Amazon turns Bee into two carry-anywhere devices: a week on the wrist, three weeks on the phone

Bee personal AI expands from one device to two hardware forms, one wearable and one magnetic phone attachment, covering carry-on-body and carry-with-phone settings. The earlier Bee Pioneer was among the original personal AI products to offer more than audio transcription. Both new devices keep Bee's function of turning conversations into summaries, to-dos, reminders and insights; Frontier targets people on the go and Eclipse targets those who want one less thing to remember.

Both devices use the same interaction to let users control when audio is processed: a button and a visible indicator light. Bee has carried the principle of user-controlled audio processing and a visible indicator since day one. Frontier can be set to turn off automatically after a period without sound, such as 5 or 15 minutes; Eclipse has one on-off button and lights up while summarizing a conversation.

Privacy handling is uniform across the devices: audio is never stored, cannot be accessed or shared, and users are prompted to get permission from call participants when recording starts. Bee is positioned to turn conversation details into useful information rather than keep the original audio. Audio is processed in real time then deleted, customers control their data, and each recording start prompts permission from call participants.

The report covers two Bee devices Amazon released and their described features and specifications: Frontier is worn on the wrist or clipped to clothing, is IP67 water resistant, runs up to seven days, charges over USB-C and has dual microphones with noise filtering; Eclipse is about the size of a credit card, attaches magnetically to a phone and runs more than 21 days per charge. Both turn conversations into summaries, to-dos, reminders and insights, and can connect to Alexa+ for hands-free summaries and to-dos or to Amazon.com for personalized shopping.

The source is a product introduction and gives no data on summary accuracy in real use, recognition in noisy settings or measured battery performance. Privacy statements say audio is processed in real time then deleted, is never stored and cannot be accessed or shared, and that each recording prompts permission from call participants; these are presented as the releaser describes them.