Microsoft Patents a System That Tracks Document Activity and Connects Coworkers Who Need Each Other
Microsoft has filed a patent for a system that watches how people interact with documents, builds a web of those interactions, and uses that web to automatically pair up coworkers who are working on similar problems or need each other's help.
What Microsoft's document activity graph actually does
Right now, when you share a document with your team, most apps tell you who viewed it or edited it, and that's about it. You don't know that a colleague across the building spent three hours on the same topic last week, or that someone on another team could answer your question in five minutes.
Microsoft's patent describes a behind-the-scenes system that builds a kind of map of document activity. When you open, edit, or comment on a file, that action gets recorded in a structured way alongside the topics that document covers. Other people can then subscribe to your activity, the way you'd follow someone on a social network, seeing what you've been reading or working on.
The most practical part: the system can spot when one person has a task they need help with, and match them with another person who has the skills or knowledge to help. Instead of you sending an email to six people hoping someone knows the answer, the system already knows who does.
… uses these structures to match users that can help with a task with users that need help with a task.
Translation: The system connects people who need assistance with coworkers who have the right expertise.
How the document graph links users, files, and tasks
The patent describes a structure Microsoft calls a document graph. Think of it as a database that doesn't just store files, but stores the relationships between files, the people who touched them, and the actions those people took.
Each user action (opening a document, editing it, sharing it) gets represented as a specific node or edge in this graph. The system also extracts the topics a document covers, so it can aggregate what a team has collectively been reading or working on, even across many different files.
Two features stand out in the claim language:
- Activity subscriptions: One user can subscribe to another's document activity, getting a feed of what that person has been engaging with, similar to following a colleague's reading list.
- Help-matching: The graph can flag when a user is stuck on a task and automatically identify another user in the organization whose past document activity signals they can help. The system then surfaces that connection.
The patent covers the data structures used to represent all of this inside the graph, not just the interface a user would see. That means it's aimed at the underlying engine that would power features in something like Microsoft 365 or SharePoint.
What this means for Microsoft 365 collaboration
For anyone who works in a large company, the daily friction of not knowing who knows what is real and persistent. You duplicate work, you miss experts, you send emails that bounce around for days. A system that passively maps expertise from document behavior and routes people toward each other could shave hours off that kind of organizational waste.
Microsoft's steady investment in productivity-graph technology suggests this fits into a longer effort to make Microsoft 365 a layer that understands work context, not just stores files. Whether this specific architecture ships in a product is another question, but it points toward a future where your office suite understands your organization's knowledge map and puts it to work for you.
That makes this Microsoft's 403rd filing in our Microsoft coverage since May, adding a new idea alongside one on private ad targeting and one on hiding AI reasoning.
The practical promise here is that your company's software would stop asking you to maintain a skills profile nobody keeps current, and instead watch what documents you actually open, edit, and share to figure out what you know. That inference engine is the core of what Microsoft is describing.
For a regular user, this would show up as a moment like: you're stuck on a budget question in Teams, and the system surfaces a colleague who has been deep in similar spreadsheets all month, someone you wouldn't have thought to ask. The failure it prevents is the familiar one where the person who could answer your question in five minutes never finds out you needed them.
Whether this ever becomes that visible feature depends on decisions far outside the patent itself, but the underlying logic is sound and the problem it solves is one that costs organizations real time every day.
There are more where this came from
We read every patent application Big Tech publishes and send you the ones worth knowing. Plain English, free, every week.
The drawings
18 drawing sheets from US 2026/0267929 A1 · click any drawing to enlarge
Want this weekly breakdown for a company we don't cover? Patentlyze Pro →