Andrew Shuman, corporate vice president for Cortana discusses how the smart assistant has pivoted since it was originally introduced five years ago.
At Microsoft Ignite 2019 in Orlando, TechRepublic’s James Sanders spoke with Cortana corporate vice president Andrew Shuman about the ways Cortana is evolving for business use. The following is an edited transcript of the interview.
Andrew Shuman: If I think about where Cortana is now, we’ve really been on a journey the last year to think harder about how an assistant can be truly helpful for you–the end user–in your day and day out life. And so we focused a lot on thinking about places where we knew a lot about users already.
Thinking a lot about the M365 users–the several hundred million users where we understand who they communicate with, the projects they’re working on, the documents they’re processing, the tasks that are on their list, how they spend their time. And leveraging that signal to be truly intelligent and truly helpful, I think is the key focus we’ve had for Cortana now. Because if you think about it, you need to know a lot about a user to be able to be assistive. The best real life assistants know you well, and that’s where we’ve really taken Cortana now.
SEE: Choosing your Windows 7 exit strategy: Four options (TechRepublic Premium)
I think one of the things that we’re really trying to emphasize today is to talk about M365 and how it becomes much more intelligent, and knows you really well. There are great features across the suite that really demonstrate that. There are instances where we’ll choose to call a feature Cortana because we think it helps to be a personified assistant, one that you can perhaps address by voice, or address directly in an email, or get an email from where it’s clear that it’s Cortana.
We see that that’s a useful mechanism, and a thing that kind of brings home the possibilities of Cortana the assistant–that’s really important. One of the features we’re announcing today that I’m especially proud of is called Play Your Emails. It happens to be Cortana under it and you see the Cortana icon, but we really want to set up a clear promise to the end user that says, ‘What is this feature going to do?’
Instead of that problem we sometimes have with assistants where you don’t quite know what to ask them, we want to deliver on that experience and say, no, ‘This is about playing your email.’ And we’re really intelligent about understanding the email and understanding how to do a great job of reading it out to you, understanding the subtleties of your organization, saying, ‘It’s a mail from your manager,’ or ‘It’s a mail that mentions you.’
These kinds of details really help in that personalization part of the puzzle, but also help set a good promise for the user. The biggest thing that we share in common with both Microsoft search, other intelligence features across and M365, the Microsoft graph, is that we’re really leveraging natural language understanding of these incredibly valuable documents. And we’re doing it in a completely eyes off fashion.
So we understand the trillions of items that are encompassed by your emails, your documents, your SharePoints, your calendar items, your tasks. All of that shares a component that understands the natural language and then can do things like you see in Bing where we answer questions about your company, like the Pet Policy at Microsoft was one example we showed today.
But also things like Cortana that can understand, ‘They’re asking you to follow up on an item,’ or, ‘Here’s the pre-read for the meeting that you’re about to attend.’ Those all encompass a similar kind of substrate of data and knowledge that we want to be really powerful for all of these experiences.
One of the experiences we announced today is how you can reply to an email and add Cortana on the CC line, and help them find time to schedule that group of people together. That’s an experience that’s been about three years in the making. We’ve been working with a lot of tough enterprise partners and customers to work closely with them on how we tune the model and explain the model.
Some of those deployments have upwards of 30,000 meetings being scheduled per month this way. And really giving them a lot of time back because those are multi-turn, tough negotiations of people’s busy schedules and calendar items. So that’s been a great example where it’s really helped people where they are today in managing their time.