# Best programming language for coding with AI agents | Peter Steinberger and Lex Fridman Source: https://www.youtube.com/watch?v=mPaEgPwJJho Recap page: https://rapidrecap.app/video/mPaEgPwJJho Generated: 2026-02-17T13:32:28.502+00:00 --- ## Quick Overview The best programming language for coding with AI agents is highly dependent on the specific task, but Rust is highly regarded for its performance and ecosystem, while Python and TypeScript are also strong contenders depending on deployment environments and agent complexity. **Key Points:** - The speaker (Peter Steinberger) recently attended many iOS conferences because his background is in that area, but now he is pivoting towards AI agents. - The speaker notes that while he used to build mockups in Swift/SwiftUI, he does not like the syntax of Go, which he used previously for agents. - For building the core logic of AI agents, Rust is considered a really good choice due to its performance and ability to handle low-level system integration efficiently. - Python is excellent for general knowledge tasks in AI and is often used for the model direction/inference part, but it is not ideal for deployment on Windows or for high-performance agent logic. - For web-based agents, the combination of JavaScript/TypeScript is strong, especially considering Atwood's Law (any application that can be written in JavaScript will eventually be written in JavaScript). - The speaker found that building agents in Go made it difficult to deploy on Windows, contrasting with Rust which handles cross-platform deployment better. - The TIOBE Index for February 2026 shows Python at #1 (21.81%), followed by C (11.05%) and C++ (8.55%), with Go ranked 16th (1.23%) and showing a decline. ![Screenshot at 00:42: The TIOBE Index for February 2026 is displayed, showing Python at rank 1 with 21.81% rating, highlighting the current landscape of popular programming languages.](https://ss.rapidrecap.app/screens/mPaEgPwJJho/00-00-42.jpg) **Context:** This segment features a discussion between Lex Fridman and Peter Steinberger on the Lex Fridman Podcast, focusing on the selection of optimal programming languages for developing modern Artificial Intelligence (AI) agents. Steinberger, who has a background in iOS development, shares his evolving perspective on which languages offer the best balance of performance, ecosystem support, and deployment flexibility for agent-centric software development, contrasting languages like Go, Python, Rust, and TypeScript. ## Detailed Analysis The conversation centers on choosing the right programming language for building AI agents. Peter Steinberger, transitioning from an iOS background, expresses a preference for languages that offer high performance and good cross-platform capabilities, particularly noting issues with Go's Windows deployment. He argues that while Python is suitable for the 'model direction' or inference part of AI due to its vast ecosystem and knowledge base, it might be too slow for core agent logic or low-level system integration where performance is critical. Rust emerges as a strong candidate for the core logic because it provides excellent performance and resilience, making it easier to handle complex system interactions. TypeScript/JavaScript is highlighted for agents that live primarily in a web context, referencing Atwood's Law which suggests JavaScript dominance in applications. Steinberger notes that while he dislikes Go's syntax, he acknowledges that the choice ultimately depends on the specific domain and required integration points, suggesting that a mix of languages might be necessary, but Rust is currently favored for robust agent backends. ### Programming Language Preferences for AI Agents - Peter Steinberger prefers Rust for core agent logic due to performance and ecosystem advantages over Go; dislikes Go's syntax and deployment limitations on Windows. ### Role of Different Languages in AI - Python excels at general knowledge and model inference tasks; TypeScript/JavaScript is preferred for web-based agents, aligning with Atwood's Law. ### Agent Development Philosophy - The best language depends on the problem domain; building high-performance, low-level agent components benefits from systems languages like Rust. ### TIOBE Index Snapshot (Feb 2026) - Python leads at 21.81% (down 2.08%), C is second (11.05%, up 1.22%), and Go is ranked 16th (1.23%, down 1.03%), indicating a dynamic landscape. ### Personal Experience & Evolution - Steinberger transitioned from iOS (Swift/SwiftUI) and found previous experience with Go challenging for cross-platform deployment, reinforcing the need for language flexibility. ![Screenshot at 00:03: Peter Steinberger speaking into a professional microphone during the podcast interview.](https://ss.rapidrecap.app/screens/mPaEgPwJJho/00-00-03.jpg) ![Screenshot at 00:10: A visual overlay of the Lex Fridman Podcast branding appears briefly at the start of the clip.](https://ss.rapidrecap.app/screens/mPaEgPwJJho/00-00-10.jpg) ![Screenshot at 00:42: The TIOBE Index for February 2026 is displayed, showing Python at rank 1 with 21.81% rating, highlighting the current landscape of popular programming languages.](https://ss.rapidrecap.app/screens/mPaEgPwJJho/00-00-42.jpg) ![Screenshot at 02:10: A slide illustrating Atwood's Law: "Any application that can be written in JavaScript will eventually be written in JavaScript - Jeff Atwood."](https://ss.rapidrecap.app/screens/mPaEgPwJJho/00-02-10.jpg) ![Screenshot at 03:24: Lex Fridman listening attentively while Steinberger gestures while explaining his preference for Rust.](https://ss.rapidrecap.app/screens/mPaEgPwJJho/00-03-24.jpg)