Max Headbox
A local voice AI agent runs on Raspberry Pi with Ollama models, custom automation tools, and spoken confirmation for sensitive actions.
Max Headbox is a local AI agent for people who want a voice-controlled assistant that can carry out actions on a Raspberry Pi. It processes voice and runs models on your own hardware without relying on cloud AI providers.
- Voice commands: Wake-word detection uses Vosk, and faster-whisper transcribes speech locally.
- Custom actions: You can extend the agent with JavaScript tools to automate tasks and interact with Pi hardware.
- Action confirmation: Tools marked as dangerous ask for a spoken yes or no before the agent executes them.
- Visual feedback: A browser interface uses an animated character to show when the assistant is processing a request.
The project uses Ollama with Gemma models, including gemma3:1b and gemma4:e2b. Ollama can run on the Pi itself or on another device on your local network.
Max Headbox is open source under GPL-3.0. It has been tested on Raspberry Pi 5 models with 8GB and 16GB of RAM, and needs about 6GB of available memory for the language models. Voice input requires a microphone. Active cooling is recommended; the screen and case are optional.
Logotyper
