mastered a single mode — they’re the ones who’ve built a wide enough range that they can meet the moment.

For me, that meant doing the uncomfortable work of understanding my own defaults: where I naturally lead from, where I overindex, and where I have to consciously compensate. The goal was never to become someone else’s version of a great leader. It was to develop a clear, honest model of my own — and keep refining it against the feedback the environment gives you, whether you ask for it or not.

3. The Supercomputer That’s No Longer Enough

Management consulting became one of the most influential industries in the world by building something remarkable: the highest-performing human organizations that had ever existed. The model was straightforward in theory and brutally difficult in practice — recruit the most intellectually capable people you can find, train them relentlessly, put them under extreme pressure, and deploy them at the world’s most complex problems.

In an era before sophisticated enterprise software, before real-time data infrastructure, before AI, this was the answer. If you needed to process ambiguous information at scale, synthesize it quickly, and turn it into a recommendation a CEO could act on — you needed a room full of exceptional human beings working harder and smarter than anyone else. Consulting firms were, in effect, the supercomputers of their day. The hardware was human.

That model worked extraordinarily well. And it is no longer sufficient.

The shift isn’t that smart, hard-working people have become less valuable. It’s that the center of gravity in any operating model is moving. Technology — and specifically AI — is no longer the tool in the consultant’s hands. It is increasingly the engine around which the work itself must be designed. The analysis that once required a team of six working through the weekend can now be scaffolded in hours. The synthesis that justified a months-long engagement can be compressed. The value that was locked inside human cognitive effort is being redistributed.

The consulting firms that thrive in the next decade won’t be the ones that add AI to their existing model. They’ll be the ones willing to ask a harder question: if machine intelligence sits at the center of the operating model, what does the work actually look like? What do we rebuild? What do we let go?

That redesign is not incremental. It requires the same willingness to draw the first box in a chaotic room — except this time, the room is the industry itself.

Looking Forward

I remain a product of those twelve years. The principles of intellectual honesty, rigorous logic, and genuine empathy for the people on the other side of every problem — those survive every technology cycle. I’m not worried about them.

What I’m excited about is the rebuilding. The frameworks, the operating models, the ways of structuring hard problems — all of it is up for reconsideration. That’s not a threat. That’s the most interesting strategic moment in a generation.

I’m grateful for the twelve years that built me. Now I’m here for what comes next.