Everyone's Wrong About AI Moats — Lessons from Inside Early Google
In this episode of Very True, Alex reconnects with his friend of fifteen years, Michael Stoppelman, recorded 7,000 miles apart — Alex in Israel, Michael in the Bay Area. Michael's path runs from a Purdue computer science and biology double-major, through Google's earliest, wildest pre-IPO days building the click-fraud detection systems that got him promoted from post-sales support into engineering, to becoming the SVP of Engineering who joined his brother Jeremy at Yelp and helped scale it into a public company. Today Michael is a Venture Partner at Verissimo Ventures, an active angel investor across hundreds of companies, and the builder of CC Marvin, an AI chief-of-staff for investors.
Alex and Michael pull the thread between two boom eras twenty years apart: Google circa 2003 figuring out whether clicks, CPA, or something else entirely was the real business model, and today's AI labs wrestling with token pricing, seat pricing, and usage-based pricing. They map Anthropic's enterprise-and-coding strategy onto Google's old Android and YouTube playbook, debate whether frontier "omniscient" models will always beat cheaper specialized ones, and ask the harder question underneath it all — is software engineering as a career dissolving, or just being rebuilt around a new killer skill: product design? They close on what's actually durable in tech — chokepoints, brand, and distribution — and what founders and workers should do about the next wave.
Episode Highlights
Alex and Michael pull the thread between two boom eras twenty years apart: Google circa 2003 figuring out whether clicks, CPA, or something else entirely was the real business model, and today's AI labs wrestling with token pricing, seat pricing, and usage-based pricing. They map Anthropic's enterprise-and-coding strategy onto Google's old Android and YouTube playbook, debate whether frontier "omniscient" models will always beat cheaper specialized ones, and ask the harder question underneath it all — is software engineering as a career dissolving, or just being rebuilt around a new killer skill: product design? They close on what's actually durable in tech — chokepoints, brand, and distribution — and what founders and workers should do about the next wave.
Episode Highlights
[03:17] Life Inside Early Google: Michael recounts working in "Building Pi" during Google's pre-IPO days, riding Segways with Larry and Sergey, and building the click-fraud detection tools that got him promoted from support into engineering.
[07:19] From Google to Yelp: How the same "defend against fake signals" instinct that shaped Google's ad business led Michael to join his brother Jeremy's scrappy local-search startup.
[15:55] Anthropic's Android Moment: Michael draws a direct line from Google's Android and YouTube acquisitions to Anthropic's enterprise-and-coding land grab with Claude, and why Claude Desktop is the new "connection with the user."
[29:20] The CRUD vs. Frontier Split: Why most software will run on cheap open-weight models like GLM, while a shrinking slice of high-stakes work (satellites, bikes, power plants) still demands the most expensive frontier intelligence.
[40:22] The Hype/Reality Gap: If AI were truly omniscient, why hasn't it rewritten the Linux kernel or shipped a faster C compiler? Michael's reality check on how far coding agents have — and haven't — come.
[51:34] Displacement or Entrepreneurship Boom?: The pair debate job-loss anxieties, why the government's own measurement systems can't be trusted, and why Michael thinks agentic AI nets out to more founders, not fewer engineers — closing with a plug for his new project, CC Marvin.
Full Chapter List
[00:53] Welcome & Reconnecting Across 7,000 Miles
[01:42] From Purdue Biotech Dreams to Google's Early Days
[03:17] Inside Building Pi: Larry, Sergey, and Pre-IPO Google Culture
[05:31] The All-Star Roster: PMs and Colleagues Who Became Famous VCs
[07:19] Defending the Click: Why Fraud Was an Existential Threat
[09:43] Jumping to Yelp: Cracking Local's Emotional Intelligence Problem
[12:45] 2003 vs. 2026: The Business Model Question Nobody's Solved
[15:55] The Android Playbook: How Anthropic Is Building Its Enterprise Moat
[21:27] Coding as the Universal Translation Layer
[24:41] Software Engineering, Dissolved and Rebuilt by Agents
[27:36] Are Coding Agent Margins Actually Durable?
[29:20] The Coming Bifurcation: CRUD Apps vs. Frontier Intelligence
[33:59] Beyond Coding: Specialized Models for Biology, Physics, and Earthquakes
[38:13] Why Product Design, Not Code, Is the New Killer Skill
[40:22] The Gap Between AI Hype and Compilers That Still Haven't Improved
[42:39] What's Actually Durable: Chokepoints, Brand, and Distribution
[47:48] Adapting to the Agentic Era Without Losing Your Job
[51:34] The Entrepreneurship Boom & a CC Marvin Plug
[01:42] From Purdue Biotech Dreams to Google's Early Days
[03:17] Inside Building Pi: Larry, Sergey, and Pre-IPO Google Culture
[05:31] The All-Star Roster: PMs and Colleagues Who Became Famous VCs
[07:19] Defending the Click: Why Fraud Was an Existential Threat
[09:43] Jumping to Yelp: Cracking Local's Emotional Intelligence Problem
[12:45] 2003 vs. 2026: The Business Model Question Nobody's Solved
[15:55] The Android Playbook: How Anthropic Is Building Its Enterprise Moat
[21:27] Coding as the Universal Translation Layer
[24:41] Software Engineering, Dissolved and Rebuilt by Agents
[27:36] Are Coding Agent Margins Actually Durable?
[29:20] The Coming Bifurcation: CRUD Apps vs. Frontier Intelligence
[33:59] Beyond Coding: Specialized Models for Biology, Physics, and Earthquakes
[38:13] Why Product Design, Not Code, Is the New Killer Skill
[40:22] The Gap Between AI Hype and Compilers That Still Haven't Improved
[42:39] What's Actually Durable: Chokepoints, Brand, and Distribution
[47:48] Adapting to the Agentic Era Without Losing Your Job
[51:34] The Entrepreneurship Boom & a CC Marvin Plug
Links & Resources CC Marvin: https://ccmarvin.com/
Verissimo Ventures: https://verissimo.vc/
Michael Stoppelman on LinkedIn: https://www.linkedin.com/in/michaelstoppelman/
Alex Oppenheimer on LinkedIn: https://www.linkedin.com/in/alex-oppenheimer/
Verissimo Ventures: https://verissimo.vc/
Michael Stoppelman on LinkedIn: https://www.linkedin.com/in/michaelstoppelman/
Alex Oppenheimer on LinkedIn: https://www.linkedin.com/in/alex-oppenheimer/
About Very True
Hosted by Alex Oppenheimer, Very True by Verissimo Ventures explores the honest, unvarnished stories of founders and the real problems they are solving. We look past the hype to find the truth in technology and entrepreneurship.
Creators and Guests
Guest
Michael Stoppelman
Former SVP of Engineering at Yelp, now an angel investor with a portfolio of over 400 companies.
