OpenAI has put the people behind Codex closer to the center of its product machinery, a change that could shape how ChatGPT, developer tools and the company’s browser ambitions get packaged for users.
Wired reported on May 15 that OpenAI is combining ChatGPT, its Codex coding agent and its developer-facing API into a single core product team. OpenAI told Wired that Codex is playing a growing role in its consumer and enterprise products, which are adding the ability to carry out digital tasks for users without constant step-by-step direction.
The personnel move matters because it shifts authority toward the team responsible for an agentic coding product, rather than keeping ChatGPT’s consumer product sensibility as the obvious center of gravity. Wired reported that Thibault Sottiaux, OpenAI’s head of Codex, has been chosen to lead the company’s core product and platform teams.
According to Wired, Sottiaux helped turn Codex into one of OpenAI’s fastest-growing products. He is also among the executives overseeing OpenAI’s planned “super app,” which Wired described as an effort to bring Codex, ChatGPT and the company’s Atlas web browser into one desktop application.
Criticism from the product side
Daring Fireball, reacting to the reshuffle, argued that OpenAI chose the wrong internal model to elevate. Its criticism was blunt: Codex, in that view, has become too broad and hard to understand, while ChatGPT had been stronger as a focused consumer product.
The critique is less about whether coding agents matter and more about product discipline. Daring Fireball argued that OpenAI should have pushed ChatGPT’s clearer product approach into Codex, rather than giving Codex leadership broader influence over ChatGPT and related platform work.
That is an opinion, not a reported internal assessment from OpenAI. OpenAI’s stated rationale, according to Wired, is that Codex increasingly underpins products aimed at both individual users and companies. The company appears to be organizing around software that can take actions across apps and workflows, rather than around chat as a standalone interface.
Daring Fireball also compared the direction to Anthropic’s Claude, saying OpenAI’s leadership is focused on copying it. The criticism singled out Claude’s overlapping product concepts, including Extensions, Plugins, Capabilities, Skills and Connectors, as an example of how AI products can become hard to explain when every new ability gets its own label.
The confirmed change is narrower: OpenAI has reorganized ChatGPT, Codex and its API under one core product structure, and Sottiaux has been given a larger mandate. The risk, as critics see it, is that a product known for broad technical capability may make ChatGPT feel less coherent. The bet, as OpenAI described it to Wired, is that Codex is becoming central enough to justify putting it closer to everything else.
This story draws on original reporting from Daring Fireball.
