Not since Magic Leap has a “next-generation” hardware company been so hyped while showing so little. Everyone in the tech world has been freaking out about this new pocket protector thing that wants to “replace your smartphone.” It’s called the “Humane AI Pin.” As far as we can tell, it’s a $700 screenless voice assistant box and, like all smartphone-ish devices released in the last 10 years, it has some AI in it. It’s as if Google Glass had a baby with a pager from the 1990s.
It’s a voice assistant box, so that means it has a microphone and speaker. There’s no hot word, and it’s not always listening, so you’ll be pressing a button to speak to it, and you’ll get a response back. There’s also a camera, and because you’re expected to mount this on your clothing at chest level via a magnetic back piece, you’ll be creepily pointing a camera at everyone the whole time you’re using it. It claims to be “screenless,” but it has a pretty cool 720p laser projection system that seems to function as a fine monochrome screen that projects a smartwatch-like UI onto your hand. It shows some super basic UI elements, like a circular media player or a scrolling wall of text. A few hand gestures, like tapping your fingers together, will let you interact with it.
Despite claiming to be able to replace a smartphone, the Humane AI Pin is going back to the Dark Ages and not supporting any apps. We’ve seen so many devices live and die by their app ecosystems, and the matter-of-fact quote from the presentation was, “We don’t do apps.” You’ll be locked into whatever features and services Humane has built into the Android-based “Cosmos” OS. So if you want to play music, it needs to be from Tidal, a service with 0–2 percent market share, because that’s who the Humane people have partnered with. It’s unclear if there is any other third-party functionality other than that. Humane’s “Cosmos” page shows logos for Slack and then logos from Microsoft and Google, which could mean anything.
Not having a screen, or at least not prioritizing the laser projector screen, means you’ll be doing a lot of work to understand what the pin is trying to tell you. There are two different lights on the device—a front one and a top one—that each blinks five or six different colors that all communicate some kind of state, so that’s 11 color/location combinations to keep track of. Without a touchscreen, input is also an esoteric affair, with seven tap or swipe gestures you can perform on the front of the pin for things like answering a phone call and changing media tracks. Rather than just seeing and tapping things on a screen, the interaction guide reads like you’ll be learning a second language. As much as it looks like a pager, a one-line text output on top of the device would have gone a long way for status communication.
As for the hardware specs, this is an aluminum and glass box that runs some kind of eight-core, 2.1 GHz Qualcomm processor. It comes with 4GB of RAM and 32GB of storage. You keep it on your shirt with a magnetic clip that goes on the inside of your shirt, and while this back part is a low-profile magnet, there’s also a “battery booster” back that is fatter and will wirelessly transfer electricity through your shirt. With no power-hungry screen, this is probably very light on battery usage. It comes with a battery booster and an exact copy of the AirPods charging case, which will store and charge the main unit.
I suspect the processor model is not listed because it’s a cheap one. It’s hard to know exactly how much processing the AI pin is doing. The spec sheet mentions “accelerated on-device AI,” but the presentation says AI responses are “streamed,” so presumably, it’s not doing much voice processing. They also save a ton of processing power by not needing to keep up with a high-fidelity display. Considering that the cheapest Apple Watch is surely faster than this, must pay for an expensive display, and still comes in at $300, it’s hard to see where the $700 price tag comes from.
Yet another thing that does voice commands
I guess we’ll start with what Humane prioritizes: voice commands. You can press the front touch button and start talking to get a response. It’s the same form factor as a Star Trek communicator.
Humane’s website recaps some of the functionality in a hushed, awestruck video that seemingly is unaware of the Google Assistant, Siri, Alexa, Cortana, Bixby, Nuance, and every other voice thing that has existed for years. Some of the response times shown in the official video are not great, with the pin taking several seconds to come back with a reply. At one point the presenters ask, “When is the next eclipse, and where is the best place to see it?” The presenters, while waiting for a response, then have enough time to explain what’s going to happen and say, “This is an AI browsing the web, grabbing knowledge from all over the Internet.” Then the response finally comes back. The Google Assistant answers the same question in less time, with a helpful map from nasa.gov showing visible eclipse locations. The video seems to always have cuts between the questions and responses, so it’s unclear how much of this is happening in real life. It also sounds like voice responses only work while you’re online.
“Online” means this thing has an always-on T-Mobile connection that costs $24 a month. That subscription is “required” by the way, so even though this can also connect to Wi-Fi, you’ll still need to pay the monthly fee. Presumably some of the money goes to Humane to pay for voice and AI compute costs. As miserable as that sounds, note that the voice assistants that aren’t charging a monthly fee are all huge money losers. When you’re not a trillion-dollar tech giant, this is the financial reality of voice and AI processing.
While basic questions like that are not very impressive, OpenAI is a partner on the device, so some of the demos were of some pretty advanced voice functionality. One command, “Catch me up,” would summarize unread messages from your email and instant messaging, just like a secretary. On one hand, that sounds like a neat feature if you’re so incredibly busy you can’t read your incoming messages from friends and co-workers, but on the other hand, it would be frustrating to have a boss who doesn’t actually read the emails you send them. Thankfully, you can also have the messages appear on your hand via the projector.
You can also have the AI search through your messages, which sounds incredible. “What’s the gate code that Andrew sent me?” results in it reading back a number. So, the AI has access to all your messages and can sift through them with a fuzzy voice search to find what you want. Again, though, this doesn’t support apps, so you’re hoping Andrew’s message came through whatever chat service this supports. As far as we can tell, that means SMS, Slack, and maybe Gmail and Outlook. If it was via any of the world’s most popular instant messaging platforms, like iMessage, WhatsApp, or Facebook Messenger, you’re out of luck.
You can dictate messages, and after hearing the robot voice read your text back to you for confirmation, you can even have ChatGPT (presumably) rewrite what you’ve spoken with a command, like, “Make me sound more excited.” There are also the usual basics like phone calls and playing music, and, like most smartphones, it can also do a back-and-forth voice-to-text translation.
It has a laser screen! pew pew
While this is the 10 millionth thing in the world that does voice commands, the laser display system sounds totally rad. This is a 720p, 25 mW laser beam scanning projection system that pumps out graphics in a monochrome teal color. It’s a class 2 laser, so just like a barcode scanner, it’s safe for accidental exposure but can damage your eyes if you stare at it.
Stick out your palm in front of the device, and the laser will kick on, showing a minimal user interface. The UI this puts out could not be closer to a smartwatch: a small display focused on a single task, with big buttons, three or four lines of text, and minimal controls usually laid out in a top-bottom-left-right configuration. If we just say “exactly a smartwatch UI,” you’ve understood everything there is to know about the UI. The media player has a circular progress bar, with previous and next to the left and right; pause at the bottom; and three lines of text for the artist, title, and timestamp. The message view is a scrolling-heavy affair: Each line of text is about three or four words, and you get five lines of text in a rectangular box.
You can interact with the UI via hand gestures detected by the camera. You can’t tap on your palm with your other hand, so picking something in one of the four cardinal directions involves a tilt gesture. Have you ever played Super Monkey Ball? So, imagine there’s a marble on your palm, and tilting your hand will make the ball roll off in that direction—that’s how you select a button. Tapping your thumb and index finger together will trigger the function.
Closing your hand will bring up the “Home Screen,” which again is extremely smartwatch-like. The main screen is the time, one line of marquee text about incoming notifications, and a button at the top that says “nearby.” Tapping your fingers together will scroll through the bare-bones pages of the home screen. Its weather feature shows a very simple icon of the current weather (a sun, cloud, etc.) and two giant digits for the current temperature. The only other page was the date, which showed only “Sun 15 Oct.” Tilt your hand to the top button for “nearby,” which in the video showed “Theater district” and what looked like three cardinal direction buttons that were not legible.
On paper, the display specs and UI don’t match up. The 720p resolution of the projector is double that of most smartwatches, and your palm is bigger than a smartwatch, yet the UI here shows dramatically less data than you’d get on an Apple Watch. Presumably, Humane is concerned about the visibility limitations of the projection system, so everything is very big and simple. At many, many points in the official professionally shot video, the display is not legible despite the hand it’s projected on being perfectly in focus. For instance, I have no idea what the “nearby” screen does because it was shown on camera yet never in focus. I guess the laser display has a fixed focus and a small sweet spot.
Repeating the camera mistakes of Google Glass
The one thing that separates this from just being a smartwatch is the camera. You’ll be constantly pointing a 13 MP sensor at everyone and everything, and after remembering the intense hatred Google Glass got for doing that, the inclusion of a camera is a big negative in a lot of people’s eyes. Humane seems to slightly understand this with a “Trust light” that will light up if it’s recording, but the people you’ll be making uncomfortable with this don’t know about the light, and the light next to a webcam camera still doesn’t stop people from putting tape over it.
Just like Google Glass, you’ll get to use the camera like a fancy Go-Pro and do hands-free recording of whatever you’re doing. A double tap with two fingers will also take a picture.
The camera can recognize objects, so you can hold things out in front of your chest and give a voice command. Some of the examples seemed implausible, like one back-and-forth where the user holds up a book: “How much is this online?” robot voice: “This is $28 online.” “Great, buy it.” And that was the end of the demo. The real world is far more complicated than that. Buy it from whom? From where? When will it get here? Is this a real product or a scam? Is it new or used? Paper or hardback? The robot voice didn’t even confirm that it recognized this as a book, nor did it read a title. The demo had such little confirmation of what was happening that you could be purchasing any random rectangular-shaped object for $28. Between this, filtering notifications, and offering information recaps, the general “Let AI take the wheel” simplification of everything is something you’ll have to be comfortable with to use the AI Pin.
Another camera and voice demo was nutrition tracking, where you can hold up some food, ask about the nutrition facts, and then awkwardly announce to everyone in earshot, “OK, I’m going to eat it.” Your pager will then keep a running total of health goals for protein or calories.
Not having an app ecosystem still means all your data has to go somewhere, so there is a web portal called “.Center”—the first character in that name is a period—that will house what looks like photos, notes, history, and nutritional data from your voice assistant.
I’ve got to ask: Why wasn’t this just a smartwatch? Some of the OpenAI-powered responses are pretty neat, but there’s no reason not to have that just show up on a screen or be read aloud by a smartwatch. A standalone smartwatch would also be a nice middle ground between “offline” and being a constantly scrolling TikTok zombie—plus, it wouldn’t come with all the ultra-creepy camera problems. Everything in the smartphone world has claimed to be AI-powered for years, but no one has really rethought things now that the ChatGPT-style large language models are so much more capable. Starting with the lack of a real screen and no apps just seems extremely limiting.
Even if you find a device like this interesting, not having an app store feels like a death sentence. Right now, it’s completely unclear what services the Humane AI pin can interact with, and it feels like that list will only be about five items long. So far, the market has proven that basically no one wants to switch core services for some random piece of hardware. Orders for your chest-mounted voice assistant start November 16.
