SAFE
The core of this job is walking into a boiler room, crawlspace, or occupied apartment and diagnosing why something broke, then fixing it with hand tools in a space no robot can navigate — replacing motors and belts, patching drywall, resetting HVAC controls, chasing leaks, rekeying doors. AI can absorb the paperwork layer around it: work-order triage, parts lookup, PM scheduling, manual search, and first-pass fault suggestions from sensor data. The exposure here is not displacement but compression — fewer techs covering more square footage because scheduling and diagnostics get smarter, and pay stays flat because the trade has no license gate.
Dipped in 2020, then grew past where it started.
Median pay $39,080 → $49,590 +1.5% 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
+3.8% 1,629,700 → 1,692,100 on the projections basis
Hard to automate, and growing
The work resists current AI and the BLS projects +3.8% more of these jobs by 2034. Note that safe does not mean well paid — several of the fastest-growing resistant occupations are among the lowest paid on the register.
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.
~159,800 openings a year on average, including replacing people who leave.
HandymanHandy ManMaintenance ManPool TechnicianTrouble ShooterBuilding EngineerBuilding MechanicBuilding RepairerMechanical ExpertRepair TechnicianMaintenance PersonMaintenance WorkerMechanics HandymanPlumbing InstallerMechanical AdjusterMechanical HandymanElectrical InstallerMaintenance EngineerMaintenance MechanicMechanical InstallerFacilities TechnicianMaintenance AssociateMaintenance RepairmanMaintenance Journeyman
Holding it up: embodiment . Weakest point: liability shield .
Tasks largely resist digitisation Swapping a seized condenser fan motor, snaking a floor drain, or shimming a sagging door frame each require an unrepeatable sequence improvised on the spot from what the failure actually looks like — the 17 rather than 20 reflects that the work-order intake, PM checklist generation, and parts cross-referencing genuinely do move to software.
Hands-on in uncontrolled environments You work above ceiling tile, inside live electrical panels, in 100-degree mechanical rooms and unheated crawlspaces, on ladders, in tenants' occupied units — none of it a fixed workcell, all of it requiring force, reach, and touch simultaneously, which is why this sits at the top of the band rather than mid.
Certification preferred, not legally required There is no general maintenance license in any state; you may hold an EPA 608 for refrigerant handling or a boiler operator card, and permit-triggering electrical or plumbing work gets handed to a licensed trade — so the certifications you carry are real but narrow slices, not a gate around the job itself.
Meaningful discretion You decide on the spot whether a 22-year-old rooftop unit gets another compressor or a capital-replacement recommendation, whether a smell means shut the building's gas off, and whether a repair is safe to leave overnight — genuine calls with money and safety attached, though a supervisor or engineer signs off on anything above a spend threshold.
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 (17/20 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 (6/20) is whether the law requires a licensed human to sign. Trust premium (12/20) is whether buyers specifically pay for a person. Judgment and accountability (13/20) 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 31 of this occupation's 67 points (46%).
Embodiment (19/20) 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.
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 83/100, still SAFE.
Insurer-driven documentation requirements: property carriers (and habitability litigation in cities with strong tenant law, e.g. NYC HPD violations, Chicago, LA) conditioning coverage on a named, credentialed individual signing off on life-safety inspections — fire doors, egress, backflow, elevator machine rooms, Legionella water management plans under ASHRAE 188. Water management plan compliance is already pushing named-responsible-person structures in healthcare facilities via CMS memo QSO-17-30.
Genuine two-tier structure exists here: the routine tier is PM checklists, filter changes, lamp relamping, parts lookup, and paperwork. If AI and remote monitoring strip that tier, what remains is undocumented legacy equipment, intermittent faults, and retrofit work in buildings whose as-builts are wrong — the part where the sensor data lies and someone has to open the wall. Watch for PM ratio shifting from majority-scheduled to majority-corrective in facilities benchmarking data.
Expansion of the existing trade-license carve-outs that already bind this work at the edges: states adding a mandatory limited-scope license or registration for building maintenance techs doing refrigerant handling (EPA 608 already required), boiler tending (already licensed in NY, MD, MA for pressure vessels above thresholds), and electrical work above the 'like-for-like replacement' exemption. If a state board eliminated the maintenance-worker exemption in the electrical or mechanical code — the way some states have narrowed handyman exemptions by lowering the dollar threshold — a licensed, personally-liable signature would be required for a large share of daily tasks rather than a minority.
Task-mix shift plus the accountability that follows it: when CMMS/AI triage absorbs work-order routing and first-pass fault suggestion, the remaining human call is the one the model flagged ambiguously — shut the chiller down mid-summer or run it to failure, red-tag the equipment, evacuate the wing. If facilities orgs formalize this as a named shutdown/red-tag authority (as hospital and data-center operations increasingly do), the role owns consequential calls explicitly rather than implicitly.
Limited and specific: tenant-facing and occupied-space work, where residents and hospital staff want a known named tech in the unit. Union contracts (SEIU 32BJ in NYC commercial buildings, IUOE Local 94) that specify in-house staffing minimums rather than vendor dispatch preserve this. Broad consumer willingness to pay extra for a human here is weak, because the buyer is usually a property manager optimizing cost per square foot.
The limit. The binding constraint is the buyer, not the technology. Property managers and REITs purchase this labor as a cost line per square foot, so compression — fewer techs covering more portfolio — proceeds regardless of how physically irreducible the work is. Embodiment at 19 is near ceiling and cannot help further. Only a real license gate changes bargaining power, and the trade has resisted one for a century precisely because owners lobby against it; partial refrigerant, boiler, and backflow credentials are the realistic path, not a general maintenance license.
| New York-Newark-Jersey City, NY-NJ | 108,700 | $59,550 +20% |
| Los Angeles-Long Beach-Anaheim, CA | 47,890 | $56,520 +14% |
| Chicago-Naperville-Elgin, IL-IN | 45,710 | $58,020 +17% |
| Dallas-Fort Worth-Arlington, TX | 39,740 | $48,860 -1% |
| Houston-Pasadena-The Woodlands, TX | 32,050 | $46,710 -6% |
| Miami-Fort Lauderdale-West Palm Beach, FL | 31,890 | $46,290 -7% |
| Phoenix-Mesa-Chandler, AZ | 26,350 | $48,930 -1% |
| Atlanta-Sandy Springs-Roswell, GA | 25,190 | $49,630 +0% |
| Kahului-Wailuku, HI | 1,350 | $63,980 +29% |
| San Francisco-Oakland-Fremont, CA | 18,550 | $63,630 +28% |
| San Jose-Sunnyvale-Santa Clara, CA | 7,750 | $63,220 +27% |
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 67. 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.