Some thought starters:
It seems to me the easiest way to think about “structuring” the DAO is to consider which agents should have which responsibilities and how reputation or tokens are weighted in decision-making. For instance: will the DAO wallet be accessible to particular designated agents (more of a ‘qualitative’ model or to agents of a certain reputation or higher – more of a ‘quantitative’ model.) Also: does there need to be some sort of secondary voting system for the DAO wallet? Such as approval from 2/3 agents who sit on a sub-DAO within the larger DAO (not unlike a multisig)… The important thing here is to start thinking in terms of reputation, agents, schemes, and constraints: or rather, who are the actors and what are the rules enabling or prohibiting them from interfacing with the DAO’s treasury/upgradability?
As far as interfacing with “centralized” systems – in relation to the DAO stack the roadmap is planning external integrations with work software such as Slack or Telegram for Q3 2019; that being said I think there’s immense value of building a dApp that can interface between DAOstack reputation systems and Google Docs, for instance. In terms of real life application, it’d be great to have it be X reputation in a DAO gives Y access to a particular Google Doc folder. I imagine people will likely propose these types of ‘bridges’ be made early on in the life of the Genesis DAO so-as to enable the exponential growth of the wider ecosystem.