■ MODERATE RISK ■ Construction & Trades
AI will not replace cartwrights, because the internal combustion engine already did — a century ago. The handful who remain serve heritage, museum, and film markets that exist precisely because the work is done by human hands.
“You build wheels for horse-drawn vehicles. AI isn't your biggest problem.”
Our AI replacement risk score — how we score jobs
A modern cartwright — and there are maybe a few dozen working professionally across the English-speaking world — builds and restores wooden wheels and horse-drawn vehicles: shaping elm hubs on a lathe, mortising spokes, bending felloes, and shrinking a heated iron tyre onto the assembled wheel so it clamps the whole structure tight as it cools. Clients are museums, breweries with ceremonial dray teams, film productions, carriage-driving enthusiasts, and Amish and Mennonite communities that still use the vehicles daily.
Here's the thing about automating a trade this size: nobody will. The total addressable market wouldn't cover the cost of developing a wheel-building robot, and CNC machining — which genuinely could cut spokes and felloes — misses the point, because heritage commissions and museum conservation standards typically require traditional methods and hand tools. The value of a restored 19th-century wagon lies in authentic technique; a laser-cut spoke is a defect, not an efficiency. Where technology helps, it's mundane: cartwrights use bandsaws and lathes like any woodworker, photograph and document restorations digitally, and find their global client base through a website.
The real threats are demographic, not algorithmic. The trade survives on a thin pipeline of apprenticeships and heritage-craft training programs, and organizations that track endangered crafts have listed wheelwrighting as critically endangered — meaning the risk is that skills die with the last generation of masters, not that a machine underbids them. Our score of 31 reflects a strange kind of safety: the job is too small, too manual, and too authenticity-bound for AI to bother with, but it was never a growth industry. You will not lose this job to a robot. You may struggle to find someone to leave it to.
Automatability: our editorial assessment of current and near-term AI capability
There is no AI disruption timeline here — the disruption happened around 1920 and the trade has already contracted to its irreducible heritage core. The genuine risk through 2040 is skills extinction: masters retiring without apprentices. Demand from museums, film, carriage sport, and plain communities is small but stable, and the authenticity premium that sustains it is, if anything, strengthened by an AI-saturated economy.
Barely, and that's the point. A small number of professional wheelwrights and cartwrights work worldwide, serving museums, film productions, carriage-driving sport, brewery dray teams, and Amish communities. Heritage-craft organizations classify the trade as critically endangered — not because of technology, but because too few apprentices are learning it before the current masters retire.
Technically yes — CNC machines could cut spokes and felloes today. Commercially, no one cares: the market is tiny, and its most valuable segment, museum conservation and heritage restoration, explicitly requires traditional hand methods. A machine-made wheel fails the authenticity test that justifies the price. This is one trade where 'handmade' isn't marketing, it's the specification.
Through heritage-craft apprenticeships, specialist courses run by craft guilds and heritage organizations, or by training under one of the few working masters — often after a foundation in general woodworking or carpentry. Expect to travel for training and to build a client network across museums, collectors, and driving enthusiasts. The barrier isn't competition; it's finding someone to teach you.
A narrow but real one. Historic vehicles need ongoing conservation, film and TV need authentic props, carriage sport continues, and plain communities use working vehicles daily. Because so few people can do the work, the ones who can stay booked. It will never scale — but it also can't be offshored, automated, or prompt-engineered.