My home
Welcome home
Home overview
Awaiting Home Assistant
Listening for available device signals.
Needs your attention
No scan yetBefore you leave View devices
Recent activity View history
How this works
Observes existing Home Assistant device states. Open a signal to see its source and time. No emergency predictions or automatic physical actions.
Available
Start a session to protect your focus.
Start a focus session
For later
0 itemsVoice ActionsCommands and helper actions NABU can understand0
Cameras
Live view and your private recording library.
A real desktop.
Right inside NABU.
Open windows, manage files, run applications and use your Raspberry Pi with a mouse and keyboard. Nothing is simulated.
Checking desktop service…Install the graphical desktop
This feature needs an XFCE desktop and TigerVNC on the Raspberry Pi. Install the included package over SSH once, then come back here.
cd ~/NABU-5.3.0 && bash install-graphical-desktop.shThe installer uses a local-only Unix socket and asks you to create a separate VNC password. It does not replace Raspberry Pi OS.
Your routines
Make it yours.
Services
OpenAI
AI, speech recognition and Nabu's voice. Your API key stays on the Nabu backend.
Home Assistant
NABU uses Home Assistant only as the smart-home device and satellite layer.
How Nabu sounds
Nabu voice
This voice is used in the web app and by Nabu / Voice PE through the NABU Voice Bridge.
More voice preferences
Nabu / Voice PE
NABU Voice Bridge
Home Assistant captures audio from Voice PE, then sends STT → conversation → TTS through NABU. The OpenAI API key never needs to be stored in Home Assistant.
Checking Home Assistant…Context
Weather
Saved permanently for Nabu. Weather questions use this location automatically unless you name another place.
Location & presence
Optional. Used for Home/Away context while Nabu Home is open.
AdvancedModels, privacy mode and diagnostics⌄
Local AI
Run a model on your own network
Use an Ollama model for text chat, or keep cloud intelligence with an optional local fallback. Saved NABU memory stays on this device. Voice and live sessions still need separate offline speech components.