As I understand it, nowadays OS shells do provide something like that, by having a smart indexing and search tool available. A "super app" seems to be another way to provide that consolidation, but is there more to it than indexing?
A user is basically having a chat with their device. The UI is the language available for the user to form queries. The more flexible the language, the more the user may express complex queries. But users don't really want to learn a complex language, they want something either "intuitive" or matching their interests from the beginning (ie, opening up the search possibilities, or narrowing down to a usable set of targets). Because intuition is only made from analogies of accumulated knowledge, the former can only be constructed by having the app capable of understanding (possibly a subset of) the user language, which leads to Siri and the like. The later by reducing the scope of the super app to a template of interests, which could either fixed (mainstream topics), or determined from a set of existing models by having the user go through a questionnaire at app install to narrow down their interests to a manageable subset.
I think that for now the key factors are:
(1) consolidation of information, because the user doesn't want to jump through each of their accounts or data sources to find what their looking for.
(2) automatic connection of different services (banking, online markets, and perhaps many other use cases); this is only relevant for services which are not equipped with a protocol to deal with a given form of connectivity. Perhaps this is a generalization of (1).
(3) social interaction and controlled sharing of data
(4) security
Ultimately, the UI would have to become some sort of programming language to really be able to fulfill all of the users wishes, but natural languages are a poor fit for this task, and because of this the feature set of these apps will always be limited, unless somehow Siri and al learn to ask the user to be more precise when there are ambiguities (not unlike the dialogue that exists between a programmer and a compiler). Also, speaking to your app is not really a secure way of interacting with it, unless you're alone when doing it, and you never leave your device.