Navigating Your Career in the Age of AI

The world has gone crazy over AI. And I mean that literally—just look at the stock market. The "Magnificent Seven" tech giants accounted for more than half of the S&P 500's 25% return in 2024. Strip out tech entirely, and the index returned just 11%. These seven companies now make up 35% of the S&P 500—up from 12% a decade ago. And here's the stat that really puts it in perspective: their combined market cap went from 10% of U.S. GDP in 2014 to 60% in 2024.

Trillions of dollars are betting that AI will reshape everything. And nowhere is that reshaping more visible than in software engineering.

AI Can Code. Like, Really Code.

I'm not talking about autocomplete on steroids. AI agents can now reason through problems, understand complex codebases, and write production-quality code at a level that matches—or exceeds—many senior engineers. That's not hype. That's what I've seen firsthand after spending the last year coding with AI nearly every night.

There's a lot to be excited about here. And honestly, a lot to be scared about too.

But one thing is certain: AI is the future. In many ways, it's already here. So the real question becomes—what do we do as professionals?

The Resistance Is Real

At least in my experience, I've seen a surprising amount of pushback from engineers themselves when it comes to using AI. The reasons I hear most often:

• "I love coding. I don't want to give it up."
• "I don't trust the AI."
• "It produces sh*t code."
• "How can I learn if it's doing everything?"
• "What's the point? We'll just be given more work to do."

I get it. These aren't irrational concerns. There's real identity wrapped up in being a craftsman. And there's real fear about what happens when the craft gets automated.

But here's the thing: this is such a tectonic shift in the way we think and work that without truly immersing yourself in it, you won't be able to see what's coming. Let me give you an example.

A Sizing Exercise

If I asked you—how long would it take to migrate a fairly static blog off Squarespace onto your own home-rolled infrastructure? Then layer on features like email campaigns, Google authentication, commenting, email notifications, and an admin portal to manage all of it.

If you engineer for a living, take a minute to actually size that in your head.

I'll leave my own estimate—and what it actually took me—at the bottom of this post.

The Mindset Shift

So what does it take to not just survive but thrive in this new era?

Blend multiple roles. AI doesn't just let you wear more hats—it rewards those who do. When I'm building now, I'm not just an engineer—I'm also a designer, a product manager, and in some cases, a founder.

Here's the part nobody talks about: AI has gotten so fast that coding is no longer the bottleneck. Decisions are.

When you can ship a feature in two hours, you suddenly need to know what to build next. You need to make the design call. You need to decide if the tradeoff is worth it. You need product judgment—right now, not after a meeting next week. If you can only wear one hat, you'll spend more time waiting than building. And you'll wonder why AI isn't producing the 10x results everyone promised.

This applies within engineering too. Frontend, backend, infrastructure, data—the lines are blurring. The full stack engineer who can move across all of it will run circles around the specialist who has to hand off and wait.

The engineers who will thrive aren't the ones who prompt the best. They're the ones who can move fluidly across disciplines and make good decisions fast enough to keep up with the machine.

That's the new flow state. And most people can't access it yet.

Become a master orchestrator. The skill isn't just prompting anymore—it's knowing when, where, and how to apply AI agents. Understanding their limits. Being creative about where they fit. Directing them like a conductor, not just a typist.

Solve real problems. This is the part that separates demos from products. Anyone can spin up something cool in a weekend. But can you productionize it? Iterate on it? Make sure actual users are benefiting? That's the work that still matters—and arguably matters more now that execution is cheap.

Go back to being a white belt. A lot of current best practices are going to become irrelevant. That's uncomfortable if you've spent years mastering them. But the people who thrive will be the ones willing to unlearn and rebuild from first principles.

My Own Tension

After spending nearly every night coding with AI, I've found that building has gotten dramatically easier. But here's the irony—I now have a harder time figuring out what problems are actually worth going after.

Distribution, or lack thereof, is giving me serious pause. I've shipped a few apps. Some got users. But the question of whether to start my own company versus join another one keeps nagging at me. Building is one thing. Finding the right thing to build—and getting it in front of people—is another game entirely.

If you're thinking about going solo, you'll probably hit this wall too. The barrier to building is gone. The barrier to distribution is not.

The Bottom Line

If you plan to be marketable for the next decade, being effective at leveraging AI is the single most important skill you can develop on your own. Not optional. Not "nice to have." Essential.

Over the last 15 years, software engineers have been financially rewarded at an outsized pace compared to almost any other profession. I believe the next wave of high earners will be those who are effective at building and orchestrating AI agents—regardless of their original job title.

The engineers who resist will find themselves outpaced—not necessarily by AI itself, but by the engineers who figured out how to work with it.

Sizing answer:

A year ago, this would have taken me 4-5 weeks of nights and weekends. I don't touch DevOps or third-party integrations often, so there'd be a lot of Googling involved. With AI, it took 6 hours across 2 nights—and by "hours," I mean the gaps between burping sessions and diaper changes.

4-5 weeks down to 6 hours. That's the delta. That's why this matters.