EXPOSED
The hands-on core — clamping and gluing up carcasses, fitting doors and drawers to a fraction of a millimeter, hand-sanding and spraying finish, scribing to an out-of-plumb wall — is not something current robotics does at cost. The real pressure is not language AI but CAD-to-CNC workflows: nesting software, automated panel saws, edgebanders and boring machines now absorb the cutlist, layout, and drilling work that used to be skilled bench time, and AI design tools shrink the drafting and quoting hours further. No license protects the trade, and imported RTA casework keeps compressing the low-custom end.
Most of this decline happened after 2021 — it is not the pandemic dip.
Median pay $35,790 → $46,680 +4.3% in real terms
This line is counted by the Bureau of Labor Statistics — the one figure on this page that isn't a judgement of ours. Headcount moves on demand, offshoring, demographics and the business cycle, and automation is one term among several, often not the loudest.
So a falling line is not evidence that AI did it, and a rising one is not evidence that it won't. Both happen in this register: some occupations resist automation and shrink anyway, others are highly automatable and keep growing. The marked year is 2020.
BLS projection, 2024–2034
-1.6% 86,000 → 84,700 on the projections basis
Shrinking, but not obviously because of AI
The BLS projects -1.6% by 2034, but at 51/100 this work is only moderately exposed — not the profile of a job current AI can simply do. Occupations shrink for many reasons, and the score does not point at automation as this one's cause.
Different clocks. The score is what current AI could do to this work today. The projection is how many of these jobs will exist in 2034. Everything between the two — how fast employers actually adopt, whether demand grows in the meantime — is why they can point opposite ways without either being wrong.
~8,100 openings a year on average, including replacing people who leave.
FramerInlayerBow MakerCarpenterHarp MakerWoodworkerChair MakerHand CarverCabinetmakerChair MenderGuitar MakerViolin MakerCabinet MakerFrame BuilderStock CheckerBoat Oar MakerCabinet WorkerChair MechanicWood MachinistWood MechanistWood Ski MakerAccordion MakerCabinet BuilderFrame Carpenter
Holding it up: embodiment . Weakest point: liability shield .
Tasks largely resist digitisation CNC has swallowed the flat-panel work — nesting, dados, 32mm system holes — but reading grain direction before a glue-up, planing a twisted door flat, scribing a face frame to a wall that's 6mm out over 2.4m, and rubbing out a finish still take a body at a bench, which is why this sits at 14 and not 18: a big slice of your day (cutlists, layout, boring) already left for the machine.
Hands-on in uncontrolled environments You work in a shop full of dust, solvent vapor, and unpredictable stock — a board that pinches the blade, a sheet of 19mm melamine that needs two people to lift onto the slider — and half the job finishes on site in someone's kitchen, kneeling on a subfloor, shimming a run of uppers; only the drafting and CNC programming keep it off 20.
No licence, no signature requirement There is no state license to make cabinets: no exam, no bonded status, no continuing education, and if the installing contractor holds a license it is theirs, not yours — a 2 rather than 0 only because shop-level OSHA obligations on guarding and finish-booth ventilation attach to somebody.
Meaningful discretion You decide how to handle a wide slab that will move seasonally, whether to notch or add a filler when the site measure disagrees with the drawing, and when material is bad enough to reject — real calls with real money in scrapped stock, but they happen inside a shop drawing someone else approved, so the ambiguity stops well short of the 14+ band.
The verdict above describes this occupation as a whole. Almost nobody does the typical version of a job — tick what's actually in your week and see how your own mix sits.
Your task mix speaks to task resistance (14/20 here) — how much of the day's work current AI already does. That is the dimension the boxes above are about.
It cannot move the other three. Liability shield (2/20) is whether the law requires a licensed human to sign. Trust premium (9/20) is whether buyers specifically pay for a person. Judgment and accountability (9/20) is whether the role exists to own consequential calls. Those are facts about the occupation's standing, not about which tasks are in your week — a paralegal who does only trial exhibits still holds no licence. Together they are 20 of this occupation's 51 points (39%).
Embodiment (17/20) is also a property of the work rather than the worker, but we don't tag individual tasks as physical or not, so the picker can't tell you anything about it. That's a limit of this tool, not a claim.
Did we get the list right? Tell us what's missing — the tasks are written from the outside, and you're reading this from the inside.
Sheet Metal Workers SAFE
The moves above are yours to make. This is the other half: what would have to change in the world for the occupation itself to score higher. None of it is in any one person's gift, but it is where the floor actually comes from. Scores here are not a one-way ratchet. Only two of the five dimensions — task resistance and embodiment — track what machines can do. The other three track law, what buyers will pay for, and who is answerable, and those move in both directions, often in response to the same pressure AI creates. If every lever below landed, this occupation would score around 62/100, still EXPOSED.
As nesting/CNC absorbs cutlists, boring and edgebanding, the remaining paid hours concentrate in the tier machines don't touch: site measurement of out-of-square historic openings, scribing and fitting, veneer matching, hand-applied and repair finishing, and hardware tuning. Watch for shops re-billing as 'installation and finishing' rather than 'fabrication' — the routine tier leaving raises the share of the day that is genuinely resistant.
A verifiable provenance mark for hand-built casework — analogous to the Architectural Woodwork Institute's Quality Certification Program but extended to a consumer-facing 'bench-built / not RTA' label, or state-level furniture origin labeling similar to existing 'Made in Vermont' and Amish-cooperative marks. If specifiers write AWI Premium Grade plus a shop-fabrication requirement into public and institutional bid documents, the premium becomes contractual rather than sentimental.
Rises where the cabinetmaker becomes the party who signs off on shop drawings and field dimensions as the responsible fabricator — some commercial millwork subcontracts already require the shop to stamp approved shop drawings and own dimensional conflicts. Broader use of that clause makes the bench lead the owner of consequential ambiguity calls, not just an executor of someone else's file.
Little headroom — already 17. Would only rise if the work mix shifts further toward on-site retrofit in occupied buildings (scribing to settled framing, stair and built-in installation in historic housing), which no cell-based robot addresses. Growth in historic-tax-credit rehab work (National Park Service Part 2 approvals requiring in-kind millwork replication) pushes this way.
The limit. The binding constraint is capital, not capability: even with every institutional lever pulled, the CNC-equipped shop with two skilled finishers underprices the eight-bench shop. Score can improve while headcount falls.
| Los Angeles-Long Beach-Anaheim, CA | 2,630 | $47,960 +3% |
| Dallas-Fort Worth-Arlington, TX | 2,370 | $38,760 -17% |
| New York-Newark-Jersey City, NY-NJ | 2,020 | $60,240 +29% |
| Atlanta-Sandy Springs-Roswell, GA | 1,950 | $45,120 -3% |
| Elkhart-Goshen, IN | 1,830 | $46,240 -1% |
| Chicago-Naperville-Elgin, IL-IN | 1,420 | $56,410 +21% |
| Minneapolis-St. Paul-Bloomington, MN-WI | 1,370 | $49,650 +6% |
| Phoenix-Mesa-Chandler, AZ | 1,200 | $45,690 -2% |
| Jonesboro, AR | 40 | $66,860 +43% |
| Norwich-New London-Willimantic, CT | 50 | $66,040 +41% |
| Burlington-South Burlington, VT | 70 | $64,130 +37% |
We have no reported case of a named organisation automating this occupation. Not one deployment, not one announcement.
That is worth saying out loud next to a score of 51. The verdict above is about what the work exposes — what current AI could do to these tasks. It is not a claim that anyone has done it. For this occupation those two things have come apart completely: the capability argument is on this page, and the evidence column is empty.
Has AI actually changed your work? One tap, anonymous, and the running tally is public. Nothing else is asked of you.
Rather than check back: get the digest and we'll tell you what changed — or watch a single occupation from its own page.