■ HIGH RISK ■ Personal Services
The floors, yes; the job, not entirely. Autonomous scrubbers are eating the biggest, most repetitive chunk of janitorial hours, but the messy, varied, fix-it parts of the work stay stubbornly human.
“Cleaning robots handle floors. But unclogging the third-floor toilet? Still you.”
Our AI replacement risk score — how we score jobs
Janitorial work is less one job than a bundle of them: vacuuming and scrubbing floors, emptying trash, cleaning restrooms, restocking supplies, wiping touchpoints, responding to spills, and handling the small maintenance calls — a leaking faucet, a jammed towel dispenser, the aforementioned toilet. Commercial cleaning companies bill by the square foot, and floor care is the single largest block of hours in most contracts. That's the target painted on this occupation's back.
Autonomous floor scrubbers and vacuums from Tennant, Brain Corp-powered machines, and a dozen startups already roam supermarkets, airports, and warehouses at night, mapping the building and logging their coverage for the client. They don't call in sick and they produce a cleanliness report. Restroom sensor systems now tell dispatch which stalls need attention instead of a person checking hourly. For big-box and campus environments, that converts a crew of five into a crew of two plus machines — which is why our risk score sits at 66 rather than something comfortable.
What survives is everything irregular. Robots handle open floor beautifully and everything else badly: stairs, cluttered offices, biohazard cleanups, high dusting, window details, and the endless minor variability of a building full of humans doing gross, unpredictable things. There's also the trust factor — janitors work unsupervised in sensitive spaces, notice the propped fire door and the burned-out exit sign, and act as a building's low-key eyes and ears. The role that survives looks like a mobile problem-solver supervising a fleet of floor robots, not a person pushing a mop for eight hours.
Automatability: our editorial assessment of current and near-term AI capability
Floor-care automation is already deployed at scale in retail and logistics, and cleaning contractors are under constant cost pressure to extend it. Expect janitorial headcounts per building to fall noticeably through the late 2020s, with the remaining roles skewing toward supervision, detail work, and light maintenance by around 2030. Small offices and older, cluttered buildings will automate slowest — robots hate stairs and character.
Not completely, but they're taking the hours. Floor cleaning is the biggest slice of janitorial labor and it's the slice robots do best, so crews are shrinking even where the occupation survives. The tasks left over — restrooms, spills, clutter, repairs, anything involving stairs — remain human for the foreseeable future. Fewer janitors, doing the harder parts.
Safer than office data-entry, less safe than plumbing. Demand for clean buildings isn't going anywhere, but productivity per worker is rising as machines absorb floor care, which means fewer positions per building. Workers who can handle specialized cleaning, machine oversight, or light maintenance will find steady work; mop-only roles will keep thinning.
Anything a robot can't do or a client won't trust it with: biohazard and medical-grade cleaning certifications, floor restoration, minor plumbing and electrical fixes, and operating the cleaning robots themselves. Reliability and trustworthiness matter too — janitors work alone in sensitive spaces, and that trust is not a feature you can download.
Turnover and math. Janitorial turnover is brutal, night-shift hiring is hard, and a scrubber that runs six hours a night pays for itself quickly on large floors while generating coverage reports clients like. Robots don't fix the whole labor problem — someone still cleans the restrooms — but they fix the most expensive part of it.