AI Summary
5 min readIn 2024, Thibault “Tibo” Sottiaux joined OpenAI and immediately began sprinting on what would become Codex, the company’s coding agent. But the project’s real roots go deeper. At DeepMind, Tibo helped build an internal chatbot a full year before ChatGPT launched — a tool researchers used to tinker with early, barely coherent language models. That experience of building tooling that makes others faster became a career-long theme. On The Pragmatic Engineer, Tibo explains how Codex was designed from first principles, why it was built in Rust despite the models being weak at Rust, and how the team thinks about the shifting boundary between what the harness does and what the model does.
Why Codex was built in Rust and made open source
Continue reading the full summary in the app — free to try.
Read Full Summary →Free • No credit card required
Never miss an episode of The Pragmatic Engineer
Get every new episode summarized in your inbox — free, ~5 minutes to read.
No spam. Unsubscribe anytime.
What you'll learn
- 1 (00:00) **Introduction and Background** - Host Gergely introduces the episode and guest Thibault (Tibo) Sottiaux, covering his role at OpenAI and the topics to be discussed.
- 2 (02:10) **Getting Into Tech and Early Career** - Tibo shares his origin story, from being a bored kid in a small village to studying applied mathematics and founding a supply chain optimization startup.
- 3 (06:22) **Google, DeepMind, and the Internal Chatbot** - Tibo describes his time at Google (working on a canceled web-speed project, then Google Maps) and his move to DeepMind, including building an internal chatbot precursor to ChatGPT.
- 4 (11:40) **Why Join OpenAI?** - Tibo explains why he left a comfortable position at Google/DeepMind to join OpenAI, driven by mission, talent density, and the desire for direct impact.
- 5 (14:22) **The Origins of Codex** - Tibo details how Codex started as an internal tool to help OpenAI’s own researchers code faster, training models on the company’s Python codebase.
- 6 (17:24) **Why Build Codex in Rust?** - The team chose Rust for the harness despite the model not being strong on Rust at the time, prioritizing correctness, security, and a clean separation between the agent and the product.
- 7 (20:12) **Why Open Source Codex?** - Tibo explains the decision to open source the CLI, SDK, and app server, a unique move among major AI labs, and discusses the real benefits and downsides.
+ Full timestamped outline available in the app
Show Notes
Brought to You By:
• turbopuffer – a vector and full-text search engine built on object storage. It’s fast, cheap, and extremely scalable
• Antithesis – verify your system’s correctness without human review or traditional integration tests – and avoid bugs or outages.
• Entire – Git hosting, rebuilt for the agentic era. Every agent session, prompt and tool calls: stored in your repo.
—
Tibo Sottiaux is one of the engineers who created Codex, and today, he heads up the Core Products & Platform org at OpenAI which also includes Codex. He’s also one of the most public faces of Codex due to his frequent – and generous – usage reset announcements, like this one yesterday.
In this episode of the Pragmatic Engineer Podcast, Tibo and I discuss how Codex was built and continues to be iterated upon. We explore why the Codex CLI is written in Rust and was released as open source, how the harness and models have evolved, and why Codex supports models from multiple providers.
Tibo also shares details about how the OpenAI team uses Codex throughout the software development lifecycle, including code reviews, maintenance, and system rearchitecture. We look into how AI is lowering the cost of changing code – and some interesting side effects of this – the merger of ChatGPT and Codex, and also how Tibo uses the tools in his own work.
—
Timestamps
00:00 Intro
07:21 Working at Google
12:41 What drew Tibo to OpenAI
15:19 The early days of Codex
18:20 Why Codex was built in Rust
21:15 Why Codex is open source
25:50 Codex plays nice with other models: why?
32:09 How the harness works
36:44 Harness and model improvements
41:19 The SDLC behind Codex
46:39 Code reviews at Codex
52:09 Maintenance and architecture
56:43 How AI tools expand what engineers can do
1:02:30 The Merge: ChatGPT + Codex
1:07:16 How Tibo uses Codex and ChatGPT
1:10:44 Advice for engineers who want to work in AI
—
The Pragmatic Engineer deepdives relevant for this episode: