Privacy-first local companion design

Keeping raw conversation audio off the cloud is less a nice privacy feature and more the core trust wedge that makes an always listening companion socially usable at all. In practice, Friend keeps only a short local audio buffer on the pendant, then uses the phone app to turn speech into supportive text nudges, which separates it from memory wearables built around long term storage, search, and retrieval of everything said.

• The real contrast is with products like Limitless, which were built so users could later search and recall past conversations. That memory use case gets more valuable when data is retained, but it also creates the strongest discomfort around being constantly recorded and around consent from everyone nearby.
  2 sacra 3 sacra
• Friend also reduces social friction by sending replies as text notifications on the phone instead of talking out loud from the device. That makes the pendant behave more like a silent emotional coach than a visible meeting recorder or sci fi assistant trying to join the conversation.
  1 sacra 2 sacra
• This design choice narrows the product into companionship rather than note taking. Plaud has shown there is major revenue in recording, transcription, and summaries for professionals, but that market rewards persistent records and workflow outputs, not short lived local buffers.
  1 sacra 4 sacra

The category is moving toward two distinct lanes. One lane is private, lightweight companion software that minimizes retention and acts in the moment. The other is work oriented memory capture that stores, organizes, and monetizes conversations as records. Friend's local only design points clearly toward the first lane.