Patrick Collison on Stripe’s Early Choices, Smalltalk, and What Comes After Coding (53 min)
ai-driven-innovation-economy ai-for-personalized-medicine ai-global-economic-shifts ai-in-workforce-disruption
- Release date: 2026-03-24
- Listen on Spotify: Open episode
- Episode description:
Michael Truell, CEO of Cursor, sits down with Patrick Collison, CEO of Stripe and an investor in Anysphere, to talk about Collison's history with Smalltalk and Lisp, the MongoDB and Ruby decisions Stripe still lives with 15 years later, why he'd spend even more time on API design if he could do it over, and whether AI is actually showing up in economic productivity data. This episode originally aired on Cursor's podcast. Follow Patrick Collison on X: https://twitter.com/patrickc Follow Michael Truell on X: https://twitter.com/mntruell Follow Cursor: https://www.youtube.com/@cursor_ai Check out everything a16z is doing with artificial intelligence here, including articles, projects, and more podcasts. Please note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see a16z.com/disclosures. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.
Summary
- 🔄 Paradigm Revival: Old ideas from Smalltalk and Lisp machines, like interactive debugging and integrated environments, are poised for AI-powered resurgence to slash iteration times.
- 🏗️ Enduring Abstractions: Early API/data model choices at Stripe and iOS demonstrate decades-long business impacts, shaping orgs via Conway’s Law and requiring careful v2 migrations.
- 📈 Productivity Puzzle: AI coding tools enhance individual workflows but show no macro productivity gains yet, with diffusion challenges and modest GDP forecasts ahead.
- 🧬 Biology Turing Loop: ARC’s read-think-write (sequencing-AI-CRISPR) aims to crack complex diseases via foundation models and virtual cells.
- 🎨 Higher-Level Coding: Future dev may evolve to intent-driven, visual UI manipulation with AI refactoring, reducing codebase burdens and enabling effortless changes.
Insights
What makes Smalltalk’s interactive debugging so superior to modern languages?
Time: 2:36 – 3:33
Category: AI-Driven Innovation EconomyAnswer: Collison describes fixing errors mid-request by inspecting stack frames, editing code, and resuming execution, turning hour-long debug cycles into seconds, a feature absent in Ruby or most mainstream tools. This underscores why powerful dev environments can outweigh language popularity, influencing choices like Stripe’s tech stack. (Start at 2:36)
Will AI shift programming from code to higher-level intents and direct UI manipulation?
Time: 13:06 – 14:19
Category: AI in Workforce Disruption, AI-Driven Innovation EconomyAnswer: Cursor’s CEO envisions AI as advanced compilers enabling less formal, intent-focused languages with visual/direct manipulation for GUIs, reducing codebase weight and mud balls in large projects. Collison agrees, hoping for runtime profiling, refactoring, and beautification to lower change costs. (Start at 13:06)
Why haven’t programming paradigms evolved much in the last 20 years despite more developers?
Time: 14:22 – 15:17
Category: AI-Driven Innovation Economy, AI in Workforce DisruptionAnswer: Patrick Collison notes a lack of experimentation at the development environment level, with ideas from the 70s and 80s like Lisp machines and Smalltalk still relevant, while AI tools like Cursor are reviving integrated environments beyond mere text editors. This matters as it highlights untapped potential for AI to transform dev productivity by blending runtime, editing, and AI seamlessly. (Start at 14:22)
How do early API and data model designs shape business success for decades?
Time: 18:47 – 21:22
Category: AI-Driven Innovation Economy, AI & Global Economic ShiftsAnswer: Collison cites iOS’s superior abstractions driving a more vibrant app ecosystem than Android despite fewer devices, and Stripe’s 15-year-old designs still defining operations, with v2 migrations akin to instruction set changes. Right abstractions endure, profoundly impacting strategy, organization via Conway’s Law, and outcomes. (Start at 18:47)
How can AI mitigate the ‘big bang’ faults of early tech stack choices?
Time: 22:22 – 25:30
Category: AI-Driven Innovation Economy, AI in Workforce DisruptionAnswer: Stripe lives with Ruby/MongoDB decisions from couch chats, requiring massive infrastructure for reliability (99.99986% uptime), while v2 APIs unify entities for customer gains but demand translation layers. AI could refactor/beautify legacy code, easing migrations and enabling n x m relationships. (Start at 22:22)
Why haven’t AI coding tools boosted economy-wide productivity yet?
Time: 40:15 – 42:39
Category: AI in Workforce Disruption, AI & Global Economic ShiftsAnswer: Despite AI adoption, recent papers show no observable gains from LLM usage, US GDP improved modestly but not exponentially or globally, as diffusion takes time and complexity. Collison references Jack Clark’s modest 0.5% annual GDP boost expectation, emphasizing need for progress studies amid contested futures. (Start at 40:15)
Can AI enable ‘programming’ human biology via a cellular Turing loop?
Time: 43:27 – 46:43
Category: AI for Personalized MedicineAnswer: At ARC, Collison’s team combines single-cell sequencing (read), transformers (think), and CRISPR (write) for foundation models on biology, targeting uncured complex diseases like cancer via virtual cells. This read-think-write completeness could unlock causal understanding of pleiotropic gene effects and environments. (Start at 43:27)