AI, Math, and the Edge of Ability | BOSS MODE #030

Is Heaven On Earth Boring? | BOSS MODE #033

Heaven is supposed to be the answer. So why does every fictional version of it — Dune's banned computers, the Culture series' bored citizens, Ready Player One's gluttony chair — read as a warning? In this episode, David Lee and Dee Deng wrestle with the limitation problem: if struggle creates diamonds, what happens when the pressure is removed? David argues that omniscient, all-powerful beings would crave only one thing — limitation — and uses Aaron Sorkin's "intention and obstacle" framework to explain why the Ready Player One gluttony chair is the more likely default than the creator-class utopia. Dee pushes back with the diminishing-returns problem: the rags-to-riches stories that built modern civilization all started in struggle, and most humans, given the choice between earning a CEO seat and being born with a silver spoon, would pick the spoon. The conversation moves through Socrates' case against democracy, why Massachusetts went from top of the country in company formations to dead last, the three layers where AI impact actually matters, and the thought experiment that crystallizes the whole problem: if someone handed you 1,000 employees tomorrow morning, would you know what to do with them? David's answer is that 9 out of 10 of the top-10%-educated people in developed nations wouldn't have a clue — which is why agentic AI is so hard to wrap heads around, and why the education system needs to change before anything else does. Plus: the Musk v. Altman trial, why Codex is suddenly viable, and a working theory on managing technical debt when you're vibe-coding at non-human speed. Chapters 00:00 Mother's Day, Arrival, and Dune's nuclear option on computers02:42 The Hollywood AI canon: why every fictional future ends badly04:43 What would omniscient beings actually crave?05:50 Aaron Sorkin's "intention and obstacle" as a theory of being human07:11 The matrix monologue and Ready Player One's gluttony chair08:47 Sales commissions, fuck-you money, and the limits of drive10:57 Why most people are conditioned to consume, not create13:51 Where does pressure come from in abundance?15:18 From physical coercion to economic coercion to what comes next17:32 Peacock feathers in the abundance era18:59 Most people will pick the VR chair, not the creator path21:34 Identity politics as a luxury of developed nations22:56 Socrates, persuasion over thought, and democracy in abundance24:46 Three buckets where AI is already cutting costs27:00 The three layers of AI impact30:30 Why we're beating curiosity out of children32:58 Ray Dalio's debt cycle and the next superpower question35:30 Chips, Taiwan, and the wrong levers38:00 Massachusetts: first to dead last in company formations41:00 Why small business is the messiest, most consequential layer43:00 Net job creation is hiding the truth44:20 Could AI adoption end up like blockchain?47:25 Crypto as a ledger problem49:32 The 1,000-employees thought experiment52:25 Open briefs and the muscle of setting direction53:50 Musk v. Altman: the trial and the meta-battle57:11 Kevin O'Leary, Utah, and the data center backlash58:46 The Larry Page birthday party and the founding of OpenAI1:02:42 Mira Murati's new lab and the OpenAI exodus1:03:08 Why Codex is suddenly good1:05:00 Grok Build, the Cursor deal, and three coders at once1:08:03 The Mac Mini as underrated AI workhorse1:09:39 Vibe coding's technical debt and the spec-first workaround1:14:08 Contextual retrieval, agentic search, and graph RAG1:16:56 Obsidian + QMD: minimum viable semantic search Subscribe for new Boss Mode episodes. If you've run the 1,000-agents thought experiment on yourself and have an answer — drop it in the comments. David Lee https://www.vanka.ai & https://www.x.com/davidyhlee Dee Deng https://www.righthookdigital.com & https://www.x.com/realdoseofdee

Manage Or Be Managed | BOSS MODE #032

What does it actually take to thrive in a world where Microsoft can wipe out a $5B legal AI company with a single Word feature, where Uber drivers are unknowingly training the robots replacing them, and where the gap between "IC" and "manager" is no longer a career step but a survival skill? In this episode, David Lee and Dee Deng dig into the management skill that's quietly become table stakes for everyone: directing the work of agents, human and AI alike. Dee unpacks the course he just built for new managers and why the leap from individual contributor to people manager mirrors the leap every knowledge worker now faces. David lays out his "leveling thesis" — the three buckets crushing the middle class (housing, transportation, food), why technology is already collapsing two of them, and why the techno-optimist version of abundance is dangerously vague about who it's actually for.We also get into the conviction problem: how do you place 6-to-12-month bets in a market where Harvey-killers ship overnight? David's answer borrows from Wayne Gretzky and his own two decades of being "too early." Dee's answer is messier and probably more honest: you process out loud, you use AI as a tireless thought partner, and you accept that the puck moves faster than it used to.If the last episode was about the gap between Basement Dweller and Creator Class, this one is about the muscles you build to make the crossing.Chapter Index00:08 Heads down on humans: building a course for new managers01:05 The IC-to-manager gap: when work stops living in your laptop01:53 Managing humans vs managing AI agents (and how transferable the skills are)03:39 Day one context loading: why onboarding logic applies to LLMs too04:48 Asian Neo, Hollywood-grade output, and the director's chair from your chair06:11 Course and Chill: why coordinating humans still makes 2+2=507:20 The unsolved problem: getting agents (human and AI) to compound08:30 The Gary Tan thread: housing, transportation, food, and the cost of living10:48 Why transportation is the easiest bucket and Boston is the hardest place13:00 Food, FSD, and human apathy in the supply chain15:11 RFK on Joe Rogan: lasers, precision hydration, and the strawberry whisperer18:43 The Leveling Thesis: why "techno-optimism" is too vague to be safe21:21 "People" and "us": who exactly does abundance accrue to?22:42 The agent-of-my-intent framework: human or AI, the skill is the same25:30 Sam Altman's Dubai trail-off: the meaning deficit nobody is solving27:48 Frederick Taylor's ghost: the education system still optimizes for compliance28:18 The bee parable: compliant, coordinated, and easy to exploit30:59 What does a company look like when every member is high-agency?32:13 Exclusive vendors in employee clothing: the new executive contract33:30 Head-mounted cameras in India and Ubers in SF: training your replacement35:43 The head-in-the-sand problem: bandwidth, overwhelm, and survival38:00 Stop chasing features, start tracking trajectory40:24 Why this is individual responsibility — and why that doesn't let us off the hook41:17 A week in David's life: living 12 months ahead instead of 15 years ahead46:40 X, community notes, and why the news business model is structurally corrupt49:33 "I feel so behind": the universal refrain and what to do about it50:10 Gretzky, white space, and the knowledge graph bet that's now consensus53:02 Conviction in a market where Harvey becomes a Word feature overnight55:17 AI as a thought partner you can never wear down57:01 Disagreeing with Claude without hurting its feelingsConnect with your hosts:David Lee https://www.vanka.ai & https://www.x.com/davidyhleeDee Deng https://www.righthookdigital.com & https://x.com/realdoseofdeeSubscribe for new Boss Mode episodes. If you're building toward high-agency work — solo, in a small team, or inside a larger org — drop a comment with where you're placing your 12-month bet. We read them.

Basement Dweller to Creator Class | BOSS MODE #031

Anthropic is dropping bombs every week. Claude Design, Claude Routines, Claude Code, and a 20x Max plan that lets a single human run circles around teams of ten. Meanwhile only 26% of Americans say they're favorable toward AI, a 20-year-old just threw a Molotov cocktail at Sam Altman's house, and Australia banned social media for under-16s. Something is breaking, and something is being built.In this episode, David Lee and Dee Deng dig into the real story underneath the headlines: that for the first time in human history, scaled intelligence is no longer the exclusive privilege of the few. We walk through the five levels of AI usage (chatbot, copilot, agentic, agentic swarms, recursive agentic swarms), why Anthropic is structurally different from OpenAI and Google, why the K-12 system is a lie that strips kids of agency, and whether the future tilts toward more net creators or more couch potatoes plugged into Ready Player One.If you have ever wondered what Boss Mode actually means at the species level, this is the episode.Chapter Index00:00 Anthropic on a tear: Claude Design, Routines, and Google's DesignMD01:17 Two vectors of scaled intelligence (and why both used to be expensive)02:22 MCP, tool calling, and the on-ramp for the mass market04:12 Was Claude actually nerfed? Bugs, cash flow, and competitive pressure05:30 The 26% problem: America's AI PR crisis and the romance of the working man07:00 Creators as newscasters: AI slop, discovery, and performative curiosity10:00 Short-form video, X, and community notes as social course-correction12:30 The $20, $100, and $200 Claude tiers, and the spectrum of AI expertise15:48 Plan mode, context management, and the green-to-fluent curve17:00 Why Anthropic has to make money (and OpenAI, Google, and xAI don't)20:00 The shift from doomerism to frothiness in the AI zeitgeist21:33 Boss Mode for everyone: the few-at-top, many-at-bottom collapse23:05 Can the human mind even envision a non-hierarchical society?25:57 Mental slavery, ingrained hierarchy, and "someone needs to do something"27:03 To be human is to create: agency, empowerment, and the better world order30:00 The Altman attack, zealotry, and the cost of hopelessness36:00 Skynet kid: the irony of needing AI to fight AI37:13 Abundance economics, food prices, and who actually benefits from automation39:00 The five levels of AI usage and what they unlock42:30 Net consumers vs net creators: where does the distribution curve land?47:00 The filmmaking analogy: 5D Mark II, YouTube vs Vimeo, and democratized tools53:00 Potatoism: when the means get cheaper, does the will disappear?55:35 Ready Player One, the obesity epidemic, and engineered hedonism58:00 The K-12 lie: why we beat agency out of kids and how to stop1:00:30 YouTuber dreams, Australia's social media ban, and what "creator" should mean1:04:09 Buckle up: why the next three years will feel violently fast ConnectSubscribe for new Boss Mode episodes. Drop a comment with what you're building, what you're stuck on, or which level of AI usage you're operating at this week.

AI, Math, and the Edge of Ability | BOSS MODE #030

AI, Math, and the Edge of Learning Min Lee is CEO of Prep Anywhere, an elite math prep academy. A year ago, he was spending half a million dollars on developers and drowning in legacy code. Today, he's rebuilt the entire app from scratch using Claude Code — not a single line of the old code remains. David & Min get deep into what vibe coding actually looks like when you're building a real product, not a toy. Then the conversation shifts to what Min is really solving for: keeping students at the edge of their learning ability. David connects this to the broader crisis in education — factory schools that beat curiosity out of children, a social contract built on lies, and why the bifurcation of society will be defined not by wealth but by who directs AI versus who is directed by it. CHAPTERS0:00 — Intro: Min Lee and Prep Anywhere0:33 — From October Skeptic to Full Vibe Coder3:01 — Half a Million in Dev Costs, Zero Lines of Code Remaining4:26 — Min's Background: Double Geek (Math + CS from U of T)6:19 — Canadian R&D Credits: SHRED and IRAP for Startups9:02 — The First Aha Moment: Refactoring Legacy Code in Two Days11:18 — Why Under-Specifying Kills Vibe Coding Projects13:15 — Treat It Like an Experiment: Failure Is Nearly Free Now14:47 — Iterative Spec Building With AI17:40 — The Stack: Claude Code, Gemini for Second Opinions, Codex for Cost19:04 — Why Anthropic's Tokens Burn Faster (and the Financial Reality)20:40 — Sub-Agents: Managing a Swarm While You Sleep22:18 — The House of Cards Problem With Overnight Agent Runs24:03 — The Five Levels of AI Usage26:17 — Open Mythos: An Open Source Interpretation of Anthropic's Approach27:10 — Using Thinking Tokens to Measure Cognitive Load29:34 — Capturing Every Stroke: Real-Time Learning Analytics30:01 — The Harvard Study: Low-Tech Note-Taking vs. Digital32:26 — Why iPad + Stylus Is the Sweet Spot34:05 — Teaching at the Edge of a Student's Ability37:41 — South Korea's Hagwon Model and Why 70% Are Just Memorizing39:47 — Personalized Education Is Now Affordable (In Theory)42:18 — The Thought Exercise: Why No Financial Model Can Serve Every Student44:45 — The Male Brain, Development, and Knowing When You're Being Stretched45:52 — Defining Intelligence: Input, Synthesize, Output47:11 — Kids Who Can't Think Critically Will Create Junk With AI48:25 — The Bifurcation: Who Directs AI vs. Who AI Directs50:32 — AI Is Like Money: It Reveals Your True Character51:39 — The Five Levels Applied: Boss Mode as a Societal Skill53:07 — From Slave Labor to Digital Labor: The Biggest Shift in History56:23 — 99 Out of 100 Students Are Using AI Wrong58:04 — Curiosity Has Been Beaten Out of Our Children1:00:13 — Social Intelligence and Stretching the Mind1:02:16 — The Tight Feedback Loop: Addictive and Exhausting1:04:06 — Exercise, Human Connection, and Managing the AI Addiction1:06:07 — Should Kids Use AI? When and How1:10:33 — The Evolutionary Fork: Humans Are Becoming a Bifurcated Species1:12:26 — Young Children Need to Touch Grass, Make Music, Play Sports1:14:04 — Metacognition: Observing How You Learn1:16:08 — Video Analysis as a Gateway to Systems Thinking1:17:00 — Book Recommendations: George Mumford's The Mindful Athlete and Unlocked Boss Mode Podcast with David Lee and Min Lee https://www.prepanywhere.comLearn more about Vanka: https://www.vanka.ai Connect with David:https://x.com/davidyhleehttps://instagram.com/davidyhleehttps://www.linkedin.com/in/davidyhleeConnect with Min:https://www.linkedin.com/in/min-lee-2812a419/ #BossModePodcast #VibeCoding #ClaudeCode #EdTech #MathEducation #AI #OwnershipEconomy #PersonalizedLearning #FutureOfEducation #Entrepreneurship #Vanka #FlowState #EdgeOfAbility #GeorgeMumford #Unlocked #MindfulAthlete

From Searchable to Transactable | BOSS MODE #028

From Searchable to Transactable Jimmy Gibson has spent 20 years helping businesses grow through digital marketing as the leader of Thrive Agency. In this episode, David and Jimmy explore what happens when AI doesn't just change how people search — it changes whether people search at all. The conversation starts with the current state of SEO: 60% of searches now end in zero clicks, LLM referrals convert at 4.5x the rate of traditional Google search, and Gartner projects organic search traffic will drop 50% by 2028. Jimmy breaks down what businesses should be doing right now to stay visible in AI-driven discovery. Then David pushes the conversation forward. If AI search is about making your business recommendable, the next phase is making it transactable — exposing API endpoints so AI agents can book appointments, check inventory, and complete purchases on behalf of consumers. The CEO doesn't Google the office supplies. Someone else handles it. Soon, every consumer will operate the same way. They also dig into Jimmy's five finger test for authentic content, the case for public knowledge graphs, why the volume of content no longer matters but your point of view does, and how AI is democratizing the apprenticeship that most people never had access to. CHAPTERS0:00 — Intro: Jimmy Gibson and Thrive Agency1:48 — Building a Brand Before Running Ads2:42 — Paying for Ads Is the Price of Not Being Remarkable4:11 — AI Search: What's Actually Happening to Website Traffic6:18 — The Search Results Page Has Become More Useful8:40 — AI Search Is Already Dead (Here's What David Means)10:00 — The CEO Analogy: From Searching to Delegating12:25 — Exposing API Endpoints to AI Agents13:25 — Most Businesses Haven't Even Opened ChatGPT15:16 — Recommendable vs. Transactable18:36 — Structured Data, Schema Markup, and What AI Is Parsing20:12 — From Recommendable to Transactable: The Full Picture23:13 — Not Everything Needs to Be Agentically Transacted25:41 — The Public Knowledge Graph: Radical Honesty at Scale28:30 — The Five Finger Test for Authentic Content32:50 — It's Getting Harder to Fool AI34:49 — Third-Party Evidence: Reviews, Podcasts, LinkedIn, Reddit37:25 — Being Brave Enough to Share Opinions When You're Rational39:51 — The Five Levels of AI44:58 — Agentic Swarms and the Oversight Question45:23 — Why College Grads Are Losing to AI Chatbots48:44 — The Social Contract Was a Lie50:38 — The Ownership Economy52:23 — Claude Just Built Me a Design Curriculum in Minutes54:35 — AI as the Infinitely Patient Teacher57:00 — South Korea's EBS: Democratizing Knowledge Before Khan Academy1:00:47 — Break Bread Together1:01:27 — Meeting Neurodivergent Learners Where They Are1:04:02 — What Educators Think About AI in Schools1:05:46 — Why Writing Still Matters and How AI Can Teach It Better1:10:54 — Closing: AI Search Isn't Dead Yet, But What's Next Is Close Boss Mode Podcast with David Lee and Jimmy Gibson Learn more about Thrive Agency: https://www.thriveagency.comLearn more about Vanka: https://www.vanka.ai Connect with David:https://www.linkedin.com/in/davidyhlee https://www.x.com/davidyhlee https://www.instagram.com/davidyhlee Connect with Jimmy: https://www.linkedin.com/in/jimi-gibson/ #BossModePodcast #SEO #AISearch #AgenticAI #Entrepreneurship #FutureOfSearch #OwnershipEconomy #ClaudeCode #DigitalMarketing #FiveLevelsOfAI