Error message

Notice: Undefined index: content in clean2014_block_view_alter() (line 631 of /var/domains/

The human brain organizes reality by associating topics. When we think about the world around us, we do not think in a highly structured manner. Instead, our thinking is much more fluid and ‘messy’—we think of things by association of categories that are more strongly or weakly correlated with one another. 

Being aware of this distinction is very important in software design. All too often, systems are designed to present information in a very hierarchical and structured form. The user moves from ‘a’ to ‘b’ to ‘c’ to ‘d’. This is fine for software designers, but unfortunately, human brains don’t follow this structured order. Our thinking is more likely to move from ‘b’ to ‘d’ then ‘c’ and back to ‘a’. Technology that reflects this reality, that fits around how humans actually organize and interact with information, can be said to be ‘humanized’.

The Microsoft Graph is a machine learning engine that is principally used by Microsoft in Delve, a ‘personal assistant’ which sits inside of Office 365. The Graph is especially useful because it is able to notice patterns in usage of Microsoft’s services. And, now, Microsoft have made an API available that allows external developers to utilize this machine learning power in their own solutions. This is a big deal, because the Graph API can be used to present information in the way the brain really works: by topic, rather than app or process. Why is this so profound? Because it can be used to help humanize the digital experience.

What is the Microsoft Graph?

The Microsoft Graph is a very sophisticated machine learning engine. It ‘learns’ about an individual user—about their job title, their diary, what events they attend, what messages they read, and what groups they are a member of—and monitors how they’re using Microsoft’s cloud services (i.e. Office 365). It is then able to find patterns between the data that it collects. The Graph is very smart—it maps out relationships between people and content, discovers who accessed, modified, shared or liked a piece of content, and also finds links between people who those individuals interact with via email, Skype for Business or Yammer.

Microsoft have now made an API available that lets you access this information via a range of calls which can be used by third party developers, for example:

  • WorkingWith call
    This call surfaces information about the people the user works with most often.

    Example: bring alerts to the user based on the activities of their most frequent contacts.
  • TrendingAround call
    This call surfaces information to the user about documents that their frequent contacts are currently working on.

    Example: when a manager logs in every morning, they can immediately see whether their team finished writing a report.

Showing what’s meaningful for the user

Traditionally, IT has been designed in a very ‘mechanical’ way—think of the hierarchical folders and files in most document libraries, where the user has to logically drill down through a structure to reach a particular document. Yet, this is the opposite of how humans actually think—i.e., finding patterns and associations between different pieces of information that are more or less similar in content. As we continually improve and extend Collage—our topic-driven, content-mapping engine—we have been making great use of the Microsoft Graph API to help design technology that works the same way that people think.

Collage is a sidebar which sits inside the user’s Outlook interface—since email is the place people spend most of their day at work. For all content that arrives in your inbox, Collage shows a range of contextual information. And, we can use the Graph to support this:

  • Show what’s meaningful
    For something to be meaningful, it needs to be contextual, related by association to what the user is doing at a certain time. We use the Microsoft Graph API in Collage to do just this. When you receive, say, an email from Greg about the Praxis project, you also want to be shown other related documents—other emails from Greg about the same topic, documents in SharePoint related to the Praxis project. Collage does just this.
  • Think by association, work by association
    We recognize that people think by making links between topics, and so we designed Collage to support this. Each topic Collage surfaces is tagged with multiple relevant labels. For instance, on reading an email about the Praxis project, the user might want to click on that ‘label’ and see all content related to the project. Then, they might click on another ‘label’ related to the Praxis project, such as the name of the client, and find out all the latest content related to that client.
  • Recent experience
    Another key factor about the brain is that it is most heavily impacted by recent experiences. We use the Graph API to bring information to users which is both contextual and recent. There’s rarely much value in surfacing data about a client that is five years old, and so we use the Graph API in Collage to only show information that is most likely to be relevant.


Collage does not just work with Microsoft products—the true aim of what we’re doing is to bring all relevant information to users from all the different apps they use. However, the Graph API is certainly one useful tool when it comes to supporting our goal of putting the user at the center of the digital experience because it makes it especially easy to bring information to them in a way that matches how they think—by context, by association and from recent experience.

See more details in this Slideshare presentation.  


Ram Tagher
Product Manager