Generalist in the age of AI
tl;dr I spent 15+ years feeling like a fraud because I knew a bit of everything and mastered nothing. Turns out, in the age of AI, that's exactly the profile that ships the fastest.
Somewhere around 2003 I was running a Gentoo shell server on an ADSL line, debugging bare-metal firmware one day and writing questionable JavaScript the next. I could navigate pretty much any part of a stack, but I never felt like a "real" engineer. Good at a lot, great at nothing. Every time a new framework dropped, the gap between me and the specialists seemed to grow.
Fast-forward to 2026. I've been building products for over fifteen years - Python, Go, Ruby, Django, FastAPI, Rails, cloud infra on AWS and GCP, IoT, energy systems, you name it. I've touched hundreds of frameworks and deployment patterns. And suddenly that "problem" of knowing a little about everything feels like a superpower.
Here's what changed: AI filled in the gaps.
The deep trench work - the boilerplate, the repetitive implementation, the stuff that used to take a specialist two days - now lives inside the AI copilots sitting on my desk. My job isn't to memorize every API or master every language. My job is to orchestrate. Describe the outcome, lay down constraints, and let an AI agent tear through the boring parts.
I can spin up a PoC backend, a basic mobile UI, and customer-facing copy in a single morning. Not because I'm some genius - but because I know enough about each domain to direct traffic instead of digging every ditch myself. That same generalist instinct that once felt like dilution now lets me compose teams of humans, scripts, and AI tools without losing the plot.
And it works both ways. A heat pump installer who knows every screw torque can now use AI to draft contracts and handle customer emails. A performance engineer can let AI write the docs while they chase the hard bugs. Depth still matters - but AI dramatically reduced the cost of jumping lanes. The rare skill now is knowing which lane to jump to, when, and how to integrate the results.
This isn't dabbling. Being a modern generalist means disciplined breadth plus the ability to go deep enough to brief AI properly, set clear success criteria, and judge whether the output is actually good.
In practice, it comes down to three things.
Portfolio breadth. Keep two or three domains in "pretty competent" shape - product + ops, firmware + cloud, design + growth, whatever your combo is. Use AI to refresh the edges instead of relearning everything from scratch.
Orchestration muscle. Write tight specs. Define "done." Push agents - human or AI - toward that target. Half the battle is asking the right question in the right way.
Context switching with judgment. Know when to hand off to a specialist versus when to let AI brute-force it. Generalists make that call faster than anyone because they've seen enough of the landscape to know what good looks like across domains.
If you've been apologizing for being "good at a lot of things" - stop. AI flattened the depth curve. The people who understand systems end-to-end, connect disciplines, and can brief a model better than anyone are the ones accelerating entire teams right now.
Pick the next lane to sample. Write down what "great" looks like. Let AI carry the heavy lifting. This is the decade where generalists stop explaining themselves and start conducting.
