I'm a big fan of Philip Van Allen's work when he was at ArtCenter about animistic design of AI systems.

I put together an animistic character design and roleplay workshop for smart homes at SXSW:

https://uxdesign.cc/how-does-the-roomba-really-feel-about-dog-poop-c590bcfb8834

Also, I did a more recent article about thinking through how LLMs that are engineered to act (via animism) as smart home devices might behave:

https://uxdesign.cc/a-smart-home-is-one-that-talks-to-itself-58bb9222d893

Would love to hear your thoughts!