Thibault Sottiaux built Codex for developers. Now he runs OpenAI's consumer super app bet, after the company shuttered Sora, a science AI platform, and other bets to consolidate.
OpenAI spent the last year building a constellation of AI products and then tore most of them down. The company has shuttered its stand-alone video app Sora, shut down an AI platform aimed at scientists, and watched many of the executives who led those products walk out the door. What remains, by design, is a single bet: a "super app" that folds ChatGPT and the Codex development environment into one personalized agent for personal and professional work. The engineer now running that bet, Thibault Sottiaux, told WIRED he was appointed OpenAI's head of core products last month, with oversight of both products.
The promotion lands at an awkward moment. Sottiaux's reputation comes from Codex, where his user base was developers and AI researchers. ChatGPT, the surface he now also runs, serves a consumer audience an order of magnitude larger than the developer base that adopted Codex. Sottiaux acknowledged the shift in his interview with WIRED, describing the combined product as a chance to build "the world's best personal agent that deeply understands what humans care about" and to make it "delightfully proactive." The catch is that he now has to make good on that promise for users whose tolerance for an agent acting on their behalf is nothing like a developer's tolerance for a tool that fails loudly.
The organizational picture around Sottiaux is unusual. According to WIRED, Sottiaux now reports directly to Greg Brockman, who is currently responsible for all of OpenAI's product teams. The executive who would normally sit above a consumer product of this size, Fidji Simo, OpenAI's CEO of AGI deployment, is on medical leave. That reporting line is not a footnote. It puts the super-app bet in the hands of an engineering leader who, by his own account, holds "enormous sway" over the strategy, with Brockman as his day-to-day boss during a period when the company is publicly committed to a single product surface.
The decision to collapse the product line is itself the news. WIRED's reporting details how OpenAI closed down several stand-alone efforts, including Sora and the science AI platform, to refocus on the super app. Many of the executives who led those shuttered teams have since left the company. Sottiaux framed the consolidation as sharpening focus, not retreat. Whether it reads that way depends on what the super app actually does for a real user tomorrow, and that is the part neither Sottiaux nor OpenAI has yet spelled out.
The pressure behind the bet is harder than the rhetoric suggests. OpenAI faces a competitive field that includes Google and Anthropic shipping competing agent features on a weekly cadence, and the consolidation of products into a single surface narrows the room for the company to be wrong. Sottiaux's own description of the work, as quoted by WIRED, is that building a personalized agent that acts on a user's behalf is "mildly terrifying." That is an honest framing of a category problem OpenAI has not solved: knowing when a proactive agent should act, when it should ask, and when it should stay out of the way is not a feature flag, it is the product.
What to watch next is concrete. The first test is whether the ChatGPT and Codex surfaces begin to share a single identity, account, and memory layer, or whether they remain two products glued together at the login screen. The second is whether OpenAI can name, in product terms, what "delightfully proactive" means for a user who did not ask for an agent in the first place. The third is whether the engineers and researchers who left after the shuttered products are replaced by people who signed up to build one app, or by people who would have preferred to keep building several. Sottiaux now owns all three questions, and the answer to any one of them will tell readers more about OpenAI's next phase than another round of "super app" headlines.