Meta's AI Pendant Leak Proves Nobody Knows Where Wearable AI Belongs

This is a product category that does not know what problem it is solving.

On May 29, Reuters reported, citing internal documents obtained by The Information, that Meta plans to begin testing an AI pendant within the next year. The device, developed by the recently acquired startup Limitless, is designed to record and summarize conversations throughout the day. It sits on your chest. It listens to everything. It is, according to Hypebeast, which reviewed the leaked roadmap, slated for internal testing in spring 2027. This is not a product announcement. It is a product leak. And the difference matters, because leaked roadmaps reveal what a company wants to build. They do not reveal whether anyone should buy it.

It is worth revisiting what happened to Humane, because every AI hardware startup and every trillion-dollar platform company now sketching pendants and pins on a whiteboard is building in the crater Humane left behind. The Ai Pin launched in April 2024 at $699 with a $24 monthly subscription. It had a camera, a microphone, a laser projector that painted text onto your palm, and a suite of AI-powered features that promised to replace your smartphone. It was reviewed poorly. It sold worse. By February 2025, less than a year after launch, the company had pulled the product from sale entirely. Liliputing reported in May 2026 that the remaining units had been reduced to hobbyist hack projects, repurposed as cheap Android devices by tinkerers who bought them at fire-sale prices. HP acquired the intellectual property and folded pieces of the technology into HP IQ, a workplace collaboration platform that has nothing to do with replacing your phone and everything to do with会议室.

Humane's failure was not a failure of execution. It was a failure of premise. The Ai Pin assumed that people wanted a second device that was less capable than the one already in their pocket, that they would pay a subscription for the privilege, and that a laser projector on the palm was a reasonable interface for anything other than a tech demo. None of these assumptions survived contact with a human being who had already paid off a smartphone. The lesson of the Ai Pin is not that AI hardware cannot work. It is that AI hardware works only when it does something the phone cannot do, or does it in a place the phone cannot go. Glasses clear that bar. Pendants and pins, so far, do not.

The Glasses That Work and the Pendant That Might Not

Meta's smart glasses are the one bright spot in this entire category, and the numbers are beginning to show it. The Ray-Ban Meta glasses, now in their second generation, have sold in the millions. On June 23, 2026, Wired reported that Meta launched its first self-branded smart glasses, dropping the Ray-Ban co-branding and cutting the entry price to $299. The hardware is essentially identical to the Ray-Ban models: a 12-megapixel camera, open-ear speakers, a five-microphone array, and Meta AI accessible via voice. What changed was the price, the branding, and the distribution strategy. Meta is now selling these through its own channels, in three frame styles, one of which was co-designed with Kylie Jenner.

The glasses work because they do not ask the wearer to adopt a new behavior. They are sunglasses that happen to take photos and answer questions. The camera is good enough for social media. The speakers are good enough for podcasts. The AI is good enough for basic queries and object recognition. Nobody buys them to replace a phone. Nobody buys them expecting a productivity revolution. They are an accessory. That modesty is what makes them the only AI wearable that has shipped in volume without becoming a punchline. Demand is strong enough that Meta now has a credible hardware roadmap stretching into 2027, with multiple new models including "supersensing" glasses that add health-tracking sensors, according to the leaked documents reviewed by Hypebeast.

The pendant is not modest. The pendant, as described in the leaks, is designed to record and summarize everything you say and everything said to you throughout the day. It is an always-on audio recorder hanging on your chest, feeding into Meta's AI infrastructure, producing meeting notes, conversation summaries, and memory aids. This is an extraordinarily ambitious product definition. It is also a privacy grenade with the pin already pulled. The Limitless acquisition, which brought the pendant technology into Meta, was a bet that ambient recording can be normalized the way camera glasses were normalized. But the social contract around an always-on chest microphone is fundamentally different from the social contract around glasses that blink a little LED when recording. One has a visible signal. The other has your boss asking whether you are recording this conversation.

Who Is This For, Exactly

Every AI wearable needs to answer one question before it answers anything else: what person, on what day, in what situation, reaches for this device instead of the phone in their pocket. For the glasses, the answer is clear enough. Someone walking their dog. Someone at a concert who wants a hands-free photo. Someone cooking and needing a timer set without washing flour off their hands. The glasses slot into gaps the phone leaves open. The pendant does not have gaps. The pendant competes directly with the Voice Memos app, with the Notes app, with the phone in a shirt pocket running Otter.ai. It asks the user to charge a second device, wear a second device, and trust a second device with every word spoken in its vicinity. The utility must be enormous to justify the friction. Nothing in the leaked documents suggests Meta has found that utility.

The competitive landscape makes this even harder to read as anything but a supply-side push. Apple is reportedly developing its own AI wearable pin with cameras and microphones, according to a TechRepublic report from January 2026. Microsoft showed Project Solara concept devices at Build 2026, including a wearable AI badge for office workers using AI agents, TechRepublic reported on June 3. Samsung is developing smart glasses to compete with Meta's Ray-Bans. Snap continues to iterate on Spectacles. Google has partnered with Warby Parker on AI-powered eyewear. Every major platform company now has a wearable AI project. Almost none of them have articulated why a normal person should care.

This is not the usual problem of early-adopter hardware being rough around the edges. This is the problem of a product category being built around a capability, ambient AI, that has not yet been matched to a daily behavior that billions of people already perform. Smartphones succeeded because they absorbed behaviors people already had: calling, texting, taking photos, checking email, browsing the web. Smartwatches succeeded because they absorbed notifications and health tracking, behaviors that were already happening on phones but benefited from wrist-level access. AI pendants are trying to invent a behavior, always-on conversational memory, that has no existing analogue in consumer life. That is harder than any chip-design problem or battery-life constraint.

Smartphones are the go-to devices for personal computing today. However, that could soon change with AI and wearable devices such as smart and augmented reality glasses., Francis Sideco, Forbes contributor, June 3, 2026

The Forbes piece that quote comes from, published June 3, makes the optimistic case: AI is the catalyst that will finally push wearables into the mainstream because it gives them something to do that feels magical rather than gimmicky. The argument has merit. The Ray-Ban Meta glasses without AI would be a novelty camera on your face. With AI, they are a camera that can identify plants, translate signs, and answer questions about what you are looking at. The AI transforms the hardware from a capture device into an interaction device. But the pendant does not have a camera. It has only a microphone. And the AI use cases for an audio-only device that is not in your ear are remarkably thin.

Meta clearly knows this, because the same leaked documents describe a "Wearables for Work" subscription service aimed at enterprises. That is the pivot that makes sense. An AI pendant that records and summarizes meetings is a productivity tool for knowledge workers. It is Otter.ai in hardware form, sold to companies with compliance departments that can evaluate the privacy implications. As a consumer product, it is a device that records your dinner conversations and stores them on Meta's servers. Those are two different products with two different adoption curves, and the enterprise version is the only one that has a clear buyer.

The form factor itself may be a dead end, regardless of the AI inside it. The history of body-worn technology is littered with devices that were technically impressive and socially radioactive. Google Glass failed not because it did not work but because people did not want to be in a room with someone wearing it. The Humane AI Pin failed because it asked too much and delivered too little. Bluetooth headsets succeeded because they solved a specific, narrow problem and did it invisibly. The pendant sits somewhere between the headset and the pin. It is visible but not obtrusive. It solves a problem but one that most people do not currently believe they have.

Where does the pendant end up in eighteen months? The most likely outcome, given the track record of this category, is that it ships in limited quantities to developers and enterprise customers, generates a wave of privacy criticism, and settles into a niche. The second most likely outcome is that it never ships at all, absorbed into the glasses roadmap as an audio module or a companion accessory. The least likely outcome is that it becomes a mainstream hit. That would require Meta to solve a problem no other company has solved: convincing hundreds of millions of people that the convenience of an AI memory is worth the cost of wearing a microphone on their chest, charging it every night, and trusting Meta with the audio of their lives. Nothing in the leaked roadmap, and nothing in the public track record of this category, suggests that problem is close to being solved.

Glasses are the real story. The $299 price point Wired reported on June 23 is the clearest signal yet that Meta believes it can sell AI wearables at scale, not as developer kits or luxury experiments but as consumer electronics. At $299, the glasses cost less than AirPods Pro. They cost less than an Apple Watch Series 10. They are priced to move, and the evidence suggests they are moving. The pendant, whenever it arrives, will have to justify a price somewhere in the same neighborhood while delivering value that is harder to demonstrate in a thirty-second ad. Cameras on your face are easy to show. Summaries of conversations you already had are not.

The AI hardware category is not a fraud. The glasses prove that. But it is a category that rewards specific, observable utility and punishes grand claims about replacing the smartphone. Every company now sketching a pendant or a pin on a roadmap is betting that ambient AI is the next camera: a sensor that starts out creepy and becomes indispensable. The bet may eventually pay. But the Humane AI Pin is still warm in its grave, and nobody has yet shown that a microphone on a lapel is worth more than the phone already in your hand.