I'm skeptical of any hardware play here - As Marques Brownlee has said a lot when trying new ai hardware, "Smartphones are kinda OP."
Phones have gotten really good through incremental improvement and keeping some backwards compatibility so they were always useful at every point in their evolution. It's hard to imagine another device for ai that isn't just better served as an app.
I'm (very) skeptical too, though if you were going to use this device (app) heavily, and it wanted to be always on, and power hungry, it needs more permissions than "an app" might allow, and it needs more battery too. Most likely it would still connect with the phone, but it would be like external GPS is for digital cameras - does one-ish function and does not drain the battery of the main device.
So maybe what they'll really get everyone to do is walk around with a second battery that bypasses some permissions, in terms of always listening or always using a camera. "Just an app" can never do that.
Funny enough people stopped buying external GPS for digital cameras because... your phone can do that now, as an app
I had to ask chatgpt what OP meant there. It had overpowered as in "gaming culture, where an overpowered (OP) character, weapon, or ability is unusually strong"
I'm not sure they are overpowered for real life uses like filming video or playing games - people still pay like $1000 to get a good one because the cheap ones don't perform as well. Which is a problem for Humana pin like devices because the cameras etc will probably be rubbish compared to the the ones people have on their phones.
Marques meant that, they are very good at the broad category of a daily carry consumer device [0]. If you are competing against them, they are heavily favored to win.
If you want to make a consumer device, as all these ai hardware companies like humane professed, you are competing against smartphones which have kinda evolved like a desire path, smooshing together all the previous personal electronics in a general form factor.
[0] https://www.youtube.com/watch?v=TitZV6k8zfA if it’s helpful context, he has similar conclusions in his rabbit r1 review. I am quoting him because I think he’s saying it more clearly than the same vibe I get, that these devices are borne out of a company objective to own a platform and not a user need that necessitated a new platform.
I suspect this is the only way they can differentiate it from a phone, and it will be pointless.
Voice-to-text is already as good as it needs to be now, and most people barely use it because unless you are driving, a keyboard is better. I don't want others to hear what I'm searching for - or for the device to be always on, listening. A keyboard of some sort seems unavoidable.
As soon as you put a keyboard on it, then it needs a screen. As soon as you have those two things you pretty much have a phone - and why would anyone want a second device that's basically just another phone that could have just as easily been an app on their existing phone.
It'll need to android based if they want any one to use it as otherwise they'll trying to start from scratch, and why would anyone use something that doesn't have access to the 27 million existing app on the android app store. (see windows mobile)
With an always on mic that can pick up a wake word, that's aesthetic and discreet, you can have a Jarvis-like access to AI. The problem is I can't see how they would sell everyone on carrying around another brick along with our phone. We may be looking at a suite of devices that are more like clothing, such as a lanier mic that you clip on, a lapel pin, something that wraps around and sits behind your collar like a clip-on tie, small buttons that act as cameras, and finally a backback (Jetpack!?) that acts as the "brick".
If we want to get totally out there, then just start selling clothes that incorporate all of that. It can be the early form of the cybernetic suit. We haven't even explored why we can't just have computers in the thick part of our sneakers, which would be completely unobtrusive compared to what we've got going on now days.
If it's just a "brick", then why not just be a phone? I'd expect the brick to do some amazing things like know where I am in the room, recognize hand gestures, voice, and so on, but again, we're almost back to a phone.
---
But the answer might quite simple. We haven't had a brick that has 30 days+ of battery power and always online. Such a brick doesn't have to do anything but pass along API calls to OpenAI servers and get back API results. Business-wise, it makes the most sense. If a brick like that can be a discreet part of my clothing, that would be perfect (thinking belt buckle like a Power Ranger).
Phones have gotten really good through incremental improvement and keeping some backwards compatibility so they were always useful at every point in their evolution. It's hard to imagine another device for ai that isn't just better served as an app.