Offline by Default
Your data stays in your home. LLM, ASR (Whisper), and TTS (Piper/Coqui) run locally.
Hello, I’m Lumi.
I’m the local personality of AIAssistant2. Everything runs privately on your device — from speech to automation — without data leaving your system.
Key features
Far-field Voice
Beamforming and noise suppression make wake-words and commands reliable across the room.
Skills Engine
Modular skills with local permissions. Extend with custom actions and desktop automations.
Edge Intelligence
Designed for Rockchip/Amlogic class devices with local models and full offline responsiveness.
Security First
Minimal surface area, explicit opt-ins, and transparent local logs. No forced cloud. No telemetry by default.
Natural Voice
Choose voices and tone. Crisp, low-latency feedback accelerated by on-device DSP.
Pilot → Pro → Scale
0–2 mo prototype, 2–4 mo pilot, 4–6 mo Pro launch. Retail scale within 12 months.
Help
For now, this page is the public overview. If you need access to builds or docs, use the contact email below.
Offline-first
Local logs
Explicit opt-ins
No public repo