The Job Sh*t Show

Uber burned through its entire 2026 AI budget in four months and still can't tie the spend to results. Starbucks pulled its AI inventory agent. Ford hired back 350 engineers to clean up after automated inspection. So what's actually going on?I think most of us are stuck in mode one and mode two thinking — using AI to automate tasks or do things faster ourselves — when the real payoff is in mode three: changing what work is worth doing at all. I walk through the steam-engine-to-electric-motor shift, the productivity J-curve, and two examples that show what mode three looks like in practice: IKEA turning 8,500 call center workers into a $1.25B design consulting business, and DeepMind's AlphaFold going from 200,000 human-predicted protein structures to 200 million.Also — quick heads up. This is the last Job Shit Show episode you'll catch here on Spotify/Apple. We're folding it into Great Mondays going forward, so you can still catch it there, plus we'll keep posting video on Substack.And if mode three thinking is something you want to work through with me directly, I'm hosting a workshop called AI Together on September 9th. Sign up at aitogether.greatmondays.com.About The Job Sh*t ShowThe Job Sh*t Show is an investigation into how finding, keeping, and doing work is changing faster than anyone is admitting — and what that actually looks like for real people on the ground. The Sh*t List is a paid Substack for leaders who want to stay ahead of what AI is doing to the workplace — and actually do something about it. Every other week, subscribers get a new AI Manager Skill: one practical, field-tested move for leading your team better in an AI world.

My friend Lindsey Caplan buys professional services. She also sells them. Recently she opened a deliverable from someone she's paying real money, and found AI output staring back at her.What she landed on is something I keep coming back to. We've confused what we actually pay professional services people for. It was never really the output, even if it looked that way for years. It's the judgment, the pattern matching, the thing someone builds by doing this work long enough to see what you can't. I said something similar in a recent interview. Domain expertise is the new oil. Information is everywhere now. Knowledge is a utility. What's rare is someone who has seen enough of the pattern to tell you what it means.The disclosure question is where it got messy for us, and I've been here before. Five years ago it was whether your web guy only builds on WordPress. Now it's whether the person you hired is quietly routing your project through AI without telling you. Lindsey's take isn't that AI is the problem. It's that nobody's saying up front where they're using it and where they're not, and that silence is what kills the trust.Here's where I landed by the end of the conversation: AI was never built to help you do less work. It's not a shortcut, and treating it like one is exactly what's breaking down in the industries where time was always the product being sold.About The Job Sh*t ShowThe Job Sh*t Show is an investigation into how finding, keeping, and doing work is changing faster than anyone is admitting — and what that actually looks like for real people on the ground. The Sh*t List is a paid Substack for leaders who want to stay ahead of what AI is doing to the workplace — and actually do something about it. Every other week, subscribers get a new AI Manager Skill: one practical, field-tested move for leading your team better in an AI world.

LinkedIn is noisy, frustrating, and often feels like a job search dead end—so why are we still using it? In this episode, Josh unpacks the tension we all feel about the platform, exploring what still works (profiles, networks, discovery) and what definitely doesn’t (spam, false leads, and the chaos of too much noise). Plus, a glimpse into how job seekers are starting to reclaim control—one Slack or Discord community at a time.About The Job Sh*t ShowThe Job Sh*t Show is an investigation into how finding, keeping, and doing work is changing faster than anyone is admitting — and what that actually looks like for real people on the ground. The Sh*t List is a paid Substack for leaders who want to stay ahead of what AI is doing to the workplace — and actually do something about it. Every other week, subscribers get a new AI Manager Skill: one practical, field-tested move for leading your team better in an AI world.

Managing an AI agent is like speed running my worst management habits. I get frustrated, I start micromanaging, and I catch myself acting like the boss I used to complain about.Boris Berenberg, founder of Zero AI, told me we're turning everyone into a manager now. There's no such thing as a non managerial role anymore.Jack Dorsey disagrees. He called middle managers "an information routing protocol," cut 4,000 jobs at Block, and said he wants all 6,000 employees reporting straight to him. If a manager is just a router, sure, a computer can replace that. But that's the part of the job that sucks anyway.Here's the part that doesn't. Gallup says managers account for 70% of the variance in employee engagement. Google's own research found coaching was the top trait of their best managers, and technical skill was dead last. We just never train for it.There's also a hard cap. One person can really know maybe eight people well. That's why good teams stay small. It's slow. It's also the whole job.So middle management might be the last job that still requires you to give a shit. Now that we're all about to manage a team of bots, the question is whether we show up as a coach or a router.About The Job Sh*t ShowThe Job Sh*t Show is an investigation into how finding, keeping, and doing work is changing faster than anyone is admitting — and what that actually looks like for real people on the ground. The Sh*t List is a paid Substack for leaders who want to stay ahead of what AI is doing to the workplace — and actually do something about it. Every other week, subscribers get a new AI Manager Skill: one practical, field-tested move for leading your team better in an AI world.

Josh's instinct is always to reach for a matching disaster from history. He thinks that's a trap, because by the time you pick the analogy, you've already picked the answer.He found the proof in a New York Times episode his wife sent him. The chief economics correspondent Ben Casselman laid out two futures for AI. One matched the internet, which displaced travel agents slowly enough that most just aged out. The other matched the China shock, which flattened entire manufacturing towns overnight.Josh traced where these comparisons actually come from. Mary Daly at the San Francisco Fed reaches for electricity, because for a central banker the moral has to be patience. Journalist Brian Merchant reaches for the Luddites, because the real history is nine years of factory workers petitioning Parliament before anyone smashed a loom. Neither side is lying. They're just telling you what they think their job is by which wreck they pick.Josh built a timeline of these disruptions for a talk in Seattle and checked every date against a source. Three of the most repeated stories turned out to be folklore. He kept them in anyway, labeled, because a myth people keep reaching for still tells you something true about what they're scared of.Follow The Job Sh*t Show investigation at https://joshlevine.substack.comAbout The Job Sh*t ShowThe Job Sh*t Show is an investigation into how finding, keeping, and doing work is changing faster than anyone is admitting — and what that actually looks like for real people on the ground. The Sh*t List is a paid Substack for leaders who want to stay ahead of what AI is doing to the workplace — and actually do something about it. Every other week, subscribers get a new AI Manager Skill: one practical, field-tested move for leading your team better in an AI world.