AI wearables expand from voice notes to companion devices
AI wearables are arriving in several forms, from pendants and wrist-worn recorders to rings and compact handhelds. Prices in the current group range from Bee’s $49.99 pendant to Sandbar’s $299 Stream ring, while products from Friend, Limitless, Omi, Plaud and Rabbit focus on tasks including transcription, reminders, summaries and conversational assistance.
The devices illustrate two broad approaches. Some continuously capture audio and use AI to turn conversations into searchable material or personal context. Others require a deliberate action before recording. Their companion apps, subscription plans and phone connections are as important to the experience as the hardware itself.
Voice capture is the central feature
Bee can be clipped to clothing or worn like a fitness band. It records what it hears, learns routines and preferences, and produces reminders and notes. Its iOS companion app costs $19 per month and provides questions, daily takeaways and chronological conversation transcripts. Bee also includes a mute button. Amazon acquired the startup in July 2025 and has introduced the $74.99 Bee Frontier and $99.99 Bee Eclipse, which retain the core functionality in wrist-or-clip and phone-mounted formats. Users can choose to connect Bee to Alexa.
Limitless, formerly Rewind, sells a $99 pendant that continuously listens and transcribes meetings, calls and conversations with consent. Its app includes 10 hours of AI features each month, while unlimited access costs $29 per month. Plaud’s $159 NotePin similarly records audio to a phone in real time and uses AI transcription and summarisation. It includes 300 monthly transcription minutes, rising to 1,200 minutes on the $8.33-per-month Pro plan. Plaud is also taking preorders for the $179 Note Pro.
Omi, priced at $89, can answer questions, create to-do lists, summarise conversations and help schedule meetings. It runs conversations through ChatGPT to retain personal context and can be worn as a necklace or attached to the side of a user’s head with medical tape. The breadth of designs shows that the category is still testing where and how people will accept an assistant that captures spoken context.
Companions and task-oriented hardware
Friend’s $129 pendant is positioned as an emotional-support companion. It connects to a phone through Bluetooth, recognises tone and mood, and can send proactive messages. The company’s Friend 2.0 update adds a built-in speaker and a new voice capability intended to give the device a more distinctive personality. The product has also drawn criticism, including backlash to its New York City subway advertising campaign.
Meta has announced the Muse Charm, its first AI wearable, though it has not yet shipped. The pocket-sized device can be carried or attached to a keychain and is expected to include a miniature display for Jolly, the digital avatar representing Meta’s Muse assistant. Users are expected to interact by voice and ask it to complete tasks. Nirva, founded by former Meta Reality Labs employees, is taking a personal-growth approach: its planned $249 pendant or bracelet turns speech into a journal designed to surface patterns in emotions, communication and habits. Orders are planned for November, with early-user shipments targeted by year-end.
Different controls, different operational questions
Not every device depends on continuous listening. Sandbar’s Stream ring has a microphone that is off by default. Users press and hold a touch-sensitive panel to record notes, speak with an assistant in the companion app or control media playback. The ring is expected this summer in silver and gold finishes. Rabbit’s $199 R1 takes a handheld route, combining a touchscreen and rotating camera with functions such as booking flights, ordering meals and controlling apps. A software update added expanded AI features, including Creations for building tools and games.
For businesses considering these products, the practical issue is not simply which assistant offers the longest feature list. Teams need to match the device’s recording model, consent requirements, mobile-app dependency, transcription allowance and subscription cost to the setting in which it will be used before making it part of meetings or everyday workflows.