Census · version v3.2-2026-09-24

The automatability census

Which knowledge work passes a structural hand-over screen, task by task, role by role and industry by industry: work whose shape would let it be handed to software. It is not a claim about what AI can do, and not a forecast. Every verdict was judged by AI models, not by people, and a task can pass the screen and still be beyond the systems of the day.

A task passes when at least two of three AI models each find three things: whether it was done right can be told within hours, a check that already exists settles it, and a single failure is cheap. Work that needs a body is left out, and so are sign-offs a person is accountable for: approving, authorising, certifying or assigning others' work. A task without three valid scores is not called and never passes. The rule was written down and frozen before any score in this version existed, but it was shaped after seeing earlier usage data, so tests against that data are post hoc.

The firmest figure is the part all three models pass. It sits beside every total on these pages. Each model's own total is shown too, because they disagree a great deal.

A small part of the knowledge payroll passes the screenChart

5.6%

passes the screen

$389B · all three $158B

11.8%

waits only on a check

$827B

20.0%

needs a body

$1.4T

62.7%

held for more than a missing check

$4.4T

to its left, all three models pass

The strip is the whole $7.0T of knowledge payroll scored. From the census (v3.2-2026-09-24), a screen scored by three AI models, not a claim about what AI can do and not a forecast. All three models pass $158B, which is 2.3% of the whole. The last part is everything else: mostly work with no existing check that is also slow to judge or costly when wrong; work whose most-given reason is a missing check but on which the models split over whether anything else holds it; and a small part that is a sign-off a person answers for.

The numbers

Part of the knowledge payroll	Payroll	Share
passes the screen	$389B	5.6%
waits only on a check	$827B	11.8%
needs a body	$1.4T	20.0%
held for more than a missing check	$4.4T	62.7%
Of what passes, all three models pass	$158B	2.3%

For most work that fails the screen, the reason given most often is that no existing check settles itChart

First, left out whatever the vote would say

Needs a body

$1.4T · 20.0%

A sign-off a person answers for

$133B · 1.9%

Then the vote on the other $5.5T: the reason given most often for the payroll that fails, as a share of all knowledge payroll

Can it be told within hours whether the work was done right?

$3.1T · 44.6%

Does a check that already exists settle it?

$4.8T · 69.3%

Is a single failure cheap?

$1.4T · 19.5%

What is left passes

Yes to all three questions, by two of the three models

$389B · 5.6% · all three $158B

left out before the votethe reason the models that voted no gave most oftenpasses the screento its left, all three models pass

Every bar is a share of the $7.0T of knowledge payroll scored. From the census (v3.2-2026-09-24), a screen scored by three AI models, not a claim about what AI can do and not a forecast. Each task that fails carries the reason the models that voted no gave most often, set by the census's own code; where reasons tie, the task counts under each, so the middle bars overlap and do not add up. A question a task also fails, but less often, is not counted, so each bar is a floor for that question.

The numbers

Step	Payroll	Share of knowledge payroll
Left out: Needs a body	$1.4T	20.0%
Left out: A sign-off a person answers for	$133B	1.9%
Goes to the vote	$5.5T	78.1%
Most often given: Can it be told within hours whether the work was done right?	$3.1T	44.6%
Most often given: Does a check that already exists settle it?	$4.8T	69.3%
Most often given: Is a single failure cheap?	$1.4T	19.5%
Passes the screen	$389B	5.6%
All three models pass	$158B	2.3%

The payroll that passes depends on which model judges itChart

All three models pass

$158B

Passes the screen: two of the three models

$389B

Claude Sonnet alone

$696B

Gemini 3.1 Pro alone

$400B

GPT-6 Sol alone

$244B

the rule: two of the three models passall three models passone model alone

Of $7.0T of knowledge payroll scored. From the census (v3.2-2026-09-24), a screen scored by three AI models, not a claim about what AI can do and not a forecast. The census's rule counts a task when at least two of the three models pass it. Each model alone would pass a different amount, which is why the part all three pass is shown beside every total.

The numbers

Who judges	Payroll that passes
All three models pass	$158B
Passes the screen: two of the three models	$389B
Claude Sonnet alone	$696B
Gemini 3.1 Pro alone	$400B
GPT-6 Sol alone	$244B

The rule

How strict should the screen be?

Every threshold in the rule is a judgement. Each step below loosens one condition, and the payroll that passes grows as it does. The census uses the step marked in the table. The condition that moves the figure most is who may check: the steps that also count a qualified person's judgement take in work a person still reviews.

Each looser step passes more payroll, and the biggest jump comes from letting a person's judgement count as the checkChart

can be checked instantly, by an existing check, and a failure costs up to about $1k

$115B · all three $4.6B

can be checked instantly, by an existing check, and a failure costs up to about $10k

$172B · all three $78.0B

can be checked within hours, by an existing check, and a failure costs up to about $10kthe rule used: an existing check settles it, with no person's judgement

$389B · all three $158B

can be checked within hours — by an existing check or an expert — and a failure costs up to about $10kalso counts work that only a qualified person's judgement settles

$1.3T · all three $697B

can be checked within days — by an existing check or an expert — and a failure costs up to about $10k

$1.9T · all three $1.1T

can be checked within days — by an existing check or an expert — and a failure costs up to about $100k

$2.4T · all three $1.6T

passes at that step of the ruleto its left, all three models pass

Steps run from the strictest rule to the loosest; the step in bold is the rule the census uses. From the census (v3.2-2026-09-24), a screen scored by three AI models, not a claim about what AI can do and not a forecast. At every step the models vote again, and the part all three pass is marked inside the bar.

The numbers

Rule	Passes the screen	All three pass
can be checked instantly, by an existing check, and a failure costs up to about $1k	$115B	$4.6B
can be checked instantly, by an existing check, and a failure costs up to about $10k	$172B	$78.0B
can be checked within hours, by an existing check, and a failure costs up to about $10kthe rule used: an existing check settles it, with no person's judgement	$389B	$158B
can be checked within hours — by an existing check or an expert — and a failure costs up to about $10kalso counts work that only a qualified person's judgement settles	$1.3T	$697B
can be checked within days — by an existing check or an expert — and a failure costs up to about $10k	$1.9T	$1.1T
can be checked within days — by an existing check or an expert — and a failure costs up to about $100k	$2.4T	$1.6T

Part I · roles

What happens to a job

Each role's share moves with the rule: the stricter and looser rules re-vote every task, and each model alone gives its own share. Differences between similar roles are within the models' noise. Open a role to see every task, whether it passes, which models pass it, and why.

What the screen says of each kind of job, best paid firstChart

Management1.0%

Legal0.3%

Computing and mathematics6.6%

Healthcare practitioners1.1%

Architecture and engineering2.0%

Business and finance6.7%

Science1.6%

Arts, design and media3.5%

Education2.8%

Community and social service3.5%

Sales11.1%

Office and administrative support16.6%

passes the screenwaits only on a checkneeds a bodyheld for more than a missing checkto its left, all three models pass

Shares of each group's payroll; the figure beside a name is the share that passes. From the census (v3.2-2026-09-24), a screen scored by three AI models, not a claim about what AI can do and not a forecast. Groups are the occupation code's major groups, ordered by average pay; the census has no field for seniority, so this reads kinds of job, not rungs within one. A task passes when at least two of the three models pass it, and the tick marks the part all three pass.

The numbers

Kind of job	Passes the screen	All three models pass	Waits only on a check	Needs a body	Held for more than a missing check
Management	1.0%	0.6%	3.0%	8.3%	87.6%
Legal	0.3%	0.1%	6.1%	14.0%	79.6%
Computing and mathematics	6.6%	0.9%	15.1%	10.9%	67.4%
Healthcare practitioners	1.1%	0.2%	2.1%	39.9%	56.9%
Architecture and engineering	2.0%	0.5%	9.9%	21.1%	67.0%
Business and finance	6.7%	1.8%	11.6%	6.3%	75.3%
Science	1.6%	0.4%	9.1%	31.6%	57.7%
Arts, design and media	3.5%	0.8%	28.7%	24.1%	43.7%
Education	2.8%	1.9%	20.6%	26.9%	49.7%
Community and social service	3.5%	0.3%	9.2%	8.1%	79.3%
Sales	11.1%	5.2%	15.6%	30.2%	43.1%
Office and administrative support	16.6%	8.5%	27.7%	27.0%	28.7%

Finance

Role	Share that passes (stricter and looser rules)	Passes the screen	All three pass	Observed AI use
Accountants and Auditors	9.1% (0.0% strict, 49.4% loose)	$12.3B	$0	34.8%
Appraisers and Assessors of Real Estate	20.1% (0.0% strict, 64.5% loose)	$424M	$361M	—
Appraisers of Personal and Business Property	8.0% (8.0% strict, 64.2% loose)	$169M	$169M	—
Bill and Account Collectors	12.9% (0.0% strict, 65.2% loose)	$1.0B	$699M	29.9%
Billing and Posting Clerks	46.0% (0.0% strict, 73.2% loose)	$9.5B	$5.5B	19.2%
Bookkeeping, Accounting, and Auditing Clerks	43.9% (6.2% strict, 70.1% loose)	$33.3B	$8.7B	31.0%
Budget Analysts	0.0% (0.0% strict, 37.9% loose)	$0	$0	6.7%
Cost Estimators	7.9% (0.0% strict, 14.9% loose)	$1.4B	$0	0.0%
Credit Analysts	18.6% (16.1% strict, 66.2% loose)	$1.2B	$0	16.9%
Credit Counselors	18.5% (16.8% strict, 69.6% loose)	$272M	$72.5M	23.4%
Financial Examiners	0.0% (0.0% strict, 30.0% loose)	$0	$0	4.3%
Financial Managers	1.4% (0.0% strict, 4.9% loose)	$2.0B	$2.0B	39.1%
Financial Risk Specialists	0.0% (0.0% strict, 33.3% loose)	$0	$0	26.5%
Financial Specialists, All Other	4.8% (0.0% strict, 45.4% loose)	$560M	$0	22.0%
Financial and Investment Analysts	0.0% (0.0% strict, 26.9% loose)	$0	$0	57.2%
Gambling Cage Workers	10.1% (4.1% strict, 23.9% loose)	$51.8M	$0	6.5%
Loan Interviewers and Clerks	11.1% (0.0% strict, 72.6% loose)	$951M	$0	20.2%
Loan Officers	15.8% (3.0% strict, 62.2% loose)	$3.9B	$736M	18.6%
Medical Records Specialists	11.8% (5.9% strict, 52.9% loose)	$1.2B	$613M	66.7%
New Accounts Clerks	2.3% (0.0% strict, 50.0% loose)	$42.0M	$0	0.0%
Payroll and Timekeeping Clerks	39.4% (0.0% strict, 76.4% loose)	$3.5B	$1.8B	4.9%
Personal Financial Advisors	0.0% (0.0% strict, 59.0% loose)	$0	$0	35.0%
Procurement Clerks	38.7% (4.0% strict, 79.5% loose)	$1.1B	$553M	24.4%
Tax Examiners and Collectors, and Revenue Agents	17.0% (2.5% strict, 72.4% loose)	$615M	$0	2.8%
Tax Preparers	11.4% (0.0% strict, 35.3% loose)	$489M	$0	12.8%
Tellers	21.9% (13.3% strict, 38.3% loose)	$3.0B	$1.1B	2.3%

Sales

Role	Share that passes (stricter and looser rules)	Passes the screen	All three pass	Observed AI use
Advertising Sales Agents	7.1% (0.0% strict, 52.5% loose)	$501M	$501M	14.8%
Agents and Business Managers of Artists, Performers, and Athletes	6.1% (0.0% strict, 27.5% loose)	$138M	$0	11.8%
Demonstrators and Product Promoters	5.9% (0.0% strict, 31.9% loose)	$131M	$0	7.9%
Door-to-Door Sales Workers, News and Street Vendors, and Related Workers	8.7% (8.7% strict, 45.5% loose)	$4.0M	$4.0M	17.6%
First-Line Supervisors of Non-Retail Sales Workers	10.6% (0.0% strict, 37.2% loose)	$2.3B	$1.5B	22.9%
Insurance Sales Agents	16.3% (6.6% strict, 59.5% loose)	$6.1B	$0	31.9%
Models	27.2% (13.0% strict, 27.2% loose)	$19.7M	$0	0.0%
Real Estate Brokers	8.2% (0.0% strict, 17.0% loose)	$354M	$0	26.1%
Real Estate Sales Agents	5.9% (0.0% strict, 47.3% loose)	$776M	$0	28.3%
Sales Engineers	5.9% (0.0% strict, 40.1% loose)	$426M	$0	32.2%
Sales Managers	0.0% (0.0% strict, 23.1% loose)	$0	$0	4.3%
Sales Representatives of Services, Except Advertising, Insurance, Financial Services, and Travel	20.0% (6.7% strict, 73.3% loose)	$19.3B	$12.9B	—
Sales Representatives, Wholesale and Manufacturing, Except Technical and Scientific Products	6.5% (3.4% strict, 32.9% loose)	$6.7B	$0	62.8%
Sales Representatives, Wholesale and Manufacturing, Technical and Scientific Products	15.1% (5.9% strict, 61.0% loose)	$5.0B	$2.5B	27.1%
Securities, Commodities, and Financial Services Sales Agents	5.0% (0.0% strict, 51.3% loose)	$2.6B	$1.1B	44.1%
Telemarketers	29.8% (27.6% strict, 100.0% loose)	$660M	$660M	28.5%
Travel Agents	26.5% (13.9% strict, 52.8% loose)	$716M	$716M	40.5%

Customer support

Role	Share that passes (stricter and looser rules)	Passes the screen	All three pass	Observed AI use
Customer Service Representatives	27.3% (16.8% strict, 89.2% loose)	$33.8B	$20.8B	70.1%
Interviewers, Except Eligibility and Loan	6.9% (0.0% strict, 90.8% loose)	$474M	$0	38.5%
Receptionists and Information Clerks	10.0% (0.0% strict, 63.4% loose)	$3.7B	$0	43.4%
Reservation and Transportation Ticket Agents and Travel Clerks	15.4% (6.6% strict, 50.6% loose)	$926M	$926M	24.7%

IT & software

Role	Share that passes (stricter and looser rules)	Passes the screen	All three pass	Observed AI use
Computer Network Architects	3.6% (0.0% strict, 43.2% loose)	$866M	$0	19.9%
Computer Network Support Specialists	3.9% (0.0% strict, 63.6% loose)	$448M	$0	28.7%
Computer Occupations, All Other	5.0% (1.3% strict, 47.4% loose)	$2.5B	$1.0B	31.1%
Computer Programmers	13.8% (0.0% strict, 92.6% loose)	$1.5B	$0	74.5%
Computer Systems Analysts	2.8% (0.0% strict, 36.7% loose)	$1.6B	$0	27.6%
Computer User Support Specialists	5.0% (5.0% strict, 61.3% loose)	$2.3B	$2.3B	46.9%
Computer and Information Research Scientists	0.0% (0.0% strict, 7.4% loose)	$0	$0	34.0%
Computer and Information Systems Managers	0.0% (0.0% strict, 24.3% loose)	$0	$0	15.6%
Database Administrators	23.4% (0.0% strict, 77.8% loose)	$1.8B	$0	33.1%
Database Architects	17.1% (0.0% strict, 63.5% loose)	$1.5B	$0	57.9%
Information Security Analysts	0.0% (0.0% strict, 50.1% loose)	$0	$0	48.6%
Network and Computer Systems Administrators	3.7% (3.7% strict, 23.8% loose)	$1.2B	$0	33.7%
Software Developers	7.5% (0.0% strict, 40.2% loose)	$18.0B	$0	28.8%
Software Quality Assurance Analysts and Testers	13.9% (0.0% strict, 82.1% loose)	$3.0B	$620M	51.9%
Web Developers	3.2% (0.0% strict, 69.8% loose)	$243M	$0	48.0%
Web and Digital Interface Designers	8.3% (1.7% strict, 58.4% loose)	$1.0B	$205M	24.9%

Admin

Role	Share that passes (stricter and looser rules)	Passes the screen	All three pass	Observed AI use
Brokerage Clerks	3.4% (0.0% strict, 46.9% loose)	$87.8M	$0	17.4%
Correspondence Clerks	12.8% (4.4% strict, 53.5% loose)	$17.0M	$0	27.9%
Court, Municipal, and License Clerks	4.9% (4.9% strict, 77.5% loose)	$423M	$423M	10.2%
Credit Authorizers, Checkers, and Clerks	16.4% (10.2% strict, 70.1% loose)	$89.3M	$55.3M	13.0%
Data Entry Keyers	42.5% (14.4% strict, 56.0% loose)	$2.3B	$790M	67.1%
Desktop Publishers	18.1% (18.1% strict, 93.8% loose)	$31.7M	$21.9M	46.4%
Eligibility Interviewers, Government Programs	12.1% (3.9% strict, 70.1% loose)	$1.0B	$324M	6.0%
Executive Secretaries and Executive Administrative Assistants	0.0% (0.0% strict, 69.0% loose)	$0	$0	23.4%
File Clerks	12.7% (6.0% strict, 32.5% loose)	$410M	$410M	15.7%
First-Line Supervisors of Office and Administrative Support Workers	9.8% (5.6% strict, 51.3% loose)	$10.4B	$10.4B	18.6%
Hotel, Motel, and Resort Desk Clerks	34.0% (6.1% strict, 68.2% loose)	$3.1B	$1.6B	17.3%
Legal Secretaries and Administrative Assistants	0.0% (0.0% strict, 60.7% loose)	$0	$0	0.0%
Mail Clerks and Mail Machine Operators, Except Postal Service	2.2% (2.2% strict, 6.8% loose)	$52.6M	$0	0.0%
Medical Secretaries and Administrative Assistants	0.0% (0.0% strict, 69.8% loose)	$0	$0	36.2%
Office Clerks, General	6.0% (6.0% strict, 52.4% loose)	$6.9B	$6.9B	45.0%
Office Machine Operators, Except Computer	19.5% (0.0% strict, 19.5% loose)	$172M	$41.6M	2.4%
Proofreaders and Copy Markers	22.6% (22.6% strict, 82.6% loose)	$39.7M	$39.7M	17.8%
Secretaries and Administrative Assistants, Except Legal, Medical, and Executive	10.0% (0.0% strict, 57.5% loose)	$8.3B	$5.0B	45.3%
Statistical Assistants	36.8% (29.6% strict, 80.7% loose)	$101M	$72.2M	51.0%
Switchboard Operators, Including Answering Service	4.8% (0.0% strict, 38.6% loose)	$67.1M	$0	38.6%
Telephone Operators	19.7% (0.0% strict, 35.4% loose)	$26.1M	$6.5M	0.0%
Word Processors and Typists	21.5% (21.5% strict, 51.1% loose)	$363M	$226M	23.8%

Retail front line

Role	Share that passes (stricter and looser rules)	Passes the screen	All three pass	Observed AI use
Cashiers	10.4% (3.9% strict, 28.7% loose)	$10.4B	$3.9B	8.5%
Counter and Rental Clerks	33.0% (20.2% strict, 71.8% loose)	$5.7B	$3.5B	20.5%
First-Line Supervisors of Retail Sales Workers	7.2% (0.0% strict, 33.0% loose)	$4.2B	$2.6B	26.3%
Gambling Change Persons and Booth Cashiers	18.6% (6.6% strict, 27.6% loose)	$149M	$52.5M	0.0%
Parts Salespersons	22.4% (14.3% strict, 42.0% loose)	$2.5B	$1.7B	0.0%
Retail Salespersons	7.2% (3.5% strict, 42.3% loose)	$10.1B	$5.3B	32.2%

Operations

Role	Share that passes (stricter and looser rules)	Passes the screen	All three pass	Observed AI use
Administrative Services Managers	0.0% (0.0% strict, 24.8% loose)	$0	$0	0.0%
Advertising and Promotions Managers	0.0% (0.0% strict, 18.6% loose)	$0	$0	17.3%
Business Operations Specialists, All Other	9.8% (3.9% strict, 36.7% loose)	$10.2B	$7.7B	18.5%
Construction Managers	0.0% (0.0% strict, 16.9% loose)	$0	$0	11.9%
Emergency Management Directors	0.0% (0.0% strict, 29.7% loose)	$0	$0	0.0%
Entertainment and Recreation Managers, Except Gambling	4.9% (0.0% strict, 42.2% loose)	$153M	$0	—
Facilities Managers	0.0% (0.0% strict, 15.5% loose)	$0	$0	13.5%
Farm Labor Contractors	17.5% (0.0% strict, 17.5% loose)	$3.4M	$0	0.0%
Farmers, Ranchers, and Other Agricultural Managers	0.0% (0.0% strict, 15.0% loose)	$0	$0	0.0%
Food Service Managers	1.9% (0.0% strict, 18.0% loose)	$327M	$327M	0.0%
Funeral Home Managers	7.3% (0.0% strict, 39.4% loose)	$81.6M	$0	1.1%
Gambling Managers	5.0% (2.3% strict, 54.0% loose)	$23.2M	$0	0.0%
Industrial Production Managers	1.5% (0.0% strict, 20.7% loose)	$463M	$0	1.3%
Lodging Managers	0.8% (0.0% strict, 25.9% loose)	$23.2M	$23.2M	12.2%
Managers, All Other	2.9% (0.5% strict, 29.1% loose)	$2.8B	$1.1B	6.9%
Meeting, Convention, and Event Planners	0.0% (0.0% strict, 36.7% loose)	$0	$0	10.2%
Natural Sciences Managers	0.0% (0.0% strict, 32.0% loose)	$0	$0	6.1%
Personal Service Managers, All Other	20.4% (8.6% strict, 40.8% loose)	$129M	$26.4M	12.4%
Postmasters and Mail Superintendents	0.0% (0.0% strict, 66.6% loose)	$0	$0	0.0%
Project Management Specialists	10.0% (0.0% strict, 45.0% loose)	$10.9B	$0	—
Property, Real Estate, and Community Association Managers	4.0% (0.0% strict, 15.1% loose)	$948M	$0	16.5%

Supply chain

Role	Share that passes (stricter and looser rules)	Passes the screen	All three pass	Observed AI use
Cargo and Freight Agents	23.4% (9.6% strict, 67.9% loose)	$1.3B	$1.1B	1.7%
Couriers and Messengers	0.0% (0.0% strict, 18.0% loose)	$0	$0	4.1%
Dispatchers, Except Police, Fire, and Ambulance	20.2% (0.0% strict, 92.6% loose)	$2.3B	$0	22.6%
Logisticians	12.1% (2.6% strict, 38.1% loose)	$2.5B	$730M	15.7%
Meter Readers, Utilities	10.3% (7.5% strict, 37.1% loose)	$111M	$80.5M	0.0%
Order Clerks	44.3% (36.7% strict, 73.3% loose)	$1.7B	$1.5B	7.5%
Postal Service Clerks	2.8% (2.8% strict, 28.8% loose)	$132M	$0	0.0%
Postal Service Mail Carriers	6.9% (0.0% strict, 13.0% loose)	$1.4B	$0	0.0%
Postal Service Mail Sorters, Processors, and Processing Machine Operators	0.0% (0.0% strict, 2.4% loose)	$0	$0	0.0%
Production, Planning, and Expediting Clerks	32.1% (14.6% strict, 78.7% loose)	$7.4B	$3.4B	9.3%
Public Safety Telecommunicators	0.0% (0.0% strict, 39.6% loose)	$0	$0	4.7%
Shipping, Receiving, and Inventory Clerks	7.0% (7.0% strict, 57.2% loose)	$2.7B	$2.7B	0.0%
Transportation, Storage, and Distribution Managers	2.8% (0.0% strict, 24.6% loose)	$686M	$272M	9.6%
Weighers, Measurers, Checkers, and Samplers, Recordkeeping	19.8% (2.8% strict, 29.8% loose)	$441M	$223M	0.0%

Education

Role	Share that passes (stricter and looser rules)	Passes the screen	All three pass	Observed AI use
Adult Basic Education, Adult Secondary Education, and English as a Second Language Instructors	0.0% (0.0% strict, 45.1% loose)	$0	$0	6.4%
Agricultural Sciences Teachers, Postsecondary	5.1% (5.1% strict, 33.4% loose)	$42.1M	$42.1M	0.0%
Anthropology and Archeology Teachers, Postsecondary	5.8% (5.8% strict, 45.1% loose)	$31.4M	$31.4M	2.4%
Architecture Teachers, Postsecondary	7.3% (7.3% strict, 39.9% loose)	$73.4M	$73.4M	0.0%
Archivists	0.0% (0.0% strict, 22.7% loose)	$0	$0	13.5%
Area, Ethnic, and Cultural Studies Teachers, Postsecondary	5.9% (0.0% strict, 51.8% loose)	$65.0M	$65.0M	0.0%
Art, Drama, and Music Teachers, Postsecondary	9.7% (6.9% strict, 40.6% loose)	$964M	$688M	21.4%
Atmospheric, Earth, Marine, and Space Sciences Teachers, Postsecondary	9.0% (0.0% strict, 52.3% loose)	$116M	$116M	4.0%
Biological Science Teachers, Postsecondary	6.2% (0.0% strict, 20.1% loose)	$335M	$335M	29.1%
Business Teachers, Postsecondary	10.6% (10.6% strict, 38.7% loose)	$986M	$585M	30.7%
Career/Technical Education Teachers, Middle School	0.0% (0.0% strict, 32.7% loose)	$0	$0	0.0%
Career/Technical Education Teachers, Postsecondary	5.6% (0.0% strict, 26.1% loose)	$426M	$426M	15.6%
Career/Technical Education Teachers, Secondary School	5.2% (0.0% strict, 35.6% loose)	$379M	$0	0.0%
Chemistry Teachers, Postsecondary	5.7% (5.7% strict, 41.4% loose)	$118M	$118M	21.9%
Communications Teachers, Postsecondary	8.6% (8.6% strict, 32.0% loose)	$226M	$226M	21.1%
Computer Science Teachers, Postsecondary	6.4% (6.4% strict, 40.1% loose)	$243M	$243M	24.1%
Criminal Justice and Law Enforcement Teachers, Postsecondary	9.1% (9.1% strict, 36.8% loose)	$103M	$103M	0.0%
Curators	0.0% (0.0% strict, 10.4% loose)	$0	$0	41.2%
Economics Teachers, Postsecondary	4.5% (4.5% strict, 25.8% loose)	$74.7M	$74.7M	0.0%
Education Administrators, Kindergarten through Secondary	0.0% (0.0% strict, 25.0% loose)	$0	$0	5.3%
Education Administrators, Postsecondary	4.1% (0.0% strict, 12.0% loose)	$892M	$0	6.4%
Education Teachers, Postsecondary	5.8% (5.8% strict, 34.3% loose)	$269M	$269M	26.2%
Education and Childcare Administrators, Preschool and Daycare	0.0% (0.0% strict, 12.5% loose)	$0	$0	4.0%
Elementary School Teachers, Except Special Education	0.0% (0.0% strict, 27.4% loose)	$0	$0	10.3%
Engineering Teachers, Postsecondary	4.1% (4.1% strict, 36.5% loose)	$197M	$197M	35.9%
English Language and Literature Teachers, Postsecondary	6.8% (6.8% strict, 29.7% loose)	$366M	$366M	36.2%
Environmental Science Teachers, Postsecondary	7.1% (7.1% strict, 37.4% loose)	$50.3M	$50.3M	2.3%
Family and Consumer Sciences Teachers, Postsecondary	7.4% (7.4% strict, 37.2% loose)	$16.4M	$16.4M	0.0%
Farm and Home Management Educators	0.0% (0.0% strict, 17.4% loose)	$0	$0	0.0%
Foreign Language and Literature Teachers, Postsecondary	9.7% (0.0% strict, 51.0% loose)	$180M	$180M	24.0%
Forestry and Conservation Science Teachers, Postsecondary	6.9% (0.0% strict, 29.1% loose)	$9.4M	$9.4M	2.1%
Geography Teachers, Postsecondary	7.0% (0.0% strict, 54.9% loose)	$22.9M	$0	0.0%
Health Specialties Teachers, Postsecondary	7.0% (7.0% strict, 26.9% loose)	$2.2B	$2.2B	30.4%
History Teachers, Postsecondary	6.4% (0.0% strict, 48.4% loose)	$118M	$118M	9.9%
Instructional Coordinators	0.0% (0.0% strict, 27.8% loose)	$0	$0	30.4%
Kindergarten Teachers, Except Special Education	2.0% (0.0% strict, 18.8% loose)	$154M	$0	0.0%
Law Teachers, Postsecondary	7.4% (7.4% strict, 52.5% loose)	$237M	$237M	0.0%
Librarians and Media Collections Specialists	0.0% (0.0% strict, 50.5% loose)	$0	$0	20.3%
Library Assistants, Clerical	10.7% (7.4% strict, 25.8% loose)	$324M	$248M	9.8%
Library Science Teachers, Postsecondary	4.8% (4.8% strict, 34.3% loose)	$16.5M	$0	1.8%
Library Technicians	17.4% (11.3% strict, 35.8% loose)	$546M	$478M	5.4%
Mathematical Science Teachers, Postsecondary	10.8% (0.0% strict, 46.5% loose)	$489M	$489M	43.0%
Middle School Teachers, Except Special and Career/Technical Education	3.8% (0.0% strict, 33.4% loose)	$1.7B	$0	29.7%
Museum Technicians and Conservators	9.1% (0.0% strict, 19.5% loose)	$61.4M	$0	0.0%
Nursing Instructors and Teachers, Postsecondary	4.9% (4.9% strict, 27.7% loose)	$315M	$0	19.1%
Philosophy and Religion Teachers, Postsecondary	7.2% (7.2% strict, 27.0% loose)	$133M	$133M	22.0%
Physics Teachers, Postsecondary	6.2% (6.2% strict, 30.1% loose)	$89.1M	$0	0.0%
Political Science Teachers, Postsecondary	6.2% (6.2% strict, 46.0% loose)	$113M	$113M	0.0%
Preschool Teachers, Except Special Education	0.0% (0.0% strict, 26.7% loose)	$0	$0	0.0%
Psychology Teachers, Postsecondary	4.0% (4.0% strict, 41.2% loose)	$154M	$0	32.2%
Recreation and Fitness Studies Teachers, Postsecondary	7.8% (7.8% strict, 54.3% loose)	$86.4M	$86.4M	11.4%
Secondary School Teachers, Except Special and Career/Technical Education	0.0% (0.0% strict, 34.1% loose)	$0	$0	29.0%
Self-Enrichment Teachers	4.7% (0.0% strict, 48.6% loose)	$773M	$773M	6.6%
Social Work Teachers, Postsecondary	4.7% (4.7% strict, 42.7% loose)	$51.6M	$51.6M	0.0%
Sociology Teachers, Postsecondary	7.2% (7.2% strict, 34.5% loose)	$86.9M	$86.9M	0.0%
Special Education Teachers, All Other	0.7% (0.0% strict, 17.0% loose)	$21.1M	$0	2.6%
Special Education Teachers, Middle School	0.0% (0.0% strict, 21.8% loose)	$0	$0	13.4%
Special Education Teachers, Preschool	0.0% (0.0% strict, 23.3% loose)	$0	$0	0.6%
Special Education Teachers, Secondary School	0.0% (0.0% strict, 30.9% loose)	$0	$0	9.9%
Substitute Teachers, Short-Term	0.0% (0.0% strict, 33.8% loose)	$0	$0	—
Teaching Assistants, Postsecondary	0.0% (0.0% strict, 44.3% loose)	$0	$0	10.0%
Tutors	13.2% (0.0% strict, 45.2% loose)	$1.1B	$510M	40.8%

People

Role	Share that passes (stricter and looser rules)	Passes the screen	All three pass	Observed AI use
Compensation and Benefits Managers	2.1% (0.0% strict, 32.8% loose)	$60.9M	$60.9M	0.0%
Compensation, Benefits, and Job Analysis Specialists	0.0% (0.0% strict, 30.8% loose)	$0	$0	6.5%
Human Resources Assistants, Except Payroll and Timekeeping	17.0% (8.5% strict, 93.3% loose)	$791M	$0	40.5%
Human Resources Managers	3.9% (0.0% strict, 16.7% loose)	$1.3B	$0	0.0%
Human Resources Specialists	12.4% (9.6% strict, 77.2% loose)	$9.0B	$3.0B	40.3%
Labor Relations Specialists	3.1% (0.0% strict, 53.1% loose)	$188M	$188M	0.0%
Training and Development Managers	0.0% (0.0% strict, 20.9% loose)	$0	$0	37.8%
Training and Development Specialists	3.2% (3.2% strict, 50.9% loose)	$1.0B	$1.0B	27.9%

Clinical

Role	Share that passes (stricter and looser rules)	Passes the screen	All three pass	Observed AI use
Acupuncturists	0.0% (0.0% strict, 30.0% loose)	$0	$0	0.0%
Anesthesiologists	0.0% (0.0% strict, 6.3% loose)	$0	$0	0.0%
Athletic Trainers	0.0% (0.0% strict, 14.8% loose)	$0	$0	5.3%
Audiologists	0.0% (0.0% strict, 23.4% loose)	$0	$0	0.0%
Cardiologists	0.0% (0.0% strict, 16.0% loose)	$0	$0	—
Cardiovascular Technologists and Technicians	0.0% (0.0% strict, 41.4% loose)	$0	$0	0.0%
Chiropractors	0.0% (0.0% strict, 26.6% loose)	$0	$0	0.0%
Dental Hygienists	2.1% (2.1% strict, 17.3% loose)	$427M	$0	0.0%
Dentists, General	0.0% (0.0% strict, 1.9% loose)	$0	$0	3.1%
Dermatologists	0.0% (0.0% strict, 28.4% loose)	$0	$0	0.0%
Diagnostic Medical Sonographers	4.4% (0.0% strict, 23.6% loose)	$351M	$0	0.0%
Dietetic Technicians	0.0% (0.0% strict, 15.0% loose)	$0	$0	24.5%
Dietitians and Nutritionists	0.0% (0.0% strict, 41.3% loose)	$0	$0	13.3%
Emergency Medical Technicians	0.0% (0.0% strict, 8.3% loose)	$0	$0	0.0%
Emergency Medicine Physicians	0.0% (0.0% strict, 15.6% loose)	$0	$0	—
Exercise Physiologists	0.0% (0.0% strict, 29.5% loose)	$0	$0	0.0%
Family Medicine Physicians	1.1% (0.0% strict, 32.8% loose)	$314M	$0	0.0%
General Internal Medicine Physicians	0.0% (0.0% strict, 1.2% loose)	$0	$0	8.4%
Genetic Counselors	0.0% (0.0% strict, 35.6% loose)	$0	$0	1.2%
Health Information Technologists and Medical Registrars	12.5% (0.0% strict, 43.8% loose)	$325M	$162M	30.6%
Health Technologists and Technicians, All Other	3.0% (1.7% strict, 35.6% loose)	$294M	$0	4.5%
Healthcare Diagnosing or Treating Practitioners, All Other	2.0% (0.0% strict, 22.8% loose)	$64.4M	$0	2.2%
Healthcare Practitioners and Technical Workers, All Other	1.0% (0.0% strict, 22.6% loose)	$25.7M	$0	0.0%
Hearing Aid Specialists	0.0% (0.0% strict, 0.0% loose)	$0	$0	0.0%
Licensed Practical and Licensed Vocational Nurses	3.4% (0.0% strict, 7.5% loose)	$1.4B	$0	0.0%
Magnetic Resonance Imaging Technologists	1.2% (0.0% strict, 24.0% loose)	$43.0M	$0	3.9%
Medical Dosimetrists	0.0% (0.0% strict, 0.0% loose)	$0	$0	—
Medical and Clinical Laboratory Technicians	0.0% (0.0% strict, 3.0% loose)	$0	$0	—
Medical and Clinical Laboratory Technologists	1.7% (0.3% strict, 13.0% loose)	$189M	$28.5M	—
Medical and Health Services Managers	0.0% (0.0% strict, 17.5% loose)	$0	$0	6.6%
Neurologists	0.0% (0.0% strict, 45.8% loose)	$0	$0	0.0%
Nuclear Medicine Technologists	0.0% (0.0% strict, 14.6% loose)	$0	$0	5.1%
Nurse Anesthetists	0.0% (0.0% strict, 0.0% loose)	$0	$0	0.0%
Nurse Midwives	0.0% (0.0% strict, 13.3% loose)	$0	$0	5.2%
Nurse Practitioners	0.0% (0.0% strict, 18.9% loose)	$0	$0	9.4%
Obstetricians and Gynecologists	1.8% (0.0% strict, 26.9% loose)	$98.5M	$0	6.9%
Occupational Therapists	0.0% (0.0% strict, 29.0% loose)	$0	$0	0.8%
Ophthalmic Medical Technicians	0.0% (0.0% strict, 10.5% loose)	$0	$0	0.0%
Ophthalmologists, Except Pediatric	0.0% (0.0% strict, 17.1% loose)	$0	$0	0.0%
Opticians, Dispensing	16.5% (0.0% strict, 48.5% loose)	$654M	$0	0.0%
Optometrists	0.0% (0.0% strict, 0.0% loose)	$0	$0	0.0%
Oral and Maxillofacial Surgeons	0.0% (0.0% strict, 0.0% loose)	$0	$0	0.0%
Orthodontists	0.0% (0.0% strict, 17.2% loose)	$0	$0	0.0%
Orthopedic Surgeons, Except Pediatric	0.0% (0.0% strict, 0.0% loose)	$0	$0	0.0%
Orthotists and Prosthetists	0.0% (0.0% strict, 20.3% loose)	$0	$0	0.0%
Paramedics	0.0% (0.0% strict, 7.7% loose)	$0	$0	0.0%
Pediatric Surgeons	0.0% (0.0% strict, 11.1% loose)	$0	$0	0.0%
Pediatricians, General	0.0% (0.0% strict, 23.5% loose)	$0	$0	0.0%
Pharmacists	0.0% (0.0% strict, 12.1% loose)	$0	$0	9.0%
Pharmacy Technicians	4.1% (0.0% strict, 23.0% loose)	$897M	$897M	7.3%
Physical Therapists	0.0% (0.0% strict, 29.3% loose)	$0	$0	1.7%
Physician Assistants	0.0% (0.0% strict, 7.2% loose)	$0	$0	0.0%
Physicians, All Other	0.6% (0.0% strict, 17.5% loose)	$446M	$0	3.0%
Physicians, Pathologists	0.0% (0.0% strict, 4.5% loose)	$0	$0	15.8%
Podiatrists	0.0% (0.0% strict, 0.0% loose)	$0	$0	0.0%
Prosthodontists	0.0% (0.0% strict, 0.0% loose)	$0	$0	0.0%
Psychiatric Technicians	0.0% (0.0% strict, 2.7% loose)	$0	$0	0.0%
Psychiatrists	0.0% (0.0% strict, 32.0% loose)	$0	$0	0.0%
Radiation Therapists	4.4% (4.4% strict, 16.2% loose)	$88.4M	$0	0.0%
Radiologic Technologists and Technicians	4.0% (0.0% strict, 35.7% loose)	$714M	$0	0.0%
Radiologists	6.9% (0.0% strict, 18.6% loose)	$692M	$0	19.9%
Recreational Therapists	0.0% (0.0% strict, 23.7% loose)	$0	$0	0.0%
Registered Nurses	0.9% (0.0% strict, 8.8% loose)	$2.8B	$0	5.9%
Respiratory Therapists	0.0% (0.0% strict, 13.8% loose)	$0	$0	0.0%
Speech-Language Pathologists	0.0% (0.0% strict, 25.9% loose)	$0	$0	0.0%
Surgical Assistants	0.0% (0.0% strict, 7.6% loose)	$0	$0	0.0%
Surgical Technologists	0.0% (0.0% strict, 5.9% loose)	$0	$0	0.0%
Therapists, All Other	0.3% (0.0% strict, 34.6% loose)	$3.4M	$0	4.0%
Veterinarians	1.2% (0.0% strict, 4.3% loose)	$138M	$138M	9.3%
Veterinary Technologists and Technicians	0.0% (0.0% strict, 10.6% loose)	$0	$0	0.0%

Marketing

Role	Share that passes (stricter and looser rules)	Passes the screen	All three pass	Observed AI use
Broadcast Announcers and Radio Disc Jockeys	5.5% (0.0% strict, 67.1% loose)	$89.9M	$0	6.3%
Court Reporters and Simultaneous Captioners	26.4% (11.2% strict, 57.1% loose)	$246M	$0	34.3%
Editors	8.0% (0.0% strict, 38.6% loose)	$639M	$0	24.6%
Fundraisers	2.5% (0.0% strict, 40.0% loose)	$183M	$0	4.7%
Fundraising Managers	0.0% (0.0% strict, 16.3% loose)	$0	$0	—
Interpreters and Translators	15.8% (15.8% strict, 83.2% loose)	$504M	$0	43.0%
Market Research Analysts and Marketing Specialists	3.8% (0.0% strict, 31.9% loose)	$2.8B	$955M	64.8%
Marketing Managers	6.3% (0.0% strict, 19.1% loose)	$4.2B	$4.2B	31.9%
News Analysts, Reporters, and Journalists	0.0% (0.0% strict, 73.5% loose)	$0	$0	21.0%
Public Relations Managers	0.0% (0.0% strict, 30.0% loose)	$0	$0	23.2%
Public Relations Specialists	0.0% (0.0% strict, 50.8% loose)	$0	$0	45.3%
Technical Writers	10.4% (10.4% strict, 80.8% loose)	$435M	$435M	47.5%
Writers and Authors	2.9% (2.9% strict, 72.1% loose)	$114M	$0	24.6%

Claims & underwriting

Role	Share that passes (stricter and looser rules)	Passes the screen	All three pass	Observed AI use
Claims Adjusters, Examiners, and Investigators	11.2% (0.0% strict, 67.1% loose)	$2.7B	$439M	8.2%
Insurance Appraisers, Auto Damage	0.0% (0.0% strict, 82.2% loose)	$0	$0	0.0%
Insurance Claims and Policy Processing Clerks	35.1% (0.0% strict, 85.5% loose)	$4.1B	$0	14.7%
Insurance Underwriters	0.0% (0.0% strict, 32.3% loose)	$0	$0	6.3%

Engineering

Role	Share that passes (stricter and looser rules)	Passes the screen	All three pass	Observed AI use
Aerospace Engineering and Operations Technologists and Technicians	0.0% (0.0% strict, 25.9% loose)	$0	$0	0.0%
Aerospace Engineers	5.0% (5.0% strict, 10.4% loose)	$476M	$0	7.5%
Agricultural Engineers	0.0% (0.0% strict, 42.1% loose)	$0	$0	0.0%
Architects, Except Landscape and Naval	0.7% (0.0% strict, 55.5% loose)	$77.1M	$0	7.8%
Architectural and Civil Drafters	0.5% (0.0% strict, 71.0% loose)	$38.9M	$0	0.0%
Architectural and Engineering Managers	0.0% (0.0% strict, 15.4% loose)	$0	$0	3.1%
Bioengineers and Biomedical Engineers	6.1% (0.0% strict, 13.6% loose)	$148M	$0	13.3%
Calibration Technologists and Technicians	0.0% (0.0% strict, 40.0% loose)	$0	$0	—
Cartographers and Photogrammetrists	5.6% (0.0% strict, 60.4% loose)	$55.2M	$0	8.0%
Chemical Engineers	0.0% (0.0% strict, 23.6% loose)	$0	$0	0.0%
Civil Engineering Technologists and Technicians	0.0% (0.0% strict, 70.5% loose)	$0	$0	7.8%
Civil Engineers	0.0% (0.0% strict, 14.0% loose)	$0	$0	0.8%
Computer Hardware Engineers	7.2% (0.0% strict, 25.5% loose)	$849M	$849M	14.5%
Electrical Engineers	0.0% (0.0% strict, 39.3% loose)	$0	$0	5.9%
Electrical and Electronic Engineering Technologists and Technicians	0.0% (0.0% strict, 37.8% loose)	$0	$0	2.0%
Electrical and Electronics Drafters	7.0% (7.0% strict, 68.8% loose)	$105M	$71.5M	8.9%
Electro-Mechanical and Mechatronics Technologists and Technicians	1.2% (0.0% strict, 24.6% loose)	$12.0M	$12.0M	0.0%
Electronics Engineers, Except Computer	6.6% (0.0% strict, 52.8% loose)	$798M	$0	10.0%
Engineering Technologists and Technicians, Except Drafters, All Other	5.6% (0.0% strict, 15.2% loose)	$272M	$0	0.0%
Engineers, All Other	5.0% (0.0% strict, 27.0% loose)	$904M	$77.3M	6.6%
Environmental Engineering Technologists and Technicians	0.0% (0.0% strict, 37.7% loose)	$0	$0	10.3%
Environmental Engineers	0.0% (0.0% strict, 22.8% loose)	$0	$0	3.6%
Health and Safety Engineers, Except Mining Safety Engineers and Inspectors	0.0% (0.0% strict, 17.4% loose)	$0	$0	0.0%
Industrial Engineering Technologists and Technicians	2.9% (0.0% strict, 25.3% loose)	$147M	$0	0.0%
Industrial Engineers	1.0% (0.0% strict, 33.8% loose)	$350M	$350M	3.7%
Landscape Architects	0.0% (0.0% strict, 40.0% loose)	$0	$0	0.0%
Marine Engineers and Naval Architects	0.0% (0.0% strict, 27.8% loose)	$0	$0	3.6%
Materials Engineers	3.3% (0.0% strict, 18.5% loose)	$84.6M	$0	0.0%
Mechanical Drafters	0.0% (0.0% strict, 78.5% loose)	$0	$0	0.0%
Mechanical Engineering Technologists and Technicians	1.7% (0.0% strict, 42.8% loose)	$43.1M	$0	6.8%
Mechanical Engineers	1.0% (0.0% strict, 19.9% loose)	$301M	$0	8.1%
Mining and Geological Engineers, Including Mining Safety Engineers	0.0% (0.0% strict, 17.6% loose)	$0	$0	0.0%
Nuclear Engineers	0.0% (0.0% strict, 13.2% loose)	$0	$0	0.0%
Petroleum Engineers	0.0% (0.0% strict, 27.6% loose)	$0	$0	0.0%
Surveying and Mapping Technicians	8.6% (0.0% strict, 62.4% loose)	$255M	$0	0.0%
Surveyors	6.9% (0.0% strict, 46.2% loose)	$275M	$0	0.2%

Procurement

Role	Share that passes (stricter and looser rules)	Passes the screen	All three pass	Observed AI use
Buyers and Purchasing Agents, Farm Products	11.0% (0.0% strict, 39.2% loose)	$1.4B	$0	—
Purchasing Agents, Except Wholesale, Retail, and Farm Products	7.2% (0.0% strict, 46.9% loose)	$941M	$941M	—
Purchasing Managers	15.3% (0.0% strict, 39.2% loose)	$1.8B	$886M	20.0%
Wholesale and Retail Buyers, Except Farm Products	4.5% (4.5% strict, 25.3% loose)	$585M	$585M	—

Social services

Role	Share that passes (stricter and looser rules)	Passes the screen	All three pass	Observed AI use
Child, Family, and School Social Workers	7.3% (0.0% strict, 36.6% loose)	$1.8B	$342M	0.7%
Clergy	9.2% (0.0% strict, 23.9% loose)	$354M	$0	11.2%
Community Health Workers	0.0% (0.0% strict, 26.5% loose)	$0	$0	0.0%
Directors, Religious Activities and Education	0.0% (0.0% strict, 20.2% loose)	$0	$0	15.3%
Educational, Guidance, and Career Counselors and Advisors	5.5% (0.0% strict, 43.4% loose)	$1.3B	$0	11.8%
Health Education Specialists	11.6% (0.0% strict, 29.2% loose)	$473M	$0	13.8%
Healthcare Social Workers	0.0% (0.0% strict, 28.6% loose)	$0	$0	9.2%
Marriage and Family Therapists	0.0% (0.0% strict, 54.0% loose)	$0	$0	0.0%
Mental Health and Substance Abuse Social Workers	0.0% (0.0% strict, 22.3% loose)	$0	$0	0.0%
Probation Officers and Correctional Treatment Specialists	0.0% (0.0% strict, 35.9% loose)	$0	$0	0.0%
Rehabilitation Counselors	0.0% (0.0% strict, 35.9% loose)	$0	$0	0.0%
Social and Community Service Managers	0.0% (0.0% strict, 11.3% loose)	$0	$0	18.1%
Social and Human Service Assistants	0.9% (0.9% strict, 49.1% loose)	$175M	$0	0.0%

Data & analytics

Role	Share that passes (stricter and looser rules)	Passes the screen	All three pass	Observed AI use
Actuaries	0.0% (0.0% strict, 1.1% loose)	$0	$0	5.4%
Data Scientists	13.4% (0.0% strict, 67.2% loose)	$3.9B	$1.1B	46.1%
Mathematical Science Occupations, All Other	14.5% (7.2% strict, 76.4% loose)	$44.2M	$22.0M	47.9%
Mathematicians	10.5% (0.0% strict, 10.5% loose)	$26.7M	$0	42.4%
Operations Research Analysts	0.0% (0.0% strict, 55.6% loose)	$0	$0	42.9%
Statisticians	3.6% (0.0% strict, 51.9% loose)	$114M	$0	21.1%

Creative

Role	Share that passes (stricter and looser rules)	Passes the screen	All three pass	Observed AI use
Art Directors	0.0% (0.0% strict, 34.9% loose)	$0	$0	21.0%
Athletes and Sports Competitors	0.0% (0.0% strict, 7.2% loose)	$0	$0	0.0%
Audio and Video Technicians	8.0% (5.4% strict, 36.0% loose)	$308M	$124M	1.7%
Broadcast Technicians	16.2% (0.0% strict, 41.9% loose)	$210M	$0	2.0%
Camera Operators, Television, Video, and Film	0.9% (0.9% strict, 31.6% loose)	$16.9M	$16.9M	16.5%
Choreographers	0.0% (0.0% strict, 14.9% loose)	$0	$0	8.0%
Coaches and Scouts	0.0% (0.0% strict, 8.4% loose)	$0	$0	0.0%
Commercial and Industrial Designers	0.0% (0.0% strict, 26.4% loose)	$0	$0	4.4%
Craft Artists	0.0% (0.0% strict, 24.8% loose)	$0	$0	5.4%
Fashion Designers	0.0% (0.0% strict, 27.4% loose)	$0	$0	5.7%
Film and Video Editors	5.7% (5.7% strict, 60.6% loose)	$131M	$131M	22.3%
Fine Artists, Including Painters, Sculptors, and Illustrators	0.0% (0.0% strict, 23.4% loose)	$0	$0	35.6%
Floral Designers	0.0% (0.0% strict, 30.1% loose)	$0	$0	0.0%
Graphic Designers	6.1% (6.1% strict, 86.1% loose)	$892M	$0	36.7%
Interior Designers	0.0% (0.0% strict, 62.1% loose)	$0	$0	0.0%
Lighting Technicians	0.0% (0.0% strict, 12.5% loose)	$0	$0	—
Merchandise Displayers and Window Trimmers	3.7% (0.0% strict, 29.9% loose)	$274M	$0	0.0%
Music Directors and Composers	6.5% (2.3% strict, 60.9% loose)	$63.1M	$22.7M	3.7%
Photographers	6.7% (6.7% strict, 28.6% loose)	$174M	$174M	19.5%
Producers and Directors	4.0% (0.0% strict, 37.0% loose)	$641M	$111M	9.2%
Set and Exhibit Designers	0.0% (0.0% strict, 44.7% loose)	$0	$0	0.0%
Sound Engineering Technicians	11.1% (11.1% strict, 26.2% loose)	$99.2M	$60.0M	0.0%
Special Effects Artists and Animators	0.0% (0.0% strict, 71.1% loose)	$0	$0	35.7%
Umpires, Referees, and Other Sports Officials	11.0% (2.9% strict, 33.9% loose)	$74.2M	$19.3M	0.0%

Research

Role	Share that passes (stricter and looser rules)	Passes the screen	All three pass	Observed AI use
Agricultural Technicians	4.0% (0.0% strict, 29.5% loose)	$18.6M	$0	0.6%
Animal Scientists	0.0% (0.0% strict, 14.5% loose)	$0	$0	0.0%
Anthropologists and Archeologists	0.0% (0.0% strict, 11.1% loose)	$0	$0	13.5%
Astronomers	3.3% (0.0% strict, 10.7% loose)	$6.7M	$0	38.5%
Atmospheric and Space Scientists	3.8% (0.0% strict, 31.5% loose)	$33.9M	$0	3.8%
Biochemists and Biophysicists	0.0% (0.0% strict, 4.6% loose)	$0	$0	7.6%
Biological Scientists, All Other	3.1% (0.0% strict, 30.0% loose)	$179M	$0	24.5%
Biological Technicians	8.5% (8.5% strict, 27.8% loose)	$363M	$0	6.0%
Chemical Technicians	0.0% (0.0% strict, 26.4% loose)	$0	$0	31.5%
Chemists	0.0% (0.0% strict, 22.4% loose)	$0	$0	26.1%
Clinical and Counseling Psychologists	0.0% (0.0% strict, 32.5% loose)	$0	$0	5.9%
Conservation Scientists	2.5% (1.1% strict, 26.1% loose)	$47.0M	$19.7M	0.0%
Economists	0.0% (0.0% strict, 30.5% loose)	$0	$0	24.2%
Environmental Science and Protection Technicians, Including Health	4.0% (0.0% strict, 27.4% loose)	$77.6M	$0	14.4%
Environmental Scientists and Specialists, Including Health	0.8% (0.0% strict, 35.5% loose)	$56.3M	$0	5.5%
Epidemiologists	0.0% (0.0% strict, 18.0% loose)	$0	$0	0.0%
Food Science Technicians	18.7% (0.0% strict, 32.3% loose)	$134M	$0	20.4%
Food Scientists and Technologists	0.0% (0.0% strict, 0.0% loose)	$0	$0	0.0%
Forensic Science Technicians	0.0% (0.0% strict, 17.7% loose)	$0	$0	0.0%
Forest and Conservation Technicians	5.8% (0.0% strict, 13.7% loose)	$101M	$0	2.2%
Foresters	0.0% (0.0% strict, 3.0% loose)	$0	$0	0.0%
Geographers	8.8% (0.0% strict, 47.1% loose)	$11.0M	$0	11.0%
Geological Technicians, Except Hydrologic Technicians	11.7% (0.0% strict, 34.0% loose)	$62.4M	$0	22.9%
Geoscientists, Except Hydrologists and Geographers	1.2% (0.0% strict, 42.4% loose)	$28.9M	$0	4.3%
Historians	0.0% (0.0% strict, 43.3% loose)	$0	$0	2.8%
Hydrologic Technicians	12.5% (0.0% strict, 18.8% loose)	$20.3M	$0	—
Hydrologists	0.0% (0.0% strict, 9.8% loose)	$0	$0	7.5%
Industrial-Organizational Psychologists	0.0% (0.0% strict, 33.1% loose)	$0	$0	15.8%
Life, Physical, and Social Science Technicians, All Other	1.8% (0.0% strict, 50.0% loose)	$79.1M	$0	9.6%
Materials Scientists	0.0% (0.0% strict, 19.2% loose)	$0	$0	18.4%
Medical Scientists, Except Epidemiologists	0.0% (0.0% strict, 0.0% loose)	$0	$0	3.8%
Microbiologists	0.0% (0.0% strict, 4.5% loose)	$0	$0	9.7%
Nuclear Technicians	2.4% (0.0% strict, 5.0% loose)	$13.5M	$13.5M	0.0%
Occupational Health and Safety Specialists	0.0% (0.0% strict, 18.3% loose)	$0	$0	0.0%
Occupational Health and Safety Technicians	4.4% (0.0% strict, 27.6% loose)	$83.8M	$0	0.0%
Physical Scientists, All Other	2.1% (0.0% strict, 42.5% loose)	$56.9M	$0	3.8%
Physicists	0.0% (0.0% strict, 50.0% loose)	$0	$0	27.3%
Political Scientists	0.0% (0.0% strict, 17.2% loose)	$0	$0	45.2%
Psychologists, All Other	0.0% (0.0% strict, 28.4% loose)	$0	$0	0.0%
School Psychologists	0.0% (0.0% strict, 28.2% loose)	$0	$0	4.7%
Social Science Research Assistants	15.6% (0.0% strict, 70.1% loose)	$299M	$299M	43.9%
Social Scientists and Related Workers, All Other	0.0% (0.0% strict, 43.6% loose)	$0	$0	3.3%
Sociologists	0.0% (0.0% strict, 20.3% loose)	$0	$0	38.3%
Soil and Plant Scientists	0.0% (0.0% strict, 0.0% loose)	$0	$0	5.1%
Survey Researchers	12.3% (0.0% strict, 65.3% loose)	$61.3M	$0	43.2%
Urban and Regional Planners	7.8% (3.9% strict, 48.8% loose)	$294M	$147M	9.6%
Zoologists and Wildlife Biologists	0.0% (0.0% strict, 29.9% loose)	$0	$0	6.1%

Legal & compliance

Role	Share that passes (stricter and looser rules)	Passes the screen	All three pass	Observed AI use
Administrative Law Judges, Adjudicators, and Hearing Officers	15.5% (4.3% strict, 42.3% loose)	$307M	$84.1M	30.5%
Arbitrators, Mediators, and Conciliators	5.1% (5.1% strict, 51.4% loose)	$31.9M	$31.9M	24.3%
Compliance Officers	3.1% (0.0% strict, 53.2% loose)	$1.1B	$259M	12.1%
Judges, Magistrate Judges, and Magistrates	0.0% (0.0% strict, 24.8% loose)	$0	$0	31.1%
Judicial Law Clerks	4.1% (0.0% strict, 32.7% loose)	$37.4M	$0	17.0%
Lawyers	0.0% (0.0% strict, 8.6% loose)	$0	$0	16.7%
Paralegals and Legal Assistants	0.0% (0.0% strict, 43.1% loose)	$0	$0	29.3%
Title Examiners, Abstractors, and Searchers	2.3% (2.3% strict, 90.2% loose)	$63.8M	$63.8M	2.2%

Executive

Role	Share that passes (stricter and looser rules)	Passes the screen	All three pass	Observed AI use
Chief Executives	0.0% (0.0% strict, 23.3% loose)	$0	$0	3.3%
General and Operations Managers	0.0% (0.0% strict, 11.1% loose)	$0	$0	13.8%
Legislators	0.0% (0.0% strict, 13.3% loose)	$0	$0	0.0%

Strategy

Role	Share that passes (stricter and looser rules)	Passes the screen	All three pass	Observed AI use
Management Analysts	0.0% (0.0% strict, 38.5% loose)	$0	$0	24.3%

Part II · functions

What to automate

By business function. Work waiting only on a check is work that at least two of the three models each find quick to judge and cheap to get wrong, with no existing check to settle it; whoever builds a reliable check for that work would bring it inside the screen. More work than this has a missing check as the reason given most often, but there the models split on whether it is also slow to judge or costly when wrong.

In most functions, more work waits on a check than passesChart

Customer support22.4%

Claims & underwriting14.8%

Finance12.7%

Supply chain12.1%

Retail front line10.1%

Sales9.5%

Procurement9.3%

Data & analytics8.7%

Admin7.8%

People7.4%

IT & software5.3%

Operations5.1%

Marketing4.3%

Creative3.1%

Social services3.0%

Education2.7%

Engineering1.8%

Research1.6%

Clinical0.9%

Legal & compliance0.7%

Executive0.0%

Strategy0.0%

passes the screenwaits only on a checkneeds a bodyheld for more than a missing checkto its left, all three models pass

Shares of each function's knowledge payroll, ordered by the share that passes, which is printed beside the name. From the census (v3.2-2026-09-24), a screen scored by three AI models, not a claim about what AI can do and not a forecast. The tick marks the part all three models pass. Work waiting only on a check is work at least two of the three models find quick to judge and cheap to get wrong, with no existing check to settle it.

The numbers

Function	Passes the screen	All three models pass	Waits only on a check	Needs a body	Held for more than a missing check
Customer support	22.4%	12.5%	49.8%	15.8%	12.0%
Claims & underwriting	14.8%	1.0%	10.0%	6.6%	68.7%
Finance	12.7%	3.7%	8.4%	9.2%	69.7%
Supply chain	12.1%	5.9%	10.0%	33.1%	44.7%
Retail front line	10.1%	5.2%	15.7%	51.0%	23.2%
Sales	9.5%	4.1%	12.4%	10.5%	67.7%
Procurement	9.3%	4.7%	6.5%	6.1%	78.1%
Data & analytics	8.7%	2.5%	18.4%	0.0%	73.0%
Admin	7.8%	6.1%	27.9%	27.8%	36.4%
People	7.4%	2.5%	16.5%	0.6%	75.5%
IT & software	5.3%	0.6%	13.2%	9.7%	71.8%
Operations	5.1%	1.8%	10.5%	16.1%	68.3%
Marketing	4.3%	2.6%	15.1%	4.5%	76.0%
Creative	3.1%	0.7%	26.6%	34.2%	36.1%
Social services	3.0%	0.3%	8.5%	7.1%	81.4%
Education	2.7%	1.7%	19.0%	24.5%	53.8%
Engineering	1.8%	0.5%	9.8%	21.0%	67.4%
Research	1.6%	0.4%	9.1%	31.6%	57.7%
Clinical	0.9%	0.1%	2.3%	37.4%	59.4%
Legal & compliance	0.7%	0.2%	6.8%	14.5%	77.9%
Executive	0.0%	0.0%	0.4%	7.4%	92.3%
Strategy	0.0%	0.0%	0.0%	13.5%	86.5%

Function	Knowledge payroll	Passes the screen	All three pass	Share that passes (stricter and looser rules)	Waits only on a check
Finance	$608B	$77.0B	$22.3B	12.7% (1.6% strict, 41.4% loose)	$51.3B
Sales	$482B	$45.7B	$19.9B	9.5% (3.2% strict, 46.2% loose)	$59.6B
Customer support	$174B	$38.9B	$21.8B	22.4% (12.2% strict, 82.4% loose)	$86.4B
IT & software	$677B	$36.0B	$4.1B	5.3% (0.6% strict, 42.4% loose)	$89.4B
Admin	$432B	$33.8B	$26.3B	7.8% (3.6% strict, 56.9% loose)	$121B
Retail front line	$328B	$33.0B	$17.0B	10.1% (4.2% strict, 38.0% loose)	$51.6B
Operations	$508B	$25.9B	$9.1B	5.1% (0.9% strict, 31.2% loose)	$53.5B
Supply chain	$170B	$20.5B	$9.9B	12.1% (5.2% strict, 46.3% loose)	$17.0B
Education	$556B	$14.9B	$9.7B	2.7% (1.4% strict, 31.2% loose)	$106B
People	$167B	$12.5B	$4.2B	7.4% (5.1% strict, 54.3% loose)	$27.6B
Clinical	$1.1T	$9.9B	$1.2B	0.9% (0.1% strict, 14.5% loose)	$23.9B
Marketing	$213B	$9.2B	$5.6B	4.3% (0.5% strict, 33.6% loose)	$32.1B
Claims & underwriting	$45.9B	$6.8B	$439M	14.8% (0.0% strict, 64.5% loose)	$4.6B
Engineering	$295B	$5.2B	$1.4B	1.8% (0.2% strict, 29.5% loose)	$28.8B
Procurement	$51.4B	$4.8B	$2.4B	9.3% (1.1% strict, 37.6% loose)	$3.3B
Social services	$135B	$4.1B	$342M	3.0% (0.1% strict, 34.4% loose)	$11.5B
Data & analytics	$46.9B	$4.1B	$1.2B	8.7% (0.1% strict, 58.0% loose)	$8.6B
Creative	$92.3B	$2.9B	$658M	3.1% (1.7% strict, 39.9% loose)	$24.6B
Research	$125B	$2.0B	$480M	1.6% (0.4% strict, 25.1% loose)	$11.3B
Legal & compliance	$203B	$1.5B	$439M	0.7% (0.1% strict, 22.1% loose)	$13.9B
Executive	$535B	$0	$0	0.0% (0.0% strict, 12.3% loose)	$1.9B
Strategy	$101B	$0	$0	0.0% (0.0% strict, 38.5% loose)	$0

Part III · industries

Industry by industry

A task gets the same verdict wherever it is done; industries differ in whom they employ. Each industry's figure is its own staffing times the share of each role's work that passes.

Industry (code)	Payroll, all workers	Passes the screen	All three pass	Share of payroll	Waits only on a check
Computer Systems Design and Related Services 541500	$298B	$18.8B	$5.4B	6.3%	$38.1B
Management of Companies and Enterprises 551100	$319B	$17.6B	$6.8B	5.5%	$30.6B
Credit Intermediation and Related Activities (5221 and 5223 only) 5220A1	$173B	$13.7B	$5.0B	7.9%	$21.9B
Colleges, Universities, and Professional Schools 611300	$260B	$13.0B	$8.3B	5.0%	$42.6B
Agencies, Brokerages, and Other Insurance Related Activities 524200	$114B	$12.7B	$2.8B	11.1%	$13.8B
Local Government, excluding Schools and Hospitals (OEWS Designation) 999300	$392B	$10.5B	$4.9B	2.7%	$25.8B
Management, Scientific, and Technical Consulting Services 541600	$192B	$9.5B	$4.0B	4.9%	$18.6B
Accounting, Tax Preparation, Bookkeeping, and Payroll Services 541200	$97.4B	$9.3B	$2.4B	9.5%	$7.0B
General Medical and Surgical Hospitals 622100	$516B	$8.9B	$2.8B	1.7%	$20.9B
Insurance Carriers 524100	$116B	$8.9B	$2.5B	7.7%	$13.0B
Federal Executive Branch (OEWS Designation) 999100	$238B	$8.8B	$3.4B	3.7%	$15.9B
Employment Services 561300	$191B	$8.3B	$3.3B	4.3%	$15.5B
Elementary and Secondary Schools 611100	$547B	$8.0B	$2.4B	1.5%	$73.3B
Real Estate 531000	$120B	$7.2B	$2.6B	6.0%	$10.6B
General Merchandise Retailers 455000	$126B	$7.0B	$3.6B	5.5%	$13.0B

Every other industry, by payroll that passes

Industry (code)	Payroll, all workers	Passes the screen	All three pass	Share of payroll	Waits only on a check
Offices of Physicians 621100	$299B	$6.5B	$2.4B	2.2%	$14.6B
Software Publishers 513200	$87.4B	$5.9B	$1.9B	6.8%	$11.8B
Food and Beverage Retailers (4451 and 4452 only) 4450A1	$117B	$5.8B	$2.8B	5.0%	$8.2B
Merchant Wholesalers, Durable Goods (4232, 4233, 4235, 4236, 4237, and 4239 only) 4230A1	$104B	$5.6B	$2.1B	5.4%	$8.3B
Architectural, Engineering, and Related Services 541300	$167B	$5.5B	$1.5B	3.3%	$20.0B
Automobile Dealers 441100	$83.2B	$5.4B	$2.9B	6.4%	$7.8B
Securities, Commodity Contracts, and Other Financial Investments and Related Activities 523000	$161B	$5.2B	$2.1B	3.2%	$23.9B
Business Support Services 561400	$35.0B	$5.1B	$3.1B	14.6%	$9.1B
State Government, excluding Schools and Hospitals (OEWS Designation) 999200	$158B	$5.1B	$2.0B	3.2%	$13.6B
Professional and Commercial Equipment and Supplies Merchant Wholesalers 423400	$69.1B	$4.7B	$2.1B	6.9%	$9.2B
Traveler Accommodation 721100	$86.2B	$4.4B	$2.3B	5.1%	$5.0B
Scientific Research and Development Services 541700	$123B	$4.4B	$1.4B	3.6%	$10.3B
Building Equipment Contractors 238200	$177B	$4.4B	$1.8B	2.5%	$6.9B
Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services 518200	$58.7B	$4.2B	$1.5B	7.1%	$7.7B
Nondepository Credit Intermediation 522200	$43.8B	$3.5B	$1.2B	8.1%	$5.6B
Telecommunications 517000	$54.7B	$3.5B	$1.7B	6.5%	$6.4B
Office Administrative Services 561100	$49.0B	$3.4B	$1.5B	7.0%	$5.2B
Building Material and Supplies Dealers 444100	$54.3B	$3.4B	$1.6B	6.3%	$6.6B
Other Professional, Scientific, and Technical Services 541900	$61.1B	$3.1B	$1.4B	5.0%	$6.4B
Machinery, Equipment, and Supplies Merchant Wholesalers 423800	$54.9B	$3.0B	$1.2B	5.4%	$4.1B
Health and Personal Care Retailers 456100	$59.7B	$2.8B	$1.5B	4.7%	$6.1B
Wholesale Trade Agents and Brokers 425100	$40.6B	$2.7B	$930M	6.6%	$4.3B
Junior Colleges 611200	$49.5B	$2.6B	$1.7B	5.2%	$7.4B
Gasoline Stations 457100	$33.1B	$2.5B	$1.1B	7.7%	$3.2B
Merchant Wholesalers, Nondurable Goods (4244 and 4248 only) 4240A1	$63.8B	$2.5B	$894M	4.0%	$3.7B
Merchant Wholesalers, Nondurable Goods (4242 and 4246 only) 4240A2	$39.6B	$2.4B	$1.1B	6.1%	$4.7B
Residential Building Construction 236100	$63.8B	$2.4B	$900M	3.7%	$3.8B
Legal Services 541100	$138B	$2.3B	$1.0B	1.7%	$12.9B
Warehousing and Storage 493100	$102B	$2.3B	$1.4B	2.2%	$3.3B
Truck Transportation 484000	$91.6B	$2.3B	$933M	2.5%	$3.1B
Clothing and Clothing Accessories Retailers 458100	$33.8B	$2.2B	$1.1B	6.5%	$5.0B
Individual and Family Services 624100	$135B	$2.1B	$843M	1.6%	$5.8B
Advertising, Public Relations, and Related Services 541800	$44.2B	$2.1B	$1.0B	4.8%	$7.3B
Aerospace Product and Parts Manufacturing 336400	$56.1B	$2.1B	$699M	3.7%	$3.6B
Restaurants and Other Eating Places 722500	$401B	$2.0B	$985M	0.5%	$3.3B
Freight Transportation Arrangement 488500	$17.7B	$2.0B	$1.3B	11.5%	$2.0B
Nonresidential Building Construction 236200	$74.7B	$2.0B	$474M	2.7%	$3.4B
Web Search Portals, Libraries, Archives, and Other Information Services 519200	$28.5B	$2.0B	$593M	6.9%	$3.5B
Postal Service (Federal Government) 491100	$40.4B	$1.9B	$263M	4.8%	$2.2B
Outpatient Care Centers 621400	$93.9B	$1.9B	$670M	2.1%	$5.7B
Offices of Dentists 621200	$79.9B	$1.9B	$682M	2.4%	$6.4B
Offices of Other Health Practitioners 621300	$84.5B	$1.9B	$765M	2.3%	$7.0B
Merchant Wholesalers, Nondurable Goods (4241, 4247, and 4249 only) 4240A3	$33.9B	$1.8B	$693M	5.2%	$2.6B
Navigational, Measuring, Electromedical, and Control Instruments Manufacturing 334500	$40.8B	$1.7B	$545M	4.3%	$3.4B
Services to Buildings and Dwellings 561700	$98.3B	$1.7B	$945M	1.8%	$3.0B
Automotive Parts, Accessories, and Tire Retailers 441300	$24.6B	$1.7B	$1.0B	6.9%	$1.3B
Business, Professional, Labor, Political, and Similar Organizations 813900	$36.7B	$1.7B	$818M	4.6%	$4.1B
Automotive Repair and Maintenance 811100	$53.4B	$1.7B	$842M	3.1%	$2.3B
Machinery Manufacturing (3331, 3332, 3334, and 3339 only) 3330A1	$54.0B	$1.6B	$636M	3.0%	$3.0B
Other Amusement and Recreation Industries 713900	$62.3B	$1.6B	$754M	2.6%	$3.6B
Media Streaming Distribution Services, Social Networks, and Other Media Networks and Content Providers 516200	$29.4B	$1.6B	$508M	5.5%	$4.9B
Electronics and Appliance Retailers 449200	$18.8B	$1.6B	$957M	8.5%	$2.5B
Newspaper, Periodical, Book, and Directory Publishers 513100	$23.3B	$1.5B	$566M	6.2%	$4.2B
Travel Arrangement and Reservation Services 561500	$12.5B	$1.5B	$1.1B	11.7%	$2.0B
Semiconductor and Other Electronic Component Manufacturing 334400	$37.2B	$1.4B	$530M	3.8%	$2.6B
Other Miscellaneous Retailers 459900	$23.0B	$1.4B	$688M	6.0%	$2.7B
Home Health Care Services 621600	$86.2B	$1.4B	$522M	1.6%	$2.6B
Rental and Leasing Services (5322, 5323, and 5324 only) 5320A1	$20.6B	$1.3B	$699M	6.5%	$1.8B
Electric Power Generation, Transmission and Distribution 221100	$43.0B	$1.3B	$607M	3.1%	$2.7B
Other Schools and Instruction 611600	$27.9B	$1.3B	$864M	4.8%	$4.9B
Sporting Goods, Hobby, and Musical Instrument Retailers 459100	$19.9B	$1.3B	$704M	6.6%	$2.9B
Foundation, Structure, and Building Exterior Contractors 238100	$64.4B	$1.3B	$540M	2.0%	$2.0B
Scheduled Air Transportation 481100	$53.1B	$1.3B	$1.0B	2.4%	$1.8B
Furniture and Home Furnishings Retailers 449100	$20.3B	$1.2B	$616M	6.2%	$2.6B
Motion Picture and Video Industries 512100	$28.9B	$1.2B	$485M	4.2%	$4.4B
Motor Vehicle and Motor Vehicle Parts and Supplies Merchant Wholesalers 423100	$22.8B	$1.2B	$584M	5.3%	$1.5B
Building Finishing Contractors 238300	$52.3B	$1.2B	$512M	2.3%	$1.8B
Nursing Care Facilities (Skilled Nursing Facilities) 623100	$77.4B	$1.2B	$291M	1.5%	$1.8B
Automotive Equipment Rental and Leasing 532100	$10.0B	$1.1B	$645M	11.0%	$1.2B
Miscellaneous Store Retailers (4594 and 4595 only) 4590A1	$15.6B	$1.0B	$509M	6.4%	$2.0B
Other Specialty Trade Contractors 238900	$50.6B	$985M	$409M	1.9%	$1.5B
Medical and Diagnostic Laboratories 621500	$23.9B	$977M	$373M	4.1%	$1.6B
Investigation and Security Services 561600	$47.3B	$912M	$435M	1.9%	$1.4B
Pharmaceutical and Medicine Manufacturing 325400	$30.5B	$906M	$364M	3.0%	$1.8B
Printing and Related Support Activities 323100	$19.8B	$893M	$393M	4.5%	$1.7B
Medical Equipment and Supplies Manufacturing 339100	$24.0B	$801M	$341M	3.3%	$1.6B
Other Support Services 561900	$17.3B	$783M	$391M	4.5%	$1.4B
Fabricated Metal Product Manufacturing (3321, 3322, 3325, 3326, and 3329 only) 3320A1	$29.2B	$776M	$325M	2.7%	$1.3B
Plastics Product Manufacturing 326100	$33.3B	$764M	$338M	2.3%	$1.2B
Educational Support Services 611700	$18.5B	$751M	$323M	4.0%	$2.2B
Fabricated Metal Product Manufacturing (3323 and 3324 only) 3320A2	$31.0B	$731M	$277M	2.4%	$1.2B
Chemical Manufacturing (3251, 3252, 3253, and 3259 only) 3250A1	$30.3B	$715M	$287M	2.4%	$1.4B
Computer and Peripheral Equipment Manufacturing 334100	$12.6B	$701M	$245M	5.6%	$1.3B
Utility System Construction 237100	$42.5B	$697M	$241M	1.6%	$1.1B
Apparel, Piece Goods, and Notions Merchant Wholesalers 424300	$10.7B	$657M	$297M	6.1%	$966M
Other Ambulatory Health Care Services 621900	$21.6B	$637M	$298M	2.9%	$1.6B
Motor Vehicle Parts Manufacturing 336300	$32.5B	$631M	$266M	1.9%	$1.1B
Other Miscellaneous Manufacturing 339900	$17.2B	$626M	$248M	3.6%	$1.3B
Couriers and Express Delivery Services 492100	$52.5B	$615M	$370M	1.2%	$1.1B
Other Motor Vehicle Dealers 441200	$9.9B	$611M	$334M	6.2%	$864M
Personal Care Services 812100	$32.6B	$591M	$182M	1.8%	$1.6B
Social Advocacy Organizations 813300	$17.8B	$589M	$240M	3.3%	$2.0B
Nonmetallic Mineral Product Manufacturing 327000	$24.9B	$585M	$216M	2.3%	$881M
Continuing Care Retirement Communities and Assisted Living Facilities for the Elderly 623300	$43.3B	$567M	$198M	1.3%	$1.2B
Grantmaking and Giving Services 813200	$15.1B	$561M	$227M	3.7%	$1.9B
Support Activities for Air Transportation 488100	$16.0B	$539M	$332M	3.4%	$770M
Religious Organizations 813100	$10.8B	$524M	$153M	4.9%	$1.3B
Specialized Design Services 541400	$11.9B	$518M	$173M	4.4%	$2.8B
Commercial and Industrial Machinery and Equipment (except Automotive and Electronic) Repair and Maintenance 811300	$15.8B	$503M	$244M	3.2%	$709M
Drycleaning and Laundry Services 812300	$10.6B	$501M	$294M	4.7%	$602M
Civic and Social Organizations 813400	$15.4B	$500M	$235M	3.2%	$1.3B
Technical and Trade Schools 611500	$10.3B	$488M	$320M	4.7%	$1.2B
Radio and Television Broadcasting Stations 516100	$9.2B	$475M	$174M	5.2%	$2.3B
Beer, Wine, and Liquor Retailers 445300	$6.6B	$468M	$207M	7.0%	$686M
Furniture and Related Product Manufacturing (3371 and 3372 only) 3370A1	$17.4B	$464M	$191M	2.7%	$767M
Natural Gas Distribution 221200	$11.6B	$454M	$218M	3.9%	$919M
Lawn and Garden Equipment and Supplies Retailers 444200	$7.9B	$448M	$230M	5.7%	$840M
Specialty (except Psychiatric and Substance Abuse) Hospitals 622300	$26.3B	$448M	$154M	1.7%	$1.2B
Special Food Services 722300	$32.7B	$446M	$224M	1.4%	$756M
Machine Shops; Turned Product; and Screw, Nut, and Bolt Manufacturing 332700	$20.4B	$444M	$188M	2.2%	$705M
Gambling Industries 713200	$11.0B	$439M	$171M	4.0%	$559M
Residential Intellectual and Developmental Disability, Mental Health, and Substance Abuse Facilities 623200	$30.8B	$425M	$168M	1.4%	$1.3B
Beverage Manufacturing 312100	$17.9B	$422M	$144M	2.4%	$738M
Highway, Street, and Bridge Construction 237300	$28.4B	$422M	$125M	1.5%	$614M
Shoe Retailers 458200	$6.4B	$421M	$219M	6.5%	$964M
Converted Paper Product Manufacturing 322200	$16.6B	$421M	$186M	2.5%	$673M
Community Food and Housing, and Emergency and Other Relief Services 624200	$12.4B	$418M	$150M	3.4%	$1.1B
Motor Vehicle Manufacturing 336100	$21.9B	$414M	$136M	1.9%	$762M
Other Personal Services 812900	$17.2B	$406M	$188M	2.4%	$847M
Jewelry, Luggage, and Leather Goods Retailers 458300	$6.8B	$393M	$211M	5.8%	$865M
Communications Equipment Manufacturing 334200	$8.1B	$388M	$121M	4.8%	$735M
Chemical Manufacturing (3255 and 3256 only) 3250A2	$11.9B	$387M	$170M	3.3%	$611M
Business Schools and Computer and Management Training 611400	$6.5B	$369M	$195M	5.7%	$1.0B
Other Food Manufacturing 311900	$14.7B	$366M	$152M	2.5%	$488M
Support Activities for Mining 213100	$19.0B	$366M	$156M	1.9%	$579M
Vocational Rehabilitation Services 624300	$12.4B	$362M	$162M	2.9%	$983M
Museums, Historical Sites, and Similar Institutions 712100	$9.5B	$361M	$164M	3.8%	$972M
Child Care Services 624400	$37.9B	$350M	$152M	0.9%	$2.4B
Promoters of Performing Arts, Sports, and Similar Events 711300	$10.5B	$339M	$155M	3.2%	$872M
Bakeries and Tortilla Manufacturing 311800	$15.7B	$328M	$149M	2.1%	$487M
Remediation and Other Waste Management Services 562900	$11.3B	$323M	$140M	2.8%	$564M
Animal Slaughtering and Processing 311600	$25.4B	$308M	$135M	1.2%	$419M
Waste Collection 562100	$12.0B	$305M	$154M	2.5%	$497M
Electrical Equipment Manufacturing 335300	$10.4B	$293M	$113M	2.8%	$541M
Other Electrical Equipment and Component Manufacturing 335900	$11.1B	$289M	$117M	2.6%	$516M
Oil and Gas Extraction 211100	$12.8B	$281M	$80.4M	2.2%	$567M
Electronic and Precision Equipment Repair and Maintenance 811200	$6.5B	$278M	$134M	4.3%	$521M
Spectator Sports 711200	$13.5B	$277M	$134M	2.1%	$798M
Death Care Services 812200	$6.7B	$265M	$99.8M	4.0%	$435M
Performing Arts Companies 711100	$6.6B	$256M	$126M	3.9%	$910M
Commercial and Service Industry Machinery Manufacturing 333300	$6.6B	$252M	$88.8M	3.8%	$459M
Psychiatric and Substance Abuse Hospitals 622200	$16.7B	$251M	$95.9M	1.5%	$851M
Metalworking Machinery Manufacturing 333500	$10.1B	$246M	$93.5M	2.4%	$457M
Other Wood Product Manufacturing 321900	$12.1B	$246M	$102M	2.0%	$378M
Facilities Support Services 561200	$10.9B	$244M	$105M	2.2%	$463M
Taxi and Limousine Service 485300	$4.9B	$234M	$51.6M	4.8%	$383M
Ship and Boat Building 336600	$9.4B	$213M	$85.6M	2.3%	$326M
Petroleum and Coal Products Manufacturing 324100	$10.5B	$205M	$74.3M	1.9%	$387M
Support Activities for Road Transportation 488400	$5.9B	$204M	$68.1M	3.4%	$298M
Dairy Product Manufacturing 311500	$9.4B	$202M	$84.6M	2.2%	$252M
Rail Transportation 482100	$15.0B	$201M	$86.2M	1.3%	$346M
Support Activities for Crop Production 115100	$14.8B	$198M	$91.0M	1.3%	$269M
Amusement Parks and Arcades 713100	$8.0B	$193M	$87.7M	2.4%	$450M
Agents and Managers for Artists, Athletes, Entertainers, and Other Public Figures 711400	$3.5B	$191M	$36.6M	5.5%	$177M
Fruit and Vegetable Preserving and Specialty Food Manufacturing 311400	$9.0B	$189M	$76.1M	2.1%	$235M
Fuel Dealers 457200	$4.2B	$181M	$96.3M	4.3%	$308M
Engine, Turbine, and Power Transmission Equipment Manufacturing 333600	$7.2B	$180M	$65.2M	2.5%	$397M
Motor Vehicle Body and Trailer Manufacturing 336200	$9.0B	$176M	$80.7M	2.0%	$283M
Deep Sea, Coastal, and Great Lakes Water Transportation 483100	$3.1B	$176M	$109M	5.6%	$275M
Other Heavy and Civil Engineering Construction 237900	$9.8B	$175M	$56.1M	1.8%	$285M
Support Activities for Water Transportation 488300	$7.5B	$174M	$95.4M	2.3%	$244M
Waste Treatment and Disposal 562200	$6.6B	$174M	$86.2M	2.6%	$288M
Other Transit and Ground Passenger Transportation 485900	$4.7B	$154M	$61.8M	3.3%	$248M
Farm Product Raw Material Merchant Wholesalers 424500	$3.8B	$153M	$61.9M	4.0%	$213M
Rubber Product Manufacturing 326200	$7.9B	$153M	$66.1M	1.9%	$246M
Other Residential Care Facilities 623900	$7.6B	$151M	$50.1M	2.0%	$382M
Book Retailers and News Dealers 459200	$2.4B	$150M	$78.4M	6.3%	$328M
Water, Sewage and Other Systems 221300	$4.0B	$149M	$77.0M	3.7%	$239M
Coating, Engraving, Heat Treating, and Allied Activities 332800	$6.9B	$147M	$67.6M	2.1%	$229M
RV (Recreational Vehicle) Parks and Recreational Camps 721200	$2.9B	$141M	$72.8M	4.9%	$237M
Nonscheduled Air Transportation 481200	$6.1B	$139M	$68.5M	2.3%	$189M
Personal and Household Goods Repair and Maintenance 811400	$4.2B	$131M	$66.1M	3.1%	$219M
Land Subdivision 237200	$3.0B	$130M	$38.4M	4.3%	$206M
Independent Artists, Writers, and Performers 711500	$3.1B	$126M	$52.8M	4.1%	$628M
Local Messengers and Local Delivery 492200	$8.5B	$123M	$55.2M	1.5%	$262M
Nonmetallic Mineral Mining and Quarrying 212300	$6.5B	$122M	$50.0M	1.9%	$175M
Animal Food Manufacturing 311100	$4.3B	$122M	$51.6M	2.8%	$176M
Cut and Sew Apparel Manufacturing 315200	$3.2B	$122M	$54.8M	3.8%	$239M
Lessors of Nonfinancial Intangible Assets (except Copyrighted Works) 533100	$2.1B	$115M	$57.9M	5.5%	$217M
Sound Recording Industries 512200	$1.8B	$115M	$53.6M	6.5%	$226M
Sugar and Confectionery Product Manufacturing 311300	$4.3B	$109M	$49.1M	2.6%	$172M
Grain and Oilseed Milling 311200	$4.5B	$106M	$42.2M	2.3%	$147M
Electric Lighting Equipment Manufacturing 335100	$2.4B	$97.9M	$40.3M	4.1%	$159M
Other Support Activities for Transportation 488900	$1.9B	$97.8M	$51.0M	5.3%	$142M
Pulp, Paper, and Paperboard Mills 322100	$6.4B	$95.2M	$42.6M	1.5%	$161M
Iron and Steel Mills and Ferroalloy Manufacturing 331100	$5.5B	$91.7M	$32.0M	1.7%	$121M
Foundries 331500	$6.1B	$90.4M	$38.2M	1.5%	$157M
Monetary Authorities-Central Bank 521100	$2.5B	$88.9M	$28.4M	3.5%	$240M
Drinking Places (Alcoholic Beverages) 722400	$16.0B	$88.5M	$39.0M	0.6%	$133M
Nonferrous Metal (except Aluminum) Production and Processing 331400	$4.0B	$86.3M	$35.2M	2.1%	$134M
Other Textile Product Mills 314900	$2.7B	$84.1M	$37.3M	3.1%	$149M
Steel Product Manufacturing from Purchased Steel 331200	$3.6B	$79.9M	$33.2M	2.2%	$115M
Other Investment Pools and Funds 525900	$2.3B	$76.2M	$28.0M	3.3%	$202M
School and Employee Bus Transportation 485400	$10.1B	$74.5M	$21.4M	0.7%	$129M
Household Appliance Manufacturing 335200	$2.8B	$72.0M	$33.1M	2.5%	$121M
Insurance and Employee Benefit Funds 525100	$1.1B	$70.9M	$25.6M	6.7%	$131M
Sawmills and Wood Preservation 321100	$4.6B	$68.8M	$28.3M	1.5%	$98.2M
Florists 459300	$2.3B	$68.7M	$34.2M	3.0%	$395M
Audio and Video Equipment Manufacturing 334300	$1.4B	$66.7M	$24.1M	4.9%	$125M
Alumina and Aluminum Production and Processing 331300	$3.6B	$66.5M	$27.5M	1.8%	$100M
Veneer, Plywood, and Engineered Wood Product Manufacturing 321200	$4.2B	$64.5M	$25.4M	1.6%	$144M
Other Transportation Equipment Manufacturing 336900	$2.2B	$55.6M	$22.6M	2.5%	$107M
Manufacturing and Reproducing Magnetic and Optical Media 334600	$886M	$53.3M	$20.4M	6.0%	$106M
Textile Furnishings Mills 314100	$1.7B	$49.2M	$23.3M	2.9%	$72.2M
Metal Ore Mining 212200	$3.4B	$45.9M	$14.2M	1.3%	$100M
Urban Transit Systems 485100	$2.4B	$45.2M	$18.8M	1.9%	$73.8M
Fabric Mills 313200	$2.1B	$42.4M	$18.9M	2.0%	$62.2M
Support Activities for Rail Transportation 488200	$2.2B	$40.9M	$20.6M	1.9%	$60.1M
Other Furniture Related Product Manufacturing 337900	$1.3B	$38.9M	$18.2M	3.1%	$54.4M
Pipeline Transportation of Natural Gas 486200	$3.2B	$38.1M	$11.7M	1.2%	$89.8M
Railroad Rolling Stock Manufacturing 336500	$1.5B	$35.8M	$12.5M	2.4%	$70.3M
Textile and Fabric Finishing and Fabric Coating Mills 313300	$1.0B	$32.8M	$14.1M	3.2%	$54.0M
Logging 113300	$2.3B	$32.8M	$13.7M	1.4%	$51.8M
Support Activities for Animal Production 115200	$1.4B	$28.5M	$12.6M	2.0%	$49.3M
Seafood Product Preparation and Packaging 311700	$1.3B	$28.4M	$11.9M	2.2%	$33.7M
Scenic and Sightseeing Transportation, Water 487200	$846M	$27.4M	$15.8M	3.2%	$40.5M
Charter Bus Industry 485500	$1.2B	$24.0M	$11.0M	2.0%	$39.8M
Pipeline Transportation of Crude Oil 486100	$1.2B	$22.0M	$8.4M	1.8%	$43.2M
Inland Water Transportation 483200	$2.3B	$22.0M	$8.9M	0.9%	$37.6M
Coal Mining 212100	$3.3B	$19.8M	$7.2M	0.6%	$37.3M
Interurban and Rural Bus Transportation 485200	$672M	$19.1M	$9.4M	2.8%	$28.4M
Tobacco Manufacturing 312200	$655M	$18.1M	$7.2M	2.8%	$22.0M
Apparel Accessories and Other Apparel Manufacturing 315900	$477M	$16.3M	$7.2M	3.4%	$28.2M
Scenic and Sightseeing Transportation, Land 487100	$474M	$15.1M	$9.3M	3.2%	$24.5M
Rooming and Boarding Houses, Dormitories, and Workers' Camps 721300	$373M	$13.8M	$7.2M	3.7%	$17.3M
Other Leather and Allied Product Manufacturing 316900	$423M	$11.8M	$5.7M	2.8%	$19.6M
Fiber, Yarn, and Thread Mills 313100	$824M	$7.3M	$3.8M	0.9%	$13.5M
Other Pipeline Transportation 486900	$607M	$6.9M	$2.3M	1.1%	$12.8M
Footwear Manufacturing 316200	$381M	$6.8M	$3.5M	1.8%	$11.9M
Apparel Knitting Mills 315100	$229M	$5.1M	$2.2M	2.2%	$8.2M
Scenic and Sightseeing Transportation, Other 487900	$126M	$4.5M	$3.1M	3.5%	$7.1M
Leather and Hide Tanning and Finishing 316100	$118M	$1.3M	$687k	1.1%	$3.6M

Part III · where a rollup could start

Industries with work to hand over and firms to buy

A rollup buys many small, owner-run firms in one industry and runs them together. It needs two things: work that passes the hand-over screen, so owning the firms keeps a saving, and many firms small enough to buy. This list ranks industries by the share of their payroll that passes, times the share of their employment in firms under five hundred staff (from the Census Bureau's Statistics of US Businesses). It leaves out industries with fewer than two hundred firms of twenty to ninety-nine staff, or under half a billion dollars of payroll passing, and nonprofits and government, which are not for sale. It is a place to start looking, not a verdict: it does not say whether an owner keeps the saving or passes it to customers.

The list is the industries where work that passes and small firms to buy multiply highestChart

Share of payroll that passes the screen

Share of employment in small firms

1Freight Transportation Arrangement
2Agencies, Brokerages, and Other Insurance Related Activities
3Accounting, Tax Preparation, Bookkeeping, and Payroll Services
4Other Motor Vehicle Dealers
5Automobile Dealers
6Travel Arrangement and Reservation Services
7Apparel, Piece Goods, and Notions Merchant Wholesalers
8Other Schools and Instruction
9Business Support Services
10Office Administrative Services
11Specialized Design Services
12Wholesale Trade Agents and Brokers
13Machinery, Equipment, and Supplies Merchant Wholesalers
14Residential Building Construction
15Other Professional, Scientific, and Technical Services
16Computer Systems Design and Related Services
17Drycleaning and Laundry Services
18Printing and Related Support Activities
19Management, Scientific, and Technical Consulting Services
20Professional and Commercial Equipment and Supplies Merchant Wholesalers

on the list, with its rankconsidered, below the cutthe cut: the score of the last industry listed

Up is more of an industry's payroll passing; right is more of its employment in small firms; the score is one times the other, so the dashed line is where the list ends. The share that passes is from the census (v3.2-2026-09-24), a screen scored by three AI models, not a claim about what AI can do and not a forecast; the part all three models pass is in the table beneath. Employment in small firms is from the Census Bureau's Statistics of US Businesses. Only industries that clear the list's floors are drawn. Industries just either side of the line are not meaningfully apart.

The numbers, and every industry drawn

#	Industry	Payroll that passes	Employment in firms under 500 staff	Firms with 20–99 staff
1	Freight Transportation Arrangement 488500	11.5% ($2.0B)	50.9%	1433
2	Agencies, Brokerages, and Other Insurance Related Activities 524200	11.1% ($12.7B)	46.7%	3364
3	Accounting, Tax Preparation, Bookkeeping, and Payroll Services 541200	9.5% ($9.3B)	52.4%	4395
4	Other Motor Vehicle Dealers 441200	6.2% ($611M)	78.7%	1850
5	Automobile Dealers 441100	6.4% ($5.4B)	75.0%	9810
6	Travel Arrangement and Reservation Services 561500	11.7% ($1.5B)	40.3%	505
7	Apparel, Piece Goods, and Notions Merchant Wholesalers 424300	6.1% ($657M)	71.2%	1149
8	Other Schools and Instruction 611600	4.8% ($1.3B)	90.9%	4772
9	Business Support Services 561400	14.6% ($5.1B)	29.1%	1761
10	Office Administrative Services 561100	7.0% ($3.4B)	58.4%	3357
11	Specialized Design Services 541400	4.4% ($518M)	92.1%	734
12	Wholesale Trade Agents and Brokers 425100	6.6% ($2.7B)	59.4%	1356
13	Machinery, Equipment, and Supplies Merchant Wholesalers 423800	5.4% ($3.0B)	64.0%	5054
14	Residential Building Construction 236100	3.7% ($2.4B)	90.4%	5259
15	Other Professional, Scientific, and Technical Services 541900	5.0% ($3.1B)	63.0%	5149
16	Computer Systems Design and Related Services 541500	6.3% ($18.8B)	48.2%	8209
17	Drycleaning and Laundry Services 812300	4.7% ($501M)	64.0%	1148
18	Printing and Related Support Activities 323100	4.5% ($893M)	66.8%	2445
19	Management, Scientific, and Technical Consulting Services 541600	4.9% ($9.5B)	57.7%	7232
20	Professional and Commercial Equipment and Supplies Merchant Wholesalers 423400	6.9% ($4.7B)	39.1%	2415

Rank by score	Industry	Share of payroll that passes	Passes the screen	All three models pass	Employment in small firms
1	Freight Transportation Arrangement	11.5%	$2.0B	$1.3B	50.9%
2	Agencies, Brokerages, and Other Insurance Related Activities	11.1%	$12.7B	$2.8B	46.7%
3	Accounting, Tax Preparation, Bookkeeping, and Payroll Services	9.5%	$9.3B	$2.4B	52.4%
4	Other Motor Vehicle Dealers	6.2%	$611M	$334M	78.7%
5	Automobile Dealers	6.4%	$5.4B	$2.9B	75.0%
6	Travel Arrangement and Reservation Services	11.7%	$1.5B	$1.1B	40.3%
7	Apparel, Piece Goods, and Notions Merchant Wholesalers	6.1%	$657M	$297M	71.2%
8	Other Schools and Instruction	4.8%	$1.3B	$864M	90.9%
9	Business Support Services	14.6%	$5.1B	$3.1B	29.1%
10	Office Administrative Services	7.0%	$3.4B	$1.5B	58.4%
11	Specialized Design Services	4.4%	$518M	$173M	92.1%
12	Wholesale Trade Agents and Brokers	6.6%	$2.7B	$930M	59.4%
13	Machinery, Equipment, and Supplies Merchant Wholesalers	5.4%	$3.0B	$1.2B	64.0%
14	Residential Building Construction	3.7%	$2.4B	$900M	90.4%
15	Other Professional, Scientific, and Technical Services	5.0%	$3.1B	$1.4B	63.0%
16	Computer Systems Design and Related Services	6.3%	$18.8B	$5.4B	48.2%
17	Drycleaning and Laundry Services	4.7%	$501M	$294M	64.0%
18	Printing and Related Support Activities	4.5%	$893M	$393M	66.8%
19	Management, Scientific, and Technical Consulting Services	4.9%	$9.5B	$4.0B	57.7%
20	Professional and Commercial Equipment and Supplies Merchant Wholesalers	6.9%	$4.7B	$2.1B	39.1%
21	Advertising, Public Relations, and Related Services	4.8%	$2.1B	$1.0B	55.5%
22	Automotive Repair and Maintenance	3.1%	$1.7B	$842M	80.6%
23	Other Miscellaneous Manufacturing	3.6%	$626M	$248M	67.4%
24	Motor Vehicle and Motor Vehicle Parts and Supplies Merchant Wholesalers	5.3%	$1.2B	$584M	46.6%
25	Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services	7.1%	$4.2B	$1.5B	34.3%
26	Educational Support Services	4.0%	$751M	$323M	59.8%
27	Automotive Parts, Accessories, and Tire Retailers	6.9%	$1.7B	$1.0B	34.2%
28	Other Support Services	4.5%	$783M	$391M	52.0%
29	Traveler Accommodation	5.1%	$4.4B	$2.3B	45.1%
30	Offices of Dentists	2.4%	$1.9B	$682M	91.9%
31	Automotive Equipment Rental and Leasing	11.0%	$1.1B	$645M	19.7%
32	Commercial and Industrial Machinery and Equipment (except Automotive and Electronic) Repair and Maintenance	3.2%	$503M	$244M	65.7%
33	Building Finishing Contractors	2.3%	$1.2B	$512M	89.2%
34	Nondepository Credit Intermediation	8.1%	$3.5B	$1.2B	25.0%
35	Nonresidential Building Construction	2.7%	$2.0B	$474M	74.8%
36	Building Equipment Contractors	2.5%	$4.4B	$1.8B	79.4%
37	Building Material and Supplies Dealers	6.3%	$3.4B	$1.6B	29.6%
38	Foundation, Structure, and Building Exterior Contractors	2.0%	$1.3B	$540M	90.1%
39	Other Amusement and Recreation Industries	2.6%	$1.6B	$754M	70.1%
40	Architectural, Engineering, and Related Services	3.3%	$5.5B	$1.5B	54.3%
41	Offices of Other Health Practitioners	2.3%	$1.9B	$765M	76.7%
42	Personal Care Services	1.8%	$591M	$182M	95.1%
43	Other Specialty Trade Contractors	1.9%	$985M	$409M	88.0%
44	Motion Picture and Video Industries	4.2%	$1.2B	$485M	40.2%
45	Medical and Diagnostic Laboratories	4.1%	$977M	$373M	38.4%
46	Semiconductor and Other Electronic Component Manufacturing	3.8%	$1.4B	$530M	37.4%
47	Legal Services	1.7%	$2.3B	$1.0B	80.8%
48	Other Ambulatory Health Care Services	2.9%	$637M	$298M	45.5%
49	Elementary and Secondary Schools	1.5%	$8.0B	$2.4B	84.2%
50	Medical Equipment and Supplies Manufacturing	3.3%	$801M	$341M	37.1%
51	Services to Buildings and Dwellings	1.8%	$1.7B	$945M	69.6%
52	Offices of Physicians	2.2%	$6.5B	$2.4B	51.9%
53	Individual and Family Services	1.6%	$2.1B	$843M	68.0%
54	Scientific Research and Development Services	3.6%	$4.4B	$1.4B	30.1%
55	Navigational, Measuring, Electromedical, and Control Instruments Manufacturing	4.3%	$1.7B	$545M	25.0%
56	Plastics Product Manufacturing	2.3%	$764M	$338M	40.5%
57	Support Activities for Air Transportation	3.4%	$539M	$332M	26.8%
58	Outpatient Care Centers	2.1%	$1.9B	$670M	42.7%
59	Utility System Construction	1.6%	$697M	$241M	51.4%
60	Nursing Care Facilities (Skilled Nursing Facilities)	1.5%	$1.2B	$291M	51.0%
61	Home Health Care Services	1.6%	$1.4B	$522M	47.7%
62	Pharmaceutical and Medicine Manufacturing	3.0%	$906M	$364M	25.2%
63	Investigation and Security Services	1.9%	$912M	$435M	36.0%
64	Continuing Care Retirement Communities and Assisted Living Facilities for the Elderly	1.3%	$567M	$198M	50.6%
65	Management of Companies and Enterprises	5.5%	$17.6B	$6.8B	10.3%
66	Employment Services	4.3%	$8.3B	$3.3B	12.5%
67	Electric Power Generation, Transmission and Distribution	3.1%	$1.3B	$607M	17.3%
68	Insurance Carriers	7.7%	$8.9B	$2.5B	6.4%
69	Motor Vehicle Parts Manufacturing	1.9%	$631M	$266M	25.3%
70	Colleges, Universities, and Professional Schools	5.0%	$13.0B	$8.3B	8.7%
71	Aerospace Product and Parts Manufacturing	3.7%	$2.1B	$699M	10.5%
72	Restaurants and Other Eating Places	0.5%	$2.0B	$985M	66.4%
73	Warehousing and Storage	2.2%	$2.3B	$1.4B	11.9%
74	Couriers and Express Delivery Services	1.2%	$615M	$370M	12.6%

Part IV · businesses

Where buyers are, and what the census says

Two tests, and both must pass: does the work pass the screen, and would the owner keep the saving? The census answers the first and reasons about the second from how a business prices its work. This list sets both beside what buyers are doing: the AI roll-ups this site follows, by the trade each works in, and the purchases they have announced. Trades are ordered by how many roll-ups are in them. The roll-ups are counted worldwide and the payroll is American, so each count is split by place. Purchases come from a search of the larger buyers only, so a trade with none listed may still have some.

the saving likely goes to clients

Accounting, tax and audit

Roll-ups followed: 30 (11 in the United States, 19 elsewhere; 4 searched for purchases) · purchases found: 26

Census: $9.3B of American payroll passes the screen (9.5% of the industry's payroll), from the row “Accounting, Tax Preparation, Bookkeeping, and Payroll Services”.

Priced as billable hours. The invoice is denominated in the hours automation removes. Cut the time and you cut the bill — the saving goes to the client unless engagements move to fixed fees first. Reasoned from how the business prices its work, not measured.

The census reasons that this saving goes to clients, yet buyers are active here. Both can hold: a buyer may change how the work is priced, or may be wrong.

Who the buyers are, what they bought, and the census's reasoning

Roll-ups (bold: searched for purchases): Accrual, Alloy River, Crete, Cynch, General Intelligence, Mandrel, Minerva Intelligence, Modus, Multiplier Holdings, Numica, OLarry, Alven, Ancora, Archipel, Asevia, Atlas, Aurion, Bhub, Camada, Ceedar, Contra, Integral, LimeTax, Numbi, Ready Plan Go, Sequence, SimpleAI, Skalar, Skrooge, Taxforce

2024-04-09 · Crete announced as joining its group: Reid (Reid CPAs / Reid Tax & Advisory) (accounting firm, USA)
2025-04-03 · Archipel acquired Groupe Rey (accounting firm, France)
2025-09-08 · Crete announced as joining its group: Mauldin Vaught (accounting firm, USA)
2025-09-10 · Crete announced as joining its group: Adeptus (Adeptus Advisors Group) (accounting firm, USA)
2025-09-16 · Archipel acquired Finantis (audit and accounting firm, France)
2025-09-16 · Archipel acquired P2R (accounting and audit firm, France)
2025-09-18 · Crete announced as joining its group: CWDL (rebranded Advisent) (accounting firm, USA)
2025-10-02 · Archipel acquired Lamy Experts (accounting firm, France)
2025-10-22 · Crete announced as joining its group: TKR Accountants and Advisors (accounting firm, USA)
2025-11-11 · Crete announced as joining its group: Accuity (accounting firm, USA)
2025-11-18 · Crete announced as joining its group: Cross Financial (accounting, advisory and wealth-management firm, USA)
2025-11-19 · Integral acquired cleverlohn (digital payroll and HR services provider, Germany)
2025-11-26 · Bhub acquired Agrocontar (accounting firm (agribusiness), Brazil)
2025-12-12 · Crete announced as joining its group: Legacy Valuations & Forensics (formerly part of Empire Valuation Consultants) (valuation and forensic accounting firm, USA)
2026-01-12 · Crete announced as joining its group: Mellott & Mellott (accounting firm, USA)
2026-01-15 · Archipel acquired AVEI'A (accounting firm, France)
2026-01-15 · Archipel acquired Groupe Conseils (accounting firm, France)
2026-02-05 · Crete announced as joining its group: Bowers Advisors (accounting firm, USA)
2026-03-09 · Crete announced as joining its group: BFJ (accounting firm, USA)
2026-04-08 · Archipel acquired Avexxens (accounting firm, France)
2026-04-08 · Archipel acquired L&A Conseil & Audit (accounting and audit firm, France)
2026-04-08 · Archipel acquired Prospective & Finance (accounting firm, France)
2026-04-13 · Crete announced as joining its group: Think Advisors R&D Tax Credit Services practice (incl. Black Line Group's) (R&D tax-credit consulting practice, USA)
2026-04-16 · Crete announced as joining its group: Maxwell Advisors (formerly Maxwell & Company CPAs) (tax advisory / accounting firm, USA)
2026-06-17 · Crete announced as joining its group: Murphy, Miller & Baglieri (MMB) (accounting firm, USA)
2026-09-01 · Crete announced as joining its group: Preferred CFO (fractional CFO / outsourced finance firm, USA)

The most payroll passing the screen of any professional-services archetype here, sitting on an invoice denominated in the hours automation removes. The false positive a naive screen walks into.

Unless engagement letters move to fixed fees, the saving goes to the client.
Attest work needs a licensed signer; the signer-leverage question decides whether the licence is a moat or a ceiling.
Offshore delivery centres already took much of this arbitrage in the 2010s.

Comparables the census named to illustrate the pricing model: Ascend (Alpine Investors); Springline Advisory (Trinity Hunt); Citrin Cooperman (Blackstone, 2025).

Sources: cpatrendlines.com

Freight brokers and forwarders

Roll-ups followed: 16 (12 in the United States, 4 elsewhere; 4 searched for purchases) · purchases found: 4

Census: $2.0B of American payroll passes the screen (11.5% of the industry's payroll), from the row “Freight Transportation Arrangement”.

Who the buyers are, what they bought

Roll-ups (bold: searched for purchases): Agent Astra, Border Logic, DocUnlock, Freight Hero, Fura, Loop, Nuvocargo, Pando, PayPath, Sedona, TerraFreight, Xhipment, NexDash, Sennder, Stracker, Voyfai

2024-07-30 · Sennder acquired C.H. Robinson European Surface Transportation (EST) operations (road freight broker / forwarder, Europe (seller USA))
2025-04-23 · Nuvocargo acquired Merge Transportation (freight broker / third-party logistics provider, USA)
2025-08-18 · Loop acquired Data2Logistics (freight audit and payment provider, USA)
2025-10-01 · Loop acquired StrategIQ Commerce (shipment visibility and freight/parcel audit provider, USA)

Property and lettings management

Roll-ups followed: 13 (1 in the United States, 12 elsewhere; 3 searched for purchases) · purchases found: 18

Census: no figure. The census has one row for all of real estate, far broader than managing property for owners.

Who the buyers are, what they bought

Roll-ups (bold: searched for purchases): Lightswitch, Arbio Group, Arjile, Buena, Dwelly, Kalea, Levels, MoIn, Onekey, Ralph, Stead, Sebas, Theo

2024-09-25 · Dwelly acquired Lime Property (estate and lettings agency (Hull), UK)
2024-12-06 · Dwelly acquired Gallant Richardson (lettings agency (Colchester, Essex), UK)
2025-05-28 · Dwelly acquired Pennington (lettings agency (Suffolk), UK)
2025-07-03 · Dwelly acquired Parks Residential (lettings agency (Brighton), UK)
2025-07-29 · Dwelly acquired Number 1 Letting (lettings agency (Pontefract, West Yorkshire), UK)
2025-09-02 · Dwelly acquired HC Lettings (formerly HoldenCopley's lettings arm) (lettings agency (Nottingham), UK)
2025-12-03 · Dwelly acquired Sheldon Bosley Knight lettings business (SBK Lettings) (lettings business of an estate agency (Midlands), UK)
2026-01-20 · Dwelly acquired Brookdale Property Management (lettings / property management agency (Peterborough), UK)
2026-03-01 · Beacon Software acquired RealEstateAPI (property data API software, USA)
2026-03-04 · Dwelly acquired Goodwin Property Services (lettings and sales agency (Stamford), UK)
2026-04-14 · Dwelly acquired Eden Harper (lettings agency (south west London), UK)
2026-05-28 · Dwelly acquired Albery Tyson (lettings agency (Midlands), UK)
2026-06-04 · Dwelly acquired Elliot Oliver (lettings agency (Cheltenham), UK)
2026-06-11 · Dwelly acquired Move (estate and lettings agency (Cheltenham), UK)
2026-06-26 · Dwelly acquired Settio (real-estate investment and portfolio advisory / lettings platform, UK)
2026-07-16 · Dwelly acquired AP Morgan lettings business (lettings business of an estate agency (West Midlands), UK)
2026-08-06 · Dwelly acquired Shaws (lettings agency (Kensington, London), UK)
2026-09-01 · Dwelly acquired Paramount Properties (estate and lettings agency (West Hampstead, London), UK)

IT managed services

Roll-ups followed: 10 (6 in the United States, 4 elsewhere; 3 searched for purchases) · purchases found: 8

Census: no figure. The census row for computer systems design holds all of software services, far broader than firms that run other firms' IT.

Who the buyers are, what they bought

Roll-ups (bold: searched for purchases): e4n, Rollout, Shield Technology Partners, Titan, Treeline, Turgon, Aretea Group, Mitigata, Monkinet, Unitas

2025-06-05 · Shield Technology Partners announced as joining its group: ClearFuze Networks (IT managed-service provider, USA)
2025-06-05 · Shield Technology Partners announced as joining its group: Delval Technology Solutions (DTS) (IT managed-service provider, USA)
2025-06-05 · Shield Technology Partners announced as joining its group: IronOrbit (IT managed-service provider, USA)
2025-06-05 · Shield Technology Partners announced as joining its group: OneNet Global (IT managed-service provider, USA)
2025-08-12 · Titan acquired RFA (Richard Fleischman & Associates) (IT managed-service provider for financial firms, USA)
2025-12-15 · Shield Technology Partners announced as joining its group: Option One Technologies (IT managed-service provider, USA)
2026-03-13 · Shield Technology Partners announced as joining its group: Exigent Technologies (IT managed-service provider, USA)
2026-09-24 · Shield Technology Partners announced as joining its group: Nerds Support (IT managed-service provider, USA)

Medical practices and care

Roll-ups followed: 10 (3 in the United States, 7 elsewhere; 1 searched for purchases) · purchases found: 3

Census: $6.5B of American payroll passes the screen (2.2% of the industry's payroll), from the row “Offices of Physicians”. The row is doctors' offices only.

Who the buyers are, what they bought

Roll-ups (bold: searched for purchases): Adaptive Innovations, Meroka, Natoe, 5c Network, Archvale, Avi medical, Jutro Medical, Nila Care, Praxis Eins, Tern

2024-05-27 · Avi medical acquired iATROS (telemedicine company (digital cardiology care), Germany)
2025-10-01 · General Catalyst acquired Summa Health (non-profit hospital system (converted to for-profit), USA)
2026-05-14 · Iterative Health acquired Three cardiology research sites of NextStage Clinical Research (Beaumont, Port Arthur, Waco, Texas) (clinical-research sites, USA)

Insurance agencies and brokers

Roll-ups followed: 9 (6 in the United States, 3 elsewhere; 0 searched for purchases) · purchases found: 1

Census: $12.7B of American payroll passes the screen (11.1% of the industry's payroll), from the row “Agencies, Brokerages, and Other Insurance Related Activities”. The row also holds claims administration, which has its own entry below.

Who the buyers are, what they bought

Roll-ups (bold: searched for purchases): American Growth Insurance, Clarence, Corgi, Covu, Equal Parts, Gyde, Afori, Quartz, Synapze

2026-09-14 · Sequence Holdings acquired The Baldwin Group (insurance broker, USA)

Contact centres

Roll-ups followed: 6 (4 in the United States, 2 elsewhere; 3 searched for purchases) · purchases found: 3

Census: no figure. The census row for business support services is far broader than contact centres.

Who the buyers are, what they bought

Roll-ups (bold: searched for purchases): Crescendo, Kim, OpenCall, UnifyCX, CX-Assist, Onepilot

2024-03-05 · Onepilot acquired Eodom (contact centre (home-based agents), France)
2024-10-02 · Crescendo acquired PartnerHero (customer-support outsourcing (BPO) firm, USA)
2024-11-04 · UnifyCX acquired Collective Solution (contact centre, USA)

Wealth management and fund administration

Roll-ups followed: 6 (2 in the United States, 4 elsewhere; 1 searched for purchases) · purchases found: 0

Census: no figure. The census row for securities and investments mixes too many kinds of firm to stand for this trade.

Who the buyers are, what they bought

Roll-ups (bold: searched for purchases): Aqua, Savvy, Mav9, Palindrome, Thrivo Wealth, Ylookup

No purchase on the record was found for the buyers searched.

Legal services

Roll-ups followed: 5 (4 in the United States, 1 elsewhere; 1 searched for purchases) · purchases found: 1

Census: $2.3B of American payroll passes the screen (1.7% of the industry's payroll), from the row “Legal Services”.

Who the buyers are, what they bought

Roll-ups (bold: searched for purchases): Alix, Ellis, Eudia, ImmCore, Lawhive

2025-09-15 · Lawhive acquired Woodstock Legal Services (law firm (consultancy / fee-share model, conveyancing focus), UK)

check before buying

Claims administration

Roll-ups followed: 3 (2 in the United States, 1 elsewhere; 0 searched for purchases) · purchases found: 0

Census: $3.0B of American payroll passes the screen, from the census's own sizing of this business.

Priced as a fee per claim or per covered life. Priced per claim or per covered life, which hides the cost base — but carriers are sophisticated buyers who re-bid on a cycle and will reprice at renewal. Reasoned from how the business prices its work, not measured.

Who the buyers are, what they bought, and the census's reasoning

Roll-ups (bold: searched for purchases): Ravin, Strala, Inca

No purchase on the record was found for the buyers searched.

Per-claim pricing with a long tail before anyone knows the claim was handled well. The verification horizon that makes the work hard to automate is the same thing that stops a cheaper competitor proving they are as good.

Excludes $82bn of insurance agencies and brokerages, a different business — 72% of the code's payroll.

Does the carrier re-bid on a cycle? Sophisticated counterparties reprice every 2-3 years.
Who owns the claims history — you or the carrier? It is the only ground truth in a long-horizon domain.
Is any pricing per-FTE? That converts every efficiency gain into a revenue cut.

Comparables the census named to illustrate the pricing model: Sedgwick; Crawford & Company; CorVel; Gallagher Bassett.

check before buying

Engineering and architectural services

Roll-ups followed: 3 (2 in the United States, 1 elsewhere; 1 searched for purchases) · purchases found: 0

Census: $5.5B of American payroll passes the screen, from the census's own sizing of this business.

Priced as a percentage of construction value, or billable hours. Percentage-of-construction-value contracts are indexed to the project and keep the saving. Hourly contracts pass it on. The mix inside the firm decides. Reasoned from how the business prices its work, not measured.

Who the buyers are, what they bought, and the census's reasoning

Roll-ups (bold: searched for purchases): Ashler, Gauge Industrial Group, Attentive

No purchase on the record was found for the buyers searched.

Percentage-of-construction-value contracts are indexed to the project, not your time. Hourly contracts in the same firm are not. The mix is the whole question.

What share of the book is percentage-of-value versus time-and-materials?
Does a P.E. stamp gate the deliverable, and can one engineer stamp 10x the volume? If yes the licence is a moat; if liability attaches to the review, it is a ceiling.
Public-sector clients often mandate rate-card disclosure. That kills retention.

check before buying· sized by occupation, not industry

Medical coding and billing

Roll-ups followed: 3 (3 in the United States, 0 elsewhere; 1 searched for purchases) · purchases found: 0

Census: $5.2B of American payroll passes the screen, from the census's own sizing of this business.

Priced as a percentage of collections when outsourced; a cost centre when in-house. Vendors quote outsourced revenue-cycle work as a share of what it collects — 4–10% — which would keep the saving. But the payer on the other side is automating too, so part of any gain may be absorbed by a faster denial-and-appeal loop rather than into margin. Reasoned from how the business prices its work, not measured.

Who the buyers are, what they bought, and the census's reasoning

Roll-ups (bold: searched for purchases): Agent AI, Candid Health, Overdrive Health

No purchase on the record was found for the buyers searched.

Rule-bound, high-volume and checkable against a fee schedule — the textbook candidate for handing work over. The original plan flagged it as the likely false positive for exactly that reason: the same properties made it some of the first work sent offshore.

Payroll is medical records specialists (the occupation that includes coders) wherever they work, plus billing and posting clerks inside healthcare. Most of it sits in hospitals and physician offices rather than outsourcers, so this sizes the work, not the outsourced market. Overlaps the labs card by $0.3bn of payroll ($0.12bn passing the screen) — records and billing staff inside labs — so the two are not additive.

How much of the coding is already done offshore? Where most of it is, the US payroll here overstates what is left to capture.
Payers are automating the other side: UnitedHealth and Cigna face class actions over algorithmic denials, and Medicare began AI-assisted prior authorisation in six states in January 2026. Does faster coding collect more, or just meet faster denials?
Percentage-of-collections contracts keep the saving; per-claim and flat-fee contracts pass it on. What is the mix?

Comparables the census named to illustrate the pricing model: R1 RCM (taken private by TowerBrook and CD&R, $8.9bn, Nov 2024); Waystar (Nasdaq: WAY); Ensemble Health Partners (Apollo-backed investment at ~$12bn, announced June 2026); AGS Health (Blackstone, 2025).

Sources: carecloud.com, www.globenewswire.com, www.courthousenews.com, www.federalregister.gov

Software companies

Roll-ups followed: 3 (0 in the United States, 0 elsewhere, 3 place not stated; 3 searched for purchases) · purchases found: 24

Census: no figure. Not a service trade the census sizes: these buyers purchase software products, not firms that sell hours.

Who the buyers are, what they bought

Roll-ups (bold: searched for purchases): Beacon Software, Bending Spoons, Circeus

2024-01-10 · Bending Spoons acquired Mosaic Group (digital assets, from IAC) (mobile apps (consumer software), USA)
2024-01-24 · Bending Spoons acquired Meetup (events / community platform, USA)
2024-07-01 · Beacon Software acquired Let's Camp (campground booking software, Canada)
2024-07-31 · Bending Spoons acquired WeTransfer (file-transfer service, Netherlands)
2024-11-25 · Bending Spoons acquired Brightcove (video streaming software, USA)
2025-01-22 · Metropolis acquired Oosto (formerly AnyVision) (AI video analytics / facial-recognition software company, Israel)
2025-02-25 · Circeus acquired Aiden (AI guided-selling software, Netherlands)
2025-03-20 · Bending Spoons acquired komoot (route-planning app, Germany)
2025-03-28 · Circeus acquired FiftyPoint6 Group (back-end SaaS for luxury retailers, UK)
2025-09-02 · Circeus acquired KrakenD (API gateway software, Spain)
2025-09-10 · Bending Spoons acquired Vimeo (video hosting platform, USA)
2025-10-29 · Bending Spoons acquired AOL (web portal and email, USA)
2025-10-29 · Nuvocargo acquired Mentum (AI-agent software startup (procurement / supply-chain workflows), USA)
2025-11-04 · Beacon Software acquired PowerUp Sports (sports club registration / management software, Canada)
2025-11-04 · Beacon Software acquired Viefund (back-office software for mutual fund dealers, Canada)
2025-12-02 · Bending Spoons acquired Eventbrite (event ticketing platform, USA)
2026-04-16 · Beacon Software acquired E2E Soccer Limited (youth soccer league management software, Canada)
2026-05-18 · Bending Spoons acquired Tractive (pet GPS tracking and health monitoring, Austria)
2026-06-10 · 2X acquired Knownwell (AI-as-a-service / agentic AI engineering firm, USA)
2026-07-07 · Circeus acquired Encodian (document automation software (Microsoft Power Platform), UK)
2026-07-21 · Circeus acquired Dondy (WhatsApp / conversational commerce software, unknown)
2026-08-04 · Bending Spoons acquired Airtable (no-code database / workflow software, USA)
2026-09-10 · Bending Spoons acquired Miro (online collaboration whiteboard software, USA/Netherlands)
2026-09-17 · Beacon Software acquired Haize Labs (AI reliability / testing software, USA)

the owner likely keeps the saving

Staffing and executive search

Roll-ups followed: 3 (2 in the United States, 1 elsewhere; 0 searched for purchases) · purchases found: 0

Census: $2.6B of American payroll passes the screen, from the census's own sizing of this business.

Priced as a percentage of the placed candidate's first-year salary. The fee is a share of the placed candidate's salary. The client never sees your cost base, so a cheaper process does not give them anything to negotiate against. Reasoned from how the business prices its work, not measured.

Who the buyers are, what they bought, and the census's reasoning

Roll-ups (bold: searched for purchases): Forge AI, Starshunt, Mojo Talent

No purchase on the record was found for the buyers searched.

The fee is indexed to someone else's wage bill, so the client cannot see your cost base and has nothing to index a discount to. Sourcing, screening and scheduling are the cost; the placement fee is unchanged by how they get done.

Excludes $140bn of temporary help, which bills an hourly markup — 73% of the code's payroll. It still includes professional employer organisations (NAICS 561330), which OEWS doesn't publish separately.

Retained or contingency? Both are priced as a share of salary; they differ in who carries a failed search. Retained firms are paid in instalments whatever the outcome — safer revenue.
What share of placements come from one client? Concentration is where the fee gets renegotiated.
Does the ATS belong to the client or to you? If theirs, they own the candidate graph.

Comparables the census named to illustrate the pricing model: Korn Ferry and Heidrick & Struggles (retained search, about a third of first-year compensation); Robert Half's permanent placement segment ($440m in 2025, a percentage of annual compensation).

Sources: www.sec.gov, www.hirecruiting.com

the owner likely keeps the saving

Managed travel and booking

Roll-ups followed: 2 (2 in the United States, 0 elsewhere; 0 searched for purchases) · purchases found: 1

Census: $1.5B of American payroll passes the screen, from the census's own sizing of this business.

Priced as a commission or transaction fee on the booking value. The fee is a share of a booking the client already pays for. Servicing cost is invisible to them; the take rate is conventional. Reasoned from how the business prices its work, not measured.

Who the buyers are, what they bought, and the census's reasoning

Roll-ups (bold: searched for purchases): Gazelle Travel, The Travel Company

2026-09-29 · Long Lake acquired American Express Global Business Travel (Global Business Travel Group, Inc.) (corporate travel management company, USA)

Fee is a percentage of a notional the client is already paying. Servicing cost falls; the take rate is conventional.

Is the contract moving to a per-trip SaaS fee? That is the commoditised version.
How much revenue is supplier commission rather than client fee? Suppliers set commissions and can cut them — airlines did in the 2000s.
Self-booking share — already-automated volume is already repriced.

Comparables the census named to illustrate the pricing model: Amex GBT (acquired CWT, Sept 2025); Navan (90% transaction revenue, 7% take rate on bookings); Corporate Travel Management; BCD Travel.

Sources: www.sec.gov, investors.amexglobalbusinesstravel.com

check before buying

Advertising and marketing services

Roll-ups followed: 1 (1 in the United States, 0 elsewhere; 1 searched for purchases) · purchases found: 3

Census: $2.1B of American payroll passes the screen, from the census's own sizing of this business.

Priced as retainer or a percentage of media spend. A share of media spend is indexed to the client's budget and keeps the saving. A retainer priced off a staffing plan is indexed to your headcount and passes it on. The contract decides which. Reasoned from how the business prices its work, not measured.

Who the buyers are, what they bought, and the census's reasoning

Roll-ups (bold: searched for purchases): 2X

2024-12-19 · 2X acquired Intelligent Demand (B2B marketing / growth-strategy agency, USA)
2025-07-30 · 2X acquired Outbound Funnel (revenue-operations consultancy, USA)
2026-01-06 · 2X acquired The Kiln (go-to-market agency (Clay partner), USA)

Where the fee is a share of media spend it is indexed to the client's budget, not to your headcount. Where it is a retainer priced off an FTE plan, it is not — check which.

Is the master services agreement priced on an FTE plan? Then automation cuts revenue 1:1.
What share is production versus strategy? Production is the part that flips.
Client tenure and notice periods — agency books churn faster than the model assumes.

Comparables the census named to illustrate the pricing model: Omnicom (including Interpublic since Nov 2025); Stagwell; S4 Capital.

Sources: investor.omc.com

the owner likely keeps the saving

Diagnostic and medical laboratories

Roll-ups followed: 0 · purchases found: 0

Census: $977M of American payroll passes the screen, from the census's own sizing of this business.

Priced as a fixed fee per test, set by the payer. Price is set by the payer's reimbursement schedule, not by your cost. It holds until the schedule is next reset — and schedule cuts are the real risk, not competition. Reasoned from how the business prices its work, not measured.

The census reasons that an owner keeps the saving here, yet no roll-up this site follows works in the trade.

Who the buyers are, what they bought, and the census's reasoning

Price is set by a reimbursement schedule, not by cost. Accessioning, result entry and billing are labour; the schedule does not move when they get cheaper.

Overlaps the revenue-cycle card by $0.3bn of payroll ($0.12bn passing the screen) — records and billing staff inside labs — so the two are not additive.

What share of revenue is Medicare/Medicaid? Schedule cuts are the real risk, not competition.
Is the LIS integrated with the referring EHR? Epic in the middle is a gatekeeper on your data.
Does any test require a pathologist signature by statute? That caps throughput per licence.

Comparables the census named to illustrate the pricing model: Labcorp; Quest Diagnostics; Sonic Healthcare.

the owner likely keeps the saving· sized by occupation, not industry

Title insurance, escrow and settlement

Roll-ups followed: 0 · purchases found: 0

Census: $63.8M of American payroll passes the screen, from the census's own sizing of this business.

Priced as a premium scaled to the policy amount, at a state-set or filed rate. The premium scales with the price of the property, at a rate the state sets outright (Texas, New Mexico, Florida) or that must be filed with it. It is priced off the policy amount, not off what a given search costs, so a cheaper search gives the buyer nothing to bargain against — though a state that sets the rate can pass industry-wide savings through at its next rate hearing. Reasoned from how the business prices its work, not measured.

The census reasons that an owner keeps the saving here, yet no roll-up this site follows works in the trade.

Who the buyers are, what they bought, and the census's reasoning

The search and examination are the cost: pulling and reading the chain of recorded documents. Counties covering over 90% of the US population now accept electronic recording, so the inputs are increasingly digital while the premium stays tied to the price of the house.

No industry code isolates title and escrow — BLS files it inside legal services and insurance carriers. Payroll here is the one occupation that exists only for this work (title examiners, abstractors and searchers), so it understates the business: closers and escrow officers are filed under general clerical codes.

A regulator that sets the rate can cut it: Texas cut its promulgated title rates 10% from July 2025.
Agent or underwriter? On agent-written policies the agent keeps most of the premium — about 83% at Stewart in Q3 2025 — and the underwriter's share is contractual. Which side of that split a business sits on decides what it keeps.
Fannie Mae's title acceptance pilot drops title insurance entirely on some low loan-to-value refinances, and FHFA said in June 2026 it is expanding. If it reaches purchase loans, the premium itself is at risk.
What share of revenue is premium versus escrow and closing fees? Only the premium is tied to a set or filed rate — check how the rest is priced.

Comparables the census named to illustrate the pricing model: Fidelity National Financial; First American Financial; Stewart Information Services; Old Republic (title segment); Doma (bought by Title Resources Group, Sept 2024).

Sources: www.oldrepublictitle.com, www.sec.gov, www.housingwire.com, www.alta.org, mortgagetech.ice.com

Other trades

Roll-ups followed: 40 (11 in the United States, 8 elsewhere, 21 place not stated; 3 searched for purchases) · purchases found: 3

Census: no figure. A mix of trades with too few buyers each to compare.

Who the buyers are, what they bought

Roll-ups (bold: searched for purchases): Bridge Defense, Cavalla, Enam, Iterative Health, Longview, Metropolis, Particle Health, Production Capital, Rocketable, Roofer, Zero RFI, Antares Labs, AppliedAI, Areg, Felz, Kaikaku, Moneylab Technologies, Recycleye, Venmark, Agent Vista, AIAI Holdings (Ai²), AIXIOM, Banyan Software, Calexo (Thermal Home Services), Cast Finance, CSC Generation, Dragonfly, Ekohe Fund, Generation Tech Partners, Legion Holdings, Mitrim, Operon Capital, Orbit Holdings, Rising Tide, Squeeze, Stelao, Topwater Group, United Accountants, XGX Ventures, Zax

2024-05-16 · Metropolis acquired SP Plus Corporation (SP+) (parking operator, USA)
2025-08-26 · Crete announced as joining its group: Cabrillo Advisors (valuation and financial advisory firm, USA)
2026-01-28 · Crete announced as joining its group: Force 10 Partners (restructuring and litigation-support advisory firm, USA)

Buyers across several trades

Roll-ups followed: 7 (4 in the United States, 1 elsewhere, 2 place not stated; 4 searched for purchases) · purchases found: 0

Census: no figure. These buyers work across trades; what each bought is filed under its own trade.

Who the buyers are, what they bought

Roll-ups (bold: searched for purchases): Antidote, Long Lake, Sequence Holdings, SKROL, Timah Partners, General Catalyst, Thrive Holdings

No purchase on the record was found for the buyers searched.

Businesses with no card, and why

Transfer agency, fund administration & corporate trust. BLS publishes these only inside all of securities and investments (NAICS 523), where brokers and fund managers dominate the payroll, and no occupation exists only for this work. Any payroll figure would be borrowed from other businesses.
Credentialing verification. No industry code of its own and no dedicated occupation — credentialing staff are filed under general records and compliance roles.
Permit expediting. Filed under 'all other professional services', with no separate staffing data and no dedicated occupation.
Retail tax preparation & bookkeeping. Published inside the same code as the accounting card (NAICS 5412), so their payroll can't be separated from it. Tax preparers alone: $4.3bn of payroll; $0.5bn of their work passes the screen (11%, 0%–35% under the strict and loose rules), $0.0bn of it agreed by all three scorers.
Translation. A clean anchor, but too little passes the screen to size an archetype on. Interpreters and translators: $3.2bn of payroll; $0.5bn of their work passes the screen (16%, 16%–83% under the strict and loose rules), $0.0bn of it agreed by all three scorers.

Method

Method, and what failed

Every test the census set itself, marked by when it was written down: before the first scores, before this version's scores, or after seeing the data. Tests still waiting on the next release of usage data have no result yet.

Tests the census set itself, reported either way

Each test is marked by when it was written down. The test names and symbols are the census's own; its files define them. A test waiting on the next release of usage data has no threshold and no result yet.

Set before the first scores (17 September)

Test	Result	Needed	Outcome	Note
R1 held-out prediction of AI uptake	ρ 0.147	≥0.30	failed	baseline 0.121
R1 uptake falls as specification entropy rises	non-monotone	decreasing	failed
R1 uptake falls as the checking horizon lengthens	non-monotone	decreasing	failed
R1 easy-to-specify, quick-to-check work sees the most uptake	Copilot highest	Tool highest	failed
R3 incremental validity over importance + wage	ΔR² 0.019	≥0.03	failed	below the gate
R4 within-occupation scramble	1.31	>1.15	passed	beats 100% of 200 scrambles
Human-rated gold set, task level	not run	α≥0.67	—	occupation-level worker survey used instead

Set before this version's scores (24 September)

Test	Result	Needed	Outcome	Note
Not-called payroll within the stop rule	0.00%	≤0.5%	passed	tasks without three valid scores
Channel test on the next Economic Index release	pending	no threshold set	—	pre-committed to run once, unchanged; the within-occupation z will be reported whatever it is
AI-alone test on the next Economic Index release	pending	no threshold set	—	pre-committed to run once, unchanged; raw and within-occupation effects will be reported

Found after seeing the data

Test	Result	Needed	Outcome	Note
R1 held-out prediction, v3 rerun	ρ 0.115	≥0.30	failed	baseline 0.121
R1 easy-to-specify, quick-to-check work sees the most uptake, v3	Copilot highest	Tool highest	failed
R3 incremental validity, v3 rerun	ΔR² 0.019	≥0.03	failed	automation share ΔR² 0.021
R4 within-occupation scramble, v3 rerun	1.28	>1.15	passed	beats 100% of 200 scrambles
AI works alone more on work that passes — consumer traffic (in-sample)	66% vs 49%	t>2 (set post hoc)	passed	within occupation +13.3pp, t 7.2; n=2,306; these tasks informed the rule
AI works alone more on work that passes — enterprise API traffic	95% vs 91%	t>2 (set post hoc)	passed	within occupation +2.4pp, t 3.8; an automation-heavy population
Harder-to-specify work gets less of its AI use through the API	z -4.9	z < −2 (set post hoc)	passed	two-part model, within occupation; whether a task gets any API use at all: z -1.9. Designed after the chat result; confirmatory only on the next release
Scorers' stakes track worker-reported consequence of error	ρ 0.62/0.69/0.57	ρ>0	passed	Dept of Labor survey the scorers never saw
Stakes predict wages	z +11.05	p<0.05	passed
An existing check predicts wages	z -1.59	p<0.05	failed
Verdict agreement across three scorers (the structural rule, before the shared gates)	Fleiss κ 0.41	descriptive	—	pairwise Cohen κ sonnet–gemini 0.47, sonnet–gpt6 0.36, gemini–gpt6 0.44
τ stability across three constructions	ρ 0.96	stable	passed	lowest pairwise rank correlation of role shares across the three time-weight constructions, v3
Industry retention (cost pass-through)	not identified	placebo ≈ 0	failed	3 of 4 estimators fail a noise placebo; the 4th flips sign with start year

Do the models agree, and does the screen match observed use?

Two checks on the screen itself. Agreement: how often the three models give the same verdict, as a statistic that is one for perfect agreement and zero for chance. AI working alone: in observed use, AI works alone more often on tasks that pass than on tasks that fail, even within the same job. The census's own label says this test is in sample and post hoc.

Check	Tasks	Where the task passes	Where it fails	Gap within the same job, in points	Strength
AI working alone, in chat	2306	65.9%	49.1%	13.3	7.21
AI working alone, through programmatic access	1936	95.1%	91.0%	2.44	3.84

The census labels the chat test: in-sample, post hoc: these tasks informed the rule; confirmatory on the next release. Agreement across the three models (Fleiss κ): 0.41 over 11,176 tasks.

Each model alone, and the same model twice

In most functions the three models are far apart, and the same model is usually the highestChart

All knowledge work

Customer support

Claims & underwriting

Finance

Supply chain

Retail front line

Sales

Procurement

Data & analytics

Admin

rule 7.8% · all three 6.1%

People

IT & software

Operations

Marketing

rule 4.3% · all three 2.6%

Creative

rule 3.1% · all three 0.7%

Social services

Education

rule 2.7% · all three 1.7%

Engineering

rule 1.8% · all three 0.5%

Research

rule 1.6% · all three 0.4%

Clinical

rule 0.9% · all three 0.1%

Legal & compliance

rule 0.7% · all three 0.2%

Executive

rule 0.0% · all three 0.0%

Strategy

rule 0.0% · all three 0.0%

Claude Sonnet aloneGemini 3.1 Pro aloneGPT-6 Sol alonethe rule: two of the three passall three models pass

Each mark is a share of the function's knowledge payroll; the line spans the lowest and the highest model. From the census (v3.2-2026-09-24), a screen scored by three AI models, not a claim about what AI can do and not a forecast. Each model's own share is its verdict under the same rule and the same exclusions. The same model re-scoring the same tasks also changed some verdicts, so small gaps between rows are noise. Where a function's marks would sit on top of one another, the rule's share and the share all three pass are printed in type instead of drawn. Where nothing passes, the marks sit together at zero.

The numbers

Function	Claude Sonnet alone	Gemini 3.1 Pro alone	GPT-6 Sol alone	The rule: two of the three	All three models pass
All knowledge work	9.9%	5.7%	3.5%	5.6%	2.3%
Customer support	39.9%	22.1%	12.8%	22.4%	12.5%
Claims & underwriting	24.1%	10.6%	5.1%	14.8%	1.0%
Finance	17.5%	13.6%	4.9%	12.7%	3.7%
Supply chain	18.2%	9.6%	9.5%	12.1%	5.9%
Retail front line	9.0%	12.4%	9.3%	10.1%	5.2%
Sales	16.5%	9.1%	5.2%	9.5%	4.1%
Procurement	17.0%	7.2%	7.5%	9.3%	4.7%
Data & analytics	12.4%	16.3%	3.6%	8.7%	2.5%
Admin	11.8%	7.7%	10.6%	7.8%	6.1%
People	23.2%	7.8%	2.7%	7.4%	2.5%
IT & software	12.2%	6.6%	1.2%	5.3%	0.6%
Operations	9.0%	3.2%	4.2%	5.1%	1.8%
Marketing	5.7%	3.2%	4.1%	4.3%	2.6%
Creative	5.5%	2.0%	2.1%	3.1%	0.7%
Social services	8.3%	1.3%	2.4%	3.0%	0.3%
Education	5.9%	3.3%	2.0%	2.7%	1.7%
Engineering	5.3%	1.8%	0.6%	1.8%	0.5%
Research	4.9%	2.4%	1.0%	1.6%	0.4%
Clinical	4.5%	1.6%	0.4%	0.9%	0.1%
Legal & compliance	2.5%	1.0%	0.2%	0.7%	0.2%
Executive	0.0%	0.0%	0.0%	0.0%	0.0%
Strategy	0.0%	0.0%	0.0%	0.0%	0.0%

Model	Payroll it alone would pass
Claude Sonnet	$696B
Gemini 3.1 Pro	$400B
GPT-6 Sol	$244B

The same model re-scored a seeded sample of the same tasks under the same rubric, weeks apart and in smaller batches, and some verdicts flipped. That share is the noise floor for comparing roles. Claude Sonnet re-scored 849 tasks and changed 12.8% of its verdicts.

Which model to believe

Each model was checked against evidence it never saw: how well its ratings of stakes and of whether work can be checked match what workers themselves report in a Labor Department survey (a correlation: the closer to its maximum, the closer the match), and how much of observed AI use of the tasks it passes is AI working alone.

Model	Tasks scored	Matches workers on stakes	Matches workers on checking	AI alone where it passes the task
Gemini 3.1 Pro	11,176	0.62	0.42	65.3%
GPT-6 Sol	11,176	0.69	0.35	65.9%
Claude Sonnet	11,176	0.57	0.51	64.9%

Why keeping the saving is reasoned, not measured

Whether a business keeps the saving would show up as prices falling when labour costs fall. A sound method for finding that should find nothing when fed random noise; most of the methods tried found something anyway, and the estimates from the one that passed change sign when the first year of data moves, so none is used.

The numbers

Method (the census's name)	Result on random noise	Trusted
original	0.57	no
W	-0.0076	yes
L	0.45	no
IV	0.82	no

First year	Industry	How much of a labour-cost fall reached prices	Give or take
1997	CPAs	0.78	0.31
1997	Tax preparation	-0.022	0.17
1997	Placement & search	-0.39	0.30
1999	CPAs	-0.61	0.45
1999	Tax preparation	-0.0070	0.16
1999	Placement & search	-0.42	0.28
2002	CPAs	-0.66	0.46
2002	Tax preparation	-0.027	0.26
2002	Placement & search	-0.42	0.29
2005	CPAs	-0.68	0.46
2005	Tax preparation	-0.33	0.25
2005	Placement & search	-0.54	0.21

How hard a task is to describe: found after seeing the data

Not part of the rule. Work that is harder to describe precisely is associated with less of its AI use coming through direct programmatic access rather than chat. The model has two parts: whether a task sees any programmatic use, and how much of its use is programmatic where it sees some. A negative effect means less; the larger the strength, the less likely the pattern is chance. The first part is weak; the second is strong. It was found after seeing the data and will be run once more, unchanged, on the next release.

The numbers (post hoc (current release); confirmatory on the next)

Month	Tasks with any programmatic use	Effect on any use	Its strength	Tasks with some	Effect on how much	Strength of that effect
2026-05	1032	-0.22	-1.90	898	-0.27	-4.87
2026-04	998	-0.25	-2.17	848	-0.28	-5.09

The physical-work filter, checked both ways

Work that needs a body is left out whatever the vote says: $1.4T of payroll. Most of it would have failed the vote anyway; the filter changed the verdict on $201B, the payroll the vote would otherwise have passed. A spot check by Gemini 3.1 Pro of 60 tasks drawn from that part found 16.7% were really desk work (a ninety-five percent interval of 8.3% to 28.5%), so the payroll that passes may be $16.7B to $57.5B too low. The same check of 60 kept tasks found 0 that were really physical (an interval of 0.0% to 6.0%), so the payroll that passes may also be up to $23.2B too high. The filter counts talking with people in person as desk work, which neither check can measure.

Modelled, not measured: what the remaining work costs

If the work that stays is needed alongside the work that passes, it becomes a larger share of what a job costs. How much larger depends on how easily one kind of work can stand in for another, which economists call the elasticity; published estimates fall on both sides of the value where the effect reverses, so these figures are conditional.

Elasticity	Share of a job's cost in work that stays, before	After
0.50	93.1%	94.4%
1.00	93.1%	93.1%
1.50	93.1%	91.5%

What every figure here carries with it

It is a structural screen, not a claim about what AI can do and not a forecast. A task can pass and still be beyond the systems of the day.
Passing is not saving. The modelled saving beside the headline assumes a cost cut and how easily one kind of work stands in for another; it is modelled, not measured, and far below the payroll that passes.
The models disagree a great deal, and the same model re-scoring the same tasks changed a share of its verdicts. Lead with the part all three pass, and treat small differences between roles as noise.
The tests against observed AI use were shaped after seeing that data, so they are post hoc. They are committed to run once more, unchanged, on the next release; until then read them as associations, not proof.
The filter that removes physical work errs both ways. A spot-check by one of the models found desk work among the removed tasks, so the headline may be too low; the filter also counts talking with people in person as desk work, which the spot-check cannot measure.
Whether a business keeps the saving was not measured; every method tried either failed a basic noise test or gave estimates that change sign with the years used.
No person has judged individual tasks; the check against people is a Labor Department survey of whole jobs.
The public task lists under-describe coordination and exception handling, which pushes the total up.

Readings · version v3.2-2026-09-24

Knowledge payroll that passes the structural hand-over screen: $389B
Of which all three models pass: $158B
Knowledge payroll scored: $7.0T
Claude Sonnet alone would pass: $696B
Gemini 3.1 Pro alone would pass: $400B
GPT-6 Sol alone would pass: $244B
Agreement across the three models (Fleiss κ: one is perfect, zero is chance): 0.41
Verdicts that changed when the same model re-scored: 12.8%
Waits only on a check: at least two of the three models find it quick to judge and cheap to get wrong, with no existing check: $827B
Left out as physical work: $1.4T
Of that, payroll the vote would otherwise have passed: $201B
Left out as an accountable sign-off, and not physical: $133B
Of that, payroll the vote would otherwise have passed: $1.6B
Modelled saving (modelled, not measured: passing is not saving): $154B

Instrument

Judged by Claude Sonnet, Gemini 3.1 Pro, GPT-6 Sol; the verdict on each task is their vote, under a rule frozen before this version's scores and shaped after seeing earlier usage data. Imported as published from the census's own files and checked against their published hashes.

Download: functions.csv, industries.csv, role_industry.csv, roles.csv, tasks.csv.

Query every table in the SQL console, e.g. SELECT title, share_passes, rule_strict, rule_loose, passes_usd, agreed3_usd FROM census_roles WHERE function = 'Finance'.

How the census is imported