Every session of the EXCAR build, written down: architecture, voice, memory, the app, and the decisions behind them.
What EXCAR is, the design philosophy, and the interfaces that make everything swappable.
The streaming voice loop (Whisper → Claude → Piper), tool use, and wake mode.
Local GGUF + hybrid engine, BT tether, persistent memory, the REST bridge, device boot.
The Expo / React Native app: design system, navigation, stores, onboarding, device setup.
An on-device terminal over /exec, the draggable sphere launcher, polling robustness.
Fuel & range tracking, voice enrollment (d-vector), EAS build profiles, API contracts.
GPS via gpsd, Spotify OAuth, barge-in, audio ducking, diarization, intent gate, Pi config.
Echo guard, barge-in hardening, the addressee detector, engine-off memory summarisation.
A provenance + confidence fact store, people/patterns scaffolds, the EX app icon.
Structured now-playing + control endpoints, Connect Spotify, the Home now-playing widget.
No entries match. Try another word.