Jobs / Production and manufacturing
Can agents do production and manufacturing work?
Assemble, machine, process, inspect, and pack goods in plants and workshops.
- US jobs
- 8,565,770 (5.5% of employment)
- Typical wage
- $49K a year (employment-weighted mean of occupation medians)
- Occupations
- 114 in O*NET 31.0
- Evidence
- 0 direct benchmarks, 2 also relevant, 0 context only
The answer today
UnmeasuredNo change since 24 Sep 2025.
Rule: No direct benchmark gives a success rate.
Next step: A direct benchmark that reports how many tasks agents complete.
How much of the work is measured
0% of the work in this family is measured by direct benchmarks. Little of the work is measured (less than 20%). On 24 Sep 2025: 0%.
Work = the O*NET 31.0 work activities of the family's occupations, weighted by how important each task is and by US employment. A benchmark measures an activity when its tasks make the agent do it and its grading checks it. Method.
| Work activity (O*NET) | Share of the work | Direct benchmark |
|---|---|---|
| Monitor equipment operation | 4.3% | Not measured |
| Maintain operational records | 4.2% | Not measured |
| Read documents or materials to inform work processes | 4.2% | Not measured |
| Measure physical characteristics of materials, products, or equipment | 4.0% | Not measured |
| Clean tools, equipment, facilities, or work areas | 3.9% | Not measured |
| Operate industrial processing or production equipment | 3.6% | Not measured |
| Evaluate production inputs or outputs | 3.4% | Not measured |
| Assemble equipment or components | 3.3% | Not measured |
| Position workpieces or materials on equipment | 3.2% | Not measured |
| Maintain tools or equipment | 3.1% | Not measured |
Our reading
No direct benchmark measures production or manufacturing work, so the family is unmeasured. Two real-robot leaderboards map here only in part. On RoboChallenge Table30, Robotera's Era0 succeeds in 64.33% of tabletop rollouts, including plugging in a network cable and pressing buttons in order, with one fine-tuned model per task. RoboArena ranks generalist arm policies by pairwise human preference and gives no completion rate. Both test manipulation primitives in a lab, not factory work, and neither sets a step. No benchmark measures monitoring equipment, operating production equipment, or assembling components. A direct benchmark with a completion rate would give the family its first step.
Evidence: RoboChallenge Table30, RoboArena by Dexmal, Hugging Face, Stanford University, and UC Berkeley. Written 24 Sep 2026.
Benchmarks for this work
Also relevant
These benchmarks cover part of this family's work, or have no result yet. They do not set its step.
- 1735 EloDreamZero (dreaming_zebra) · 24 Sep 2026Unrated
Functions: Robot manipulationHousehold chores
What we do not know
- Benchmarks measure tasks, not whole jobs. A job is a bundle of tasks, judgment, and relationships.
- Capability is not adoption. A task that agents can do in a benchmark may still be done by people at work.
- Most results come from single runs. Reliability over many runs is often unknown.
Occupations in this family
Sorted by US employment (BLS OEWS May 2025). O*NET sub-occupations share the employment figure of their SOC group. Work measured = the share of the occupation's O*NET work that direct benchmarks test.
How to cite
Credit the original work first: the benchmarks listed on this page, and their institutions.
Then, if you used this page:
Can Agents Work. "Can agents do production and manufacturing work?." https://canagentswork.com/jobs/production-manufacturing/ (accessed 2026-09-24). CC BY 4.0.@misc{caw-jobs-production-manufacturing,
title = {{Can agents do production and manufacturing work?}},
author = {{Can Agents Work}},
year = {2026},
howpublished = {\url{https://canagentswork.com/jobs/production-manufacturing/}},
note = {Accessed 2026-09-24. CC BY 4.0}
}