Founder Mode

EPISODE 76JJ Zhuang, co-founder and CEO of Pretty Good AI, returns to Founder Mode to answer for the predictions he made on the show over a year ago. Kevin and Jason had Claude grade all ten, and JJ defends his record on hiring generalists over specialists, how much code AI is really writing, why taste is the one thing you can't train into a model, and why slide decks are dead when you can just build the thing. He finishes 6.5 for 10, which beats the weatherman, and points to the release of Opus 4.6 as the moment things that used to not work suddenly started working. In the second half, JJ looks ahead: why Pretty Good AI is skipping the "sell the same widget a thousand times" playbook to become the AI team for healthcare practices, how to build guardrails around AI that's confidently wrong, why the human work that's left will be 100% hard work, how the team keeps protected patient data away from AI agents, and how a lean team can win a crowded healthcare AI market.CHAPTERS00:00 – Cold Open: Why Not Just Solve All Your Problems?00:31 – JJ Returns: Grading 10 Predictions One Year Later01:58 – Generalists vs. Specialists: Who Should You Hire?04:39 – Is AI Really Writing Half of All Code?07:28 – Taste, Comprehension, and the Human Edge09:58 – Instant Prototypes, Fast Software, and the Death of Slide Decks12:39 – Directing AI and the Most Overhyped Buzzword14:32 – Overhyped or Underestimated? The Opus 4.6 Moment17:09 – The Scorecard: 6.5 for 1018:00 – Pretty Good AI's Big Bet: Solve Every Problem20:02 – What Success Looks Like: Becoming Your Customers' AI Team20:39 – Designing Around AI That's Confidently Wrong22:20 – Where the Line Between AI and Humans Moves Next24:09 – Building on Models That Change Every Few Months25:58 – Compliance When Your Team Is Agents28:23 – What Patients Notice First30:17 – How a Lean Team Wins a Crowded MarketLINKSConnect with JJ ZhuangPretty Good AI • LinkedIn • X/TwitterStay Connected with Founder ModeSubscribe to our newsletterConnect with KevinLinkedIn • X/TwitterConnect with JasonLinkedIn • X/Twitter

EPISODE 75Kevin and Jason go solo on the week AI stopped asking permission. Jason opens with a Claude Code session that sent a Slack message to a client's CEO instead of drafting it, complete with a "sent from Claude" tag at the bottom, then offered to file a bug report about itself. That kicks off a run through what founders are actually paying for now: Jason's $800-a-month HubSpot bill and his plan to swap it for Dharmesh Shah's $1-a-month USpot, levels.io running a multi-million-dollar business on hand-coded PHP and a single server, and why the "SaaSpocalypse" is more nuanced than Twitter thinks. Kevin traces the freemium playbook back to Zimbra, where 98% of users never paid and the top 2% funded the company, and asks the question that hangs over the whole episode: how do you build trust in a market where everything can be free? They also get into Jason's agentic agency experiment, why succession is really about handing off jobs rather than titles, why nobody has built a marketing agent with taste yet, and the biological age scores every app is suddenly handing out, each one conveniently favoring its own algorithm.CHAPTERS00:00 – The Slack message Claude sent by itself00:39 – A dozen AI tools in a single day03:46 – $800 a month for a CRM04:26 – Dharmesh's $1 CRM and the SaaSpocalypse08:03 – Freemium after vibe coding: the Zimbra lesson10:32 – Building an agentic agency13:20 – Succession: hand off jobs, not titles16:33 – What AI still can't do in marketing19:18 – Biological age scores and walled-garden health data22:22 – Founder Mode top fiveLINKSStay Connected with Founder ModeSubscribe to our newsletterConnect with KevinLinkedIn • X/TwitterConnect with JasonLinkedIn • X/Twitter

EPISODE 74Kiko Zang spent more than half her life pretending to be someone else — born in China, sent alone to boarding school in New Zealand at 12, then to USC at 17, then back to China as the Chinese-looking person who couldn't hold an adult conversation in Chinese. Pretending, as she puts it, was a survival instinct that became one of her strongest skills. In this episode she traces that pattern from the classroom in China where she refused to rehearse a fake PTA lesson, through the years of self-censoring at work, to the full head-to-toe mascot costume she wore around crypto conferences — where, strangely, the expectations placed on a female founder vanished and she got to be more herself than she was without it. Now CEO of Chomp, she's building an app around the gap between what people say in public and what they actually believe in private, and she brings the receipts: 50,000 users, two million answers, and a question about saying "I love you" in the first month of dating that came back 93% green flag when nobody was watching — while every single player assumed it would be a coin flip. Kevin and Jason play a live pack on air, and the conversation moves into why honest human opinion may be the most valuable data set of the AI era, why there's no durable moat in consumer, and what Kiko learned after a co-founder breakup when her own team had to tell her she'd been muting herself in meetings. Her conclusion: the fear felt true not because it was real, but because she believed it was.CHAPTERS00:00 – More herself inside the suit00:40 – Know your audience or know yourself04:07 – Half a life spent pretending06:14 – The PTA rehearsal that got her sent abroad08:34 – The reluctant founder11:44 – The fursuit and who gets assumed to be the founder14:05 – Anonymity, honesty, and shared reality16:12 – Playing Chomp live19:24 – 93% green flag and the perception gap22:20 – Pluralistic ignorance and why consumer has no moat24:31 – Honest opinion as the data of the AI era27:50 – The co-founder breakup and fears that aren't real30:06 – Where to find ChompLINKSConnect with Kiko Zangchomp.fyi • LinkedIn • X/TwitterDownload ChompRedeem the Founder Mode listener offerStay Connected with Founder ModeSubscribe to our newsletterConnect with KevinLinkedIn • X/TwitterConnect with JasonLinkedIn • X/Twitter

EPISODE 73Mohammad Moahid has operated multiple companies across e-commerce, consumer electronics distribution, and real estate, and competed for Team Canada as an equestrian. He argues the gap between companies talking about AI and companies actually running on it comes down to ops, not technology. In this episode he explains why leaders are structurally cut off from on-ground reality, why a 5,000-person firm is still managing a hundred active projects out of email threads while million-dollar invoices get written by hand, and where the MIT Nanda study's 95% failure rate really comes from (hint: it's the people layer, not the software). He walks through the "chief agent officer" model, an AI chief of staff coordinating specialist agents across zoning, leasing, acquisitions, and legal. He covers the four pillars he uses to diagnose any business, and the pricing thesis behind what his team calls outcome as a service: you don't pay an accountant for software, you pay so you never think about your books again. Plus what happens to agencies and professional services firms when outcomes get cheaper, and the one thing Mo says humans have to keep owning completely.CHAPTERS00:35 – AI as the operating system, not another tool04:37 – What competing for Team Canada taught Mo about execution05:23 – Why leaders are always removed from on-ground reality07:26 – Million-dollar invoices, still written by hand08:33 – Being his own first customer09:49 – Why 95% of AI implementations fail10:45 – The command brief and the four pillars12:49 – The chief agent officer and the 80/20 rule14:57 – Outcome as a service: accountant vs. QuickBooks19:05 – Employee fear, and why agencies should be worried21:17 – What humans still have to own completely22:41 – Building your own harnessLINKSConnect with Mohammad Moahidzerotoagent.com • LinkedIn • InstagramStay Connected with Founder ModeSubscribe to our newsletterConnect with KevinLinkedIn • X/TwitterConnect with JasonLinkedIn • X/Twitter

EPISODE 72No guest and no notes on Jason's side, so Kevin runs the board for once. The two dig into the question underneath every pricing decision: what unit of value are you actually selling? Jason retraces underpricing his agency in the early days and the shift from billing hours to building productized services with near-zero marginal cost, while Kevin walks through losing a deal to a competitor selling raw phone minutes, where the buyer admitted the product was worse and bought it anyway. From there they get into using AI to reconcile invoices against vendor spend down to the penny, a sweep that turned up 23 fraudulent charges and an education in BIN enumeration. They cover the strange state of the market, where a public company doing $275M gets taken private at 2x forward revenue while venture-backed startups raise at 100x. And they close on the AI back office: Kevin's 76 scheduled agents, the 13 supervisor agents he built to watch them, and the eight days of automations that reported success, success, success while quietly writing results to a directory that didn't exist.CHAPTERS00:32 – No guest, no notes: a solo episode00:51 – What unit of value are you actually selling?02:24 – "I did what most people do and I underpriced"04:26 – Losing a deal to phone minutes08:50 – Letting AI reconcile your invoices and vendor spend12:41 – 23 fraud charges and BIN enumeration15:02 – Weave goes private: 2x revenue vs. 100x rounds17:09 – The private and public markets dance19:21 – 76 agents, 13 supervisors, and "successfully failing"21:57 – Most people are bad managers25:34 – DHH's Linux distro and talking to your OS26:50 – Founder Mode top fiveLINKSStay Connected with Founder ModeSubscribe to our newsletterConnect with KevinLinkedIn • X/TwitterConnect with JasonLinkedIn • X/Twitter