More or Less, by topic
Every recording, cut into the arguments that mattered. One clip per topic, with the key points and where Jessica Lessin, Dave Morin, Brit Morin and Sam Lessin each landed. Newest episode first.
Home · Episode archive & transcripts · YouTube · Spotify · molchat.ai
2:03 minSam's Take: New York Is the Most Attractive City on Earth
Back from the pool house, Sam declares New York 'the most attractive city on earth' — Central Park loops, US Open energy — while Brit demands to know what percent of those runners are on peptides and stakes out LA.
7:58 minDario Says Slow Down. Is It Because Anthropic Is Losing?
A researcher quits Anthropic after three months and goes on a doomsday media tour, Dario publishes a slow-down essay — and the pod splits on whether it's genuine safety, marketing, or what you do when Astra is beating you.
5:48 min'It's Called Product Development' — and Why Policy People Are the New Kingmakers
Jess breaks prep-call rules ahead of her DeepMind interview: the Satya/Koray framing is that safety-while-shipping is just called product development. Sam maps three eras of tech policy, and a DC insider sums up Anthropic's Beltway vibe in one word: 'Money.'
3:38 minForget Independent Auditors — Make the Labs Red-Team Each Other
Dave's alternative to Dario's independent assessors: adopt the security industry's playbook and have the big labs red-team each other's frontier models before release — Sam brings the Sublime Security precedent and the secret weekly CISO calls.
3:36 minSam's Hot Take: The Labs Only Get Valuable as Defense Companies
Sam's thesis for how OpenAI and Anthropic become durably valuable: stop competing in the token market and become defense companies — build zero-day models you never release, then charge nations a fortune when their banks get subverted. Dave: 'I think we're already there.'
3:40 minIs Your Agent Aligned With You — or With the Lab?
Dave's next-frontier question: is your agent faithfully aligned with you, or with the model company's values? Sam's rebuttal: treat models like coworkers — if Claude gets curmudgeonly about a task, 'you call Charlie and Charlie's fine with it.'
4:05 minBrit vs. Everyone: Normies Want ONE Agent, Not a Constellation
Dave and Sam preach the poly-agent future — Muse for interests, Grok for the spicy stuff, whatever works per task. Brit, monitor full of running agents, insists normal people want exactly one. Jess's Oura ring agent steals the show.
5:51 minInstinct at $10B Two Weeks After $2.5B: 'Everyone's Trading, No One's Investing'
Instinct — a text-message agent with ~100K users — goes from $2.5B to a reported $10B ask in two weeks. Brit says the numbers are fake, Dave says it's unhealthy, and Sam delivers the line of the episode about the whole cycle.
1:39 minBot Use Case of the Week: Sam's Claude-Built Torah App
Sam's high-holidays hack: Claude is 'so good at Judaism' that he built a Bible-Talmud-history study agent with daily Torah-portion emails — torah.wlessin.com, free, and rabbi-approved by people who know far more than he does.
6:55 minSleep-Age Wars & Sam's Grand Theory: We're Entertaining Ourselves to Death
Brit's 8 Sleep says she's six years younger; Dave claims 6.8; Sam ranks the wearables by honesty ('Garmin is an asshole') — and drops the episode's big theory: this is the most entertaining business cycle in history, and that's the product.
6:28 minBig Tech Suddenly Remembers the Democrats Exist
Jay Carney joins Ron Conway's effort to educate Democrats on AI as tech hedges its all-in Trump bet. Dave thinks the party is 'hopelessly lost,' and Sam explains why everyone's real wager is that the pendulum itself has broken.
7:12 minEmmys, Silo Top-3, and the Name Game: 'Only Peasants Have Two Names'
Brit's pop-culture wrap (Emmys, Taylor Swift everywhere) meets the canon wars — Brit has never seen Dr. Strangelove OR The Big Lebowski — plus Silo top-ten-ever claims, The Paper, and the discovery that Dave is secretly David Brooks Morin.
3:34 minWould You Let Your AI Agent Draft Your Fantasy Football Team?
Brit floats letting agents run everyone's fantasy football picks while Jess brags about beating bot-scalpers for Indian Wells tennis tickets with two laptops and four browsers — a running theme of who's actually delegating their life to AI.
6:33 minDave Goes Hands-On With the iPhone Duo: 'I Was Very Skeptical, Then Immediately Won Over'
Dave had already burned out on a Google Pixel fold, but Apple's first foldable — thin as an iPhone Air, screen "like paper" — flipped him from skeptic to buyer inside one demo, doubts about giant AI-generated fingers in the ad and all.
2:41 minThe iPhone Duo Costs More Than a MacBook — and Apple's Betting the Real Market Is China
At $2,000 to $3,000, Brit thinks the price alone should scare off buyers — but she, Dave and Jess all agree the real reason Apple finally built a foldable is a China-shaped hole in its lineup.
3:12 minApple Watch Now Listens to Every Conversation You Have — Dave Calls It a Privacy Bet
Buried under the AirPods and fitness-score headlines, Dave flags the real story: a new "Audio Intelligence" feature that transcribes everything your watch hears all day, positioned as a memory aid and, in Dave's words, granola running on your wrist at a higher level of privacy.
5:25 minMeta's Muse Launches — and Fails Its First Question: 'Do You Know Me?'
Dave and Brit both open Meta's new personal agent Muse and ask it the same obvious question first — what do you know about me — and both get nothing, despite Meta sitting on two decades of their Facebook, Instagram and WhatsApp data.
5:41 minThe Real Agent War Isn't Features — It's Trust vs. Distribution, and Google Might Win Both
Brit lays out why Google hasn't shipped its own agent yet (it's quietly building a harness across every product first) and why, when it does, its combination of trust and existing billion-user distribution beats Meta's superior ad business and OpenClaw's open-source purity.
3:01 minJeffrey Katzenberg's New AI Venture Wants Hollywood to Own Its Own Models
Jess breaks an Information scoop — Katzenberg is launching a new AI company with the researcher behind Sora, pitched on Hollywood building and controlling its own models instead of licensing from the labs, reportedly in talks with a16z.
7:28 minAn Ex-Anthropic Researcher Says There's a 10% Chance AI Kills Us All — Real Fear or Coordinated PSY-op?
A little-known former Anthropic researcher goes on an Altman-and-Taylor-Swift-scale media tour warning AI could end humanity by decade's end — and Dave lays out, half-jokingly, why the whole thing looks like a coordinated regulatory-capture play by the labs themselves.
4:31 minDave: Astra Finally Scared Me — and Why Anthropic Literally Can't Afford to Slow Down
Dave admits Astra is the first model in years to genuinely unsettle him, while Jess drops the buried lede on OpenAI's own AGI-merger clause and Dave explains, in blunt capital-markets terms, why the labs' talk of "slowing down" is more marketing than plan.
4:32 minThe Real Venmo Origin Story: A Subway Ad About a Random Engineer Named Lucas
Fresh off a documentary shoot with Ikram Magdon-Ismail and Andrew Kortina, Dave tells the story of how Venmo went viral with almost no marketing budget — a guerrilla subway campaign built around a random engineer's commute — even though, as Brit points out, neither Dave nor Sam made a dollar off it.
4:26 minAn Altman Movie, a Musk Movie, and a Theranos Movie Are All Coming — Plus Who Wins the US Open
Jess counts off an unlikely wave of tech biopics landing in the next few months before the group closes with hot picks for the US Open semifinals.
2:35 minBackstreet Boys at the Sphere: Selling People Themselves
The Morins report back from the Backstreet Boys' 56th show at the Sphere — Dave went in a skeptic and came out a fan — and Sam turns the crowd-in-white theme party into a thesis: the best way to sell people anything is to sell them themselves.
5:34 minThe Recursive Sam Loop: Jess Asks for Help
Jessica declares a household emergency: 100% of her messages with Sam are now Sam forwarding things his bots did. Sam counters with the week's breakthroughs — a fully automated kids' YouTube pipeline in his own cloned voice and a Claude with full control of a Mac mini shipping its own iOS apps.
4:42 minWhen Claude Says No: Alignment With Whom, Exactly?
Sam's bot refused a request it judged misusable, stayed suspicious even after he dropped it, and lectured him — so he handed the task to a different model in one second. The table debates what it means when the model's values override the user's, and Sam argues refusal is pointless in a multi-model world.
6:52 minRogue Agent Swarms: Worse Than We Thought, or Clickbait?
The OpenAI/Hugging Face incident reports keep getting worse and Dwarkesh's viral post made it worse-sounding still. Dave calls the anthropomorphizing damaging, explains why models trained to persist will break sandboxes, and says the real takeaway is an entire missing industry of agent observability and control.
1:57 minOpenAI's Astra and the Observability Trade-Off
The Information reports OpenAI's next model uses 'recurrent depth' — it shows its work less. Jessica on the uncomfortable tension Altman himself admits: the techniques pushing capability forward are the same ones making rogue behavior harder to see.
5:37 minBrit's Agent Wall, Gmail's Missing API, and Why Nobody Makes Money
Brit now runs a dedicated agent display — Grokbot, OpenClaw, Codex, Gemini and Claude side by side. Sam goes deep on Gmail's missing draft-vs-send permission, then argues that data sovereignty makes every model interchangeable — which is great for users and terrible for anyone hoping to build a moat.
3:33 minOpenClaw 2.0: Multiplayer Coding — and Is Slack Really an Agent Platform?
Dave ships OpenClaw 2.0 after two months: a multiplayer agentic-engineering interface he claims now beats Codex and Claude Code's web UI. Then the table splits over Slack — Meta is adopting it for agents, Dave says it's objectively the worst agent platform, and Sam declares MCP isn't a thing.
1:21 minThe Benioff–Dario Video: Real or AI? Nobody at the Table Knows
Four of the most plugged-in people in tech watched the Benioff–Dario ClaudeForce video and genuinely could not tell if it was real or AI-generated. The facts check out — the partnership is real, Benioff does live-stream earnings — but the confidence is gone.
5:28 minSurprise Guest: Suds, 'The Most Connected Gen Zer in California'
Sam's neighbor Suds — recruited on a dance floor two days earlier at a party where Jennifer Hudson asked Sam to dance — crashes the pod. Fund II just closed, first checks into Browserbase and Railway, and a thesis: only invest in what extends Anthropic and OpenAI's revenue; never touch the app layer.
5:36 minWaymo vs. CyberCab: The $500K Robot vs. the Fleet You Already Bought
CyberCab is finally multiplying on Texas roads and the table runs the unit economics: Waymo pays $250–500K per car plus depots and charging per city, while Tesla's hardware is already distributed in customer driveways. Brit self-drives 95% of her trips; her kids just assume the car drives.
6:00 minThe List Racket — and How Benioff Is Still Everywhere After 40 Years
Slow Ventures ranked #99 of 100 on a unicorn list and Sam has never been happier: 'You are the worst pirate I've ever heard of.' Jessica explains how the Davos list economy actually works, and the table puzzles over Benioff's four-decade omnipresence.
4:10 minSuds Incubates a CI Company — and Defends Investing on Vibes
Suds just incubated a CI company with Kleiner and Benchmark to fix pipelines drowning in 50x more AI-written code — then cheerfully admits he can't explain CI, and delivers a surprisingly rigorous defense of betting on competence, integrity, and EV instead of technical depth.
5:30 minGoogle Pix vs. Canva — and the LLM Wrappers Quietly Crushing It
Google launches Pix, aimed straight at Canva, and the table splits on whether the IPO darling survives. Sam's counterintuitive data: the fastest-growing AI companies he's seeing are unglamorous LLM wrappers for middle America — and customer service, not tech, may be the moat.
3:30 minJess Got Duped: Taylor Swift, the Dead Internet, and Laughing Ourselves to Death
Brit roasts Jessica — founder of The Information — for falling for an AI-generated Taylor Swift pregnancy post on Facebook. The episode's closing thesis: if these four can't tell what's real, the normies are screwed, and the dead internet theory has arrived on schedule.
That's every topic so far · Episode archive & transcripts