AI + a16z: Patrick Collison on Stripe’s Early Choices, Smalltalk, and What Comes After Coding

Date: March 24, 2026
Host: a16z featuring Michael Truel (Cursor)
Guest: Patrick Collison, CEO and Co-founder of Stripe

Episode Overview

In this episode, Patrick Collison explores the impact of programming paradigms and technological choices made during Stripe's inception, the enduring influence of languages like Smalltalk and Ruby, reflections on developer environments, and the future of software creation in the age of AI. The episode dives into API design, lessons from history, the slow diffusion of AI-driven productivity in the economy, and Collison’s forays into biological computation through ARC.

Key Discussion Points & Insights

1. Programming Paradigms, Early Decisions, and Smalltalk

• The Power of Non-Mainstream Languages

  • Collison shares why his first startup used Smalltalk, inspired by its advanced interactive development environment, like editing code and fixing errors mid-request for instant effect.
  • Quote: “Smalltalk is actually this extremely interesting development environment that had a lot of the aspects of Lisp that I’d really appreciated... like a fully interactive environment with a proper debugger.”
    – Patrick Collison, [01:49]

• Hiring and Transition

  • Despite concerns, hiring for Smalltalk was easy: “Smart people learn languages really quickly, so... not really a reason not to use a non-mainstream language.”
    – Patrick Collison, [04:07]

2. Early AI Exploration and Lisp

• First AI Bots and Bayesians
  • Collison recounts building a Bayesian-based MSN Messenger bot in Lisp, which fooled unsuspecting users but fell short of passing the Turing Test.
  • Paradigms of AI Programming by Peter Norvig was formative, but neural nets were largely unexplored by him due to hardware limitations at the time.
  • Quote: “It never really passed the Turing Test... but it certainly passed some weaker versions... people ended up having quite lengthy conversations with it.”
    – Patrick Collison, [05:36]

3. Esoteric Languages and Modern Developer Environments

• What Modern Environments Miss

  • The episode laments that ideas from 1980s/90s environments (Lisp machines, Smalltalk, Mathematica) haven’t fully permeated modern tools.
  • The core insight: merge text editing, runtime, and environment for seamless coding and feedback; VS Code is only a partial step.
  • Quote: “It’s just such a mistake that we have ended up with development environments where there is such a separation between the runtime, the text editing and the environment…”
    – Patrick Collison, [08:15]

• Desired Features

  • Collison advocates for richer integrations: hovering to see runtime profiling, logging, error info, and variable values from production.
  • Quote: “When I hover over a line of code, I would like to see profiling information about ... that code or that function... how the most common values that it takes on in production, these kinds of just rich, deep integrations.”
    – Patrick Collison, [09:40]

4. Visually-Oriented Programming: Praise and Skepticism

• Inventing On Principle
  • Collison credits Brett Victor’s work but is personally less inclined toward graphical/spatial approaches, preferring symbolic and lexical reasoning.
  • Quote: “I reason much more kind of symbolically and sort of lexically than I do visually and graphically.”
    – Patrick Collison, [11:16]

5. AI-Augmented Programming and the Future of Developer Tools

• Shifting Paradigms

  • Debate on whether the main surface of programming will be code in five years; consensus that higher-level specification may supplant traditional code.
  • AI tools should make programming higher-level, focusing on what should be built rather than how.

• Barriers To Change

  • Programming ecosystem “lock-in” (skills, legacy code) limits change.
  • One hope: as AI handles more, the burden of legacy code may be relieved.

• AI for Refactoring

  • Collison sees great potential for AI to beautify and refactor code; this is as important as helping generate new code.
  • Quote: “Nocturnally this thing comes up behind you and makes it all beautifully factored.”
    – Patrick Collison, [16:28]

6. APIs, Organizational Impact, and Abstractions That Last

• Importance of Early Technical Choices

  • APIs and data modeling set a company’s structure for years to come (Conway’s Law) and can even influence business strategy and outcomes.
  • Examples: iOS API longevity vs. Android; Stripe’s use of Ruby and MongoDB.
  • Quote: “The right API design, the right abstraction design ended up having just quite significant business ramifications. ...and for the first versions of iOS, many of the classes that one used were prefixed with NS... That’s a case where the API design survived for two decades or more.”
    – Patrick Collison, [18:46]

• The Stripe “Big Bang”

  • Stripe’s early use of MongoDB and Ruby was, in part, convenience and personal taste over “top-down” process.
  • MongoDB was chosen for its flexible data model, avoiding what Collison saw as the restrictiveness of SQL.
  • Quote: “So we were sitting on the couch, it's like, should we use Mongo? Yeah, fine.”
    – Patrick Collison, [25:26]

7. Stripe V2 APIs: A Case Study in Large-Scale Evolution

• API Evolution as “Instruction Set Migration”

  • Shipping V2 meant not just defining new abstractions, but painstakingly ensuring backward compatibility and coexistence for users.
  • Quote: “It’s not that useful to just define these APIs in isolation... It’s all the kind of coexistence questions that become hard. It’s hard to ship this year and we’re excited about it.”
    – Patrick Collison, [28:29]

• Best Practices

  • Ensure one “singular person” owns holistic API vision (alongside working groups).
  • Support general N:M relationships (unify everything possible).
  • Rigorously validate APIs with customer feedback and integration exercises.

8. AI’s Real-World Impact: Productivity and Economic Markers

• Lack of Measurable Productivity Gains (So Far)

  • Little evidence that LLM-driven AI has moved broad productivity numbers yet. US GDP up, but the effect is not global or exponential.
  • “The diffusion of these technologies through the economy really takes time and involves substantial complexity.”
    – Patrick Collison, [40:41]

• Adequacy of GDP as a Measure

  • Collison believes that meaningful economic impact will be reflected in GDP, without needing new core measures.
  • Quote: “In any world where what we generally think of as the economy is massively enhanced, it'll show up in GDP, I believe.”
    – Patrick Collison, [43:06]

9. Programming Biology and Foundation Models

• ARC and Virtual Cells
  • ARC’s mission is to build foundation models for biology, focusing on “read, think, write” loops in cells using advances in sequencing, neural nets, and gene editing.
  • Aim: Enable progress against complex diseases by marrying computation with biological experimentation.
  • Quote: “If you put those together, you now have the ability to, again at the kind of level of the individual cell, to read, think and to write. And this starts to really feel like a new kind of Turing loop…”
    – Patrick Collison, [43:27]

10. Who Benefits in a “Post-Coding” World?

• Winners of Low-Code/No-Code AI Era
  • Collison offers no strong bets, but acknowledges the unpredictability: “I’m struck by how many predictions have held up reasonably poorly even for people who are… well, in fact informed.”
    – Patrick Collison, [47:52]

11. What Stripe Wants from Next-Gen Tools Like Cursor

• Top Three Features Desired:
  1. Rich Runtime Integration – Surfacing profiling and operational data at the code level in the editor.
  2. Automated Refactoring/Beautification – Making codebases more elegant and adaptable post-hoc.
  3. Emphasis on Quality/Craft – Not just more software, but better software; tools should raise the bar for what “quality” means.
  • Quote: “There's obviously a concern with AI that it leads to the creation of more slope and more kind of crappy things, but not more of the best things... to ensure that the world is creating more of the best software and not just more software...”
    – Patrick Collison, [50:18]

Notable Quotes & Memorable Moments

• “It’s just such a mistake that we have ended up with development environments where there is such a separation between the runtime, the text editing and the environment…”
  – Patrick Collison, [08:15]

• “Nocturnally this thing comes up behind you and makes it all beautifully factored.”
  – Patrick Collison, [16:28]

• “APIs and data models really… end up shaping the organization. I think the strong version is that it substantially shapes your strategy and just your business outcomes.”
  – Patrick Collison, [18:46]

• “We were sitting on the couch, it’s like, should we use Mongo? Yeah, fine.”
  – Patrick Collison, [25:26]

• “In any world where what we generally think of as the economy is massively enhanced, it'll show up in GDP, I believe.”
  – Patrick Collison, [43:06]

Timestamps for Key Segments

• Smalltalk and Early Startup Choices: [01:43] – [04:07]
• Early AI/Lisp Projects: [05:00] – [07:38]
• Programming Languages & Environments: [08:15] – [10:37]
• Brett Victor and Visual Environments: [10:37] – [12:51]
• AI in IDEs, Programming Abstractions: [12:51] – [16:28]
• Stripe’s API & Tech Choices: [18:46] – [27:15]
• Stripe V2 API Migration: [27:20] – [33:30]
• AI Productivity & Economic Impact: [36:47] – [43:06]
• Programming Biology & ARC: [43:27] – [47:02]
• Wishlist for Developer Tools: [49:21] – [51:40]

Conclusion

Patrick Collison’s reflections offer a dual look: backward, to the foundational decisions and philosophies that shaped Stripe, and forward, to a reimagined future where AI plays a pivotal role not just in code generation, but in the very essence of software creation and even biology. He urges technologists to focus on the unglamorous—but pivotal—craft of API and abstraction design, calls for more artful, integrated dev environments, and offers pragmatic skepticism about the speed with which AI will reshape the economy. If the world is ready for “what comes after coding,” Collison stands squarely at the intersection of history, innovation, and caution.