From Stateless Prompts to Persistent Enterprise AI Agents
OpenAI’s acquisition of Ona marks a shift from short, stateless prompts toward enterprise AI agents that live in secure, persistent cloud execution environments and can run long-running workflows that continue working for hours or days inside customer-controlled infrastructure. OpenAI plans to fold Ona’s cloud execution and orchestration stack into Codex so agents are no longer tied to a single device or active user session. This changes how enterprises think about agent deployment: instead of a model answering one-off questions, Codex agents can manage extended tasks such as research, structured analysis or software maintenance, with progress checkpoints and human review stages along the way. The move reflects rising enterprise demand for secure AI infrastructure where agents can access tools, systems and context across time while respecting governance, audit and access control requirements that apply to production systems and sensitive data.
Codex Adoption Shows Demand for Code-Driven Automation
Codex’s rapid growth underpins the Ona deal. OpenAI reports that weekly Codex usage has passed five million people, a 400% increase from earlier in 2026, as teams apply it to research, analysis, software development and workflow automation. This surge shows how code-focused enterprise AI agents are moving beyond simple code completion toward orchestrating entire software and knowledge-work pipelines. Organizations are starting to hand Codex agents responsibilities such as automated testing, resolving software issues, addressing vulnerabilities and modernizing applications. As these tasks expand, one-off API calls are no longer enough: enterprises want agents that can hold state, remember context and continue work without being tethered to a developer’s laptop. Growing Codex adoption therefore highlights an architectural gap that Ona’s cloud execution environments are designed to fill, aligning AI model capability with long-running engineering and business workflows.
Why Secure Cloud Execution Environments Now Matter Most
Ona has spent years helping around two million developers move from fragile local setups to reproducible cloud execution environments, experience that now becomes central to Codex. OpenAI is targeting enterprises that are moving agents from pilot experiments into production workflows, where governance, compliance and secure AI infrastructure are as important as raw model performance. Ona’s customer-controlled execution model lets AI agents run in an organization’s own cloud environment, while OpenAI supplies models and orchestration. Enterprises decide where agents run, which systems they can access, how credentials are limited, what gets logged and when human review is required. According to OpenAI Core Products Lead Thibault Sottiaux, “Enterprises want powerful agents that can do real work while meeting the security and control requirements of their environments,” making secure cloud execution environments a key differentiator in enterprise AI agent deployments.
Enterprise Infrastructure Priorities: Sovereign, Secure and Agent-Ready
The Ona acquisition highlights how infrastructure priorities are changing as enterprises plan for agentic AI workloads. Instead of sending prompts to vendor-hosted black boxes, organizations want execution that runs within their own cloud execution environments, aligned with existing access control, auditing and data residency rules. Long-running workflows such as full software lifecycle management, complex internal analytics or multi-step business processes require agents that can be paused, inspected and restarted without losing context. Johannes Landgraf, Ona’s co-founder and CEO, sums up the new requirement: “Agents need more than intelligence; they need a trusted workspace.” By integrating Ona, OpenAI positions Codex as not only a model but an operational layer for enterprise AI agents—one that supports sovereign control over infrastructure while enabling agents to work continuously, collaborate with humans and access production systems under strict, transparent guardrails.
