OpenAI rolls out Codex desktop app as AI agents move beyond code generation
The new macOS app signals a shift from single-agent coding to coordinated, long-running AI workflows, with implications for how developers learn, collaborate, and build software at scale.
OpenAI has launched a new desktop application for Codex, introducing a macOS-based interface designed to manage multiple AI agents working in parallel across software projects.
The release reflects a broader shift in AI-assisted development, as tools move beyond code generation toward long-running, supervised workflows that span design, build, testing, and maintenance.
The Codex app is positioned as a control layer for orchestrating agent-based work, addressing limitations in traditional IDEs and command-line tools that were built for single-threaded development rather than multi-agent collaboration.
A command center for AI agents
OpenAI says the Codex app provides a centralized environment where developers can run, monitor, and review work from multiple agents at once. Each agent operates in its own thread, organized by project, allowing developers to switch between tasks without losing context.
The app supports isolated worktrees so multiple agents can operate on the same repository without conflicts. Developers can review changes, comment on diffs, open work directly in their editor, or allow agents to continue independently without altering local Git states. Session history and configuration are shared across the Codex CLI and IDE extension, allowing existing users to move into the desktop app without reconfiguration.
Since Codex launched in April 2025, OpenAI says developer workflows have shifted toward delegating substantial tasks to agents that may run for hours or longer. The focus, it argues, has moved from what agents can do to how humans supervise, direct, and trust them at scale.
From code generation to applied skills
The Codex app expands the system beyond writing code through a feature OpenAI calls “skills,” which bundle instructions, tools, and scripts so agents can complete structured tasks such as research, documentation, deployment, and content creation.
Skills allow Codex to connect to external services and workflows, including design tools, cloud platforms, and document formats. OpenAI says teams can create, manage, and share skills across projects, enabling more consistent delegation of work that would otherwise be difficult to define repeatedly.
The company frames this as a step toward making AI agents useful across a wider range of technical and knowledge-based tasks, not just programming, with potential implications for how developers and students learn applied problem-solving with AI support.
Automation, access, and availability
The Codex app also introduces Automations, which allow agents to run scheduled tasks in the background, such as issue triage, testing, reporting, and documentation updates. Completed tasks surface in a review queue, keeping humans in the loop without requiring constant oversight.
The app is available today on macOS for users on ChatGPT Plus, Pro, Business, Enterprise, and Edu plans, with usage included in existing subscriptions. For a limited period, Codex is also available to ChatGPT Free and Go users, alongside increased rate limits across paid plans.
OpenAI says Codex usage has grown rapidly since the launch of GPT-5.2-Codex in December, with more than one million developers using the system in the past month. The company plans to extend the app to Windows and continue refining multi-agent workflows, with a stated goal of narrowing the gap between what advanced models can do and how easily people can apply them in real-world work.
ETIH Innovation Awards 2026
The ETIH Innovation Awards 2026 are now open and recognize education technology organizations delivering measurable impact across K–12, higher education, and lifelong learning. The awards are open to entries from the UK, the Americas, and internationally, with submissions assessed on evidence of outcomes and real-world application.
