The starting point
Almost nobody chooses their office stack again because AI is coming, and that is sensible. The relevant question is the reverse: given what you already have, where does a workflow run smoothly and where does it cost extra work?
Three things differ in practice: how the connection comes about, how permissions carry through, and who carries the administration.
The route of a connection
Not every system is opened up in the same way. On our platform there are dedicated toolsets for Gmail, Google Drive and calendar, among others, on the Google side, and for Outlook mail, Outlook calendar, SharePoint and OneDrive on the Microsoft side. Within Microsoft 365, Teams is the exception: there is no dedicated connection for it, and whatever a workflow does with Teams runs through MCP or custom work. That is no judgment on Microsoft — but it does determine the lead time of that one connection and who maintains it.
In practice: a document or mail workflow on SharePoint, OneDrive, Drive, Gmail or Outlook starts from an existing connection. A workflow that posts a notification in a Teams channel or picks up a message from Teams requires an explicit choice about the route before you begin. Always check that route before you make a plan; “it is possible” and “it is a direct connection” are different statements.
Permissions
Both stacks have a well-developed permission model, and the same principle applies in both: a workflow may see what the user may see, and no more. The difference lies in where the complexity sits.
- Microsoft 365: permissions are often layered — at site level, library level, folder level and sometimes file level, with inherited and broken permissions. That is powerful, and it makes the question “who will be able to see this?” harder to answer without an inventory.
- Google Workspace: sharing is flatter and more visible, which makes it easier to get an overview. The downside is that loose share links spread easily and are therefore hard to contain.
In both cases the first step is the same and is the one most often skipped: establish what is already shared too broadly today. A workflow changes nothing about that, but it does suddenly make it visible — and that then feels like a problem the workflow caused.
Administrative load
The question that counts: who notices when a connection stops? In both stacks authorizations expire, administrators change policy and accounts lapse. Make sure every connection is registered to the organization rather than a personal account, and that there is one internal administrator who can restore it. That matters more than the choice of brand.
Mixed environments
Many organizations have both: mail in one stack, documents in the other, and a department that uses something of its own. That is workable, provided you write down explicitly for each workflow which system is the source and which is the destination. What does not work is keeping the same information in two places and letting the workflow guess which one is right.