Any voice assistant that isn’t LLM AI based is near-worthless. Hence why Alexa, Siri, Google etc have done the bare minimum task-based assistance since their inception. When you have to actively think about what you can say or get a response to then you just do the bare minimum.
A voice assistant will be good when all interactions are indistinguishable from sitting and talking to another human being. When it can understand any context or like a human, when it’s not sure what you mean exactly, and asks for clarification, then and only then will it be great.