I’ve done a thing with Kamailio and Baresip and MQTT and Linphone on my phone so when someone presses the door bell button, I get a video SIP call from “doorbell”. But other I think are doing things with HomeAssistant, go2rtc and Frigate. I just didn’t like it so went my own way. Would love to have done my own Signal client that wrapped RTSP, but it wouldn’t be allowed on the Signal network, but Linphone is ok. Video SIP is standard at least.

You can also flash a Wyze Doorbell v1 with Thingino

https://github.com/themactep/thingino-firmware/wiki/Camera:-Wyze-Doorbell-(V1)

Lots of ways self hosting ways of doing bidirectional rtsp doorbell.

Humans are going to be the weak point of any system.

I was thinking this about getting off America servers and services. More a question of digital sovereignty security. But it is all do with hacking via humans by pretending to be support staff.