I’ve always insisted my AI assistants pause before making critical decisions, a safeguard against unexpected outcomes. This cautious approach, however, often leaves them in a state of digital waiting, effectively stalled until I provide the go-ahead. The problem is, life happens – I’m often juggling multiple tasks and easily miss these silent pauses.
When actively watching the AI work, it’s easy to spot the hesitation. But when focused elsewhere, those idle moments can stretch unnoticed, the AI essentially frozen, awaiting direction. It’s a frustrating disconnect, a sense of untapped potential sitting dormant on my desktop.
OpenAI’s new “Codex pets” offer a surprisingly elegant solution. These aren’t just whimsical additions; they’re subtle, animated indicators that live on your desktop, constantly communicating the status of your AI agents. They provide a constant, visual awareness of what’s happening behind the scenes.
The real power lies in their ability to signal when your attention is needed. A small red clock appears in the pet’s thought bubble when an agent is awaiting approval, a gentle but insistent reminder. Conversely, a green checkmark signifies task completion and readiness for new instructions.
These little companions remain visible above all other windows, and can be positioned anywhere on the screen. I’ve found a spot for mine in the bottom-right corner, a constant presence without being intrusive.
Beyond their functionality, the pets are undeniably charming. You can choose from eight distinct designs, ranging from the classic blue “Codex” to a duck named “Dewey,” a rocky space companion, and even a playful nod to the infamous “BSOD.”
Summoning a pet is simple – just type “/pet” into Codex. You can then customize your companion through the Appearance menu. For those feeling creative, a skill called “hatch pet” allows you to design a unique pet inspired by your own projects.
While not a revolutionary advancement, the Codex pets are remarkably effective. They bridge the gap between user and AI, fostering a sense of collaboration and control. I now feel consistently informed about my agents’ activities, and it’s sparked a desire for similar features in other AI desktop applications.
