OpenAI Merged ChatGPT, Codex, and Its Developer API Three Days Before Google I/O. Greg Brockman Is Now Running All of Product. Here's Why the Timing Is Not a Coincidence.
On May 16, OpenAI announced that Greg Brockman (co-founder, former CTO, the person who was technically on sabbatical) is now permanently running all product strategy. ChatGPT, Codex, and the developer API are unified under one organization. The announcement landed four days before the Google I/O keynote.
That timing is a choice, not a coincidence.
What the reorg actually changed
Before this announcement, OpenAI had three distinct product groups operating with semi-independent roadmaps. ChatGPT was consumer. Codex was developer tooling. The API was the enterprise revenue engine. Each had different priorities and different customers, and when their roadmaps conflicted, which they did constantly, there wasn't a clean decision layer to resolve it.
Brockman changes that. One person owns all three surfaces, which means one person sets the priority when ChatGPT's product team wants a feature that the API team doesn't want to build, or when Codex's agentic coding direction pulls against the model deployment roadmap that enterprise customers depend on.
The practical result: the "super app" thesis becomes executable. OpenAI has been building toward a single desktop interface: ChatGPT, Codex, and the Atlas browser merged into one product. That's been described in reporting since March, but describing it and executing it are different things. A unified product org under one founder is what lets you actually ship the integration, because the tradeoffs get made in one room rather than three.
Why the timing is a strategic play
Google I/O runs May 19 and 20. The main keynote is Tuesday. Every major tech outlet is writing preview pieces, anticipation features, and speculation threads about what Google will announce. That's the news environment OpenAI chose to launch a major organizational announcement into.
This is a classic pre-show counter-positioning move. OpenAI saturates the pre-I/O news cycle with a "we're unifying everything into one platform" story, which reframes the narrative before Google can land its messaging about Gemini being the AI that runs across all your screens. By the time Tuesday's keynote airs, readers have already spent two days reading about OpenAI's structural consolidation. Google's announcements land into a conversation OpenAI already started.
It's not subtle. That's the point.
What this means if you build on OpenAI
If your stack is already OpenAI-native (you're calling the API, you're using Codex for coding tasks, your users are ChatGPT users), the unification is directionally positive for you. The model capabilities, memory, and tool access across surfaces are going to get more coherent, not less. A platform that shares context across ChatGPT, Codex, and the API is a platform with higher switching cost, which is good for continuity if you're already committed.
If you're evaluating whether to go deeper on OpenAI versus an alternative, the Brockman reorg is worth reading as a signal about where the platform is heading. The "super app" goal (browser + chat + code agent in one interface) is a significant lock-in play. If OpenAI executes it, moving a customer off the platform gets harder every month.
I've been watching how much of my own workflow has consolidated into OpenAI surfaces over the last six months. Not because I made a deliberate bet, but because the convenience accumulated. The reorg formalizes what was already happening organically.
The honest counter-take
The case against reading this as purely strategic: OpenAI has real organizational problems. The turnover at the leadership level has been constant. Brockman's return from sabbatical may be as much about stabilizing internal dysfunction as it is about executing a brilliant pre-Google move. Org charts can be changed without actually unifying product culture, and a single org chart doesn't guarantee a coherent super app ships in 2026.
There's also a real risk in the platform consolidation thesis: the more OpenAI's surfaces share context and memory, the more interesting the data governance questions get. That's a conversation enterprise buyers will push on. "What does ChatGPT know about our Codex sessions?" is not a question that currently has a clean public answer.
Watch how the developer docs change in the next 60 days. If the API team's documentation starts aligning more tightly with ChatGPT feature language, that's evidence the unification is real. If the docs stay siloed, the org change is cosmetic.
Author
Lukas
@lukcombinator