Podcast Episode Summary
The podcast episode features an interview with Alex Rattray, the founder of Stainless, a company that provides API platforms, SDKs, and documentation for developers. Alex is joined by Jennifer Lee, a partner at Andreessen Horowitz (A16Z). The conversation revolves around the evolution of APIs, the importance of SDKs and documentation, and the impact of large language models (LLMs) on API design and development.
Main Topic
The main topic of the episode is the future of APIs and the role of LLMs in shaping the way developers interact with APIs. Alex and Jennifer discuss how APIs are becoming the "dendrites of the internet," connecting different systems and enabling communication between them.
Key Points
Takeaways
Participants
Recommendations
Stainless founder Alex Rattray joins a16z partner Jennifer Li to talk about the future of APIs, SDKs, and the rise of MCP (Model Context Protocol). Drawing on his experience at Stripe—where he helped redesign API docs and built code-generation systems—Alex explains why the SDK is the API for most developers, and why high-quality, idiomatic libraries are essential not just for humans, but now for AI agents as well.
They dive into:
Timecodes:
0:00 – Introduction: APIs as the Dendrites of the Internet
1:49 – Building API Platforms: Lessons from Stripe
3:03 – SDKs: The Developer’s Interface
6:16 – The MCP Model: APIs for AI Agents
9:23 – Designing for LLMs and AI Users
13:08 – Solving Context Window Challenges
16:57 – The Importance of Strongly Typed SDKs
21:07 – The Future of API and Agent Experience
24:45 – Lessons from Leading API Companies
26:14 – Outro and Disclaimers
Resources:
Find Alex on X: https://x.com/rattrayalex
Find Jennifer on X: https://x.com/JenniferHli
Stay Updated:
Let us know what you think: https://ratethispodcast.com/a16z
Find a16z on Twitter: https://twitter.com/a16z
Find a16z on LinkedIn: https://www.linkedin.com/company/a16z
Subscribe on your favorite podcast app: https://a16z.simplecast.com/
Follow our host: https://x.com/eriktorenberg
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.
Stay Updated:
Find a16z on X
Find a16z on LinkedIn
Listen to the a16z Podcast on Spotify
Listen to the a16z Podcast on Apple Podcasts
Follow our host: https://twitter.com/eriktorenberg
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.
Jennifer Li
I love to say that APIs are the dendrites of the internet. If you think about the brain as a collection of neurons that fire together to think, you can think of the internet as a collection of servers or programs running on servers that interact to think or act. The connection between them, the synapse, is almost always an API of some kind. Every company is becoming an API company, and every API company needs an API platform with SDKs, documentation, API versioning, and more. People want to focus on their core capabilities, expose thoughtful interfaces to users, and not worry about low-level details. APIs are the connective tissue of modern software, powering everything from payments to cloud services. As APIs multiply, so do the challenges for developers—SDK quality, documentation, versioning, and now a new frontier: interfaces for AI agents.
Alex Rattray
In this episode, I join Jennifer Li to share how my time at Stripe shaped Stainless' mission to bring world-class developer platforms, SDKs, and documentation to every API company. We discuss why SDKs are often the API for developers, how the MCP Model Context Protocol reframes APIs as interfaces for large language models, and the new design challenges of treating AI agents as first-class users. It's a conversation about the future of APIs, agent experience, and building robust developer platforms.
Jennifer Li
Let's dive in. You started working on API tooling at Stripe. Can you share how that experience shaped your approach to SDK generation?
Alex Rattray
I worked on other teams at Stripe but saw how critical the API platform was. I moved to that team and redesigned the Stripe API documentation. Through that, I realized how important SDKs were. Internally, people didn't see that to developers using the Stripe API every day, the SDK is the API. They were falling behind, so I built a code generation system that turned the Stripe OpenAPI spec into great Python and TypeScript libraries. It was a fun challenge, especially the compiler aspects and library design.
Jennifer Li
After Stripe, people asked how to build great developer experiences for their APIs. What was your answer?
Alex Rattray
I realized there was no good answer. OpenAPI generator was far from Stripe's quality bar. The maintenance burden was too much for most companies. It was frustrating to tell people they needed SDKs but couldn't have them. So I started Stainless to bring Stripe-level developer platforms to every API company.
Jennifer Li
How did you approach SDK generation at Stainless, given your Stripe experience?
Alex Rattray
We focused on making client libraries robust and polished. For Python developers, it should be idiomatic—familiar and well-documented. Crafting that is a journey. I love the nitty-gritty details, like making sure documentation pops up in editors. We've moved the industry forward with our Python libraries.
Jennifer Li
How do you handle the challenge of context size with MCP and LLMs?
Alex Rattray
We generate all tools in the API but let users limit interactions to specific resources or operations. We also offer dynamic mode, exposing tools as needed. It takes more turns but saves context. We added JQ filters to handle large API responses, letting LLMs focus on relevant data without overwhelming the context window.
Jennifer Li
Some think MCP makes SDK quality less important, but you disagree. Why?
Alex Rattray
SDK quality is more important. LLMs prefer using SDKs for type checking and accuracy. Without good SDKs, LLMs make mistakes. Strongly typed, polished SDKs help LLMs focus on higher-level tasks while humans review business logic.
Jennifer Li
How do you see the role of API developers evolving with AI as a common developer?
Alex Rattray
API design will become more declarative and dry. Humans will focus on high-level business needs, while LLMs handle coding. APIs will need clear, prescriptive standards for LLMs to follow. The goal is to let humans think strategically while robots handle the less exciting details.
Jennifer Li
What lessons have you learned from collaborations with companies like OpenAI and Cloudflare?
Alex Rattray
Everyone sees how important it is to be thoughtful about exposing capabilities. Whether it's a large language model or cloud services, people want to focus on their core capabilities and not worry about low-level details. High-quality, idiomatic libraries are essential for both humans and AI agents.
Jennifer Li
Thank you, Alex. This has been insightful.
Alex Rattray
Thanks, Jennifer. It's been fun.
Jennifer Li
If you enjoyed this conversation, please rate and review the podcast. Stay tuned for more discussions on software infrastructure and AI. Remember, this content is for informational purposes only and should not be taken as legal, business, tax, or investment advice. For more details, visit A16Z.com/disclosures.