Elyse Betters Picaro / ZDNET Follow ZDNET: Add us as a preferred source on Google. ZDNET's key takeaways OpenAI makes Codex available to all developers and teams. Slack integration and SDK bring AI into coding workflows. Admin tools add visibility, control, safer automation for enterprises. OpenAI announced yesterday that Codex, its programming agent, is now "generally available." I'll deconstruct the marketing speak of "generally available" in a minute, because it's a bit of a story. But before that, let's discuss some new capabilities OpenAI introduced in Codex. They're cool. (Disclosure: Ziff Davis, ZDNET's parent company, filed an April 2025 lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.) Slack integration I have often said that coding with the help of a chatbot feels a lot like talking to a colleague in Slack. There are a number of colleagues I've worked with for years. But not only haven't we ever met, we've also never spoken using any medium other than Slack. Also: Everything OpenAI announced at DevDay 2025: Agent Kit, Apps SDK, ChatGPT, and more Talking to a chatbot over Slack is a bit of a Turing test moment. I have some engineering buddies. I've known them since way before generative AI, so I know they're actually human. But from their patterns of conversation, it wouldn't be hard to imagine that they could be AIs. So it's not that big a jump to feel as if talking to an AI is a lot like talking to a colleague, especially when doing it over Slack. Once you integrate Slack with Codex, all you have to do is ear burn Codex with an @codex mention, and you'll get results back from the machine. Also: ChatGPT wants to act more like an OS - as it transforms into an app platform Stay tuned, because I'll be setting this up in a forthcoming article and will show you all about how that works. Codex SDK The next big capability announced by OpenAI is the Codex SDK (software development kit). You can now call Codex from your code. Additionally, OpenAI announced that they've released a new GitHub Action (smart code that runs on GitHub) that makes it easy to integrate Codex into your CI/CD (continuous integration and continuous deployment) pipeline. Also: OpenAI has new agentic coding partner for you now: GPT-5-Codex Wrap your head around that for a minute. You can write code that sends a message to an AI asking it to write code, which you can then incorporate into your code. I know. Recursion: see recursion. Right? In any case, this is big because it allows Codex to be part of automated engineering workflows. I mean, you'd have to trust Codex enough to let it run autonomously as part of another workflow. But still, I can picture having it integrated with, say, a bug-reporting system. When a bug is reported, I can envision the ticket management system automatically firing off Codex to see if it can fix the bug, all without needing coder time. Also: 10 ChatGPT Codex secrets I only learned after 60 hours of pair programming with it Just, you know, be careful. Codex is astonishingly powerful, but it also is astonishingly boneheaded. In my experience, it does some really dumb stuff if it doesn't have firm human guidance. I'm concerned that the Codex SDK may be seen as a solution that obviates the need for human oversight. Don't make that mistake. New management features This came as something of a surprise to me. According to OpenAI, ChatGPT admins can edit or delete Codex cloud environments. When Codex works on your repo, it creates an isolated container to operate in while working. Also: ChatGPT can buy stuff for you now - forever changing online shopping I always thought (or assumed) that Codex deleted those containers when the session ended. But it seems that Codex keeps them around so you can inspect logs and diffs later. Of course, those containers might contain sensitive information. Well, now ChatGPT admins can remove those containers proactively. Admins can also "enforce safer defaults for local usage with the Codex CLI and IDE extension, like defining overrides through managed configuration or monitor actions taken by Codex." Also: Don't use ChatGPT voice mode if you want accuracy - here's why The company has also introduced new analytics that report on usage across interaction paradigms, as well as track the quality of code reviews. These admin features are available to those on Business, Edu, and Enterprise plans. Pricing changes I've written before about Codex resource allocation and how there is considerably more usage allotted for users of Pro plans than for users of Plus plans. That factor remains, but as of Oct. 20, OpenAI will start counting your usage of Codex cloud tasks as part of your resource allocation. For more information on Codex plans and their differences, check out this page. Also: Is ChatGPT Plus still worth $20 when the free version offers so much - including GPT-5? 'Generally available' So let's talk about the big "generally available" headline on OpenAI's announcement. The fairly obvious question is, "How could something that grew in usage by 10 times between August and September not already be considered 'generally available'?" I actually reached out to OpenAI about this very question. I've been paying $200 a month for something. It works really well. Obviously, it is available. So what do they mean when they announce it's only now becoming "generally available"? Also: ChatGPT is crushing rivals in the AI chatbot race by all measures - but for how long? A corporate spokesperson got back to me, saying: "You're correct. Codex has been in research preview, available to Plus and Pro users. Codex has already seen a lot of traction with individual developers, but for GA, OpenAI wanted to be sure Codex is available everywhere developers work (in the editor, terminal, Slack, or their own tools) and also more helpful for engineering teams. The Slack integration, Codex SDK, and admin tools were important for ensuring that." Basically, OpenAI graduated Codex from a paid preview to a full-on product. While there is clearly some marketing hair-splitting, my takeaway is one key message: the game is afoot. And, based on my intense deep dive into Codex, I can personally confirm that it's got game. Also: ChatGPT just saved me 25% off my dinner tonight - here's how Have you tried using Codex yet, either in your editor, terminal, or over Slack? Do you see yourself integrating it into your development workflow or even letting it run autonomously through the new SDK? What do you think of OpenAI's approach to making Codex "generally available"? Does this feel like a true release or just a rebranding of the preview? How much autonomy would you personally trust an AI coding assistant with in your own projects? Let us know in the comments below. You can follow my day-to-day project updates on social media. Be sure to subscribe to my weekly update newsletter, and follow me on Twitter/X at @DavidGewirtz, on Facebook at Facebook.com/DavidGewirtz, on Instagram at Instagram.com/DavidGewirtz, on Bluesky at @DavidGewirtz.com, and on YouTube at YouTube.com/DavidGewirtzTV. Get the morning's top stories in your inbox each day with our Tech Today newsletter.