OpenAI is taking its Codex software engineering agent off the laptop. At DevDay on Tuesday, the company added reusable cloud development environments that you can reach from any device, TechCrunch AI reports. It’s one of several Codex updates announced at the event, along with a refreshed CLI, a new code review experience, security tooling and a batch of API changes.
Codex could already run cloud tasks before this. What’s new is that those environments now stick around and can be configured. Until now, each cloud task ran in its own isolated sandbox. According to TechCrunch AI, OpenAI now wants developers to run Codex “wherever they need it,” whether that’s on a computer, remotely from a phone or in the cloud.
🚀 What OpenAI shipped
- Reusable cloud environments – These persistent workspaces are built so tasks start faster. They also give teams a shared setup with approved settings and permissions. That second part matters most for companies. Admins can decide once how the agent’s environment is configured, and every developer on the team works inside those rules.
- A refreshed Codex CLI with voice control – You can now start and steer tasks by voice. A new /agents view makes it easier to hand off work and track several tasks at the same time. OpenAI also improved the everyday basics: editing prompts, resuming sessions and working with worktrees (separate working copies of a repo that let you run parallel tasks without them colliding). There’s also a “cleaner” terminal interface meant to make long sessions easier to read.
- Code review inside the ChatGPT desktop app – You can bring Codex into a new review experience in the ChatGPT desktop app. It lets you read summaries, explore code changes and ask Codex about possible issues before you leave feedback on GitHub pull requests or GitLab merge requests. With automatic code reviews turned on, Codex does a first pass while you’re away from your desk.
- Codex Security Cloud – This new set of security tools can scan entire GitHub repositories on demand, on a schedule or whenever new commits land. Codex then looks into what it finds, removes duplicates and prepares fixes in the cloud, even when your laptop is closed. It also comes with access to models from OpenAI’s Daybreak Blue cybersecurity initiative, and you don’t have to file a separate application to get them.
- API updates: Decisions API and a bigger Agents API – The new Decisions API handles real-time decision-making. It uses Luna to answer a set of questions you define, picking from answers you’ve set in advance. The updated Agents API now supports computer use. It also lets you build OpenAI agents with Amazon’s Bedrock Managed Agents that run entirely on AWS.
🧭 Why it matters
The big idea is that the agent keeps working after you walk away. The cloud environments, the automatic reviews and the security scans that run with your laptop closed all push Codex in the same direction. It’s becoming a background worker, not just an assistant in your terminal.
It also puts OpenAI in more direct competition with other agentic coding tools that already run tasks in the cloud and plug into pull request workflows. Shared, permissioned environments are the kind of feature engineering leaders need before they’ll let an agent into a real codebase. Individual developers mostly care about speed. Teams care about control, and this update speaks to both.
The AWS piece is worth noticing too. Letting developers build OpenAI agents that run fully on Bedrock is a practical move. Plenty of enterprises have standardized on AWS, and this removes one reason for them to choose a different model provider.
⚠️ What we don’t know yet
The announcement leaves some gaps:
- Pricing and plans: TechCrunch AI’s report doesn’t say which ChatGPT or API tiers get the cloud environments or Codex Security Cloud, or whether they cost extra.
- Rollout timing: OpenAI describes the security tools as “rolling out.” That usually means you may not get them right away.
- Fix quality: Automated security fixes are only worth something if they’re accurate. Teams will still want a human to review anything Codex suggests before it gets merged.
🔭 What comes next
OpenAI’s direction is clear. Codex is becoming an always-on engineering teammate that isn’t tied to one machine or one session. The next real test is adoption: will teams trust persistent agents with security fixes and first-pass reviews on production code? Full details on each announcement are in the original TechCrunch AI coverage.