As a child, did you ever look up into the stars and wonder how they were connected together? I often did. And so I often think about how things are connected across different areas, whether it's inside of our universe or inside of organizations. My name is Naomi MoneyPenny. I have the privilege of leading the Project Cortex product team at Microsoft. And I want to explore a little bit about the connections that we have between expertise, topics and metadata inside of your organization. Now, when we look up at those stars and we see those pieces of light, those points of light that really make us understand where we come from and put some context of where we are.
And so when you look at that early patents, you can imagine that when the Greek philosophers and folks looking up at the stars for the first time, they started drawing images, those constellations that we see across the sky. And so it helps us to put some shape and some definition, and to understand some of the characteristics of how things are connected together. The same thing is very true inside of your organization.
When you look at sort of the pinpoints of light, those resources, the people, the knowledge, the topics, the expertise inside of your company, you can really think together how they're going to come inside of your organization and think in most connected way possible so that you understand the patents and the diagrams that they make to understand how knowledge is flowing inside of your organization. One of the key areas that I want to explore here is really thinking about curiosity. The curiosity that makes humankind look up at the stars is incredibly important inside of our own businesses as well. And this quote from Robin Sharma, "The best in business have boundless curiosity "and open minds."
Thinking about that open mind, what we call the growth mindset inside of Microsoft is really thinking about how knowledge comes into your organization. How you can look at things with a beginner's mind and have new ideas and use that curiosity to be able to explore across your company. Now, our agenda today for Project Cortex is really about understanding where we're coming from with Project Cortex, and then a little bit about the vision of where we're going. This isn't a product focus sessions. We have lots of great sessions that help to explain the content of what we're doing, give you great product demos and let you get hands on with the technology. But this is really about understanding why we're motivated to build Project Cortex and where we want to go in the future.
Now, when I think about the learning journey that we've been on, it's really been about learning from customers, learning from all kinds of organizations across the company, all kinds of organizations across the industry. And we've often times spoken to customers who are at Microsoft. We've had a wide variety of customers from probably every industry and every area, and in particular, the retail industry, manufacturing, engineering, construction, all of those areas. And when we've had these conversations, it's been really interesting to learn about what's important to them.
And there's always this interesting, I think, from looking at hard benefits and soft benefits. Hard benefits are things around automation and how can we help to get productivity out very quickly and how can we automate tasks that are repetitive inside of an organization. But then there's always this need around knowledge sharing and thinking about how does knowledge get communicated inside of your organization. And so when we talk to customers, whether they were Microsoft customers, or even the customers we talked to that have no business with Microsoft and think about the different conferences and the different venues that we talk to there, they still put up the top of their list of unsolved challenges, was really about enabling peer to peer sharing and expertise sharing across their organizations.
And so there was always this interesting blend between we want to do better management of organizational information, think about how we can put more process around areas. How can we do more efficiently contract processing, or think about form processing in our companies. And then there was this other half that was really thinking about enabling peer to peer and expertise sharing, and then thinking about how we could develop new HR policies, how could we do better training materials, how could we onboard you as a new employee, whether you were in your different role or just new inside of the company. And so there was always this equation that was going on thinking through what it meant to be part of your organization. How do I be collaborative and how do I put together my productive day to day work, and then how do I think about that knowledge sharing and contributing more generally inside of the company and then being able to build on the work of others as well. And so when I look at these different areas, it's really interesting if you reflect that this is not a new problem.
When I started my career way back when, I think that looking at knowledge management was one of the first areas that I ever worked in. And I looked at that as an interesting system of that time, because knowledge sharing, best practice sharing, we've been saying it for a long time inside of organizations. When I worked on those first projects, it was really about understanding things like expert systems, and they were rules driven systems that helped you to understand, who could I go and contact for a particular area of expertise. And those expert systems were very, very brittle.
They were very difficult to understand. They quickly ran out of answers if you tried to ask a question that didn't quite fit its categorization model. And then I saw other efforts where people were really coming from it from a soft skills perspective. People were coming from knowledge sharing and understanding, knowledge sharing and sort of really understanding where it is your talent drives, where your passion is, how do we help you communicate, how is that role of a communicator inside of your organization and the folks that like to connect through people as opposed to through, you know resources or documents or pages or sites.
And so I try to put these areas together, really helping to understand a little bit about that context. And both of these approaches are quite valid, except that the technology system that we were using to back these things up, this is 20 years ago was really, really brittle. And so with the advent of new connected technologies, new AI models that help us to actually learn and grow from the corpus of information it's presented with, we can use the best of what we need inside of people paired with AI to do some of that automation work that we find boring as people. So when I look at this area, we might be struck as part of the environment that we're in right now, and sort of looking at the new ways of working inside of your organization, maybe everybody in your company is working remotely right now, maybe for folks who have had to been furloughed, because you've got organizations that can't quite perform because you rely on a retail location or some other frontline business.
And so I like to reflect back at sort of other times when people have seen these challenges. And so this is a great quote here from Abraham Lincoln, "Thinking through that this occasion "is piled high with difficulty, "and we must rise with the occasion. "As our case is new, we must think anew and act anew." And it's really interesting to reflect back, he's got this words from 1862 facing a whole different crisis. And so we have worked through these kinds of problems before, and this is one of those areas where looking at knowledge sharing, looking at remote working, how all of these pieces come together is something that we can do together.
So let's talk about people and the AI. We think about AI really as being that automation engine. It's your partner in your productivity. It's not trying to take away from what it is that you're doing.
It's really about how we can engage it and think about the best ways for it to come to life, as opposed to you having to do a bunch of manual work. I would say when you're looking at knowledge sharing inside of an organization, oftentimes one of the biggest challenges companies have is it's really, really daunting to look at all this information inside of the company and you think it's only something would go through and categorize all of that information. And I think, okay, yeah, this would be great, right? This is a great application of AI. This is AI doing the heavy lifting. This is what computers do well.
Crunch a whole bunch of documents, find connections, try to find topics, identify from that information what truly could be knowledge inside of your company, but it really isn't alone. It's not a magic bullet, it is not any kind of a bullet. It's really just a technology that can be used to make your organization better. We really do think about this people plus AI combination as being incredibly powerful.
It means that if AI can carry us 70% of the way, it means there's still a lot of work for people to do. There's 30% of that way that really helps to put context to make sure AI hasn't totally screwed up and gotten it wrong, and then to understand a little bit more how things need to be refined and how things can be put into the connections for other people to understand. And so when I think about this journey together, this is about people and AI. Primarily about people, but this is AI being that supercharger, that workhorse in the background that can help you to identify things and then keep them up to date when you've pointed it in the right direction. Now, every company's a little bit different. When you implement these kinds of technologies, this really isn't about a tech and here it comes in and it's going to unify your business process for you.
It's really not even going to dictate your business process, which many other applications in the past have tried to do. What we want to do again, is incorporate that inside of the culture that you have. And even inside of your business, you might have lots of different cultures. You might have some areas that are very open to new technologies and exploring those, and you may have other areas that are less interested in those particular technologies. And so we've got to think about these things, how they work best inside of your company, and think about the whole aspect of culture and what does knowledge sharing mean inside of your organization.
So when we looked at Project Cortex, there's two sets of needs that really became apparent. And so we looked at the content understanding area, building on the great content services work that we have inside of SharePoint and Microsoft 365. There was some common objections when you think about I know content services and how that morphs into knowledge, and there's all this information that's coming inside of my company all the time. I don't have time to think about sharing knowledge because I'm too busy trying to get my day job done.
And so there's this other aspect as well of thinking about knowledge is power somehow inside of your company. And so if your subject matter expertise isn't being valued in the right way, then your power base is going to be eroded if you share that more generally inside of the organization. And so when we look at the content understanding area, it's really about recognizing that you have a subject matter expertise. That you are able to take control of these AI models and you can build what you need, with the goal of being able to extract what you need from the information that's coming in. Because if we can help you to automate the information extraction process, we can help you to extract the metadata that's most valuable to you, whether it's from a contract or a form or a business agreements coming in, then we can actually create time. We can give you time back to understand how you can contribute that knowledge into others.
We want to be able to connect all of that work that you're doing to process information that's coming in and think about how it connects to a workflow and automation across your company as well. Now, the important part here with content understanding is it's not just extracting that information that saves you time and saves you money right now, it's actually helping you to encode what you know into a model that could be reused across your company. And this is a really exciting area if you think about it. Something as simple as it might seem in our minds of, hey, I'm going to process a contract that's coming in.
I need to extract clauses from there. So you've got through various clauses that are important. We might have things like conditions or business rules that fire off those clauses. But really what you're doing when you build an AI model inside of Project Cortex is actually helping your organization understand the rationale that you're approaching, what's important to your particular subject matter expertise. So for example, in a contracts area, I might have a legal expert, I might have a procurement expert, I might have a project manager. All of those folks have expertise to be able to contribute into how that information gets processed.
And if I could encode that information into a model, then it's actually reusable in lots of different ways inside of my organization. So that's one half of you think about it from content understanding side and how we can help to transform and process information faster inside of your organization to give you some time to think about knowledge sharing, but while you're doing that, you're actually contributing into the knowledge that you have inside of your organization. Now, the second area is really around topic experiences. And so when I look at topic experiences, it's thinking about the objections folks have of, hey, knowledge is going to be going to live in this new or central silo. So if you've ever worked on a project in the past where we're going to create a knowledge base, and that's our centralized approach for something, and that actually gives us a whole world of problems that we're not asking for. Thinking about creating a new silo of this is where the knowledge lives, and it doesn't really live with everything else, that's the content of the collaboration and my day to day work, that creates a whole new barrier that you don't want either.
Now, one of the other areas that this really helps to overcome is thinking about people who are searching inside of your organization. Maybe they're searching for things. It's like, how do I even know what to search for if I don't know what it is that I'm searching for? So it's like I have to hit around until I find the right keywords, maybe something will help, maybe I'll reach out to a colleague, all of those areas coming together. So how do I help you to understand in the context of where you are? And then the other big area here is really about the folks who are kind of the new ways of working, I would say are the softer skills. It's like the folks who are really inside of your organization, who are less about sort of the distinct subject matter expertise and much more around connecting with people inside of your company.
So when you think about that, there's a whole group of people inside of your company who are really not being valued right now for their ability to connect to others. And you know this person. You know this person who knows everybody inside of your company, maybe even outside of your company.
If you've ever needed to call up somebody and say, hey, who's the best person to connect you to XYZ because I'm looking for this subject matter expertise, or do you know the latest from this person and be able to connect back that way, that person performs a really, really vital construct inside of the network. That ability to be able to connect and to give context to that connection is incredibly valuable. Right now inside of our organizations, we really don't value that skill other than just on a human network perspective. So when you think about topic experiences and the way that we're designing them inside of Project Cortex, it's really about how do we deliver knowledge in the context of where you are, is that everyday work that you're doing is contributing into that knowledge network. And then we also want to recognize your strengths uniquely.
So whether that subject matter expertise around a specific topic, whether you're really great at identifying all the different people and how things are connected together and putting context on that, or maybe you're even great about understanding how to explain different concepts together and showing connections between those. So maybe it's thinking about more like a mind map around the topics. Maybe you're looking at the topic graph type of experience and saying, how can I connect these topics together because I know there's a connection, to things that might seem disparate inside of your organization.
So when we look at topic experiences, it's really about delivering that benefit, all of the content, the collaboration that's going on every single day, and to be able to connect it into everyday work so that knowledge has developed in the context of where you are and is being contributed to as a normal part of your day job as well. Now I've got some experiences here that helps explain a little bit about what we're doing in the topic experiences area. Now you can see in here, we've got the construct around sewing a page here. I see a highlight that's automatically placed inside of the document. I'm able to click through, see that highlight. I see a topic card that's being generated automatically.
Now when you look at this topic card, it's really about understanding the synonyms on here, it's about understanding the acronyms importantly, like if you ever get an email or a message or you see something in a news post and you don't know what that acronym stands for, this is really what computers are great at. This is where AI really comes to the fore because it helps you to map a little bit of that context between what does this acronym mean and what does it really stand for inside of my organization. Now, there might be multiple ways that that acronym stands for things, and we have lots of technology to help with that. But there's also that people context in here that's really important. Now, when on this topic card, I see this topic description.
This topic description might have been generated automatically by extracting it out of a document, but then people can also go in there and help to curate the definition for this topic. And that's a really important piece of understanding that subject matter expert, the types of ways that you think about the context and how that knowledge will come into it, and then automatically we've helped to identify some of the people that are connected as part of this too. Now I look at these kinds of resources area in particular. The resources here are being security trimmed. Then you're only seeing the things that you have access to.
And these resources were really used by the AI essentially to help generate that topic inside of your organization. When you click through onto those pages, you're going to see the topic page. It helps to understand who's helped to edit the page as well.
So you'll see some recognition here. For example, the alternate names if they've been edited by somebody and you can also see who are the folks who are pinned on this topic. And so they may have specific roles, they could be curating this page, but they could also be the leads for different projects or programs inside of your company. And then you'll see the best of AI here with suggested people.
And those suggested people let us to understand who's contributing research resources that are relevant in this area. And so those suggested people are just suggestions. They're are folks that can come in and you can curate them, you can take them away if you need to, but it's just things that AI has found and is keeping the page essentially up to date for you. Now, the same thing is true over on the files and pages. So if you think about it, we have those two areas.
They're looking at pinned files and the suggested files and that same tension is here too. So there's a balance between what a curator can do as a subject matter expert, and then be able to come in and look at those suggested pages as well. Now, finally, this area around related topics is really important. And this is one where I think about the context of where you're coming from.
This is a great opportunity to showcase how things are connected inside of the organization, even if they might feel quite disparate to somebody else. And this context is really a great thing when you're getting up to speed with a particular topic. And so we've got a combination here to really look at how mind-mapping happens in the minds of subject matter experts like yourself, but then we also want to think about different ways that AI might be able to suggest connections between different topics as well. When we look at those two different areas, it's about understanding how can I put the plan together, how can I put this map together that helps me to really see the context of the specific topic.
Now with that confirmation and other areas, we want to be able to reach out and look at subject matter expert communities. So you'll see in here, we've got Yammer's questions and answers. It helps you to understand where communities are having ongoing dialogue, because knowledge is not static inside of your organization.
Oftentimes it's in a very tacit form and it's locked up in communities and conversations and emails and all kinds of other places. And so we're really trying to help to make that transparent inside of your organization. Of course, all with that security trimming that's very important. Now, one other area that I wanted to touch on is to really understand how you're connected to these specific topics. Now, it's really important for us to understand, we can surface up a connection and say, we think you're connected to this.
This is AI trying to sort of put some reasonableness around its decision here. It's going to suggest to you that there were topics that we're not sure really topics yet, and whether you're actually connected to them. So getting your feedback as part of the system is really important. And so this is one of the big areas that Cortex is really invested in. We look at Project Cortex, it's about understanding that feedback.
How can I help tune that feedback? How can I use everybody in the organization to understand how to tune that feedback so that the knowledge network will get better and more powerful over time. Now, in the context of everyday work, when we think about these different areas, it's really about understanding people. And oftentimes those connections, those mnemonics that we have inside of our heads are really through people. We don't remember how people send you a document, you don't oftentimes the title of the document, but you do remember who sent it to you. So you oftentimes can have search experiences that reflect that. Now we want to do the same thing with skills and expertise.
And so one of the experiences that we have as part of Project Cortex is this understanding of skills and how to surface them up automatically to you. Now, these skills are something that you need to manage as a person. It's an important balance here between people and AI.
You have the skills here, you have the ability to change your own profile. What we can do is proactively suggest that, hey, maybe you want to update your profile with these particular topics and skills, because these are things we think you're working on right now. And it's your choice whether you update that or not, but know that it does help you to become more discoverable inside of the organization. It helps people to understand the connections between what you're working on.
Maybe they don't realize all the things that you're working on or the things you might've worked on in the past. And that it helps them to bring context and then they can easily look up what that specific topic is within there. So that's how people in AI and thinking about how those systems really need to come together. Now there's another big tenant that I wanted to talk through with you.
I always think about good AI needs good IA. And so this is one of those balances that maybe over many years, if you've been in information management before, if you haven't been in information management before, one of those areas that you might think about is information architecture. Now with our move to the cloud, I think a lot of people have taken focus off of information architecture.
It used to be that we spend a lot of time on information architecture and there was whole a series of people who had careers based on this, whether you were librarian way back when, if you were in records management and you were working specifically on things like retention policies and those areas, and then maybe you were into content management and information management, where you're really looking at how documents got tagged inside of the organization. And some of that still goes on, but oftentimes with companies now, the reason that was promoting information architecture before was oftentimes because storage was something that we had to actually minimize as much as possible, 'cause it was costly for an organization, especially if you're wearing an on premises world. And now with the move to the cloud and just search goes over everything, there's really like the less emphasis I would say on information architecture. But this is one of those areas where I think it's really valuable to invest. And so we've done a lot of work in the modern managed metadata services.
You might've seen the new taxonomy services that we have Project Cortex and within Microsoft 365, and this was one of those areas where any investment that you make in information architecture will pay dividends in AI, helping to give it structure, helping it, to give seeding and actually promoting a much better experience. Again, it's that combination of people and AI coming together. So good AI needs good IA. If you have it, that's great. If you want to invest in it, if you're in that capability and able to do so, I really hardly recommend it. But if you're not able to do that, then I understand.
It's something that Project Cortex we know about. We understand that some folks will be able to invest in it. Maybe you just want to see what happens first and then think about the investment information architecture, totally get that.
But anything you could do can really promote that. So when I look at that area, you think about what we're doing in the content understanding area. So we're going to go back to that value proposition of thinking through a little bit about how we can help you to process the information that's coming in as part of a business process, whether it's contracts, maybe it's statements of work, all the kinds of things that you have to read in from documents and then be able to extract. Now the extraction piece here is really important because we want to help to simulate what a person would think or a team of people would think.
And so being able to extract what you need and then being able to stack it up in a column, maybe you're doing things like automatically applying retention policies. One of the cool things that we can do as part of that content understanding area is look at specific areas where maybe you're importing a clause and actually map it back to a managed metadata service. So if you have a taxonomy and that taxonomy might have a number of different areas in engineering, for example. So if I have engineering and I have skills around engineering, maybe I need to transpose those specific skills and transpose them into an engineering category. So for example, if I get a statement of work in and that's that statement of work says geophysical processing, and then it says civil engineering or something like that, I can actually transform that category using a synonym in taxonomy to help me to say, okay, the civil engineering and geophysical processing, they're both types of engineering. I just need to be able to extract engineering essentially so that I can categorize what the statement of work is or this contract is.
And it's really important to be able to do that because it helps us to understand how metadata connects into the information that we're working on every single day. Now with metadata, it's really interesting because a lot of people think about this and it's very structured and kind of librarian system. So if you're familiar with the Dewey decimal system and these cards that people used to have, when we tried to structure it and have a library of every piece of information, it's really, really difficult to achieve. And so we've made lots of improvements in this area to help you understand the corpus of knowledge inside of your company. But it's one of those areas where you should invest tactically inside of your organization. You should think about what's the important information that I really need to have some structure around what's that information architecture and how it connects together as part of all of this? Now, when you think about Project Cortex, I've talked about two halves and then a little bit about how these things come together.
When I think about Project Cortex, the developments around content understanding are really about unlocking the information that you have coming into your organization. How can we make you faster, more productive and efficient in the work that you're doing every single day, such that we hope you'll recoup some time to be able to do knowledge sharing and maybe contributing to communities inside of your organization. As of yours, if you're doing that, all of that collaboration data that's coming in every day, we really want to map that and have it understand knowledge inside of your organization as well. And so these two pieces of cortex really work very well together. It's really about understanding how I can process information as it comes in. I'm saving you time, I'm saving you money, and then I'm also helping you to understand how it contributes into the knowledge part as well.
When I look at that knowledge experience, it's really about delivering the information that I need in the context of where I am. So even if my job is processing something like contracts all day long, oftentimes there'll be an acronym or something else. Maybe there's a project name that I don't understand. And having the delivery of both of those things in the context of where I'm working is an incredibly valuable proposition. Now we recognize that not all of the information that you have lives inside of Microsoft 365, and far from it. We understand that connecting in other sources of data is incredibly important.
Now, whether that's in the topic area, when you think about explaining the Microsoft graph search connectors and how that information comes in, you can think about using information from Salesforce, from ServiceNow, other places as well, being able to bring in that content so we can actually run a knowledge index on top of all of that content, including what you have inside of Microsoft 365. Now the same is true on the content services side. When I look at content understanding, I'm going to get PDFs that are coming into my organization. I need to be able to extract information from there as well. Maybe I have a taxonomy where I need to be able to import a taxonomy that's an industry standard. I'm able to push that information into my system as well.
And all of this is contributing into Project Cortex into all of the information that you have inside of your organization. Now putting these two things together is really important. And it's one of the areas that we're facing a lot of challenge for our roadmap. In the brain, it's an interesting area because it's called the Corpus callosum. They put these things together.
It's really that connection point between these two halves of your brain. And so the way that we process information, the way we try to put some schemer and metadata around different areas. And then how do we think about transforming that into knowledge and things that I can use in my everyday work, maybe abstract and then build on top off.
So when you look at our roadmap, it reflects both investments in both of these areas, but it also reflects the investment that we have between these two hemispheres. Really thinking about how we're helping to automatically identify different entities that are important inside of your organization, learning from the entire knowledge base that you have inside of your organization and how we can put together content experiences that help you focus on a solution or a specific business process area. Because all of this knowledge and information is nothing without context. And so if we can bring together that context as part of a business process, we have the opportunity to deliver the content and the extraction that you need at the same time as being able to use it for knowledge and understanding how we can bring that together into your everyday work. So if that takes us to the next level to say, hey, what comes after all of this? It's really this focus of understanding expertise, topics, and metadata, and how they connect together very broadly inside of your company. If you think about that connection that happens across the organization, the most important element here is being able to create those connections and be able to create them dynamically depending on the kinds of needs you have inside of your organization.
We also want to help you think about how new knowledge is created, whether that's assembling new kinds of content and how we put those together, how we might adapt existing content for the needs that are new inside of your organization. And then we also want to think about how we tap into expertise inside of your company too, whether that's automating things like questions and answers, if we can extract it and find it from a document, but also being able to actually ping people, ping the right experts inside of your organization if we can't source the right answer automatically. Oftentimes there's lots of different ways that an answer can be correct. So it depends on the question, depends on the context. And so this is really where we want to put the best of both people and AI together. That information architecture construct is really important in here too.
So anything that you can do to help with metadata, to think about taxonomies and how to you use topics inside of your organization is incredibly valuable in the future. So I hope that you understand a little bit about this interesting journey that we've been on with Project Cortex, and how these pieces connect together really across your organization. If you look at those pinpoints of data, of knowledge, of information, of people, of expertise, of skills, of capabilities, really can help draw the patterns and lines and dots between these things and to make them a connected experience across your organization, that helps you to go faster. Now, I'll leave you with this quote from Sir Ken Robinson. "Curiosity is the engine of achievement," and I would encourage you all to keep being curious.
Thank you so much.
2020-12-20