Meet the OpenAI Engineer Leading ChatGPT’s Biggest Transformation Yet
OpenAI has tapped Thibault Sottiaux to help turn ChatGPT into a personalized AI super app. The plan hinges on folding Codex into ChatGPT and making the assistant more proactive and useful.
Intelligence analysis by GPT-5.4 Mini
OpenAI is betting that ChatGPT can evolve from a chatbot into a broad personal agent that handles everyday tasks across work and life. Thibault Sottiaux, newly in charge of core products, is now central to that push after helping build Codex into a fast-growing business.
OpenAI wants ChatGPT to become like a super helper that knows a person well, remembers things, and can do jobs like booking a table or sorting work chores. Thibault Sottiaux is the person helping build that big upgrade.
Analysis
What OpenAI is trying to build
OpenAI is overhauling ChatGPT into what it internally and publicly calls a “super app”: a personalized AI agent meant to help across many parts of a person’s life. The company’s aim, as described in the article, is not just a smarter chatbot but a system that can bring the right information or help at the right time, with more memory and more initiative.
Why Sottiaux matters
Thibault Sottiaux was recently named OpenAI’s head of core products, putting him in charge of both ChatGPT and Codex. The article frames him as a key executive because he helped build Codex into one of OpenAI’s fastest-growing revenue lines, and now he has to apply that success to a much broader consumer product with nearly a billion weekly users.
How the product may work
Sottiaux says the future ChatGPT should be “delightfully proactive.” In practical terms, the article suggests it could help with tasks like making dinner reservations, remembering dietary preferences, and filing expense reports. Under the hood, OpenAI expects Codex to power much of this by writing code, calling APIs, or browsing the web without exposing those steps to the user.
The challenge
OpenAI has tried agent-style products before, including Operator and ChatGPT Agent, but they did not gain significant adoption. Sottiaux argues those efforts came too early, before the models were reliable enough. The bigger challenge now is not only technical reliability, but also teaching regular users how to trust and use AI in new ways.
The larger bet
The article makes clear that OpenAI sees this as a way to restart growth and strengthen its position ahead of a possible IPO. But it also notes a structural risk: the company wants a universal interface that depends on outside services, platforms, and infrastructure it does not control.
Key points
- OpenAI wants to turn ChatGPT into a personalized AI super app that can handle many everyday tasks.
- Thibault Sottiaux, formerly a key Codex builder, now leads OpenAI’s core products team.
- The company plans to merge Codex into ChatGPT and use it as the engine for more agent-like behavior.
- OpenAI’s previous agent efforts did not gain much adoption, partly because the models were not reliable enough.
- The article frames the effort as a major growth bet as OpenAI competes with Google and Anthropic.
If OpenAI pulls this off, ChatGPT could become a genuinely useful everyday assistant instead of just a chat box. The article suggests that combining Codex with ChatGPT could make the product more proactive, more personalized, and more capable for both personal and work tasks.
The article also shows how hard this is: OpenAI’s earlier agent products did not catch on, and users may still not know what to do with a more capable assistant. There is also a dependency risk, since OpenAI wants a universal interface built on top of services and infrastructure it does not fully control.
