It is in-general very difficult. ->Lack of APIs ->Interface on Phones ->Access to every app which should be explicity given by the user in case of no apis ->A l

It is in-general very difficult. ->Lack of APIs ->Interface on Phones ->Access to every app which should be explicity given by the user in case of no apis ->A lot of companies have forked chromium and deployed it on cloud to automate the tasks. Basically a bot doing things on a browser in cloud with access to credentials. ->iOS is not open sourced. ->Running a hyper-optimised AOSP is difficult(Andorid) It is good in case, for an app that does everything already(a super-app), they can release a voice assistant which is confined to that app itself.