OpenAI Codex now runs persistent cloud tasks you can hand off between devices. Start a job on your laptop, shut it, and resume on your phone. The environment is not a sandbox that dies with your session.
The setup involves three concrete pieces: repository access, network configuration, and secrets management, all bundled into a reusable environment template you define once and share across your team. This is infrastructure, not a chat interface.
The part worth watching in full is how the sharing model works across a team and what the networking layer actually allows Codex to reach. Those two details determine whether this is useful for real engineering workflows or just a demo.
[WATCH ON YOUTUBE →]