OpenAI Finalizes Ona Acquisition, Boosting Codex Coding Agent

OpenAI has closed its acquisition of Ona, a German startup previously known as Gitpod that provides secure, ready-to-use cloud environments for AI agents. The deal was announced in June and formally closed in August, after which Ona's entire team joined the group responsible for developing Codex, OpenAI's coding agent.

From Gitpod to Ona

The company was founded in Germany as Gitpod and spent years building tools that moved coding from local machines to the cloud, serving more than 2 million developers along the way. In late 2025 it rebranded as Ona and rebuilt its product around AI agents instead of traditional development environments for humans.

The team has been led from the start by co-founder and CEO Johannes Landgraf. After the deal closed, Ona's entire 79-person staff moved to OpenAI and is now working on integrating the cloud sandbox technology with Codex.

A trusted work environment

At the heart of the acquisition is a problem every maker of coding agents faces: the longer an agent works independently, the harder it is to guarantee it won't do anything harmful or stop working the moment the user closes their laptop. Ona's technology offers persistent, isolated cloud environments where an agent can keep refactoring a large codebase, work through a long backlog, or fix an extensive test suite without interruption.

Agents need more than intelligence, they need a trusted place to work - Johannes Landgraf, co-founder and CEO of Ona

Ona's environments come with built-in security mechanisms designed for agents rather than humans: they block renamed applications from running by verifying hashes, prevent access to file paths containing credentials, and cut off outbound connections to known malicious servers.

Ona will help us make Codex easier to deploy safely in production environments for customers operating at the highest standards of trust and scale - Thibault Sottiaux, head of core products at OpenAI

The coding race with Anthropic

The acquisition fits into OpenAI's broader strategy against growing competition from Anthropic, whose Claude Code agent has become highly popular among companies rolling out coding automation. OpenAI is responding with investment in Codex, whose usage grew 400 percent this year and whose weekly user count topped 5 million.

Ona is already OpenAI's sixth acquisition in just the first half of 2026, compared with eight such deals for all of 2025. Earlier, in March 2026, OpenAI bought the startup Astral, and the company also recruited OpenClaw's founder to work on its own agent architecture.

What it means for businesses and developers

For IT departments considering deploying coding agents at scale, the security of long-running processes, not just the quality of generated code, will be the deciding factor. The isolation mechanisms and credential-access blocks that Ona brings address the concerns of security teams that have so far treated autonomous agents as a hard-to-control risk.

For Polish companies using Codex in their daily development work, this could mean the ability to hand the agent tasks lasting hours or days, such as large code migrations or clearing a backlog of tests, without keeping a machine running or having a human oversee every single step.

OpenAI has not given a timeline for integrating Ona's technology into the publicly available version of Codex. Landgraf's team is now working inside the Codex division to bring the cloud sandbox mechanisms into the main product, which is meant to pave the way for longer, fully autonomous agent work sessions in enterprise customer environments.