Hi everyone!
I wanted to share a small project I’ve been working on for my home setup. It’s a custom conversation agent that bridges the gap between HA’s speed and LLM flexibility.
How it works:
- It first tries to match the user’s message using HA Built-in Intents (super fast for lights, TV, etc.).
- If an intent is matched, it executes the command and then uses a Local LLM (via Ollama) to generate a natural response.
- If no intent is matched, the LLM handles the request directly.
Everything is 100% local. No cloud, no API keys.
Note: This is a “personal itch” project. I’m sharing the code as-is for inspiration or for anyone who wants to fork and improve it. I might not be able to provide active support as I’m moving on to other experiments!