AI made building easier, but engineering harder

I was hunting for a database URL.

As a product person, if you'd told me I'd be digging through Lagoon environment variables at 10pm trying to figure out why my token analyzer (something I built by myself for work) couldn't connect to anything, I would've laughed and gone back to my figuring out how to get two people to just agree on a decision. After all, nothing says fun in product management like teams that are unaligned.

But here I was. And the thing that got me there? AI.

There's this narrative floating around right now that AI is going to eliminate technical jobs. That soon, anyone can build anything. That the barriers are falling and we're entering some kind of golden age where you don't need engineers anymore.

While much of this is true, it's true to a point.

The paradox nobody's talking about

I had a conversation with one of my engineering colleagues that clarified something I'd been feeling but couldn't quite articulate. The crazy thing about AI is that because you can do so much now, it's pushing you to become more technical. Not less.

Read that again, because it runs counter to basically every AI hot take on your LinkedIn feed right now.

AI lowered the barrier for me to attempt technical work. I can spin up a token analyzer now. I can build things I couldn't have touched a year ago. But the moment I tried to do anything real with it I needed to understand database URLs, environment variables, YAML files, Kubernetes deployments. Things I've only thought about as being engineering's problem to figure out.

The tools let you try things you couldn't before. But understanding what you're doing? That's still the whole game.

My colleague put it pretty bluntly. "To have somebody non-technical write an app that we put in production is madness."

And he's right. I think about how at work we have several people building internal apps with AI tools. These people are smart, resourceful and getting a working prototype up quickly has a lot of merit. But "working prototype" and "production-ready application" are two very different things.

It's like the maker movement. I love the maker movement. But imagine a world where everything you used, your car, your electrical system, your medical devices, was built by enthusiastic hobbyists with a YouTube tutorial and a 3D printer. That's chaos.

AI coding tools are giving everyone a 3D printer. That's genuinely exciting. But it doesn't mean we suddenly don't need engineers.

Scott Hanselman — who actually is deeply technical — has been saying the same thing. AI isn't making him less technical. It's making him more technical. Because the easy stuff gets automated, and what's left is the hard stuff. The architecture decisions. The deployment infrastructure. The "why is this failing silently in production at 3am" stuff.

As soon as you start doing anything that's non-trivial, you need to know what's going on.

And here's what's wild about my own experience. The more AI helps me build, the more non-trivial things I encounter. I've gone deeper into technical territory, not shallower. The AI didn't replace the need to understand things. It just moved me to the part where understanding things actually matters.

Why this matters for product people

If you're a product person reading this, I'm not saying you should become an engineer. (I'm definitely not becoming an engineer.) But I am saying the "AI will handle the technical stuff so you don't have to" story is wrong. Or at least dangerously incomplete.

What AI actually does is give you access to the messy middle of technical work. The part where you're close enough to see how things work but far enough to realize how much you don't know. And honestly? That's where the best product thinking happens anyway.

The SaaS model exists for a reason. Standardized, maintained, professionally engineered software exists because "anyone can build it" doesn't mean anyone should build it, or maintain it, or be on call when it breaks at 2am. There's a reason we don't all run our own email servers.

I keep coming back to this analogy. AI is a power tool. Power tools are amazing. They let you do things faster, attempt things you couldn't before, build things that would've been impossible without them.

Power tools in the hands of someone who knows what they're doing? Incredible.

Power tools in the hands of someone who watched one YouTube video? That's how you lose a finger.

I'm not saying don't pick up the power tool. Pick it up. I picked it up, and it's changed how I work. But respect what you don't know. Learn the safety protocols.

The future isn't less technical. It's technical in different ways, and it's asking more of us, not less. I'm still figuring out what that means for someone like me.

But I'm figuring it out.