The Generalist's Edge: Doing IT Work in the Age of AI

The apology I used to make

For most of my career, “I’m a generalist” came out as an apology.

In IT, the word can sound like jack of all trades, master of none. You keep the systems running, you fix what breaks, you learn whatever the day demands — but you’re not the person with ten years deep in one stack. When a job posting wants a Kubernetes specialist or a senior network engineer, “generalist” feels like the thing you say when you don’t have the title they’re looking for.

I stopped apologizing for it. Not because I suddenly became a specialist — I didn’t — but because the tools changed what a generalist can do, and the math is different now.

What the job actually is

My day job is end-user and infrastructure support. The honest version of that work isn’t glamorous: keep the systems healthy, unblock people who can’t do their jobs, and quietly automate the repetitive stuff so it stops eating the week. The skill that matters most isn’t depth in any one technology. It’s triage — walking into an unfamiliar problem, figuring out what’s actually wrong, and knowing enough across enough domains to find the thread to pull.

That’s the generalist’s real job: not knowing everything, but knowing how things connect, and being able to learn the specific thing fast when you need it.

The bottleneck was always that last part — “learn the specific thing fast.” It used to mean documentation dives, forum archaeology, and a lot of trial and error on a Tuesday afternoon. That’s the part that changed.

AI didn’t make me a specialist. It made the depth available on demand.

Here’s the shift, in one sentence: AI collapsed the distance between “I don’t know this yet” and “I know enough to act.”

I’m not deep in Rust. But I shipped a Rust app, because I could ask the right questions in the moment and get just enough understanding to make the next decision. I’m not a security specialist. But I ran a serious, multi-round penetration test against my own software, because AI let me go as deep as I needed on each specific question without first spending a year becoming an expert in all of it.

That’s not the same as the AI doing the work. The judgment stayed mine — what to build, what to trust, when something smelled wrong, when to throw a whole approach out. AI is fast and tireless and completely without taste. It will help you build the wrong thing with total confidence. The generalist’s instinct for “wait, this doesn’t fit how the rest of the system works” is exactly the thing that doesn’t automate.

What AI removed was the tax on breadth. A generalist used to pay for range with shallowness — too many things to ever go deep on any of them. Now the depth is a question away, so the range becomes the asset instead of the excuse.

Why this favors the generalist specifically

A specialist’s edge is depth in a domain. That edge is real, and AI narrows it — the model can hand a generalist a competent first pass at the specialist’s domain.

But a generalist’s edge is connection — seeing how the network, the identity system, the app, the data, and the humans all interact. That edge AI doesn’t narrow. If anything it widens it, because now the person who can see the whole board can also reach into any one square on demand. The constraint was never the vision; it was the time to execute across all of it. AI is execution speed.

So the generalist who’s willing to keep learning — who treats “I don’t know that yet” as a starting line instead of a wall — gets a force multiplier aimed precisely at their weak spot.

The honest caveats

I want to be careful not to oversell this, because the overselling is everywhere and it’s exhausting.

AI is wrong a lot. It’s confidently wrong, which is worse. Using it well is mostly about knowing enough to catch it — which means the floor of competence didn’t disappear, it moved. You still have to understand the thing well enough to know when the answer is garbage. The people who get burned are the ones who mistake a fast answer for a correct one.

And there’s a real failure mode where leaning on AI keeps you permanently shallow — you get the answer, you ship, you never actually learn. I fight that by asking why, not just what, and by being honest about the difference between “I shipped this with help” and “I understand this.” On my own site I try to say plainly where I’m expert and where I’m still learning, because that distinction matters and pretending otherwise ages badly.

Where this leaves me

I’m an IT and infrastructure person who builds with AI. A generalist, on purpose. I keep systems running, I ship real things, and I learn the next one by building with it rather than waiting to feel expert enough first.

For most of my career that sentence would have felt like a confession. Now it feels like the most useful version of me I can offer — and, finally, like an advantage instead of an apology.