Defining the shift: from quick prompts to sustained AI operations
OpenAI’s acquisition of Ona marks a strategic transition in AI agents enterprise workflows, where Codex moves from short, single-session tasks toward long-running AI automation that lives inside secure, persistent cloud environments owned and controlled by customers. Instead of agents running only while a user keeps a laptop open, Ona’s infrastructure is designed so Codex agents can keep working for hours or days on complex software and knowledge work. OpenAI reports that more than five million people now use Codex weekly, a 400% increase earlier in 2026, as usage expands from code snippets to research, analysis and workplace automation. This surge and the Ona acquisition show that OpenAI Codex acquisition strategy is shifting from isolated productivity boosts to sustained enterprise operations, where continuity, auditability and cloud-native execution matter as much as raw model performance.
Codex usage climbs 400% as workflows grow longer and more complex
Codex began as a coding assistant, but its usage profile now looks broader and more operational. OpenAI says weekly Codex usage has passed five million people, representing a 400% increase from earlier in 2026, driven by software development, research, analysis and workflow automation. As teams push AI agents into more complex work, tasks stretch beyond a single sitting: running extensive test suites, resolving software issues across multiple services, modernizing legacy applications or completing multi-step business processes. These workloads may run over hours or even days and span different tools and systems. Current limitations—agents tied to a single device or browser session—make it hard to keep context and progress intact. The Ona deal responds to this shift, aligning Codex with enterprises that expect AI agents to behave less like chatbots and more like dependable, long-running collaborators embedded in their engineering and business workflows.
Secure cloud execution as the missing piece for enterprise AI agents
For many organizations, the barrier to wider AI deployment is not model capability but control. Enterprises want long-running AI automation without losing sight of where agents run, what they can access and how activity is logged. Ona’s secure cloud execution technology directly targets these requirements. Its customer-controlled execution model lets AI agents operate inside an organization’s own cloud, while OpenAI supplies the models and orchestration. That means teams can define security boundaries, manage credentials, and decide when work should pause for human review. Johannes Landgraf, Ona’s co-founder and CEO, argues that “agents need more than intelligence; they need a trusted workspace,” capturing why secure cloud execution is central to AI agents enterprise workflows. By integrating Ona’s reproducible, persistent environments, OpenAI Codex can fit into existing governance and compliance frameworks rather than forcing enterprises into vendor-managed black boxes.
Customer-controlled environments bring AI agents into production workflows
Ona’s experience with around two million developers moving workloads from laptops into reproducible cloud environments now feeds directly into Codex’s enterprise offering. Once the acquisition closes and the Ona team joins the Codex organization, OpenAI plans to support agents that run where the customer chooses, with clear policies around systems access and auditing. This is especially relevant for production software lifecycle tasks such as automated testing, vulnerability remediation, issue resolution and application modernization. In these settings, agents act over extended periods, interacting with real infrastructure and sensitive code. According to OpenAI, enterprises “want powerful agents that can do real work while meeting the security and control requirements of their environments.” By embedding secure cloud execution into Codex, the OpenAI Codex acquisition strategy positions AI agents as first-class participants in production workflows, not experimental side tools.
From short-task automation to sustained enterprise operations
The Ona acquisition signals a broader shift in how OpenAI imagines Codex in the enterprise. Early Codex deployments focused on short-task automation: generate a code block, draft a query, summarize a document. Now, OpenAI is designing Codex for end-to-end assignments that span initial request to finished outcome across long-running AI automation. With persistent cloud environments, agents can keep state, coordinate multiple tools and accept intermittent human guidance without restarting every session. Users will be able to check progress from different locations, provide new instructions and review outputs while the agent continues working in the background. This aligns Codex with the operational reality of software and knowledge work, where value comes from completing complex chains of activity reliably. The OpenAI Codex acquisition of Ona therefore marks a turning point: AI agents enterprise workflows evolve from quick interactions toward durable, governed and continuous operations.
