Codex Comes to the ChatGPT Mobile App
OpenAI has brought its Codex AI coding agent directly into the ChatGPT mobile app, turning smartphones into remote dashboards for live development sessions. Available in preview on both iOS and Android across all ChatGPT plans, the feature connects the app to Codex instances running on a Mac, a dedicated dev machine, or a remote development environment. Rather than trying to write entire applications on a touchscreen, the experience is designed as Codex coding control on the go: developers connect to an existing session and see the live state of their host machine mirrored on their phone. Active threads, current tasks, and pending prompts are all accessible. This transforms the ChatGPT mobile app into a powerful piece of mobile development tools infrastructure, letting developers stay in sync with their AI coding agent without being chained to a desk or laptop.
Remote Code Management from Your Pocket
Once a phone is paired with a machine running Codex, the ChatGPT mobile app effectively becomes a remote code management console. It streams terminal output, code diffs, test results, screenshots, and approval requests back to the device in real time. Developers can review outputs, switch between AI models, re-prompt the AI coding agent, or redirect work if Codex veers off course. The app also sends notifications when long-running tasks complete or when Codex pauses for human approval. This makes it practical to let AI handle background refactors, test runs, or bug triage while developers commute or step away. Crucially, files, credentials, and permissions never leave the host machine; only working output and status updates travel to the phone, preserving security while extending continuous oversight of AI coding workflows.
Secure Sync, Remote SSH, and Enterprise Controls
Under the hood, Codex mobile control relies on a secure relay layer that keeps trusted machines reachable from authorized devices without exposing them to the public internet. The macOS Codex app can now automatically discover remote hosts from a user’s SSH configuration, allowing Codex threads to run directly inside remote environments via Remote SSH. For enterprises, OpenAI is complementing these capabilities with programmatic access tokens suited for CI pipelines, plus custom hooks for scanning prompts or logging conversations. The company has also introduced HIPAA-compliant support for eligible workspaces using Codex locally, positioning the system for healthcare and other regulated sectors that require strict data controls. Together, these features extend Codex coding control beyond a single desktop, tying mobile development tools, remote infrastructure, and compliance workflows into a unified, policy-aware AI development stack.
A New Workflow for Supervising AI Coding Agents
OpenAI describes a new rhythm of collaboration as AI agents take on longer-running tasks, and Codex on the ChatGPT mobile app crystallizes that shift. Instead of treating Codex as a fire-and-forget tool, developers can now continuously supervise AI coding agents, approving sensitive commands, revising prompts, or adjusting priorities as work progresses. The ability to pick up active chats on a phone, respond to questions, and kick off new jobs from anywhere enables real-time oversight even when away from a primary workstation. This also puts OpenAI in direct competition with other remote code management offerings, including rival tools that added remote monitoring earlier in the year. By weaving Codex into mobile, browser, and desktop experiences, OpenAI is pushing toward an always-available development assistant that follows developers across contexts while keeping critical code and credentials anchored on trusted machines.