Patrick Collison on Stripe’s Early Choices, Smalltalk, and What Comes After Coding - a16z Podcast Recap
Podcast: a16z Podcast
Published: 2026-02-20
Duration: 53 minutes
Guests: Patrick Collison
Summary
Patrick Collison reflects on the technological choices made during the early days of Stripe, the role of API design in business success, and the potential impact of AI on economic productivity. He also shares insights on new biological technologies and the future of coding.
What Happened
Patrick Collison began his entrepreneurial journey using Smalltalk, a programming language that enhanced his ability to fix errors and inspect processes in real-time. He later worked with Lisp and Lisp dialects, creating an AI bot for MSN Messenger, showcasing an early interest in artificial intelligence and predictive models.
Stripe's foundational technologies, Ruby and MongoDB, were selected for their flexibility and object-based data storage capabilities. Collison maintains that these choices have shaped Stripe's infrastructure and necessitated the construction of robust systems to ensure reliability and fault tolerance.
Collison believes that a well-designed API can significantly influence business outcomes. Stripe's API evolution, particularly the transition from V1 to V2, underscores the importance of integrating customer feedback and maintaining interoperability with existing systems.
Despite advancements in AI, Collison notes that its impact on productivity has yet to be reflected in economic metrics like GDP. He remains optimistic about AI's potential, citing a prediction that it could enhance GDP growth by 0.5% annually.
Collison is co-founder of ARC, a biomedical research organization focused on training foundational models for biology. He is hopeful that new technologies such as CRISPR and neural networks will advance our understanding of complex diseases.
Stripe prioritizes research and development, investing heavily in software creation. Tools like Cursor are used extensively by employees to boost productivity and streamline code writing processes.
AI tools, while useful for answering factual questions, fall short in writing tasks, as their output often lacks the personal touch and specificity Collison desires. He utilizes LLMs for coding, finding them particularly effective when mediated through specific tools.
Collison expresses a desire to see a return to development environments that offer more than just text editing, emphasizing a need for innovative programming paradigms that prioritize logic and abstraction design.
Key Insights
- Patrick Collison's early use of Smalltalk, a programming language, allowed him to fix errors and inspect processes in real-time, setting a foundation for his later work in AI and predictive models using Lisp dialects.
- Stripe's choice of Ruby and MongoDB as foundational technologies was driven by their flexibility and object-based data storage, requiring the company to build robust systems to ensure reliability, a choice that continues to influence their infrastructure today.
- The evolution of Stripe's API from V1 to V2 highlights the critical role of customer feedback and interoperability, showing how thoughtful API design can drive business success by aligning with user needs and existing systems.
- Despite AI's rapid advancement, Patrick Collison observes that its productivity gains are not yet reflected in GDP metrics, though he remains optimistic about AI's potential to boost GDP growth by 0.5% annually.
Key Questions Answered
What programming languages did Patrick Collison use at Stripe?
Stripe initially used Ruby and MongoDB, chosen for their flexibility and object-based data storage, which remain foundational technologies for the company.
How does Patrick Collison view the impact of AI on economic productivity?
Patrick Collison notes that while AI has potential, its impact on productivity has not yet been reflected in economic metrics like GDP, although it is expected to contribute to GDP growth in the future.
What is ARC, co-founded by Patrick Collison?
ARC is a biomedical research organization focused on training foundational models for biology, aiming to leverage new technologies like CRISPR and neural networks to understand complex diseases.