Home Assistant smart home software can now also control devices in the home using large language models, or LLMs. This makes it possible to give more complex voice commands.

Home Assistant, like other smart home software, uses an AI-based voice assistant that understands voice commands, can perform the desired action and provide a response. Home Assistant uses its own “conversational agent” for this purpose as standard. In addition, it is possible from the beginning to have a conversation with a voice assistant based on LLM. But users were unable to use this assistant to control devices in their homes.

The company stated that this will change with the new version of Home Assistant advertisement. From now on, there is also an LLM-based chat agent, which can handle more complex commands. For example: “I’m going to a video meeting, can you make sure people can see my face?” The organization claims that the new voice assistant can understand the intent behind these types of commands and take appropriate action.

Users can choose whether to use a traditional voice assistant or a voice assistant based on the LLM. Anyone who chooses an LLM assistant can decide for themselves which LLM to use. Home Assistant already has integrations with OpenAI, the maker of ChatGPT, and Google AI. There is also the possibility of running local LLMs, but they cannot yet take over houses. Home Assistant is working with Nvidia to make this possible at a later date.

Home Assistant was founded in 2013 by Dutchman Paulus Schoutzen. Tweakers spoke to him last May during a State of Open Home broadcast.