It connects to a local LLM (Ollama), supports speech (STT/TTS), embeddings, and simple memory via RAG. The goal is to have a persistent, always-on personal assistant without sending data to the cloud.

It’s still experimental, but usable today on a Raspberry Pi or a small home server.

A short demo video is available in the README.

GitHub: https://github.com/orangekame3/mirrormate

Feedback welcome.