Microsoft 365 Copilot is coming to you, and as it does, it's important to understand how it works under the hood. What's different about this than maybe other large language model systems that you've used like Chat GPT before? And, of course, answer what IT can do to prepare for it. So under the surface, it may seem really simple.
I have this wonderful chat-like prompting that aggregate and works in my system to provide a better experience. In that instance, you might think of it as kind of like taking Chatty BT and bringing it into Microsoft 365 or bringing Bing Chat or the new Bing search experience and bringing it into Microsoft 365. It's a little different because, in both of those examples, they don't reason over your data in your organization; they don't know the connections between people and content within your organization, which is powered by the Microsoft 365 graph today.
What it essentially does is when you make a request, you may ask for support creating a draft, or you're looking for a summary before a meeting. When you make that request, what it does is it goes and looks at what they call grounding. It goes and looks at the information within your organization that you have access to.
So it's not grabbing information you don't have access to. It's then taking that information and passing that along to, yes, a large language model. Then it takes that information from the large language model, which then gives a response back, and goes back through the Graph from a grounding perspective. So it provides more enrichment, makes sure that it's tuned a little bit better for you, and then brings it back to you.
So what this means is from an experience perspective, what's actually happening is I'm only using and reasoning over the data that I have access to via the Graph and via my security. So this removes a lot of the privacy concerns or security concerns people have. The second thing that's important is that Microsoft never takes the knowledge in your Graph. That's your Graph. That's your organizational data.
They don't take that and go put it and upload it or share it with that large language model. They don't do that because that would be a huge danger for privacy and other reasons. It's only the request itself that is then going to bring back key data enrichment, specifically around language rationalization, into your Graph itself, the graph knowledge that you have, et cetera.
And then, of course, Microsoft has services that make sense of that and then bring that back to you. So this is an important distinction because, again, that cycle that it's going through each time, that's basically providing the experiences that we get. But it's based on the technology that we're already familiar with, like security controls for items and documents and sites and teams and things like that, as well as how we think about working with our data.
It's our data. It's in our tenants. It's in our services. Again, you have all the same places like Geolocation, where that data resides, et cetera, et cetera. Now, if you aren't familiar with things like Microsoft Loop and you saw some of these announcements, and you're like, oh, there's this wonderful new collaborative canvas for multiplayer by design, and that canvas, if it's working in there, where does that data store? Again, the way Loop works is it stores records sort of like files in relation to that.
So again, compliance and sensitivity and all sorts of other rules over time can be applied to that content when you're ready to do those investments. Microsoft themselves are creating a number of different ways that you can administer and manage this. They'll share more of those governance announcements soon.
Right now, they are doing a private preview with customers. So they're learning based on feedback on how to tweak and optimize some of those signals and things before they release it more broadly to a broader public preview and then eventually a larger release. That's the first thing to note underneath all this; yes, it uses something like OpenAI or ChatGPT 4, et cetera.
But what it's doing is it's rationalizing it with their own systems, in this case, Graph, as well as their own logic systems that are essentially making sense of this and putting this together. This is important, too, because while I love those other tools that are in the marketplace, large language models don't really understand the millions of different configurations and app signals that you can do within an app. So if you've ever used PowerPoint before, you can do all sorts of things like creating animations and transitions, and you can retool and realign images, and you can do all sorts of things within the app.
There are hundreds and hundreds and hundreds, thousands and thousands of commands that people run within these apps. Microsoft is doing the hard work of taking all those commands and rationalizing them so that when you ask for something in a very simple natural language way, it's getting the best results or the best responses for you, or at least giving you options to regenerate it in different ways very quickly. And so that is the Microsoft investment they're doing, not just this response app and command experience, but the fact that they have this understanding of their own tools.
And you've seen some suggestions of stuff like this in the past, right? PowerPoint has had a designer for a while that's helped people kind of make better-looking slides based on what you've already input. Now what you're seeing is instead of just creating better slides based on your input, they're creating the entire slide deck, and you can more naturally interface with that. And again, if you've never used it, at the top of PowerPoint today, you can type in natural language questions, and it'll give you different app commands and things like that.
It helps people find where they can crop an image, et cetera. If I'm not sure where in the ribbon - that bar at the top where this command is - It helps people find that this is basically, instead of just doing that, it's doing the action for you.
Again, some actions you'll, of course, be able to tweak yourself if you want to make minor adjustments or want to take it further. But this idea of really accelerating the time and work that it requires is important. So this is another reason why it's really important to understand Microsoft's place within this narrative in this journey because they are essentially enabling a lot of those more effective productivity signals and opportunities because they understand those products and how those language models might be able to interact with them in creative ways.
One last comment on preparation for a lot of organizations that I'm working with today, they are thinking about this from an It perspective, and they're saying, oh, we'll watch the roadmap when we see the announcements for it. We'll do private previews in our own organization. We'll pilot it. We'll go through these motions.
It's really important. It's very important to understand that if you're not already doing it, motions to help people understand how to use AI tools, tools that are in the marketplace and available today, and how and when they are allowed to copy content into those tools. Because if you don't know this, most of those tools allow you to upload documents and things like that so that it can reason over it and provide a more contextualized and more effective answer.
If you don't already have governance and rules and guidance and mentorship and proactive digital excellence around that today, you absolutely need to invest in there right now. Because not only is this going to be a more dominant thing and better in the future because it'll be part of your Microsoft content, but this is much better because there's less of this content exiting the organization, right? So it's better, this is going to be much better than the way people are working today because, trust me, people are using these tools today. They're just messaging and using these tools in probably a way that it's not really a big fan of in this instance.
Now they can do those things within the organization, and then you'll be able to have different types of monitoring, auditing, and support tools for it. So again, this is a good change, but it should already be at that table and should already be an active contributor in those discussions on how to do those things. And if you're not, then now is a wake-up call to really start to move forward with that because you don't want to be waiting in a few months, or it's going to come very quickly when this stuff comes out when it starts to be realized within the products.
You don't want to be on the back foot and holding your organization back because these tools change the competitive landscape considerably. Copilot is by far the most substantial change Microsoft has ever made within the way we work. And I'm really excited about how we all make the best of that and help our employees be successful.
