Why the mail still moves through people
Mailroom work looks automatable on paper. Addresses can be read by software. Postage can be calculated in a second. Sorting by destination is a solved problem for machines that handle standard envelopes.
The day, though, is mostly physical and mostly local. Someone wheels the cart to the third floor and leaves the envelopes on the right desks. Someone signs for a courier delivery, logs it, and walks the box to the person whose name is on it. Someone clears a jam in the inserting machine and restarts the run. None of that is a text problem.
The other reason is scale. This job sits in offices, hospitals, universities and warehouses with a few thousand items a day, not a few million. That is the volume where a person with a cart is cheap and a sorting line with robots is not. The share of this job’s task time our scoring puts with people is 89%, and almost all of it is work that has to happen in a building, in person.
What software does, what it assists, and what stays manual
Some tasks are already machine work. Reading a printed address and routing an item by destination is one. Weighing a package and computing the correct postage rate is another. The share of task time that software or existing machines can handle on their own is 0%.
A second slice is assisted rather than taken over. Preparing a bulk mailing goes faster when the address list is cleaned and deduplicated by software first. Tracking numbers can be read from a label and dropped into a log, with the clerk checking the ones that scan badly. The assisted share is 11%.
The rest is hands and judgment. Sorting mail that arrives with a wrong room number, a nickname or no department at all means knowing who sits where and who moved last month. Opening, inspecting and rerouting items that do not fit the normal flow, dealing with a crushed parcel, and hand-delivering to people who are not at their desks all stay with the clerk. So does loading, feeding and unjamming the mail machine itself.
What the evidence actually shows
There is no published head-to-head test of AI against a mail clerk doing this job. That is why the evidence grade on this page is D, and why no parity number is given. A grade like that means not measured, not measured and found wanting. What would settle it is a trial in a real mailroom: the same week of incoming volume handled by a current setup and by an automated one, scored on items delivered correctly, misroutes, and time to resolve an unclear address. Nobody has published that.
The task-side figure does not rest on a single study either. It comes from the O*NET task list for this occupation scored against what current tools can do, which is explained on the coverage method page. Our overall score, 82 out of 100 (higher is safer), combines that with the physical and judgment blockers; the full method sits at how our scoring works.
Labor data tells the other half of the story. The Bureau of Labor Statistics counts about 55,230 of these jobs in the United States, with median pay near $39,280 a year, and projects employment falling about 7.2% between 2025 and 2035 (BLS, 2025). That decline is not a machine doing the whole job. It is less paper mail, fewer mailrooms, and openings that are not refilled. Jobs on that path show up on our list of jobs expected to shrink.
When the picture could change
Most likely after 2037 (8 in 10 of our scenarios). What that window measures is set out on the replacement-year method page.
Two things could pull it earlier. Mobile delivery robots are the robotics tier this job depends on, and about 79% of its work is physical, so cheaper indoor robots that ride elevators and navigate corridors matter more here than better language models. The cost gap shown above is the second: the tooling for the automatable tasks is priced well below a staffed role, so each new capability has an easy business case.
Two things hold it back. Buildings are messy and private: locked floors, hot-desking, shared loading docks and items that need a signature. And volume keeps dropping, which makes employers cut hours rather than buy equipment. A shrinking mailroom rarely gets a capital budget.
What to do: if your mailroom is part of a facilities or shipping team, get your name on the parts that involve people, vendors and exceptions, not just the machine runs.
How to stay needed
Lean into the tasks that keep failing without you. First, exception handling: unlabeled, misaddressed and damaged items, and knowing the building well enough to fix them. Second, receiving and chain of custody: signing for couriers, logging accountable packages, and tracking who collected what. Third, running and maintaining the equipment, including the postage meter, scales, inserters and jams, since whoever keeps the line moving is the last person cut.
Two skills travel well. One is inventory and shipping systems, the software side of logging, labeling and reconciling deliveries. The other is plain service work: handling a frustrated person whose contract did not arrive, and writing a clear note about what happened.
Close jobs worth comparing are office machine operators, except computer, which shares the equipment side, and postal service mail sorters and processing machine operators, where the volumes and the automation already in place are very different. Couriers and messengers is the move for people who prefer the delivery half of the day. You can put any two of them side by side on the compare page.
For the wider picture, this role sits in the other office and administrative support workers family and the administrative support sector, both of which show how neighboring clerical work scores. If your own mailroom job mixes in reception, inventory or facilities duties, the job scoring quiz will reflect that mix better than any single occupation page.