Why the bench work stays with people
A damaged shoe or bag arrives with its own history, and no two are worn the same way. Before anything is cut, the worker has to judge how far a sole has pulled from the upper, whether old stitch holes will hold a new seam, and whether a heel is worth rebuilding or replacing. That judgment comes from handling the material, not from reading a file about it. People ask whether AI will replace leather workers, and the honest answer sits in how that work is actually spent.
Then comes the physical part. Resoling and reheeling, dyeing and polishing, sewing rips by hand or on a machine, cutting leather parts to a pattern, and fitting orthopedic shoes to a prescription all need steady hands on irregular material. Leather stretches, scars and takes finish differently from hide to hide. Our robotics read puts nearly all of this job in the physical column and at the dexterous humanoid tier, which is the hardest tier to build and the most expensive to run. The wider picture on that is in our guide to humanoid robots and physical jobs.
Scale matters too. Most repair work happens in small shops with one or two benches, mixed jobs and short runs. A machine built for one repeating cut is a poor fit for a day that moves from a cowboy boot to a handbag strap to a brace for a child’s shoe.
What AI does, what it helps with, and what it leaves alone
Start with the group where AI works alone. On this job’s task list, no task sits there yet; that share of task time reads 0%. Software can price a job, order materials or draft a quote, but those are shop admin rather than the tasks O*NET records for this occupation.
The assist group is thin for the same reason. No task on the list is currently marked as one AI meaningfully supports, and that share reads 0%. Pattern software and digital cutting tables exist, but they belong mostly to factory production lines and to fabric and apparel patternmakers, not to the repair bench.
That leaves the work with people. Our task split puts 100% of task time in the needs-a-human group, including hand and machine stitching of torn seams, attaching and shaping soles and heels, cleaning, dyeing and finishing leather, and custom or orthopedic fitting. On our Can AI do it? scale the job sits at 3 out of 100, where a higher number means more task time a model can handle today. How that figure is built is set out on the coverage method page.
What has actually been tested
Not much, and that is the honest position. Our evidence grade for this occupation prints as D, which is the grade we use when no study has put a machine against a qualified worker on these specific tasks. Because of that, we publish no parity number here. A grade is not a guess about quality; it is a record of what has been measured.
What would settle it is straightforward to describe. A timed trial of full resoles and seam repairs on worn, mismatched footwear, done by a robot system and by an experienced repairer, judged on fit, durability after wear and customer acceptance, would move the grade. Lab demonstrations on fresh, flat, identical material would not, because the hard part of this job is damaged and uneven stock. The scale and its grades are explained on the Is it better than a person? page, and the wider approach is set out in our methodology.
When this could change
Most likely after 2046 (8 in 10 of our scenarios). What the window measures, and how we build it, is described on the replacement year page.
Two things could pull that earlier. General-purpose robot hands getting cheaper and more reliable would attack the gripping, trimming and stitching steps that currently hold the work in place. More volume moving to automated cutting and assembly in leather goods factories would also cut the amount of hand work available, even if no repair bench is touched.
Two things hold it back. The capital cost of dexterous hardware sits far above the labor it would replace in a small shop, which is the comparison shown in the cost panel on this page. And the inputs are unpredictable: old glue, stretched uppers, mixed thread and hides that behave differently under a needle.
Head count is a separate question from capability. The Bureau of Labor Statistics counts about 7,450 workers in this occupation, with median pay near $37,800, and projects employment falling 6.5% between 2025 and 2035 (BLS, 2025). That is a slow-shrinking trade with fewer entry-level openings, not work being handed to a machine. Other occupations on that path are tracked in our list of jobs expected to shrink.
How to stay needed
Lean into the tasks the split leaves with people. Custom and orthopedic fitting to a prescription is the clearest one, because it mixes measurement, medical instruction and repeated adjustment. Full restoration of high-value boots, bags and saddles is another, since the value is in the repair being invisible. Hand stitching and finishing round it out: welted construction, edge work and color matching are still judged by eye and touch.
Two skills carry the most weight next to the bench. The first is diagnosis and quoting, meaning the ability to tell a customer what a repair will cost, how long it will last and when it is not worth doing. The second is running the shop end: mail-in repair, pricing, materials sourcing and a steady stream of trade work from retailers and orthotists.
What to do: add one repair type your area lacks, such as orthopedic adjustments or luxury bag restoration, and price it as a specialty rather than a general repair.
If you are weighing nearby work, the closest jobs are Shoe Machine Operators and Tenders, Sewers, Hand and Tailors, Dressmakers, and Custom Sewers. You can put any two of them side by side on our compare page, see the rest of the trade on the textile, apparel and furnishings family page, or look at the wider picture for other services.