VERDICT SUMMARY
The integration of Ona’s cloud orchestration technology into the OpenAI Codex ecosystem represents a fundamental shift in the utility of AI agents. By moving from session-based, ephemeral interactions to persistent, autonomous workflows, OpenAI has addressed the primary limitation of current AI assistants: the requirement for constant human supervision and active device connectivity. We rate this acquisition 4.5/5, as it provides a revolutionary framework for enterprise workflow continuity. The core value proposition lies in the ability for agents to operate for hours or even days within secure cloud environments, completing complex technical tasks independently of the user’s local machine status.
OVERVIEW
OpenAI recently announced its agreement to acquire Ona, a cloud development environment (CDE) provider formerly known as Gitpod. This 79-person startup specializes in secure cloud execution and orchestration, technologies that OpenAI intends to weave directly into its Codex ecosystem. This move comes at a time of massive scaling for OpenAI’s coding tools; Codex usage has reportedly climbed 400% in 2026, now serving approximately 5 million weekly users. The acquisition is a strategic step toward making agentic AI enterprise-friendly by solving the “ephemeral session” problem that has previously limited AI agents to short-term, reactive tasks.
The acquisition occurs against a backdrop of significant corporate expansion for OpenAI. As the company navigates a reported $852 billion valuation and prepares for a potential IPO, its focus has shifted toward building a “business-ready” suite of AI tools. Ona’s technology provides the infrastructure necessary to move AI agents out of the chat box and into production-grade cloud environments. According to reports from InfoWorld, the deal is designed to help organizations deploy agents securely, ensuring that these autonomous entities have the tools and context they need to make progress over time without constant human intervention.
Currently, OpenAI and Ona continue to operate as independent entities while the deal remains subject to regulatory approval and customary closing conditions. Upon completion, Ona Co-Founder and CEO Johannes Landgraf and his team are expected to join the OpenAI Codex organization. This integration is expected to accelerate the development of “agentic” workflows, where AI does not just suggest code but actively manages software lifecycles within controlled, persistent environments.
KEY FEATURES
Persistent Cloud Execution Environments
The most significant technical upgrade provided by the Ona acquisition is the transition from single-device sessions to long-running cloud execution. Previously, AI agents were often tied to a single active session or the user’s local device; if the user closed their laptop or the session timed out, the agent’s progress would halt. As reported by EdTech Innovation Hub, Ona’s technology enables agents to work for hours or days within persistent cloud environments. This persistence allows for the completion of time-consuming tasks that are impossible to finish in a single sitting.
OpenAI CFO Sarah Friar has emphasized that this capability allows agents to function over extended periods, moving away from a “minutes-based” interaction model. This is critical for tasks like large-scale refactoring or complex debugging, where an agent may need to run tests, wait for results, and iterate multiple times. Ona provides the “context” and “tools” necessary for the agent to maintain its state, ensuring that when a human developer checks back in, the agent has made verifiable progress rather than losing its place in the workflow.
Enterprise-Grade Security and Infrastructure Control
A primary barrier to the adoption of autonomous agents has been the concern over security and data boundaries. To address this, the integration allows organizations to run agents inside their own cloud environments. This ensures that the enterprise retains full control over the infrastructure, data, and security protocols. According to Built In, leveraging Ona’s capabilities within Codex enables teams to resolve issues and optimize applications while remaining within the organization’s existing security framework.
Thibault Sottiaux of Ona noted that the technology is designed to meet the high security and control requirements of sophisticated enterprise customers. This level of infrastructure control differentiates the new Codex agents from standard API-based tools. Instead of sending data to an external black box, the agent operates as a temporary “cloud employee” within the company’s own virtual walls. This setup mitigates risks associated with data leakage and provides Chief Information Security Officers (CISOs) with the oversight needed to authorize autonomous operations.
Autonomous Software Engineering Lifecycle
The incorporation of Ona’s technology into Codex aims to support the entire software engineering lifecycle. Beyond mere code generation, these agents are being equipped to function as autonomous AI software engineers. This includes the ability to address vulnerabilities, optimize application performance, and support complex workflows over time. According to OpenAI, the goal is to expand Codex beyond work tied to a single device, helping organizations deploy agents in full production environments.
By integrating these capabilities, OpenAI is positioning Codex as a platform for “agentic AI.” This means the tool can take a high-level instruction—such as “optimize this database for query speed”—and carry out the necessary analysis, implementation, and testing autonomously. Because the environment is persistent, the agent can handle the “wait times” associated with deployment and testing cycles, notifying the human lead only when the task is complete or if a high-level decision is required.
PROS AND CONS
Pros
- Sustained Productivity: The primary advantage is the ability for Codex to handle long-running tasks without human supervision. This effectively doubles or triples the productive hours of a development team by allowing AI agents to work through the night or during off-hours.
- Proven Reliability: Ona’s underlying technology is not experimental; it has already supported over 2 million developers in its previous incarnation as Gitpod. This provides a level of stability and reliability that is often missing from newer AI startups.
- Regulatory Compliance: The ability to run agents within a company’s own cloud boundaries is a significant pro for industries like finance and healthcare, where data residency and security are non-negotiable.
Cons
- Vendor Lock-in: Gartner has issued warnings regarding the “vendor-specific integrated stack.” By moving toward a model where OpenAI provides both the intelligence (Codex) and the environment (Ona), users may find it increasingly difficult to migrate their workflows to other providers.
- Operational Risks: CIOs and CISOs remain concerned about “mission tangents,” where an autonomous agent might deviate from its assigned task. There is also the risk of “token bill shock,” where a long-running agent generates massive costs by running in a loop without human oversight.
- Infrastructure Complexity: While persistent environments offer more power, they also require more sophisticated management compared to simple, stateless API calls.
PERFORMANCE / TESTING
The real-world adoption metrics for Ona’s technology suggest a high level of demand for persistent agentic workflows. Since the beginning of the year, there has been a reported 13x growth in Ona agent sessions. This exponential increase in production environments implies that the system is stable enough to handle scaling and that developers are finding concrete value in moving away from local, session-based development. When an agent can continue working “when a user’s laptop is closed,” it effectively decouples human presence from technical progress.
Analysis of this 13x growth suggests that the technology is moving out of the “hobbyist” phase and into core business operations. For a system to sustain this level of growth, it must demonstrate high reliability; if the persistent environments were prone to crashing or losing state, developers would revert to manual sessions. The ability to maintain a secure, persistent state across hours of operation is a benchmark of production readiness that few other AI orchestration platforms have achieved.
OpenAI has already highlighted successful deployments of this tech-stack in highly regulated sectors. These include the “oldest bank in the US” and a “large pharma company.” In these environments, performance is measured not just by speed, but by the agent’s ability to operate within strict compliance parameters while performing broad technical and knowledge-based work. The shift from pure software development into these broader categories suggests that the Ona-powered Codex is being used for more than just writing lines of code; it is being used to manage complex data and infrastructure tasks that require long-term “thinking” and execution time.
COMPARISON
The following table compares the OpenAI/Ona integrated approach against traditional AI assistant models and general cloud development environments.
| Feature | OpenAI + Ona (Integrated Stack) | Standard AI Assistants (e.g., Claude, ChatGPT) | Traditional CDEs (e.g., GitHub Codespaces) |
|---|---|---|---|
| Persistence | High; runs for days in the cloud. | Low; session-limited. | High; persistent but human-led. |
| Autonomy | Autonomous; manages tasks independently. | Reactive; requires prompt/response. | None; requires human input. |
| Environment | Secure, private cloud orchestration. | Stateless API or hosted chat. | Standardized cloud VM. |
| Best Use Case | Long-running enterprise workflows. | Quick queries and code snippets. | Manual remote development. |
When compared to current session-limited competitors like standard GitHub Copilot or Anthropic Claude sessions, the Codex/Ona combination offers a distinct advantage in workflow continuity. While Copilot is excellent for real-time pair programming, it lacks the “background agent” capability that Ona provides. Furthermore, this acquisition fits alongside OpenAI’s April 2026 acquisition of Hiro Finance. Together, these moves suggest OpenAI is building a comprehensive, business-ready AI suite that handles both the financial transactions of agents and the technical environments they inhabit.
WHO SHOULD BUY THIS
The Ona-integrated Codex is primarily targeted at large-scale financial and pharmaceutical institutions. These organizations require the high level of security provided by running agents within their own cloud infrastructure to meet regulatory compliance standards. For these users, the ability to audit the agent’s environment and control data boundaries is more important than the raw speed of the AI itself.
Additionally, this tool is highly suitable for organizations looking to automate the resolution of technical debt. Small-to-medium businesses with large, aging codebases can utilize persistent agents to work through thousands of lines of code, identifying vulnerabilities and optimizing performance without pulling their senior developers away from new feature work. If your workflow requires an AI to “stay on the job” while your team is offline, this is currently the most robust solution on the market.
FINAL VERDICT
The acquisition of Ona does not necessarily “rein in” AI agents; rather, it expands their leash by providing them with a stable, secure, and persistent place to work. By moving the execution environment to the cloud and allowing for multi-day autonomy, OpenAI has removed one of the biggest friction points in AI adoption. While risks regarding token costs and vendor lock-in remain, the productivity gains from persistent workflows are too significant for most enterprises to ignore. The transition of the Ona team into OpenAI signals a future where AI is a constant, background presence in the software lifecycle, rather than just a tool used in active sessions.
Frequently Asked Questions
How does Ona's technology enhance OpenAI Codex?
Ona provides cloud orchestration that enables persistent, autonomous workflows, allowing AI agents to continue working on complex tasks even after a user closes their local session or device.
What security features does the OpenAI and Ona integration offer?
The integration allows enterprises to run AI agents within their own secure cloud environments, ensuring full control over data boundaries and infrastructure protocols.
What are the risks of using persistent AI agents in enterprise workflows?
Potential risks include vendor lock-in, operational 'mission tangents' where agents deviate from tasks, and unexpected costs known as 'token bill shock' from long-running autonomous sessions.
