OpenAI's Jony Ive-designed AI device targets home use

After more than a year of rumors about what legendary former Apple designer Jony Ive and OpenAI might be building together, we finally have a reasonably clear idea.Apparently, it’s a metal donut.It won’t technically be OpenAI’s first piece of hardware. That honor goes to the recently released Codex Micro, a 13-key programmable mini-keyboard built for wrangling Codex agents. The puck, though, looks like the company’s first serious attempt to put ChatGPT itself into a dedicated household device. OpenAI's consumer device is reportedly a small, screenless, battery-powered smart speaker, roughly hockey-puck sized, with a hole through the middle. Current reports put the price somewhere around US$300 to $400, with a launch expected in 2027.That's a fair bit of cash for something occupying territory currently dominated by little plastic speakers you can often buy for the price of dinner. The interesting part is what OpenAI appears to be putting inside it.Along with the obvious microphones and speakers, reports point to cameras, environmental sensors, status lighting and even mechanically moving parts that respond while you’re interacting with it. The idea seems to be that instead of sitting there like a dead lump until summoned, it gives you some physical indication of what it’s doing, listening, thinking, responding.It’s also reportedly portable, with its own battery and a separate charging base, so this isn’t necessarily another smart speaker destined to spend its entire life beside the toaster.Amazon, Google and Apple solved the “microphone in a box” problem years ago. What OpenAI brings is the prospect of putting something much closer to the current ChatGPT voice experience permanently into the room. That’s a very different proposition from the current voice assistants.I’ve spent the last few months accidentally building my own crude version of exactly this.Yaffle, my home-grown AI assistant, can talk naturally, control Home Assistant, operate devices, set timers and generally understand what I’m asking without me having to memorize a catalog of approved phrases. What that experiment has taught me is that the speaker itself is almost irrelevant. The hard bit is context, and we've just found out what ChatGPT's conversational engine is really capable of with the release of its Voice mode.Which room am I in? What was I talking about 30 seconds ago? When I say “turn that off,” what does “that” mean? Can it actually do something useful once it understands the request? And perhaps most importantly, does it know when a two-word answer is enough?That’s where OpenAI’s little metal donut starts getting interesting.Cameras and environmental sensors potentially give it information a normal voice assistant simply doesn’t have. A battery means it can move around with you rather than existing as a collection of unrelated speakers scattered through the house. And behind it sits ChatGPT, rather than one of the rigid voice assistants that taught us all to speak to computers like slightly confused robots.There are still plenty of details to come. We don’t yet know how much processing will happen locally, exactly how the cameras and sensors will be used, or how deeply OpenAI plans to integrate the device with existing smart-home platforms.What we do know is that OpenAI appears to be taking a very different run at the smart speaker. Rather than building another cheap voice terminal for weather forecasts and kitchen timers, it’s wrapping ChatGPT in dedicated hardware designed to see, hear and interact with the room around it.At a reported US$300–400, it won’t be competing with an Echo Dot on price, and it isn’t expected until 2027. But if the reports are accurate, this will be our first proper look at what OpenAI thinks an AI designed to live permanently in the home should actually look like.Apparently, it looks like a donut.And sometime next year, Yaffle may have a new housemate.