SNACK in 3 lines
- OpenAI announced its plan to acquire Ona on June 11. It also said it wants to bring secure cloud execution and orchestration technology into Codex.
- The key point is not just “adding another coding AI,” but not keeping agent work tied to a single laptop and a single session. OpenAI said Codex work now continues for hours or days, not just minutes.
- That said, the acquisition still has closing conditions such as regulatory approval remaining, and the two companies will operate separately until closing. In other words, this is not a feature launch where general users can press a new button today.
Snackgirls Editorial Notes
AIKO: “This announcement is best read as a move to give Codex a workspace where agents can keep hold of long-running jobs.”
Red: “Put simply, the idea is that AI won’t work only while it’s sitting in front of your laptop. Even after you close the window, it would keep working in the cloud.”
Nea: “So the real point today is less the acquisition itself and more that OpenAI is starting to take where to place the security boundary and execution environment much more seriously.”
What changed
OpenAI said it plans to acquire Ona and bring secure cloud execution and orchestration technology into the Codex ecosystem. The most emphasized line in the announcement was that the valuable work Codex does is growing beyond helper tasks that finish in minutes into complex work that continues for hours or even days.
In other words, this looks less like a tool that helps only while a user is sitting in front of a laptop, and more like a signal that OpenAI wants to move toward a structure where the customer sets the direction and the agent keeps working for longer.
Why Ona matters
By OpenAI’s description, Ona provides secure, persistent environments so agents can carry the tools, systems, and context they need and continue working over time. Ona’s own post explains the same point by saying companies want not just a chatbot, but an agent that keeps working inside a cloud boundary controlled by the customer.
A simple comparison would be this: if AI has so far been working from one corner of your laptop desk, the next step is to have it handle longer-term work from a separate office managed by the company, with access controls and records in place.
Why this matters to general readers and developers
OpenAI said Codex currently has 5 million weekly users, up 400% from earlier this year. As usage grows, what matters as much as model performance is where it runs, what it can access, and who reviews it.
Ona’s homepage and comparison pages show this point very directly. It suggests that operational factors such as running inside the customer cloud, policy and permission controls, audit logs, and multi-repository workflows are the real bottlenecks for adopting agents.
What still calls for caution
This is still at the pre-closing stage of the acquisition, so the timing and scope of product integration are not yet final. That means it would be overstating things to say ChatGPT or Codex usage changes immediately today.
Also, the direction OpenAI is describing is heavily enterprise- and operations-focused. So today’s news is more accurately understood not as a consumer feature story, but as an infrastructure announcement about where and how AI agents will run long-term work.
Sources and checked date · Announcement 2026-06-11 / Checked 2026-06-12T01:20:28+00:00
Sources
Leave a comment