StarZero All shows Try the podcast clipper →

Apr 16, 2026 · 29m · another-podcast

What jobs are AI jobs?

Benedict Evans · 20m spoken Toni Cowan-Brown · 6m spoken
0:00 / 0:00

gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions

Technology analysts Benedict Evans and Toni Cameron Brown critique contemporary AI job displacement metrics, drawing on economic history, business frameworks, and enterprise realities to demonstrate how automation shifts value and expands markets rather than simply eliminating labor.

How this conversation actually went

Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The hosts hold 77.9% of the talking time here. How this is scored →

The hosts as informed peer 7.8 Guest teaching 2.0 Guest disagreement 1.2 The hosts pushing back 1.8
05100:0010:0020:000:00–6:11 · The hosts as informed peer 8/10 Predicting AI Impact and the Illusion of Precision Benedict establishes the episode's core premise, drawing on his background as a 1999 tech analyst to criticize numeric AI exposure scores from Anthropic and OpenAI. Toni acts primarily as a supportive sounding board, agreeing with his skepticism about false precision.6:12–12:22 · The hosts as informed peer 8/10 Unbundling Value: Jobs to Be Done and Points of Leverage Benedict uses Clayton Christensen's jobs-to-be-done framework to analyze how technology unbundles logistics from core value propositions across airlines, retail, and newspapers. Toni contributes an example from Substack newsletters, which Benedict smoothly integrates into his wider thesis.12:23–17:19 · The hosts as informed peer 8/10 Code, Execution, and the Accountant Automation Paradox Benedict dismisses the idea that automated code generation threatens SaaS moats, explaining that go-to-market execution and product definition are the real barriers. He illustrates this with the historical divergence between elevator attendants and accountants.17:19–23:41 · The hosts as informed peer 7/10 The Premium on Human Consultancy, Implementation, and Taste Toni takes an active role by drawing on her SaaS sales experience at NationBuilder, arguing that high-value software enterprise sales fundamentally depend on human consultancy rather than the tech itself. Benedict acknowledges this and connects it to the luxury goods and taste economy.23:43–29:36 · The hosts as informed peer 8/10 Real-World Corporate Priorities and Limits of Job Scoring Benedict discusses real-world enterprise priorities, demonstrating that operational maintenance often supersedes AI innovation for traditional corporations. Both speakers conclude that consulting and software moats rely on intangible human factors rather than automated artifacts like slide decks or lines of SQL.0:00–6:11 · Guest teaching 1/10 Predicting AI Impact and the Illusion of Precision Benedict establishes the episode's core premise, drawing on his background as a 1999 tech analyst to criticize numeric AI exposure scores from Anthropic and OpenAI. Toni acts primarily as a supportive sounding board, agreeing with his skepticism about false precision.6:12–12:22 · Guest teaching 2/10 Unbundling Value: Jobs to Be Done and Points of Leverage Benedict uses Clayton Christensen's jobs-to-be-done framework to analyze how technology unbundles logistics from core value propositions across airlines, retail, and newspapers. Toni contributes an example from Substack newsletters, which Benedict smoothly integrates into his wider thesis.12:23–17:19 · Guest teaching 1/10 Code, Execution, and the Accountant Automation Paradox Benedict dismisses the idea that automated code generation threatens SaaS moats, explaining that go-to-market execution and product definition are the real barriers. He illustrates this with the historical divergence between elevator attendants and accountants.17:19–23:41 · Guest teaching 5/10 The Premium on Human Consultancy, Implementation, and Taste Toni takes an active role by drawing on her SaaS sales experience at NationBuilder, arguing that high-value software enterprise sales fundamentally depend on human consultancy rather than the tech itself. Benedict acknowledges this and connects it to the luxury goods and taste economy.23:43–29:36 · Guest teaching 1/10 Real-World Corporate Priorities and Limits of Job Scoring Benedict discusses real-world enterprise priorities, demonstrating that operational maintenance often supersedes AI innovation for traditional corporations. Both speakers conclude that consulting and software moats rely on intangible human factors rather than automated artifacts like slide decks or lines of SQL.0:00–6:11 · Guest disagreement 1/10 Predicting AI Impact and the Illusion of Precision Benedict establishes the episode's core premise, drawing on his background as a 1999 tech analyst to criticize numeric AI exposure scores from Anthropic and OpenAI. Toni acts primarily as a supportive sounding board, agreeing with his skepticism about false precision.6:12–12:22 · Guest disagreement 1/10 Unbundling Value: Jobs to Be Done and Points of Leverage Benedict uses Clayton Christensen's jobs-to-be-done framework to analyze how technology unbundles logistics from core value propositions across airlines, retail, and newspapers. Toni contributes an example from Substack newsletters, which Benedict smoothly integrates into his wider thesis.12:23–17:19 · Guest disagreement 1/10 Code, Execution, and the Accountant Automation Paradox Benedict dismisses the idea that automated code generation threatens SaaS moats, explaining that go-to-market execution and product definition are the real barriers. He illustrates this with the historical divergence between elevator attendants and accountants.17:19–23:41 · Guest disagreement 2/10 The Premium on Human Consultancy, Implementation, and Taste Toni takes an active role by drawing on her SaaS sales experience at NationBuilder, arguing that high-value software enterprise sales fundamentally depend on human consultancy rather than the tech itself. Benedict acknowledges this and connects it to the luxury goods and taste economy.23:43–29:36 · Guest disagreement 1/10 Real-World Corporate Priorities and Limits of Job Scoring Benedict discusses real-world enterprise priorities, demonstrating that operational maintenance often supersedes AI innovation for traditional corporations. Both speakers conclude that consulting and software moats rely on intangible human factors rather than automated artifacts like slide decks or lines of SQL.0:00–6:11 · The hosts pushing back 2/10 Predicting AI Impact and the Illusion of Precision Benedict establishes the episode's core premise, drawing on his background as a 1999 tech analyst to criticize numeric AI exposure scores from Anthropic and OpenAI. Toni acts primarily as a supportive sounding board, agreeing with his skepticism about false precision.6:12–12:22 · The hosts pushing back 2/10 Unbundling Value: Jobs to Be Done and Points of Leverage Benedict uses Clayton Christensen's jobs-to-be-done framework to analyze how technology unbundles logistics from core value propositions across airlines, retail, and newspapers. Toni contributes an example from Substack newsletters, which Benedict smoothly integrates into his wider thesis.12:23–17:19 · The hosts pushing back 2/10 Code, Execution, and the Accountant Automation Paradox Benedict dismisses the idea that automated code generation threatens SaaS moats, explaining that go-to-market execution and product definition are the real barriers. He illustrates this with the historical divergence between elevator attendants and accountants.17:19–23:41 · The hosts pushing back 2/10 The Premium on Human Consultancy, Implementation, and Taste Toni takes an active role by drawing on her SaaS sales experience at NationBuilder, arguing that high-value software enterprise sales fundamentally depend on human consultancy rather than the tech itself. Benedict acknowledges this and connects it to the luxury goods and taste economy.23:43–29:36 · The hosts pushing back 1/10 Real-World Corporate Priorities and Limits of Job Scoring Benedict discusses real-world enterprise priorities, demonstrating that operational maintenance often supersedes AI innovation for traditional corporations. Both speakers conclude that consulting and software moats rely on intangible human factors rather than automated artifacts like slide decks or lines of SQL.

speaking balance: gold is the hosts, purple is the guest (3 minute bins)

0:00 · the hosts 76.7% · guest 23.3%0:00 · the hosts 76.7% · guest 23.3%3:00 · the hosts 92.7% · guest 7.3%3:00 · the hosts 92.7% · guest 7.3%6:00 · the hosts 98.3% · guest 1.7%6:00 · the hosts 98.3% · guest 1.7%9:00 · the hosts 59.8% · guest 40.2%9:00 · the hosts 59.8% · guest 40.2%12:00 · the hosts 99.2% · guest 0.8%12:00 · the hosts 99.2% · guest 0.8%15:00 · the hosts 73.9% · guest 26.1%15:00 · the hosts 73.9% · guest 26.1%18:00 · the hosts 27.2% · guest 72.8%18:00 · the hosts 27.2% · guest 72.8%21:00 · the hosts 99.9% · guest 0.1%21:00 · the hosts 99.9% · guest 0.1%24:00 · the hosts 66.2% · guest 33.8%24:00 · the hosts 66.2% · guest 33.8%27:00 · the hosts 85.7% · guest 14.3%27:00 · the hosts 85.7% · guest 14.3%
Sharpest disagreement ▶ 19:12 Toni rejects replacing human conversation with AI

Toni vents frustration at the reflex to offload marital and interpersonal problem solving to chatbots like Claude and ChatGPT.

 Hardest push from the hosts ▶ 13:55 Benedict dismisses autonomous marketplace agent hype

Benedict aggressively rejects viral claims that AI agents will automatically build multi-sided marketplaces like DoorDash, calling the premise naive about operational reality.

 Biggest teaching moment ▶ 17:45 Toni educates on the human services necessary in enterprise SaaS

Toni challenges pure-software thinking by recounting executive pushback at her former SaaS company, proving that hundred-million-dollar deals require high-touch human consultancy.

 The host holds their own ▶ 14:50 Benedict details 100 years of accounting automation data

Benedict demonstrates deep analytical authority by showing how accounting headcount rose across every wave of mechanization from adding machines to ERPs.
the scores for every segment, with the reasoning behind each
ChapterTopicThe hosts as informed peerGuest teachingGuest disagreementThe hosts pushing backWhy
Predicting AI Impact and the Illusion of Precision 8112 Benedict establishes the episode's core premise, drawing on his background as a 1999 tech analyst to criticize numeric AI exposure scores from Anthropic and OpenAI. Toni acts primarily as a supportive sounding board, agreeing with his skepticism about false precision.
Unbundling Value: Jobs to Be Done and Points of Leverage 8212 Benedict uses Clayton Christensen's jobs-to-be-done framework to analyze how technology unbundles logistics from core value propositions across airlines, retail, and newspapers. Toni contributes an example from Substack newsletters, which Benedict smoothly integrates into his wider thesis.
Code, Execution, and the Accountant Automation Paradox 8112 Benedict dismisses the idea that automated code generation threatens SaaS moats, explaining that go-to-market execution and product definition are the real barriers. He illustrates this with the historical divergence between elevator attendants and accountants.
The Premium on Human Consultancy, Implementation, and Taste 7522 Toni takes an active role by drawing on her SaaS sales experience at NationBuilder, arguing that high-value software enterprise sales fundamentally depend on human consultancy rather than the tech itself. Benedict acknowledges this and connects it to the luxury goods and taste economy.
Real-World Corporate Priorities and Limits of Job Scoring 8111 Benedict discusses real-world enterprise priorities, demonstrating that operational maintenance often supersedes AI innovation for traditional corporations. Both speakers conclude that consulting and software moats rely on intangible human factors rather than automated artifacts like slide decks or lines of SQL.

Statements from this episode (9)

Opinion
Evans: Anthropic and OpenAI job exposure scores are self-deception
“Because Anthropic does this thing, and OpenAI have done these things based on sort of US census data where they try and put like a numeric score job by job of exposure to AI. Which seems to me just absolutely ludicrous as an exercise in sort of self-deception …”
Benedict Evans Apr 16, 2026 ▶ 0:20
Assertion Supported
Evans: Uber's TAM Was Far Bigger Than the Taxi Market
“There was a valuation professor who notoriously did an analysis of Uber kind of 10 years ago, where he said, you know, the valuation of Uber should be based on the fact that its TAM is the size of the taxi market, and this is, I'm going to work out the size of…”
Benedict Evans Apr 16, 2026 ▶ 4:16
Insight
Evans: In-store grocery shopping remains more efficient than home delivery
“There are some cases where internet is not more efficient, like groceries, which is why Walmart is still a giant business, because it's actually more efficient to drive to the supermarket Than it is to deliver all of that product with the cold chain, and the s…”
Benedict Evans Apr 16, 2026 ▶ 7:08
Opinion
Evans: Amazon cannot provide true recommendation due to massive catalog size
“Amazon has seven, eight, 900, Amazon has seven or eight or nine hundred million schools, so they can't do recommendation.”
Benedict Evans Apr 16, 2026 ▶ 8:28
Insight
Evans: A software company's moat is go-to-market, not writing code
“I don't feel like the reason it's hard to compete with X or Y or Z company is how long it would take you to write the code to replicate the product. It's almost never the problem. It's the problem. The hard part is everything else. It's working out what the pr…”
Benedict Evans Apr 16, 2026 ▶ 12:59
Assertion Supported
Evans: US accountant headcounts rose every decade of the 20th century
“The number of accountants in the USA increased every single decade in the 20th century. As you go through adding machines and computers and mainframes and PCs and Excel and ERPs and SAS, you get wave after wave of automation and the number of accountants keeps…”
Benedict Evans Apr 16, 2026 ▶ 15:13
Insight
Cameron Brown: Enterprise Clients Won't Pay $100M for Software Without Human Consultancy
“No one's going to pay a hundred million dollars just for the technology that you can buy off the shelf. They need to understand how they can map this to the needs that they have.”
Toni Cowan-Brown Apr 16, 2026 ▶ 18:22
Assertion Contradicted
Evans: Average corporate audit costs have remained flat since 2003
“What I was wondering is, is there data on what the average company pays for an audit? And it turns out it is, there is, and it hasn't changed since about 2003.”
Benedict Evans Apr 16, 2026 ▶ 23:54
Assertion Supported
Evans: The typical large enterprise uses 400 to 500 SaaS apps
“The typical big company today has four to 500 SaaS apps.”
Benedict Evans Apr 16, 2026 ▶ 29:01
Made with StarZero

Turn any episode into a week of clips.

This entire site, over 100 episodes transcribed, diarized, checked and made playable, runs on the StarZero media pipeline. Drop in your own episode and the podcast clipper finds the moments worth sharing, cuts them, captions them, and reframes them for every feed.

Try the podcast clipper Learn more