A wearable microphone that feeds context to your personal AI

What it does Adeus is a three-piece system built around a wearable microphone that records your real-world conversations. Audio gets shipped to a Supabase backend running on your own server for transcription and storage, while a mobile or web app lets you chat with an LLM that has access to that full history. The pitch is context: instead of a generic chatbot, you get an AI that can reference your meetings, rants, and grocery lists.

The interesting bit The stack is billed as fully open source and self-hosted, which in the current wearable-AI gold rush makes it a rare specimen. The README now redirects active development to the BasedHardware “Friend” repository, suggesting this codebase is largely in stasis.

Key highlights

• Full-stack wearable: custom hardware, mobile/web app, and Supabase backend
• Self-hosted data pipeline: transcripts and audio live on your own server rather than a vendor cloud
• Context-aware chat: the app surfaces your conversation history to ground LLM responses
• TypeScript codebase with 3.4k stars

Caveats

• The README explicitly redirects active development to the Friend repo, so this appears to be in maintenance mode or superseded
• Hardware details, battery life, and physical form factor are unspecified in the README
• It is unclear from the docs how transcription is performed or which LLM providers are supported

Verdict Tinkerers who want to own their real-world conversation data should look here, though most should probably follow the README’s pointer to the actively developed Friend project instead.