In my opinion, the first thing any service should do is to own and operate their own unified interfaces and to collect and own all their own raw data. Any service used to process data should be at the back end and have no right to acquire the data. And so if they want to switch the processing entity, they can do it without the consent of the processing entity independently otherwise they will become hostages of software companies with or without ai. This includes named databases. Any basic open source database software can be used to collect the raw data.