TL;DR
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
A November 2, 2026 episode of Thorsten Meyer AI’s “Work After Work” series uses the three-sector model to trace shifts in employment from farming to manufacturing and services. Its figures show long-term changes, but the episode’s forecast that automation could become a sector doing up to 90% of work is a thesis, not an established finding.
The episode divides economic activity into three categories: the primary sector, which takes resources from nature; the secondary sector, which makes goods and builds infrastructure; and the tertiary sector, which provides services. A loaf of bread illustrates the distinction: growing wheat is primary work, milling and baking are secondary, and transport and retail are services.
Historical estimates cited in the episode put farming at 73.7% of the U.S. labor force in 1800, falling to 12.0% in 1950. The episode cites an ILO modelled estimate of 1.5% for 2025. It also says that, in the ILO’s modelled estimates, services have employed more people worldwide than farming since 2001. These figures describe employment shares, not the total amount of food or manufactured goods produced.
The episode also challenges the claim that the United States no longer makes things. It says U.S. factory employment is about 35% below its June 1979 peak, while real manufacturing output in 2025 was about 1.9 times its 1979 level. The supplied material does not identify the underlying datasets for those two figures, so their definitions and methods cannot be assessed here.
Work After Work · Episode 01 · November 2, 2026
3 Ways AI Is Changing Work Across Different Sectors
A first episode traces the long shift from farming to manufacturing and services—and frames automation’s possible future role as a thesis still to be tested.
TL;DR
The episode uses the three-sector model to describe work moving from farming toward manufacturing and services. Its figures show long-term changes, while the claim that automation could perform up to 90% of work remains the creator’s thesis, not an established finding.
Three sectors, one connected economy
Economic activity can be grouped by what work does: extract, make, or serve.
Take from nature
Work that obtains raw materials and food from the natural world. Example: growing wheat.
Make and build
Work that turns resources into goods or infrastructure. Example: milling flour and baking bread.
Move and serve
Work that provides services to people and businesses. Example: transporting and selling bread.
Resources come from the land.
Raw materials become a product.
Transport and retail provide services.
Farming’s share of U.S. work fell sharply
These figures describe the workforce share in farming, not food output.
U.S. labor force in farming
Selected estimates cited in the episode · bars use a common 0–100% scale
Fewer factory jobs can coexist with more output
The episode challenges the idea that the United States no longer makes things.
About 35% below its June 1979 peak, according to the episode.
2025 output compared with 1979, as reported by the episode.
Employment and output measure different things. The source material does not identify the datasets or methods behind these two headline figures, so their definitions cannot be assessed here.
How the sector framework took shape
The episode presents a history of economic thought, not proof that workers always move smoothly between sectors.
Allan G. B. Fisher
The New Zealand economist is credited with adding “tertiary” to the terminology in a book written in 1934 and published in 1935.
Colin Clark
His comparison of sectors and employment across countries helped establish the model as a way to describe economic change.
Jean Fourastié
He argued that services could absorb workers released by technical progress in industry—an influential argument, not a guarantee for every worker.
A software engineer employed by a carmaker may be counted in manufacturing in some statistics and services in others. Mining classifications can also vary between systems.
What the episode’s 90% claim does—and does not—say
Possible share of work automation could perform, as proposed by the creator.
A proposition, not an established projection
The supplied material gives no time-bound evidence or method supporting this figure. It should not be read as a confirmed outcome. The episode’s historical model does not settle what automation will do to occupations, wages, or working conditions.
Evidence status · Unsubstantiated in supplied materialQuestions the first episode leaves open
The historical overview sets up the subject; it does not resolve what happens next.
“Work After Work”
Thorsten Meyer AI describes the series as a daily, 60-episode project. The first installment is about seven minutes long. Its source material says the narration uses an AI voice clone of Thorsten Meyer and the images are AI-generated.
Episode 48 is planned to examine the automation thesis. No release date or further evidence for the forecast is provided.
Automation thesis · release date not specified
Four useful checks for reading the claims
What is the three-sector model?
It groups work into primary activities that take resources from nature, secondary activities that make or build things, and tertiary activities that provide services.
What does the episode report about U.S. farming?
It cites 73.7% of the labor force in 1800, 12.0% in 1950, and an ILO modelled estimate of 1.5% in 2025.
Does it establish that automation will do 90% of work?
No. The figure is presented as Thorsten Meyer’s thesis, without supporting time-bound evidence or a method in the supplied material.
Does lower factory employment mean less is made?
Not necessarily. The episode reports lower employment alongside higher real output, but its source material does not provide the underlying datasets or methods.
From historical pattern to open question
Farming’s employment share declines over time.
Employment share is not a production measure.
The 90% claim remains a thesis.
Jobs, wages, and conditions need evidence.
Employment Moved Beyond Farming
The figures offer a long view of how economic change can shift workers between sectors without eliminating the need for what they produce. Farming’s share of employment fell sharply in the United States, while food production continued; manufacturing employment also fell from its peak even as the episode reports higher real output.
That distinction matters as readers assess claims about automation and job loss. A change in employment share alone does not show whether output rose, whether particular workers found new jobs, or how gains and disruption were distributed. The episode sets up those questions for the rest of its series, but its historical model does not settle what automation will do next.
Products Worth Considering
LG UltraGear 27GX704A-B 27-inch QHD OLED Gaming Monitor
Bose SoundLink Flex Bluetooth Speaker (2nd Gen), Waterproof, Dustproof, USB‑C, Up to 12 hr Battery, Black
Samsung 34-Inch Odyssey G5 Ultra-Wide Curved Gaming Monitor
As an affiliate, we earn on qualifying purchases.
From Fisher to the Sector Model
The episode credits New Zealand economist Allan G. B. Fisher with adding “tertiary” to the terminology in a book written in 1934 and published in 1935. Economist Colin Clark compared sector and employment patterns across countries in his 1940 book, helping establish the model as a way to describe economic change.
French economist Jean Fourastié later argued in a 1949 book that services could absorb workers freed by technical progress in industry. The episode presents this as an influential argument in the history of economic thought, not as proof that every worker displaced by technology will move smoothly into service work.
The categories also have limits: how jobs are classified can vary. The episode notes, for example, that a software engineer employed by a carmaker may be counted in manufacturing in some statistics and services in others. Mining, too, is classified differently across some systems.
Limits of the Forecast
The episode’s claim that automation could become a sector of its own and perform up to 90% of work is explicitly presented as the creator’s thesis. The source material supplies no time-bound evidence or method supporting that figure, so it should not be read as an established projection.
The materials provided also do not specify the underlying sources and definitions for every employment and output figure. Sector classifications differ, and the episode’s headline figures alone cannot show how technology affects particular occupations, wages, or working conditions. Those outcomes remain unresolved here.
Series Plans Further Episodes
“Work After Work” is described as a daily series planned for 60 episodes. The first episode says episode 48 will examine the automation thesis. The source does not provide a release date for that installment or further evidence supporting the forecast.
The first episode is described as about seven minutes long. Its narration uses an AI voice clone of Thorsten Meyer, and its images are AI-generated, according to the source material.
Key Questions
What is the three-sector model?
It groups work into primary activities that take resources from nature, secondary activities that make or build things, and tertiary activities that provide services.
What does the episode report about U.S. farming employment?
It cites estimates that farming accounted for 73.7% of the U.S. labor force in 1800 and 12.0% in 1950, alongside an ILO modelled estimate of 1.5% for 2025.
Does the episode establish that automation will do 90% of work?
No. The figure is presented as Thorsten Meyer’s thesis, not as a confirmed outcome or a substantiated forecast in the supplied material.
Does lower factory employment mean the United States makes less?
Not necessarily. The episode says U.S. factory employment is about 35% below its June 1979 peak while real manufacturing output in 2025 was about 1.9 times its 1979 level. The source material does not provide the underlying datasets or methods for these figures.
Source: Thorsten Meyer AI
