Christopher Stanton on What 'Reshaped, Not Replaced' Really Means for the Trades

Few academics have studied the collision between new technology and work as closely as Christopher Stanton. He's the Marvin Bower Associate Professor of Business Administration at Harvard Business School, where he teaches in the Entrepreneurial Management unit and, for the 2025–2026 year, served as co-course head of The Entrepreneurial Manager — a required course in the MBA curriculum. His research on labor markets, personnel economics, and entrepreneurship has run in the field's most demanding journals, among them the Quarterly Journal of Economics, the American Economic Review, and Management Science, and been picked up by The Economist, The New York Times, and Bloomberg. Outside Cambridge, he's a Faculty Research Fellow at the National Bureau of Economic Research and holds fellowships with CESifo in Germany and the Centre for Economic Policy Research in the UK. His HBS elective "Managing the Future of Work" spent years on exactly the question now landing on every contractor's desk: what artificial intelligence, robotics, and digital labor markets do to the people who do the work.

Earlier this year, Stanton brought that lens to the trades directly, joining the board of advisors at Bluon, an Irvine-based HVAC data company. We wanted his read on something every shop owner is tired of hearing answered badly: what does AI actually do to a service business? His answer is less dramatic than the headlines and more useful. He doesn't think a robot is turning a screw in your customer's attic anytime soon. He does think the tool in a first-year tech's hands is about to change how quickly that first-year becomes worth sending out on a truck alone.

---

ServiceMag: You've spent 15-plus years on labor markets — freelancing platforms, the gig economy, now generative AI. What pulled you toward this in the first place, and how did the skilled trades end up on your radar?

I first got interested in labor markets because of globalization — the idea that the internet, coupled with voice and video, could change where and how people worked. This was around the time Thomas Friedman and others were writing about a flattening world, with places converging. I watched my Stanford classmates find programming talent abroad at a fraction of Silicon Valley wages, and I thought we'd see a boom in distributed work. It turned out to be more of a trickle. I've spent a lot of the years since documenting the adjustment process — what firms actually do when they adopt a new technology or change how work gets done.

The trades came onto my radar for two reasons: unfilled jobs and durability in an AI-driven world. The unfilled-jobs problem has been persistent since the mid-2010s. On durability, I sat on the advisory board of a robotics company and saw firsthand how hard and how expensive it is to automate any job with a manual, physical component that requires dexterity. So you've got demand — jobs no one's filling — and a likely supply response, as people worried about AI look for work it can't easily touch. Both point to the trades mattering a lot in an economy that's adjusting.

ServiceMag: A thread through your work is that new technology rarely wipes out a whole profession — it reshapes it. For an HVAC tech or an appliance repair owner watching the AI headlines, what does "reshaped, not replaced" actually look like on a service call?

AI isn't going to turn a screw anytime soon — not in the unstructured environments where techs do repair work. What it will do is surface context. Think of it as a mentor or a trainer sitting next to you, augmenting the skill you already have. It'll take on a lot of the support and administrative load, too, and that can make a worker more productive — productive enough, in some cases, to go out and start their own shop. The job doesn't disappear. The parts of the job that used to slow you down get smaller.

ServiceMag: Contractors have sat through years of software that overpromised, so "AI" lands as a buzzword for a lot of them. What's a fair way for a shop owner to tell whether an AI tool actually helps versus one that's just marketing?

I'd run it through three steps. First, pressure-test it on your historical workflows — support calls, past repairs. Put it in a tech's hands and ask a simple question: does it give accurate, useful information? A lot of tools fail right there. Second, run a field test, where you can actually document behavior change and workflow improvements on real jobs. Third — and this is the one people skip — process change. If the system genuinely augments your techs in the field, then you start adjusting scheduling, staffing, and inventory around the gains you're seeing. A tool that can't survive the first step isn't worth the second. A tool that clears all three is one you build the business around.

The systems I trust most tend to pair predictive AI — trained on a large body of real service calls — with a deterministic lookup that carries brand- and model-specific context. The prediction gets you close; the deterministic layer keeps it from inventing a part number. That combination is what separates a tool a tech can rely on from one that sounds smart and gets the answer wrong.

ServiceMag: The trades are short on labor, and a lot of shops are putting green techs in the truck fast. Where does AI genuinely shorten the learning curve for a first-year tech, and where does it fall short of time in the field?

I won't pretend to know where it falls short in the field — that's your expertise, not mine. But a lot of what I hear about how Gen Z learns points to just-in-time skill acquisition. That's where AI fits: getting a newer tech ready for a specific task with the right, matched content at the moment they need it. Deliver the material to where people are, when the job calls for it. This is speculative, but I'd bet long, comprehensive training programs are less effective for a generation raised on YouTube and TikTok than short, exactly-timed content is.

ServiceMag: In a lot of AI productivity studies, the newer or lower-skilled workers gain the most. Do you expect that to hold in the trades, where the gap between a first-year tech and a 20-year veteran is enormous?

It's a fair proposition, and often true — but there are also cases where the veteran benefits most. It comes down to time allocation. If your 20-year tech is spending a disproportionate share of the day ordering parts or grinding through administrative drudgery, and AI can offload some of that, he moves to higher-value work. So it isn't simply "the rookie gains and the veteran doesn't." It depends on what the tool takes off each person's plate.

ServiceMag: If you were advising a small shop — five techs, tight budget — what's the one move you'd tell them to make this year to stay competitive as these tools mature?

Experiment. Try tools, see what actually works on your calls, and then adjust your workflows to match the productivity you observe. Don't buy the pitch; buy the result you can measure. And the budget math is usually easier than owners expect — if a tool lets you squeeze one extra job per tech per week, the unit economics tend to make the case on their own.

---

Christopher T. Stanton is the Marvin Bower Associate Professor of Business Administration at Harvard Business School, in the Entrepreneurial Management unit, where he researches personnel economics, labor markets, and entrepreneurship and built the MBA elective "Managing the Future of Work." His work has appeared in the Quarterly Journal of Economics, the American Economic Review, and the Proceedings of the National Academy of Sciences, and he is a Faculty Research Fellow at the National Bureau of Economic Research. Read his HBS faculty profile.

Disclosure: Stanton serves on the board of advisors at Bluon, an HVAC data and AI company.

Related reading: Chris Buttenham on Phantom Equity and Why Contractors Lose Their Best People | Kevin Arnow on Maintenance Contracts and the Refrigerant Transition in Burbank