richard-companion
Self-hosted house companion driven by a local LLM — voice, Home Assistant, and automations you talk into existence
richard-companion is an open-source, self-hosted home AI companion (v0.1.0) that runs entirely on your local hardware using a local large language model (LLM). It integrates with Home Assistant to give you voice-driven control over your smart home, letting you create and manage automations simply by speaking to it — no cloud services, no subscriptions, and no data leaving your network. Designed for privacy-conscious home automators, it combines speech recognition, a local LLM backend, and Home Assistant's ecosystem to deliver a fully private, conversational smart home experience.
Who it's for
Pricing · free
checked 1d ago| Plan | Price | Includes |
|---|---|---|
| Free / Open Source | $0 /mo | Completely free and open source · Self-hosted on your own hardware · No cloud dependencies or subscription fees · Bring your own local LLM (e.g. via Ollama, llama.cpp, etc.) |
AI-researched pricing — verify on the official site before subscribing.
Use it for
- — Voice-controlling smart home devices via Home Assistant
- — Creating automations by describing them conversationally
- — Running a fully local, private home assistant with no cloud reliance
- — Querying the state of home devices using natural language
- — Automating routines through spoken commands processed entirely on-device
Get the most out of it
- 01Run a capable local LLM (e.g. Mistral or LLaMA 3 via Ollama) on a machine with a dedicated GPU for the best voice-response latency.
- 02Ensure your Home Assistant long-lived access token has sufficient entity permissions so richard-companion can read and control all desired devices.
- 03Start with simple single-device automations to validate your setup before building complex multi-step routines.
- 04Use a high-quality USB microphone or a wake-word-capable device to improve voice recognition accuracy in noisy home environments.
- 05Check the project's changelog and GitHub issues regularly — at v0.1.0 it is early-stage software and updates may bring breaking changes or important new features.