If you buy something using links in our stories, we may earn a commission. This helps support our journalism. Learn more. Please also consider subscribing to WIRED
I spent an entire day of CES wearing a little yellow bracelet. To the unsuspecting nearby humans, it probably looked like a fitness tracker. But the whole time, this yellow Pioneer wearable from Bee AI recorded everything around me. It wasn't storing audio like a typical recorder app, but it processed my conversations, then gave me personalized to-do lists and readable summaries of my in-person chats.
A few days before the trade show, I spoke with the founder of another new company, Omi, which was officially unveiled for the first time today. Guess what it does? Record everything around you to create an activity log, and then have AI disseminate the information to give you actionable insights and tasks from your day, almost like a personal assistant. Omi's wearable can go around your neck, but it is best worn stuck to your forehead near your temple—it has an electroencephalogram inside, and Omi claims that if you think specifically about talking to the wearable, the device will understand and perk up to receive your request.
This is the new world we're in, with artificially intelligent wearables continuously recording the world around us. Voice assistants—which first landed in speakers and on our phones, but quickly moved to our wrists and face—at least required active engagement like a tap or a wake word to activate their ability to eavesdrop. But the next wave of hardware assistants, which also includes the forthcoming Friend pendant, can absorb information passively and work in the background. They are always listening.
The wearable hardware leading this space is often cheap—Bee AI's watch is just $50, and Omi's stick-on bead is $89—but the real magic is in the software, which often requires a subscription as it taps into multiple large language models to analyze your conversations.
Bee AI
Bee AI was founded by Maria de Lourdes Zollo and Ethan Sutin. Both previously worked at Squad (Sutin was the founder), which enabled media screen sharing in video chats so people could remotely watch the same movie or YouTube video together. The company was acquired by X (back when it was called Twitter), and the pair both joined briefly to work on Twitter Spaces. Zollo has previously worked at Tencent and Musical.ly, which subsequently became TikTok.
Sutin says he explored the idea of a personal AI assistant back in 2016 when chatbots were all the rage, but the technology wasn't there yet. That's not the case anymore. The company launched its Bee AI platform last February in beta, with an active community providing feedback. It only just began selling its Pioneer hardware a little more than a week ago. (The “Bee” name plays with the idea of ambient computing, as if something buzzing around and taking in information.) You don't need the company's hardware to use Bee AI—you can just interact with the AI via the iPhone app—but Zollo says the wearable offers a richer experience as it can continuously recording all day. An Android app is on the way at the end of the month.
The wearable is simple. It has two microphones for noise isolation, and Sutin says that if you can hear the person you're speaking with in a busy environment, the wearable should be able to hear both parties as well. It can be worn as a band on the wrist or clipped to your shirt. There's an “Action” button in the center; pressing it once mutes the mics, and pressing it again enables them again. You can press and hold the button, and this action is user-configureable, so that can trigger things like processing the current conversation or awakening the “Buzz” AI assistant to ask it a question. (There's no speaker on the wearable, so answers will be spoken out through your phone.) When the mic is muted, there's a red LED. When it's recording, you'd think the green LED would be lit up, but there's nothing to indicate that this wearable is picking up everything around you.
Zollo says having a green LED all the time would impact the purported 7-day battery life of the wearable, but this omission seems like it may make Bee AI fall into a gray area of recording laws, which vary from state to state in the US. While the wearable technically isn't storing audio, you can see a full transcript of conversations, even if it's not completely accurate sometimes. Sutin says everything captured is “treated as maximally sensitive,” and he ensures me the company's business model does not include plans to monetize the collected data, and that nothing will be shared with third parties. He even says no human can see this data.
The conversations are not processed locally on the phone; Sutin says the gap is closing for edge processing, but battery life still poses a fundamental problem. So for now, your data is processed in the cloud. Which large language models are deployed by Bee AI depend on the task you want to do. There's a mix of commercial and open source models, including OpenAI's ChatGPT and Google's Gemini, plus some the company hosts itself.
Sutin says Bee AI's target demographic is people “who talk a lot for a living.” If you're sitting at a desk all day not saying anything to anyone, there's not much for Bee AI's wearable to process unless you start asking it questions. But since it's recording all the time, it can recall things from conversations you have throughout the day. The accuracy of this is a bit hit or miss because Bee doesn't necessarily know who the people around you are, but it can discern different speaking voices and arrange the transcripts of the conversations to show different speakers. You can set a name for who the other speakers are. It can also save facts about you. In the same vein, you can ask it to forget things it may have picked up if you don't want them on the record.
In the app, you can see a summary of the conversations you've had throughout the day, and at the day's end, it generates a snippet of what the day was like and has the locations of where you had these chats on a map. But the most interesting feature is the middle tab, which is your “To-Dos.” These are automatically generated based on your conversations. I was speaking with my editor and we talked about taking a picture of a product, and lo and behold, Bee AI created a to-do for me to “Remember to take a picture for Mike.” (I must have said his name during the conversation.) You can check these off if you complete them.
It's worth pointing out that these to-do's are often not things I need to do. You'll probably end up deleting most of them, but when it does get these tasks accurately, it feels a little magical. Bee AI can be connected to your Gmail, Google Calendar, and Google Contacts accounts, so you can ask it to summarize an email or what's next on your calendar, but I didn't have access to try this functionality.
Rollo says Bee AI has a freemium service model, so with just the hardware, you get basic memory recall and summarization features. You'll have to pay $12 per month to access many of the other features, including the integrations with third-party apps, which the company wants to expand.
Omi
Nikita Shevchenko began his entrepreneurial ventures at 14 when he began mining crypto and by the age of 18, he sold his first company. His latest project is Omi, a wearable you either wear as a pendant or affix to the side of your forehead using the included medical tape. If you opt for the latter, and don't mind how ridiculous it looks, Shevchenko says if you think about talking specifically to Omi, it will understand and be ready to process your request.
I haven't tried it, but he claims Omi was trained to recognize the specific brainwaves when you focus on speaking to the wearable, so instead of having to say a hot word, you just think. That interaction is only something you use when you want to engage with the device, though. At every other moment, Omi is essentially a wearable microphone capturing the conversations you have throughout the day, just like Bee AI. With that capability, it can do many of the same things, like transcribe conversations, summarize them, add events to your calendar, and translate.
Everything is processed on the paired phone and in the cloud, so this is once again not a standalone piece of hardware like the Humane Ai Pin. Shevchenko says Omi is open source but is currently trained on ChatGPT. One area where Omi differs from Bee AI is with its marketplace for third-party ideas. There are “apps” built by the community, but think of these more like mods or skills that enhance Omi's integration with everyday apps. For example, there's a Google Drive “app” you can enable to have every conversation summary stored in a Drive folder at the end of the day. These apps can be published to Omi's store and developers can choose to make it free or paid. There are already dozens of apps because Shevchenko shipped 5,000 units of an early version of Omi last year to developers.
He says eventually, Omi will be able to let users create an AI clone of themselves—one that can talk to followers on their behalf and answer questions. Users can publish these clones for free or as a subscription, earning some side income. This is already somewhat in action through Omi's Personas platform, which lets anyone create an AI clone of a Twitter personality and chat with them.
Unlike the Bee AI wearable, Omi's wearable always has a light for implied consent that it's capturing and processing conversations around you. Its battery lasts for three days and, like the Bee, it can also leave you tasks to finish based on the context of your chats. Every night, it will send an action plan for the next day. It can even mentor you; imagine leaving a job interview and getting a summary afterward with suggestions on what to do better.
Shevchenko's long-term goal is to have the Omi “read the brain” so it can figure out what you're thinking about. Unlike Neuralink's brain-computer interface, which requires an implant in the brain, Shevchenko wants to accomplish this by continually adding more electrodes to the head. He was able to have the system construct two words with a more complex wearable, so there's still a long way to go.
The Omi is available today for $89 and ships in a few weeks.
HumanPods
If you find the utility afforded by an always-listening AI-powered wearable intriguing but you're wary of the possible privacy overstep, just know that not every new AI wearable is taking the “always-on” approach.
A company called Natura Umana—an offshoot from the team behind Swiss accessory company Rolling Square—has made a pair of wireless earbuds called HumanPods. They have microphones, but you have to double-tap an earbud to turn them on and trigger the onboard AI.
Like the Omi and Bee AI Pioneer, the HumanPods are designed to be worn all day, though the battery won't last more than a day. They're not in-ears, but hang pretty well on the ears and were comfy in my limited time testing them at CES. The system also employs multiple large language models, but rather than listening to everything around you, these earbuds are designed around you talking to the AI.
And there are multiple AI avatars you can chat with. Athena is the fitness and health AI persona, and the idea is that if you connect your fitness and health wearables and apps with it, you'll be able to ask questions like what exercises you should do today. Athena will look at your health data and suggest a workout based on what data it has available, like your sleep history and heart rate. Hector is another AI persona I spoke to; he's supposed to be the “AI therapist.” I talked to him about the stress of CES, and he suggested ways to make the event less stressful, like focusing on just a few companies to talk to. (Carlo Edoardo Ferraris, the company founder, tells me Hector has disclaimers that he is not a licensed therapist, of course.)
Just as we have different people in our lives we go to for different things, Ferraris says there will be an AI persona for your specific needs. He wants to establish a marketplace where people can publish personas—like an AI therapist developed by a mental health startup.
The earbuds will arrive in the first quarter of this year, and Android support will arrive either at the same time or in the second quarter. There's no firm price for the earbuds, though it may sit around the $100 mark. Of course, there will be a subscription.
All of these AI wearables are following the disastrous launch of the Humane Ai Pin, which was one of the most high-profile tech launches of 2024. Omi createor Shevchenko believes his company is already in a better place than Humane because it's launching with dozens of apps to enhance the AI assistant's capabilities. Ferraris thinks Natura Umana's wearable will be more successful because wireless earbuds are easy to understand and the app is designed like a messaging app—things people are already familiar with.
More often than not, wearables like these overpromise and underdeliver, and these always-listening AI-powered devices may not actually prove as helpful as their founders intend. Still, these vanguards in the wearable industry are pushing us toward an era when always-listening gadgets on our wrists and heads could grow to be more helpful. A microphone that's always on could quickly be seen as the new normal. The privacy concerns that come with such a push are bound to raise alarms, and what we don't yet know is whether those alarms will be loud enough to slow this steady march.