AI is advancing at a staggering pace. We're living in an era where saying "every day is different" isn't hyperbole—it's an understatement.

Running an engineering organization of about 100 people through this transformation, what strikes me most isn't the technology itself. It's the widening gap between how different people respond to it. And that gap isn't closing. It's accelerating.

The Polarization AI Creates

The output gap between those who use AI and those who don't has become impossible to ignore. Lines of code committed, new features identified, deployment frequency, volume of documentation—polarization is happening across every measurable dimension. Same hours, same org, same problems, yet outcomes diverge by multiples.

This isn't simply about whether someone uses a tool or not. There's something more fundamental separating those who embrace the new from those who don't.

First, they're deeply curious. When a new tool appears, they touch it before they fear it. This matters because the pace of AI tooling evolution is relentless. Best practices from six months ago are already outdated. Without curiosity, you can't keep up with how tools are evolving. Curiosity is the starting point of learning, and without learning, even the most powerful tool is just an expensive toy.

Second, they reach beyond their assigned scope. They want to solve not just the problem handed to them, but the adjacent problems connected to it. They're aware that authority has boundaries. But when they judge that the barrier is technical rather than political, they use AI to leap over it. Why does this matter? Because real-world problems don't resolve neatly within one person's job description. A frontend bug traces back to a backend API design flaw. A deployment delay has its root cause in infrastructure configuration. Without the will to cross boundaries, you end up treating symptoms on repeat.

Third, they build trust through results. Even with curiosity and the will to cross boundaries, if those efforts don't produce actual outcomes, you won't earn the trust of those around you. And without trust, you never get the opportunity to tackle bigger problems. Curiosity drives learning, boundary-crossing gets you to the root of problems, and the results you produce along the way generate trust and open doors to greater opportunities. These three reinforce each other in a virtuous cycle.

These differences seem small in the moment. Over time, they create entirely different trajectories.

Vision Comes First

What we should build remains a question of vision. The fact that products succeed only when vision aligns with market reality and business context hasn't changed—not before AI, not after.

Higher productivity doesn't save a wrong vision. But here's what has changed: AI lets us validate whether a vision is right at dramatically faster speeds. Hypotheses that once took months to test can now be examined in days, sometimes hours. This changes the learning velocity of an entire organization.

What This Era Demands

When I synthesize all of this, I keep arriving at one question: what kind of person does this era actually need? Over the past year of running an org through this shift, three qualities—each learned at a different time, in a different organization—have proven themselves remarkably well. A principle born on the battlefield, a behavioral standard from the world's largest cloud company, and the cultural essence of Silicon Valley's finest engineering practice. These three are converging into what I believe defines talent in the age of AI.

Extreme Ownership

This concept comes from Jocko Willink and Leif Babin—two former U.S. Navy SEAL officers who distilled the leadership lessons they learned on the battlefields of Iraq into a universal framework. The core idea is simple: a leader takes complete responsibility for everything that happens in their domain. When failure occurs, you look inward first—what could I have done differently?—before pointing outward. Their phrase "there are no bad teams, only bad leaders" captures it perfectly. What started as a military doctrine has become one of the most practical leadership principles in organizational management.

In applying this to how I run teams, the insight is this: problems don't stay neatly within assigned domains. Authority is never fully granted to a single person. So there's a real difference between "someone who only does what they're told" and "someone who understands what's needed to solve what they've been given."

Add will to that understanding, and you get someone who builds the structure and attitude required to collaborate with whoever holds the authority. They refuse to treat it as someone else's problem. This is different from the hollow "treat the company like it's yours" ownership mantra we've all heard. To borrow Willink's framing, it's not about sentiment—it's about action. It's far more concrete and practical: I will do everything within my power to solve this problem.

Deliver the Result

This one I learned at Amazon Web Services. Among Amazon's 16 Leadership Principles, "Deliver Results" is defined as: "Leaders focus on the key inputs for their business and deliver them with the right quality and in a timely fashion. Despite setbacks, they rise to the occasion and never settle."

The key phrase here is "right quality"—not perfect quality. At Amazon, you never have the time or resources to do things the way you'd prefer. So you do them the way they need to be done. This is also the single biggest indicator that separates someone solving problems from someone seeking authority.

Does this person actually resolve issues and produce outcomes? There's a meaningful difference between consuming AI relentlessly and actually using it to deliver results. Most of the polarization shows up right here. You can burn through all the tokens in the world. If it doesn't translate into outcomes, it means nothing. Conversely, those who deliver are the ones wielding AI as a precision instrument.

Be Kind

This one I learned at Pivotal Labs. Since 1989, the culture of Pivotal Labs (now VMware Tanzu Labs) has been distilled into nine words: "Do the right thing. Do what works. Be kind." These three tenets, established by founder Rob Mee, formed the cultural bedrock that supported their practice of Agile and Extreme Programming (XP). At Pivotal, empathy was explicitly evaluated during hiring—for PMs, developers, designers, everyone. When someone struggled, they didn't shame them. They paired with them. That was what kindness looked like in practice.

Why is Be Kind the third principle, and not the first? Because of how these three relate to each other. Extreme Ownership is something a leader can practice alone—it's a personal commitment to take responsibility. But Deliver the Result almost always requires collaboration. The authority you need, the information you lack, the systems you can't access—these live in other people's hands. And getting those people to willingly collaborate with you? That's what Be Kind makes possible.

Not every visitor is a customer, and kindness isn't an obligation owed equally to everyone. The kindness I'm talking about operates on a different level.

True kindness is about your attitude toward people you don't yet know—teammates, collaborators, the person who holds the authority you need—before trust has been established. How you behave in that gap reveals who you really are. It's through building trust with the diverse range of people you encounter professionally that authority gets delegated, information gets shared, and problems get solved in the right direction. This is not optional—it's foundational.

Sometimes this looks transactional. But when it crosses the threshold into genuine trust, it becomes something far more than a transaction. There's a reason Pivotal placed "Do what works" and "Be kind" side by side—pragmatism and kindness aren't opposing forces. They're the twin pillars that build trust.

To put it plainly: these three are not independent virtues. Extreme Ownership makes you take on the problem. Be Kind builds the collaborative foundation to solve it. Deliver the Result proves you actually did. Responsibility → Trust → Results. The person in whom this cycle turns is the talent this era demands.

Highly skilled people are often weakest here, worn down by repetitive questions from others. But that's a problem that only applies when Extreme Ownership and Deliver the Result are already in place. When someone lacks the competence to build trust in the first place, their unkindness is just a symptom of incapability.

Give the $500K Engineer $250K in Tokens

This brings me to something Jensen Huang, CEO of NVIDIA, said that resonates deeply.

Jensen Huang — "Spend $250K on tokens for your $500K engineer"

I agree wholeheartedly. There is nothing more important than ensuring that the experience of burning through tokens—driven by curiosity, driven by the obsession to solve problems—translates into real results. The goal isn't to economize on tokens. It's to maximize the value each token creates.

The People You Want to Work With

If we define "talent" as someone you'd want to work alongside, then anyone who embodies those three qualities is very likely that person.

These people share common traits. They invest willingly in their own growth. They have a burning desire to get to the essence of problems. They work relentlessly to fill their own gaps. They crave more experience than anyone else around them, and they don't hesitate to expand the breadth and depth of what they know.

The age of AI is giving these people wings. Things that were invisible behind authority boundaries, domains where they had no choice but to depend on others due to learning speed and time constraints—those locks are being opened. The people experiencing explosive expansion through this unlocking are the true talent of this era.

Closing Thoughts

There have been many attempts to build startup-like teams within enterprise organizations. Some succeeded, some failed. That's still true today.

But the emergence of cloud, and the use of AI on top of it to expand individual capabilities across different roles, is becoming a means to validate the possibilities of a vision at ever-increasing speed. Among those who try, some will succeed and create value in domains others can't reach.

Grounded in a personal value—"always be someone who's needed somewhere"—I've been preparing and supplying what the organization needs for this shift. The metrics are starting to show patterns that suggest good outcomes are ahead, so I wanted to share these thoughts.

Finding ways to narrow the knowledge gap within an organization, and identifying and retaining the people who can positively influence those around them—this has never been more critical than it is right now.

In an era where AI is becoming not just a tool but the environment itself, what matters most is still the people.