December is the only month where I give myself permission to stop. Not slow down — stop. The laptop stays closed more than it's open. The terminal doesn't get touched. My wife and I cook too much food, the kids are home, and for a few weeks the house feels like it did before work swallowed everything. We bought a new car this month — a 2026 Lexus TX in Matador Red — and before anyone thinks this was some midlife-crisis impulse buy, let me explain. My son is six foot five. Six-five. This kid does not fit in normal cars. He doesn't fit in most SUVs. Every family trip involves him folding himself into the backseat like a human origami project, knees pressed against the dashboard, head tilted sideways. The Lexus TX has three rows and enough headroom for a small giraffe, which is basically what he is. This tall child is extraordinarily expensive — the grocery bill, the shoes, and now the car. But it's red and it's beautiful and he called shotgun before the dealer even handed us the keys.
The new family ride. No API required.
But even when I stop, my brain doesn't. And this December, sitting on the couch while my family watches something I'm pretending to pay attention to, I keep circling back to the same thought. My son is in his third year at Georgia Tech studying computer science. He's smart, he works hard, and he picked this major because he saw what I do and thought: I want to build things too. And now I'm watching the industry he's preparing to enter start to reshape itself in ways that make me genuinely worried. Not panicked. But worried in the way that a father worries when he can see something his kid can't yet.
Here's what I keep turning over. The companies adopting AI aren't just adding it as a tool for their engineers — they're using it to replace the need for as many engineers. An agentic coding system today can do in hours what used to take a junior developer weeks. That's not hype. I've built these systems. I know exactly how capable they are and how fast they're getting better. And the uncomfortable math is this: if a company can ship the same product with a team of ten that used to require a team of fifty, what happens to the other forty? They don't just move to another team. They move to another industry. Except every industry is doing the same math at the same time. And the jobs they'd normally redeploy into — the mid-level knowledge work, the SaaS implementations, the consulting gigs — those are exactly the jobs that AI is eating first.
I think about the hundreds of thousands of kids like my son, sitting in computer science programs across the country right now, learning data structures and algorithms and software design patterns, and I wonder: are we training them for a world that's about to change underneath them? Not in ten years. In three. Maybe two. The curriculum they're studying was designed for a labor market that assumed human programmers would always be the bottleneck. But what if the bottleneck shifts? What if the scarce resource isn't someone who can write code, but someone who can architect systems, understand business domains, and direct AI agents to do the implementation? That's a fundamentally different skill set, and most universities aren't teaching it yet.
And it's not just tech. When AI agents can handle transactions, broker deals, process claims, underwrite policies, and manage portfolios — entire layers of the economy that exist because humans are the ones moving information from point A to point B start to compress. The people in those layers are the ones buying the houses, paying the mortgages, filling the restaurants, and driving the consumer economy. What happens when their income disappears not because they did anything wrong, but because the work itself stopped requiring a human? Output keeps growing. GDP looks fine. Corporate earnings look great. But the money isn't routing through households anymore. It's routing through compute. And you can't spend compute at the grocery store.
The thing that scares me most isn't that this will happen. It's the speed. Every previous technological disruption — the printing press, the steam engine, electricity, the internet — gave society at least a generation to adapt. Workers retrained. Institutions adjusted. New industries emerged to absorb the displaced. But AI isn't moving at the speed of a generation. It's moving at the speed of a model release cycle. Every six months the capabilities double. Every year the cost halves. The displacement isn't gradual — it's compounding. And the policy infrastructure, the education system, the social safety nets — none of it is built for this speed. We're trying to absorb an exponential shock with linear institutions.
I don't say any of this to my son. Not yet. Because I don't have a clean answer, and I refuse to hand him my anxiety without a roadmap attached. What I do tell him is this: don't just learn to code. Learn to think. Learn to architect. Learn to understand why systems exist, not just how they work. Be the person who knows what to build, not just how to build it. Because the "how" is getting commoditized at a pace that would have seemed insane even two years ago. But the "what" and the "why" — that still requires a human who understands context, domain, stakes, and consequences. That's the moat. That's what AI can't do yet. And if I'm honest, I don't know how long that moat holds either.
So I sit here in December, in my quiet house, with my family around me and a shiny red car in the driveway, and I hold two truths at the same time. The first: AI is the most extraordinary thing I've ever worked with, and I wouldn't trade this moment in history for any other. The second: the world my children are inheriting is being rewritten in real time, and the people doing the rewriting — people like me — have a responsibility to think about what comes after the disruption, not just the disruption itself. December is for family. But even in December, the questions don't stop. They just get quieter. And somehow, that makes them louder.
