What the heck is “Pet,” and why is OpenAI turning Codex into Windows by introducing junk features likey clippy with zero practical value?
Why are highly paid engineers being asked to spend time on features like this when Codex still has so many serious problems? OpenAI should focus its resources on improving performance, preventing intermittent crashes, and fixing the freezing that occurs when users start typing. I have noticed significant performance problems since the latest update.
What problem is Pet supposed to solve? It appears to do absolutely nothing useful. Honestly, it feels as though Codex is being managed by someone with MBA who prioritizes engagement gimmicks but has no understanding of how developers actually use the product.
Please focus on stability, performance, and the core developer experience before introducing pointless features like this.
I personally like the pet feature and feel like it adds some creativity to my day and is actually useful. It’s a nice way to keep track of active and recent status updates on tasks while I am working on something else. It’s okay to have a little fun while you work.
The actual plushy is irrelevant. It’s a creative, smart way to keep track of your ongoing tasks without constantly switching attention to a full separate screen. That is, if you work on multiple spaces. Personally, I have like 8 virtual screens and 2 physical ones, minimum, at all times. So I find the pet pretty great, although I am relatively immune to bloatware. AI is fuelling feature overload, lol
I told Codex to edit/create a local version of the “Hatch Pet” skill that supports higher frame rate, starting new sessions, replying to plan and subagent spawning. In my case, I made different characters so I remember which pet does what (review, plan, code, read, etc.)
Why are highly paid engineers being asked to spend time on features like this when Codex still has so many serious problems?
Stumbled across this and thought I’d give some clarity. The Codex pet feature was a week 1 ship from a new joiner, I saw the feature and created the initial /hatch-pet skill (originally called codex-avatar-create) in my free time because I thought it would be cool to use imagegen to do so. Didn’t expect it to ship, and didn’t expect so many people to love it - very happy with the outcome though!
I really like the new Pet feature, especially because ChatGPT creates multiple animations/poses for the Pet — such as waving, blinking, jumping, running, and resting.
However, on Android, the Pet currently seems to appear mainly while ChatGPT is Thinking. I would love to be able to interact with my Pet more directly during a conversation.
For example:
If I tell ChatGPT, “I’m really tired today,” my Pet could curl up or appear sleepy.
If I say, “I’m going to exercise,” my Pet could jump or run.
If I say, “I finished my project!”, my Pet could wave or jump excitedly.
If I ask, “Can my Pet say hello?”, it could wave.
If I say, “Show me my Pet,” the Pet could appear on demand.
This seems especially appropriate because ChatGPT has already created multiple visual states for the Pet. It would be wonderful if the AI could select those existing animations based on the context of the conversation.
I understand that some Pet animations may currently be tied to system states such as Thinking, waiting, or completing a task. I’m suggesting an additional capability: allow the conversational AI to invoke the existing Pet states when appropriate, or allow the user to explicitly request a particular animation.
For Android users in particular, this would make the Pet feel much more like a companion and much less like an indicator that only appears while ChatGPT is processing a response.
Even a simple command system would be great, such as:
“Show my Pet.” “Make my Pet wave.” “Make my Pet jump.” “Make my Pet run.” “Make my Pet sleep.”
The animation assets already seem to exist. What appears to be missing is a way for the conversation/user to trigger them.
I’d love to see this functionality added to the Android ChatGPT app.
