e.g. if it hears the jingle of your keys (and not your flatmates) after 6pm and it hasn't heard you turn the shower on in the morning, it turns on a warm bath for you.
All-encompassing platform plays are hard to do without a solid use case to build from. However niche, you need to absolutely smash someone's problem with this platform, and once you find success there, you can go horizontal.
There is nothing wrong with solutions looking for problems: there are a zillion of them on store shelves right now. But this means that 1) you'll have to overcome an education barrier to convince people why they NEED your product, and 2) you'll be able to get much less insight about whether anyone will buy your idea.
These sorts of ideas could be bad for a new startup. It's easier to build and sell something like this if you have a larger R&D budget, a large marketing budget, or a large reach so that you can get it in front of potential buyers.
Similarly, you don't have to ask Hacker News if a pill that cures diabetes would be popular, the risk there isn't product-market-fit it is invention-risk.
How about if it listens for someone screaming for help and calls an ambulance? Assuming you can limit false positives, you could probably sell it for use with the elderly and the infirm.
Considering your example, are there any automatable bathtubs yet? Or is making such a thing part of your idea? (The cost of retrofitting such installed plumbing, and the electrocution liability, suggest to me it'd be a long, slow quest.)