AI changed what we build. Then it changed who we hire.
Fabio Hauser argues that in an AI era the job goes to the person with judgment and speed across disciplines, not the person with the senior-most title in one. My background is ethical hacking. I built our whole private AI stack this year by pointing that same mindset at the problem. Hauser's argument explains why that worked.
What Hauser is saying
Fabio Hauser's essay is called "AI changed what we build. Then it changed who we hire." The argument goes like this. AI shrinks the gap between an idea and a working version of it. Once that gap shrinks, the bottleneck moves from "can you execute" to "can you judge which thing to build, and how fast can you move between fields." The person who wins is rarely the most senior specialist. It's the younger generalist who can cross disciplines, make real decisions in the moment, and ship the thing.
Hauser is writing about tech companies. I read it as a small-business operator and the argument landed harder than he probably meant it to, because the same shift has already happened in my own life. I just hadn't had a name for it.
How a pentesting mindset built the stack
My background is ethical hacking. The core of that work is simple. When you walk into a system you've never seen before, you don't freeze because you don't know it. You figure out what it's made of. You map the pieces. You find the path through. There are a million tools in the world, and even when you encounter one you don't know, you figure it out. That's the job.
I pointed that same mindset at AI infrastructure. What does this agent framework actually do. How does it talk to a model. Where does state live. What would have to exist for these pieces to cooperate. I've written plenty of tools and scripts over the years, so the reading-code part wasn't new. What was new was the thing I could do on top: I could ask the AI itself how to improve it. That's how PAI evolved. I'd look at a rough edge, ask the model what it would do differently, try the change, and see if the system got more useful. Recursive. A year of that, and the stack runs a real business, supports a growing client roster, and keeps improving on its own weekly cadence.
I'm not saying this is easy. I'm saying that the shape of what you have to do has changed. A year ago you needed deep expertise in one track to ship. Today you need enough judgment to know what to point the tools at, and enough speed to iterate until the thing works. Those two things reward the people Hauser is describing. They reward a methodical, cross-domain mindset over a single-track title.
Why this matters for small business owners
Here's where Hauser's argument stops being about big companies and starts being about the people I meet at chamber dinners.
Small business owners have been told for years that if they want serious technology, they have to hire a full-time CTO, pay an agency by the month, or sign a six-figure SaaS contract. Two of those three are out of reach for most small operators. The third drains cash for years before it earns its keep.
What changed is not the need for a technical partner. A small business still needs one. What changed is what that partner looks like. In 2026 the right shape isn't a full-time executive on your payroll. It's an outside technical partner who builds and maintains a private AI infrastructure layer that fits your actual business, and who keeps improving it without constant attention from you. That partner plays the same role a CTO used to play for the few weeks a year you actually need that role played. You rent the outcomes, not the seat.
That is the exact shape of what Obsidian AI Labs sells. I do the technical work. I own the infrastructure side so my clients don't have to. The methodical, cross-domain mindset Hauser is describing is what this role actually rewards. Judgment about what to wire together. Enough speed to ship and correct. Enough range to know when the answer isn't more tooling, it's a different decision. Hauser is describing the conditions that make this kind of partnership viable for a small business for the first time.
The counterargument I take seriously
There's an obvious objection. Experienced specialists exist for a reason. They see problems generalists miss. They know the failure mode you haven't yet encountered. I believe that. I have watched expertise save projects that a faster but shallower person would have broken.
Hauser's claim is not that expertise is worthless. It's that the ratio changed. The old shape was one senior specialist and a lot of people to move the output around. The new shape is a small number of generalists with good judgment, using AI as the execution layer, with specialists brought in for the narrow slices that actually require them. Fewer people, better-aimed.
That matches what I see. I still lean on specialists. I have lawyers, accountants, and a handful of deeper technical friends I call when I'm in over my head. I don't have an in-house team of them. I have judgment about when to pick up the phone.
What to do with this if you run a business
If you're a small business owner reading Hauser's essay, the practical takeaway is probably not "go hire a new kind of employee." You likely can't anyway. It's more like this.
You are probably capable of more than the previous era of tooling let you show. The ground moved under you, and it moved in a direction that favors your style of thinking, not the one that required you to hire somebody fluent in a specific stack. Before you pay anyone to build something for you, spend a weekend putting an AI assistant in front of the most annoying part of your week. See how far you get.
If that's as far as you want to take it, hire someone like me to close the rest of the distance. The point is that the distance is much shorter than the old rules suggested. Hauser was writing about software hiring. I think the same reshaping is coming for how small businesses decide who to bring on, what to insource, and which problems to just solve themselves before anybody shows up to help.
The job went to the generalist with judgment. That's good news for most of us.
← Back to all posts