Sadly unlike their Ubiquiti APs, where you can host the software on any hardware (even a Pi), their doorbell requires their hardware. Last thing I want is a security system that dies with whim of some random manufacturer.
It makes for great demos in keynotes, followed by adoring media regurgitation, but actually try to set it up and nothing works as it’s meant to. The benefits outweigh the costs to me, but it’s only suitable for technically inclined folks with a lot of patience or those who are willing to outsource the whole process. An average consumer can’t just setup their lights/aircon/curtains to be voice controlled out of the box.
This has been my experience for both Google and Alexa. Google’s app is particularly slow and tedious. Routines take several seconds to load and save. Neither platform has anything resembling a proper programming module with reusable code, ability to pull in open source modules, or listening to input and using it as variables.
Unlike Alexa, you can’t even add sound effects to a routine with Google Home. This is a platform built around “smart” speakers.
The voice interaction on both platforms is just not smart or contextual. ChatGPT highlights that.
Customers are learning or have learned, the hard way, that home automation is not at all like fairy dust and unicorns.
See for example: Linus Tech Tips.