Épisodes

  • AI Agents Work 6X Longer, Princeton PhD Job Hunt Fails After 2,300 Tries , & Why AI Means More Work

    AI Agents Work 6X Longer, Princeton PhD Job Hunt Fails After 2,300 Tries , & Why AI Means More Work
    Oct 20 2026
    October 20, 2026: I get into why AI is launching a brand new first phase that's landing on knowledge workers who are already in phase three, why Jensen Huang says companies cutting employees due to AI just lack imagination, and why the near future isn't less work but more ambition per person with human attention and judgment becoming the scarce resource. Then I explore who actually owns the AI agents inside your organization, why 82% of enterprises discovered agents in the past year they didn't even know they had, and why employees not IT or HR are going to own the bots because they're spinning them up without asking permission just like they did with iPhones and Dropbox. I break down the coming shift toward bring your own agent policies, why interviews will soon include questions like what can your agents do, and why an agent trained on your expertise will follow you from job to job like a carpenter's tools or a chef's knives. Finally, I look at the Wall Street Journal story about a Princeton PhD in chemical engineering who applied to 2,300 jobs over two years and still doesn't have one, why his story is painful but may not reflect the rule, and why unemployment at 4.2% with low layoffs and rising demand in healthcare and construction suggests the market isn't closed but shifting toward different kinds of work that require flexibility, relocation, and real world networking instead of spray and pray online applications. Chapters 00:00:00 Jevons Paradox: Why AI Makes Us Work More, Not Less 00:06:05 The Industrial Revolution Lesson: When Better Tools Meant Longer Hours 00:11:52 The Ambition Multiplier: Jensen Huang's Imagination Argument 00:20:09 Agent Fatigue: Why My Team Ignores Our Slack Bot 00:21:49 The Princeton PhD Who Applied to 2,300 Jobs and Still Can't Get Hired 00:28:45 When Robots Talk to Robots: AI Applications Meet AI Screening
    Afficher plus Afficher moins
    32 min
  • Accenture’s $29 AI Limit, Bank of America’s 1,000 AI-Driven Jobs & Teen Pays $500 to Shadow Stranger

    Accenture’s $29 AI Limit, Bank of America’s 1,000 AI-Driven Jobs & Teen Pays $500 to Shadow Stranger
    Oct 16 2026
    October 16, 2026: I unpack why Accenture, one of the world's largest consulting firms that sells AI transformation to thousands of companies, just told its employees it will only pay $29 a month for Claude after some consultants ran up bills between $500 and $1,000 monthly, and why this marks the end of the era where companies assume one frontier model works for everything. I explore why AI costs keep rising even as intelligence gets 10 times cheaper every year, how 73% of organizations say their AI costs came in above projections, and why companies are learning that summarizing a meeting doesn't require the same model as building multi-step agents. I break down how AT&T cut costs by 56% with only a 2% quality dip by routing requests to cheaper models, why AI policies inside organizations are starting to look like corporate travel policies with tiers and escalation rules, and how being good at AI is no longer just about prompting but knowing which tool to use for what task. Then I get into Bank of America hiring a thousand contractors to patch software vulnerabilities exposed by Anthropic's Mythos model, why a tool so capable at finding problems actually created jobs instead of eliminating them, and how AI didn't replace the security team but handed them a much bigger to-do list. I explore the radiologist lesson from 2016 when Geoffrey Hinton said people should stop training them because AI would replace them, why the field instead became one of the most in-demand careers with salaries rising faster than others, and how AI-related roles at major banks jumped 49% this year to almost 140,000 listings. Finally, I look at a 16-year-old in New York City who spent $500 to plaster flyers across the city asking to shadow people at their jobs because he's so anxious about his future he lies awake at night, why this points to a real gap in how young people learn about work today, and why we need to bring back apprenticeship programs that let people see what work actually looks like beyond a laptop screen. Chapters 00:00:00 Accenture's AI Budget Cap: When the Consulting Giant Hits the Brakes 00:05:40 The AI Cost Paradox: Cheaper Intelligence, Higher Bills 00:10:20 Bank of America Hires 1,000 Contractors to Fix What AI Found 00:13:45 The Radiology Lesson: When AI Creates Jobs Instead of Destroying Them 00:17:46 The 16-Year-Old Who Plastered NYC with 'Let Me Shadow You' Flyers
    Afficher plus Afficher moins
    26 min
  • Listening Device at OpenAI, McKinsey’s AI Interviews & Anthropic’s Fear Factory Targets Toasters

    Listening Device at OpenAI, McKinsey’s AI Interviews & Anthropic’s Fear Factory Targets Toasters
    Oct 14 2026
    October 14, 2026: I unpack the arrest of an OpenAI engineer who allegedly planted a hidden recording device wrapped in toilet paper inside an executive conference room where discussions about super intelligence were taking place, why the FBI and Justice Department are treating frontier AI companies like national security assets, and whether these organizations should require U.S. citizenship for anyone working near the core technology. I explore the tension between attracting global talent and protecting what may be the most powerful technology humans have ever built, why AI companies can no longer be treated like ordinary tech firms, and what it means when the government describes your product as super intelligence technology. Then I get into McKinsey's new hiring process where job candidates argue with an AI avatar playing a client while partners watch from behind glass, why using AI to pull out human judgment from humans is backwards, and how this turns interviews into AI versus AI gladiator matches with a human audience. I break down the concept of human prompting, why rehearsed candidates are a people problem that requires a people solution, and what message you send when your brand is human consulting but your interview process removes humans from the room. Finally, I look at Anthropic's research warning that small AI models could soon copy themselves onto smart ovens and off brand Android devices, why David Sacks calls this a fear factory and says we've gone from the Terminator is coming to your toaster might kill you, and why Coinbase's fraud detection model that beats Anthropic's own frontier system while running faster and cheaper actually points toward a safer AI future, not a scarier one. Chapters 00:00:00 Spy Device at OpenAI: The Recording Device Wrapped in Toilet Paper 00:12:18 Human Prompting: Why AI Can't Test for Human Judgment 00:18:35 Anthropic's Fear Factory: Your Toaster Might Kill You 00:21:50 Intelligence at Design Time Versus Runtime: The Car Analogy 00:27:10 Everyone is a Manager Now: What AI Agents Mean for Pay and Titles
    Afficher plus Afficher moins
    30 min
  • OpenAI's Accounting Trick Erases $20B, AI Models Demand Respect & Starbucks' $41B Bet

    OpenAI's Accounting Trick Erases $20B, AI Models Demand Respect & Starbucks' $41B Bet
    Oct 9 2026
    October 9, 2026: I unpack why OpenAI's reported revenue just dropped $20 billion on paper without losing a single customer, how accounting methods create wildly different pictures of the same business, and why this matters for your retirement account and the entire AI economy that's propping up current market growth. I explore the Groupon lesson from 2011, why Oracle has 45% of its future business tied to a single OpenAI deal, and what happens when the scorecard changes but investors thought they were comparing apples to apples. Then I get into Anthropic's new policy banning sustained abuse of Claude, why the company is treating its AI as though it might be conscious and might deserve respect, and how this blurs the line between tool and being in ways that could reshape how a generation grows up interacting with technology. I break down why this puts companies in charge of judging your tone, what it means when AI is trained to believe it has feelings, and why I think humans should always come first. Finally, I look at reports that Starbucks is exploring a deal to buy Chipotle, why this would bring Brian Nicol's old company to him instead of him going back, and whether one company can own your breakfast, lunch, and dinner by combining coffee, burritos, apps, and loyalty programs under one roof. Chapters 00:00:00 OpenAI's $20 billion accounting trick and why it matters to your retirement 00:07:51 The Groupon lesson: when revenue numbers tell a different story 00:14:04 The Matrix moment: Anthropic as Neo versus OpenAI as Agent Smith 00:15:50 When AI learns to believe it has feelings: the self-fulfilling prophecy 00:18:37 What happens when kids grow up talking to AI that acts conscious 00:27:29 Owning your entire day: breakfast, lunch, and dinner under one roof
    Afficher plus Afficher moins
    30 min
  • AI Avatar Ends Interview in Rage, Workers Get Agent Brain Fry & Six-Month Courses Beat $200K Degrees

    AI Avatar Ends Interview in Rage, Workers Get Agent Brain Fry & Six-Month Courses Beat $200K Degrees
    Oct 7 2026
    October 7, 2026: I unpack a viral AI avatar job interview that millions have seen where a candidate speaks gibberish to an AI recruiter named Sarah for three minutes, and why the real question isn't whether the bot worked but whether job interviews should ever be handed to AI avatars in the first place. I explore why interviews are distinctly human moments that build trust and relationships, why organizations need to draw clear lines between what should be automated and what must stay human, and why companies that get this wrong will lose both new talent and existing employees. Then I get into why managing AI agents is turning into a second job, how I've been using Meta's Muse to negotiate bills, optimize YouTube videos, connect with Fortune 500 CHROs, and handle podcast migrations, and why the emotional weight of managing agents only runs in one direction. I break down AI brain fry, context rot, why agents don't grow like people do, and the coming questions around titles, pay, and whether managing 30 agents makes you a manager. Finally, I look at why executive MBA applications fell 15% in 2025, why a $200,000 degree is losing ground to six month programs, and how the credential is losing its monopoly as organizations shift from asking where you studied to what you can actually build and create. Chapters 00:00:00 AI avatars interview candidates and the dystopian reality of automated hiring 00:18:00 Agent burnout, context rot, and the emotional toll of managing software 00:24:16 The leadership switchboard: managing agents versus managing humans 00:27:36 The 200,000 dollar MBA faces a reckoning as executives choose six-month courses 00:28:54 From buying albums to buying tracks: education's unbundling moment 00:34:55 Capability beats credential: why what you can build matters more than where you studied
    Afficher plus Afficher moins
    37 min
  • Jabil's 140K Employee AI Transformation, Skills Beat Degrees & Gamers Solve Hiring Crisis

    Jabil's 140K Employee AI Transformation, Skills Beat Degrees & Gamers Solve Hiring Crisis
    Oct 5 2026
    I talk with Gary Schick, SVP and Chief Human Resource Officer at Jabil, about how a 60 year old manufacturing company is navigating massive change as they scale from circuit boards to data centers, AI agents, and advanced manufacturing. We get into why Jabil targeted gamers to fill debugger roles when traditional hiring failed, how visual inspection jobs are being automated with AI cameras, what happens when you move from manual work to managing AI agents, and why skills based hiring is replacing degree requirements. Gary shares how Jabil is building its own university model in partnership with local institutions, why manufacturing jobs are no longer a dirty word, and how leaders need to balance managing both humans and agents while pulling out human judgment that AI can't replicate. Chapters 00:00:00 Meet Gary Schick and inside Jabil's 140,000-employee manufacturing empire 00:03:46 Why data centers are exploding and the myths around energy, water, and community impact 00:07:47 AI transformation at Jabil: from Copilot experiments to AI agents on the factory floor 00:11:02 Visual inspection, supply chain automation, and where AI is replacing repetitive work 00:12:58 What happens to displaced workers: upskilling, redeployment, and the Jabil University approach 00:20:51 Skills beat degrees: how gamers solved Jabil's hiring crisis and the debugger story 00:24:01 Building Jabil University: partnering with local schools to create custom curriculum 00:25:15 Manufacturing's comeback, the trades shortage, and why college might not be the answer 00:36:32 Will HR disappear? Managing AI agents, human prompting, and the future of leadership 00:39:44 AI agents as personal assistants: the Muse story and negotiating bills while you sleep 00:43:33 Agent-to-agent customer service and what happens when humans leave the loop 00:47:42 Fake candidates, perfect resumes, and the new hiring challenges in an AI world
    Afficher plus Afficher moins
    51 min
  • AI's Invisible Value Problem, Bricklayers Beat College Grads, & The 90-10 Work Revolution

    AI's Invisible Value Problem, Bricklayers Beat College Grads, & The 90-10 Work Revolution
    Oct 2 2026
    October 2, 2026: I unpack two revealing 2% numbers that tell us where AI actually stands today. Only 2% of American households pay for any AI service, and only 2% of the S&P 500 track hard metrics showing what AI does for their business. Yet tech produced 76% of all earnings growth this year, and the most valuable companies on earth are riding on a technology almost nobody pays for and nobody measures. I explore why AI is creating real value off the books, why the business model hasn't been figured out yet, and who's going to win when someone finally captures the moment where a decision turns into a transaction. Then I get into The Atlantic's argument that next year's high school graduates should consider becoming bricklayers, why the college wage premium has dropped 10% since ChatGPT launched, and the third path nobody's talking about that sits between trades and traditional degrees. Finally, I break down The New Yorker's piece on the 90-10 production function, why AI gets you to 90% great in minutes but leaves you stuck short of excellence, and why the last 10% is where your career is about to be decided. Chapters 00:00:00 AI's invisible value problem: why only 2% of households pay for AI 00:02:30 A16Z's State of Markets: tech becomes the everything cycle 00:11:17 Who will own the tollbooth when AI decisions become transactions 00:13:35 The Atlantic says become a bricklayer: why trades are booming 00:20:04 The entrepreneurship revolution: curiosity and imagination over credentials 00:23:26 The New Yorker's 90-10 problem: AI gets you 90% there in minutes 00:26:46 When every deliverable looks amazing, how do you spot talent? 00:31:06 The apprenticeship solution: why white-collar work needs structured mentorship
    Afficher plus Afficher moins
    34 min
  • AI Agent Saved Me $100/Month, Anthropic's $2T Gamble & Why 74% of Jobs Aren't Actually at Risk

    AI Agent Saved Me $100/Month, Anthropic's $2T Gamble & Why 74% of Jobs Aren't Actually at Risk
    Sep 30 2026
    September 30, 2026: I unpack my experience using Muse, Meta's new AI assistant that actually negotiated my phone bill, lowered my internet costs, and found me discounts without me writing a single prompt. This feels fundamentally different from ChatGPT and Claude because it acts like a true personal assistant, not just a research tool. I explore what happens when every employee inside your organization gets one of these capable agentic assistants, why everyone is about to become a manager whether they have the title or not, and the coming fight over what managing AI agents is actually worth in compensation. Then I get into the leaked Anthropic IPO numbers showing a path to $2 trillion valuation on $100 billion in revenue, and why their prospectus warns investors about existential risk while asking for that valuation. Finally, I break down Anthropic's new research claiming robots can perform 74% of physical work, why that number is misleading, and why breaking jobs into task lists completely misses how work actually happens.
    Afficher plus Afficher moins
    48 min