 Hello and welcome, my name is Shannon Kemp and I'm the Executive Editor of Data Diversity. We'd like to thank you for joining the current installment of the Monthly Data Diversity Smart Data Webinar Series with host Adrienne Bolz. Today, Adrienne will discuss leverage the IoT to build a smart data ecosystem. Just a couple of points to get us started. Due to a large number of people that attend these sessions, he will be muted during the webinar. For questions, we'll be collecting them via the Q&A in the bottom right-hand corner of your screen. Or if you'd like to tweet, we encourage you to share how it's a question via Twitter using hashtag Smart Data. If you'd like to chat with us and with each other, we certainly encourage you to do so. Just click the chat icon in the top right for that feature. And as always, we will send a follow-up email within two business days containing links to the slides, the recording of this session, and additional information requested throughout the webinar. Now let me introduce to you our series speaker for today, Adrienne Bolz. Adrienne is an industry analyst and recovering academic providing research and advisory services for buyers, sellers, and investors in emerging technology markets. His coverage areas include cognitive computing, big data analytics, the Internet of Things, and cloud computing. Adrienne co-authored Cognitive Computing and Big Data Analytics, published by Wiley in 2015, and is currently writing a book on the business and societal impact of these emerging technologies. Adrienne earned his BA in psychology and MS in computer science from SUNY Binghamton and his PhD in computer science from Northwestern University. And with that, I will give the floor to Adrienne to get us started. Hello and welcome. Great. Thanks, Shannon. It's always fun to do these, and it seems like the year has gone by pretty quickly here as we do the last one for this year, anyway. What I'd like to do today, I'm going to go into here, is look at the impact of the IoT, the Internet of Things, the Internet of Everything, as it works with what I'm calling modern AI and cognitive computing to provide new business opportunities. So we're going to do things just a little bit differently today than some of the other webinars. We tend to focus a lot on specific technologies and apps companies and different markets. Today's going to be more about the business strategy implications of adopting an IoT first, or an IoT and cognitive approach to building your businesses or growing your businesses. So to do that, knowing that we have a pretty diverse audience, I mean, typically we get people from many countries in different roles and different industries. I want to do some level setting. I'm going to talk about conventional strategy and how businesses make plans for competing in their market. Look at a couple of common frameworks for strategy, and then I'm going to take a quick stroll down the memory lane. I have a series of slides that are labeled flashback. To show some things from when I was a semi-serious academic teaching strategy in an MBA program. Look at what's changed, what's stayed the same, but then take that into the actual implications when we have access to the internet or things as it exists today and as we expect it to grow over the next several years to decade. How does the combination between IoT and modern AI, and I'll explain in a little more detail what I mean by modern AI, rather than just conventional artificial intelligence, then I'm going to give what I consider to be five rules for creating business value with modern AI and the IoT. So let's start by taking a look at SWOT analysis or Strengths, Weaknesses, Opportunities and Threats. This is a pretty common model when you're trying to understand a business to decide where you go, where you invest your money, how you hedge your bets. When you think about it, pretty much all of business is about managing risks and identifying opportunities and then backing them with capital. So we look at the SWOT model, looking internally, we're looking at what are our own competitive advantages, do we have resources, capital data? I'm going to emphasize data here because at the end of the day, as we go from bricks and mortar to e-commerce to cognitive commerce to IoT-enabled cognitive commerce, if you will, it's really all about the data and data is a form of capital and it's the way we deal with people. So looking internally to understand how the business is today, assuming you're not starting from nothing, what are the opportunities? Can we create new products? Can we create partnerships, mergers and acquisitions? Do we have some strengths based on our geographical presence? Do we have some limitations on the flip side, the external focus? When we think of the weaknesses, in general, there's sort of a mirror image there where our competitor is stronger but we also have specific weaknesses that may be independent of competitors, maybe perhaps an overall dependency on a natural resource that scarce or could be disrupted, that's a weakness. And that also ties in with threats like disruption in the business, disruption in your supply chain, displacement by an alternative. So this is just one of the models. And what I want people to think about and have this kind of in the back of your mind as we go through today is for every new type of data that we will have access to through the IoT and for every way that we can add value to that data through cognitive computing and machine learning and sort of narrow it down a little bit with the different technologies for different opportunities. How does it fit back to this overall, are we in the right business? Do we have the right product mix and where do we go from here? So the next model that I want to spend just a couple of slides on, Michael Porter's Five Forces, which really kind of shaped the way business schools taught strategy for many years. A number of things have been updated, but I wanted to start with the basics here. And in the abstract, when we were thinking about what to present, we talked about creating an ecosystem. And this is really sort of the central diagram to the ecosystem when you think about it. Porter looked at strategy from an industry point of view. So you could draw a diagram like this and let's say you're studying insurance as an example of an industry. I know we have a lot of people from the insurance industry that participate in these webinars. So to look at your position, you have to look at the whole chain. You have your suppliers that are providing you with either physical goods or intellectual property or capital, whatever it is. You have the people that are buying your services. You have the threats, if you will, of substitute products or services. There can be some disruption. I'll talk about that in a second and how the IoT is really changing each of these relationships with the industry competitors. And then the idea of potential entrants, new companies that can get into it. If you think about the companies that you use all the time, for example, perhaps Google for search or Facebook for communications. These are companies that weren't around 20 years ago and now it's hard to imagine our daily life without them. So we're going to look, as this comes in, in terms of whatever industry you're in, your business may change industries. The popular term today is to pivot. We're going to develop something completely new. You may be a company like GE that makes everything from appliances to locomotives. So for each of the specific sub-industries, if you will, you want to do this sort of analysis. And the reason I'm starting with this is because as we look at the kinds of data that we can have today because of the Internet of Things, because of a sensor-based infrastructure, if you will, and what we can do with that data and what we can predict based on that data using predictive analytics and machine learning and some of the modern AI techniques, it changes the way we look at our ecosystem and companies that used to provide us with maybe physical goods now perhaps the value of the information that we're getting from them is actually more useful to us or higher value than the physical goods themselves because we can start to spot trends and that's what I want to look at. So within the Five Forces as a basic model, just given a couple of examples here for each of those relationships. So the threat of new entrants, how do you have a competitive advantage? Is it because you have economies of scale? You're already there and you've built up a network, for example? I mentioned Facebook. There was no Facebook when I started teaching in business schools. Now it's so entrenched it would be hard to create an alternative even though the technology isn't difficult to replicate, if you will, but there are these network effects of having the platform and there's switching costs when somebody is used to using it. So just to have you thinking in those terms, the traditional approach to strategy looks at relationships between these entities, your relationship between the supplier and your own business. But today, the relationship that matters the most in many cases is the data relationship. And so if we look at last slide in this section, when Porter was writing about this as a way of evaluating overall markets and he did a lot of work in specific industries and that was followed by some of the stuff that he may be familiar with in the high-tech industry, writing about creative disruption and looking at how individual companies tend to fail when they can get disrupted from the outside. Supporters' generic strategies were to lead by cost to differentiate products and to focus on a product niche. Back in the days when I was a product manager for a software product or a suite of software products, dealing with our company management, it was always, do you want it fast? Do you want it cheap or do you want it high quality? And you could sort of pick two out of three to optimize, but that meant that you were giving up on one of those. So do you want to be the cost leader? And if you look in here, the issue today is that so much of the value that our customers perceive, people are actually paying the bills or paying them indirectly in the case of users to whom our partners sell advertising, that cost leadership is a much more difficult thing to deal with unless you're dealing with manufacturing. So the sub-strategies create the barriers, and most of these have actually changed based on the availability of data and the movement of data. So a lot of what was talked about as strategic advantage in the past was based on the idea that you would have access to information or access to data that the buyer didn't have. And so for the next couple of slides, this is where I have the flashbacks, I decided to just leave them in the form that I used them back in 2000 when it was a PowerPoint, just a novice. The reason I'm doing it here is to show what was predictable, and this is going back now, 16 years, 17 years I started teaching this class at NYU in 1999, to look at some of the things that we knew were coming, but also to see how the structure of the different industries has changed and where we go further. Now that we've got a full-fledged Internet of Things, if you will. So back then, I was talking about the IP-enabled future, and IP was, I actually talked about IP over IP, it was intellectual property over Internet protocols. The idea was that going beyond the e-commerce, which was selling everything that we already sold, selling it online, there were things that were coming that we're seeing today that kind of changed the way we thought about business and our relationship between a customer and a business or between businesses that collaborated within a supply chain to serve customers. And here the one point on the slide is just that the example used here was proprietary software and home PC to control thermostats. And today, it's certainly much more advanced. You can go to Home Depot, you can go to online, Amazon, wherever, and you can buy thermostats that communicate without requiring as much in the way of proprietary signals. Because we're going into much more open standards. And so these things can be mixed and matched, which wasn't, it was certainly predictable in 2000, but it wasn't the norm. People differentiated by trying to hold on to that. Today it's going to be about the data, not about the proprietary piece. So one of the questions that I used to ask is where could you put this intellectual property? And this is before we talked about the Internet as the Internet of Things. But it's been obvious for a very long time that devices of all kinds would be Internet protocol enabled. So talking back then about real-time monitoring price control based on temperature and volume, where this has gone, and I have to laugh here because it says home heating, warm up your home from your pilot on the way home. We happen to have a family with teenagers. So the pilot I'm talking about today would be the Honda pilot. But when I wrote this slide, it was a Palm pilot, which was the latest in devices. So now almost any sort of device that you have is going to be enabled enough to be able to interact with the others. And that's a key distinction from where we were just a couple of decades ago. This one I just wanted to mention because back in, let's say, 16 or 17 years ago, the different companies were trying to have their own payment aggregators. And this has changed the balance of power in terms of the customer relationship. And we see things like PayPal, for example. My son was having trouble getting home from school yesterday and I had to transfer cash from my business account to my business PayPal to my personal PayPal to his PayPal. It's all relatively seamless. It's not necessarily as fast as you would want. There's still some frictions along the way. But all of these things have come to pass. So the relationship with the customer may have moved from the bank to PayPal in this case or to Uber. And now we have to start to look and say, well, if everybody has all these devices and they all talk to each other, how do we distinguish ourselves? And create some real business value. So I just laugh because this is what we're talking about today, how to make money by enabling devices and focusing on the data. And that's something that a lot of people are looking at this as, oh, it's a great new opportunity. It's something that we've been working on for a couple of decades. And what I'm going to show you today is that even though some of the terminology is new, even though some of the players are new, the concepts are really fundamentally evolutionary rather than revolutionary in terms of IOT enabling the business supply chain and the relationships between customers and businesses and between businesses within an ecosystem. So the point of this one, and some of these slides I'm going to skip over pretty quickly. If you have any interest, I can give you more context offline. And there's at least a dozen slides that will be included in the deck that Shannon sends you afterwards that I'm not going to get to today that have some of the specific technologies mentioned. But the point here is that when we looked at e-commerce as a change in the way businesses worked with their ecosystem, it was all information-based. The thing that was being exchanged was information. And all the red arrows here are just to indicate that today every one of these relationships can benefit from both IOT, some sensors, and I'll talk about that again in just a minute, but also about modern AI. And the simple way to look at it is the IOT gives us more data, and the modern AI and cognitive computing gives us more insights based on that data. And I'm going to skip that and just go to the idea of the missing middle now. One of the interesting things at the sort of the turn of the century, if you will, the e-commerce revolution was companies being disintermediated. And so instead of having multiple layers between the producer of a commodity or good and the person that was actually using the value, some of those could be disintermediated when the person could get more information. And the key here was that if the good wasn't something that was needed immediately, people would rather shop online and be able to disintermediate the retailer. Today, with drones and one-hour delivery service in Manhattan, it's going a step further. But what we want to do is look at how we get from where we were to where we are today to where we're going to be as we can anticipate the requirements based on some IOT and some modern AI technology. And here this one was just the idea of the new intermediaries. We got rid of the old ones, and of course we had to add some new ones. So the last of the flashbacks is that even nearly two decades ago, the emerging technologies that we now take for granted were out there, and that's just been refined, things like voice and speech and voice over IP. All of this is now at the point where we take it for granted. And some of the things I'm going to be talking about today as being leading edge and perhaps leading edge two or three years from now are just going to be as archaic in most of your minds as voice over IP. So let's look at how does the IOT, either of things, actually change strategy or how we look at strategy? It all starts with sensors and ends with insights. So I did a webinar on this a couple of months ago on just the IOT and the technologies. So we're not going to get into a lot of detail on the actual technologies, but just the key things that you need to think about in terms of strategy. So sensors, any kind of a device that detects the presence of a signal and reports a signal. It's event driven. You get something based. It is sensing, just like a human. You sense a change in your environment. And the sensor itself may be stationary or mobile. So you may have a sensor that travels with you or that travels in a vehicle. You can have your car certainly has lots of sensors. You may have sensors in your phone. It will get into more detail there. But there are also sensors that are stationary and are looking for events in that area. So a sensor that detects people or cars passing through an intersection. You could have just the one sensor, let's say at the intersection that's tracking visually cars going by, something, an easy pass lean sensor, for example, where you could have, in the case of the easy pass, you've got a sensor in your car that's communicating with the sensors as you're passing by. Stationary and mobile sensors. And this is building the huge network. That leads me to another term that we're going to have to draw out a lot, which is stream or streaming data. Any idea here is if the data is always moving and we've got great quantities, we're talking about big data. In general, I'm more concerned with fast data than the volume. The question is, are we going to divert that flow to be able to do something with it? Do we need to capture, analyze and then release the data? Or can we do it as it passes through? And this is a diagram that I've used once before. Just to get you thinking about where the analysis can happen, are we aggregating the data and then doing the analysis or with the modern infrastructure for streaming analytics, we can basically analyze data as it's passing through. So here the data is flowing on the edges, but the queries can be anywhere. Something that's as opposed to the data comes in, it's analyzed, and then it goes out. We have things now, it's working with a company recently that their client does real-time advertising decisions. And they're talking about the time from people bidding on advertising space on video sources to the actual ad being placed is so small and the transaction is happening so fast. That the data can't be captured and then run through. Okay. So there are a number of these platforms. We're not going to go into the specifics of any one of them. Just put it up here to show that you do have kind of the usual suspects, if you will, the Cisco's and Intel's and IBM's of the world. But you've also got companies that are known for building infrastructure on open platforms like Red Hat, as an example, and C3, which started out as very focused on energy monitoring and is migrated to build a general-purpose IoT platform. There are references to some of these in the supplementary slides that you'll get with the tech. Now, so that's kind of the quick overview of the pieces of the IoT that are relevant to changing your strategy. Here, I want to get into the contribution of modern AI. The reason for this slide is to show where I make the distinction. So AI for the last 50 or 60 years has been looking at opportunities to create machines that exhibit some of the characteristics of human intelligence, if you will. So it's artificial intelligence. Today, it's often AI is being used for augmented intelligence. But whether it's relatively intelligent image processing to natural language processing to gaining insights using logical reasoning, that's all pretty standard fare for AI. When we talk about it today, though, the area of machine learning has almost become synonymous with AI. But machine learning is really, I show it as something that's separate, but there's an overlap. Others have shown it as a subset of conventional AI. I don't really believe that or subset of AI. I don't believe it is. When we look at machine learning, we're really looking at systems that can improve their performance on some task without reprogramming. So it's based on the data that machine learns or the system learns based on experience with data. And it's the most typical way it's done today is with models of neural networks that are heavily statistically or probabilistic systems that learn from identifying patterns. And the third component that I include when I talk about modern AI is big data because many of the things that we had a fundamental understanding for what we wanted to do 20 or 30 years ago. We just didn't have the data or the access to the data or the ability to manage data fast enough to use it in real problem solving. And today we do. So those are the three components that I would include in modern AI. And where I just make one more distinction in cognitive computing, we're looking at sort of the subset of modern AI that is focused on four things. Understanding, which is a whole area by itself, learning, which would include machine learning. Reasoning, which can be a combination of top, down, bottom up, inductive, deductive, abductive reasoning. And then I had a fourth component. Many people ignore this and just look at the first three. But that's planning because if you're dealing with a cognitive system and you try to do something that's going to have general intelligence, if you will, as people, I would say a kind of a higher order of intelligence than others in the animal kingdom. We look at things. We have a model of the world that we would deal with and we plan ahead based on that. We're not always in reactive mode. So if you have those four things, and then in this diagram, you can see the top half is dealing with human interaction with a cognitive system. The bottom half is machines. And it's the bottom half that's most relevant for the rest of the discussion today, which is when you have machine input coming from sensors, that's where the IoT fits in here. Got to get through whatever big data management systems that we need to handle the data and put it in a form that the cognitive system can handle it. And that's going to be, you know, how to spark all the open stack, open source stack of tools to get it in there. You can also have data coming in from the bottom, if you will, and going out through the top. So you get all this data coming in and then the system is going to analyze it and produce narratives or visualization reports for the people. I can go the other way. You can have input from people that ends up going to the machines. But that's kind of the context. So now we've got sensors and systems that are producing information. The value of that information is being identified or enhanced by modern AI. And now I want to spend a few minutes looking at where does this fit with the business models and what are the opportunities. So the combined impact, if you will, of the IoT and modern AI, three areas that I want to look at and we'll go right into the five rules after this. So when everything is connected, what happens is you have new sources of data. So there's data from my refrigerator. There's data from my phone. There's data from my car, within my car. Not my car, because I happen to drive a 40-year-old car, but that's a whole different story. But if you're driving a modern high-end car, there's a lot of systems in there that talk to each other, but they also can communicate via their infrastructure to the rest of the world. So you can have a system within your car that internally is broadcasting your speed, or you can have a sensor externally, like a police radar system. That's a sensor that's capturing your speed. But you can also have within that your engine is transmitting or could be transmitting or could be batched up for later. Information about its performance, about the last time the oil was changed, about temperatures. You could have individual tire sensors that are providing information. And some of this is going to be issues in terms of privacy and security, but all of these have the potential. Basically any physical device has the potential to be sensor-enabled and communicate. The question is, just because we can, do we want to? For the moment, I'm going to assume that we do, because it's a tech audience. So intelligence and the sensors, the intelligence can be distributed. It can be something that's local to the device, so we can have a machine learning system that's within your automobile that's looking at it. And then it may or may not communicate synchronously or asynchronously with your car's manufacturer. Let's say you're driving a Volkswagen and you take your car in for service, give them the key, and based on the key, they can get information about what you've been doing since the last time you were there. That's one approach. The other approach is you can have a system that is communicating in real time and you're getting that information is being used to either sell you new services, to predict failure, et cetera. So it may not be that important if you're driving your Volkswagen to work from back, but if you're driving your Boeing plane across the ocean, you probably don't want to wait until you get halfway across the ocean to know that the system is predicting an engine failure. And if we look at the amount of data that's being produced in real time on jet engines, Formula One cars, it's one of these things where putting them together, we have all sorts of new opportunities. And what we're going to see in the next couple of slides is that all of this translates to new opportunities, even with a much more mundane world of retail, of professional services. But it all starts with technologies that were developed in large part for very sophisticated requirements. So here's the five rules and I'm going to go through the examples of each. And although I'm putting this as all about customers, you could substitute for any of these what do the other entities in your supply chain want? What do they need? How can you work together? But I'm going to focus on it from a company to a customer because I think that's a simpler way to start today. So we're going to look at each of these, know your customers, know what they want, know when they want to change, know where they are and anticipate opportunities. But for each one, I want you to think as we go through it and say, do you have or can you capture streaming data that's going to change your relationship with your customer as you're trying to know them? Do you have information about your product, your supply chain or the customer that can improve your business performance? And if not, where can you get it? I'm going to look at some possibilities there. And finally, can you create value from new analysis of data that's available to everybody? Because one of the competitive advantages way back in the SWOT analysis, your strength, could be the proprietary data. But there's a lot of data out there that's either free or freely available, everything from news feeds to current weather to historical weather to weather prediction to economic analysis. That is certainly valuable. I mean, weather data is very valuable for a lot of industries. That's why IBM got involved in buying the weather company. But in general, something that anybody can buy, then there's going to be one or two companies that are going to get good benefit out of selling that information. But the real value comes and what do you do with it if you can combine it with other data sources or combine it with your proprietary algorithms that allow you to get better insights from it. So let's look at these five from it. So knowing your customer, and here's where I want to just kind of give this model for doing the analysis. You need to be able to track financial services, particularly in banking, there's some know your customer regulations. They're based on the idea that you have to know who's putting money through your system, if you will, and what's the source of funds, things like that. What I'm talking about here for know your customers is in order to be able to provide goods and services that will maximize the business relationship with a customer, which is generally profit oriented. You can collect demographics, standard information, your age, ethnicity, zip code and all that good stuff, psychographics, behavior. But today, those things don't tend to change very quickly from moment to moment, for example. Your age changes predictably if we have you in a database and we have your date of birth, we can calculate your age at any given time. Whether there's a life changing event, marriage, divorce, having a baby, etc., all of those things we can track. But what we're getting to today is being able to use sensors, use emotionally detecting, if you will, technology that's going to change how we interact with that customer right now. If you're looking at a customer relationship and saying, okay, somebody has just bought a car for me or a refrigerator, I know that they're going to need another one. If it's a car and they've taken a three year lease, I know that they're going to anticipate another car in three years. If they bought the car, I don't necessarily know when they're likely to replace it. But if I have access to all this other information, and if I have either based on, here I show that, you know, how is this information captured? If I have information about them based on their use of a cell phone, their credit history, their travel history, I can start to put all of that together. And these profiles that we're building on people are either from opt-in information at the time to become your customer, or it can be generated based on observation, or we can be purchasing it. We can derive this based on using the data and then using technologies like the analytics and machine learning to derive new information about that person. And so it's really kind of this combination that creates more value to you as an enterprise about your customer. But the next step in this, if you will, and this isn't really a hierarchy, but you would need to have that information before you could get into looking at what your customers want. You may actually get an indication of what your customer wants by asking them. And certainly, you know, surveys have been around for a very long time. Today, the general approach is to try and figure out what they may want based on what they've already done. And in this case, we're still dealing with things like natural language processing. And in general, I'm talking NLP, and I don't further qualify it. I'm talking about natural language understanding. So we may be looking at what they say they want, and that could be – this is less sensor-oriented. It's more looking at things like social media, for example, or looking at natural language understanding. Here, we're also getting into the same sort of profile that we're building. But if we're trying to understand what they want, part of it today is by looking at what their peers want. And I'm sure we're all familiar with getting recommendations. You know, customers who bought X also bought Y. But behind the scenes, the combination of looking at what's been expressed as an intent and the actual behavior – behaviors is a more accurate predictor of what they're going to want in the future. So now we may be looking at this together to see what they want. The next thing is to take all of that and add information when the wants change, and the wants change based on changes in the profile. This is a famous case about six years ago, I think it was, where Target started sending out special offers to women based on their prediction that they were pregnant. It was a case where these offers went to a teenage girl and her father got all upset and started going after Target and it turned out that Target was right. It was just perhaps not a very smart thing to just send these out without understanding the context of who it was going to. And that was based on different things, different behaviors, different buying patterns that indicated with high confidence that this young woman happened to be pregnant. But you get that information, you can use that information much more effectively because you have all this peer information and that allows you to make those correlations. So that's when the wants change. And that can be with or without the IoT, but with the IoT as we have more devices, we have more data, and as we have more data, all other things being equal, more data allows us to get more refinement with our machine learning with our cognitive systems. So now the next one, which to me is particularly fascinating is knowing where the customers are or where they will be and where they are is pretty easy based on a couple of technologies, in particular if you're using, if you have someone who has opted in with a mobile phone app, you have access to a lot of location information. I'm going to show some of the sensors that will enable you to do that. But if you go to a mall and you've got systems there that are looking at your phone, whether it's a Wi-Fi, you know, if you've ever logged into Wi-Fi at Starbucks or Panera or one of those, you've basically given them information about yourself that allows them to know any time that you're in the vicinity. But going to the next level, we have devices that will be sensor oriented for facial recognition. And if you take that from facial recognition and voice recognition and then the emotion analysis that I talked about a little earlier, you start to combine it. Not only do you know where they are, you know what they're experiencing, what their motivations are to some extent. But then you can start to build this profile. And based on that, it's one thing to know where they are right now, and that's important. You have things like Cisco, for example, has their enterprise mobility services platform. And so if you're a retailer, you could use that, give your customers the ability to get free Wi-Fi every time they're in the store. They come in and you have information. Now you have the rest of the profile. You can start to make special offers. You can know, based on the sensors, you know where they are, you know where they are within the store. It's just like, you know, the big thing 15 or 20 years ago, early in the days of e-commerce, was tracking your virtual progress through an online site and then looking for things like abandoned shopping carts. I got an email last night from Home Depot saying, well, we noticed you put something in your cart. You didn't buy it. What they hadn't figured out was that I actually went to the store and bought it. So, you know, you have online and offline that are not always coordinated. But here, if you have enough information about the person in their profile, you have enough information about their behavior, and you have these peer groups and you start to look at that, we can now start to predict where they're going to be, verify it, but prepare offers that are more complex than just someone shows up at the store and you see that they've been there for 20 minutes in the hardware section and now they start to leave and it's a geofenced, it's proximity-based. Do you, you don't wait until they get home to make an offer, you know where they are and you know which direction they're heading and that's all sensor-based. Now we can start to say, you know what, I'm going to call your bluff. You're looking for something. I'm going to give you a 20% off coupon if you buy it now before you leave the store. So, that's where it's heading. And now the last one here is anticipating opportunities. And this again is looking at an opportunity with one person largely based on a change in some of your understanding. So you have an opportunity with one person or one category of person, one category of customer based on changes around them. So you can figure based on the analytics if somebody is an early adopter, a vast follower, whatever category they fit into for a particular type of product or service, and now you can create a new opportunity. Going back to that slide I had for my year 2000 strategy class at Stern, one of the things I talked about then that wasn't really feasible but it is now is you've got a Coca-Cola machine. You know what the inventory is. You know what the going rate is for a can of Coke in your area. But you create an opportunity if you can now look at that and go, you know what, I have six cans left and I know that the temperature is rising and that when the temperature rises, a certain class of people are going to pay more. You change the opportunity for a return on that good. And maybe it's not a big difference if you're talking about six cans of Coke, but if you're talking about something at a retail chain, you've got a thousand stores and now you know that there's some event that's happening or that people like this person have done it. It's all about understanding the implications of the data and where it's going in the future. So to bring it back to reality, if we just take a sort of the givens that we want to be able to get more information. Sensors are an important part of the way we're going to get that information. We want to make smart choices based on it. That's going to be based on modern AI techniques and our learning and our reason. What data do your customers already produce? What do you have? And this is just a list of the sensors that are in the iPhone today. So you've got an accelerometer. You've got one for ambient light, a barometer, location. So I already mentioned what you can do with geolocation, whether it's Wi-Fi or cellular, you can start to push offers to them. Or maybe it's more passive. You just track for information. You start to look for patterns. So okay. If you're a sports team, for example, one of the NBA teams has started to look at season ticket holders and they scan the sites to see when a season ticket holder is selling their seats. And how often they're actually using themselves. Well, now you can look at this and say, okay, based on prior behavior, this person is not likely to renew because they're starting to sell a higher percentage of their seats. But maybe by using geolocation within the phone and the barometer as an example, you can start to look for what are the correlations. And then start to derive what are the conditions under which they're likely to, let's say, an MLB season ticket holder where you've got 81 games for your home team. Well, if you're selling half of those, maybe you would rather have – you've bought two seats, but maybe you would rather have four seats of the games that you go to. You can start to create these special offers based on looking at all this data. And again, this can be stuff that you're getting directly if you have an app. And I would venture to guess that many of you have like a flashlight app on your phone. You probably didn't read the fine print when you downloaded that free app. Well, the reason it's free is because they're getting all this information that they can turn into a profile and use for other purposes. So all of this is already out there. It's just a question of how do you get access to it? How do you leverage it in a smart way? Here's another one. I've used this one before. This is a site called thingful.net, and they're attempting to build a search engine for the Internet of Things. And so in this particular one, I did the snapshot when I was giving a talk at the Seaport Hotel in Boston. I just went online and searched, and it showed all the devices, the IoT devices that were registered or that were detected by – and in this case, one of them that's giving down information was the bike rack at the hotel where I was staying. And I could see how many spaces there are on the bike rack and how many were open. Now, if you start to combine that with knowledge that you have – maybe you have actual knowledge that your customer is a bicyclist, or maybe you can infer that based on the fact that they're averaging eight miles an hour instead of walking speed or driving speed between two points, you can start to create value for them by combining just this data and the interpretation, the analysis, the predictive power of modern AI to really look at what are they going to want next. So that's how those two fit together. So there are all these sources – and this is just taking it a little bit further out, and I think this is actually my hometown here. So the issue is there's no shortage of data. It's adding value to it. But depending on your relationship, you really need to start looking at what sorts of sensors can you add to the systems that you already have. And at each step, just as we would try and optimize the supply chain going between builder and logistics and distribution, at each step, look for places where data is being created but not captured. Maybe you need to be in that capture business. Look for opportunities to leverage the data that's already out there. There's a lot of sensor-based information that's available – freely available and available free from most municipal governments today. The question is just how do we look at it and how do we look at it differently so that we can create value where other people don't see it? And I close by just saying that my own interest in this and where I'm going is we're actually creating some case studies and use-cake archetypes for industries right now. We'll broaden it. But insurance, banking and financial services, healthcare and pharma. And if you're in those industries and you're interested in getting more information or participating in the research, talking about what you're doing, I would love to communicate with you and happy to send you some of our results. So we'll wrap it up with this is How You Can Reach Me, and this is the list of topics we have for next year. And I'll hand it over to Shannon to see if there are any questions that we can address now. Or if we don't get to you today, give them to Shannon and I'll get to you when she sends them to me. So thanks, everyone. Adrienne, thank you so much for this great presentation. And like you said earlier, I can't believe it's December already, the last one in the series for the year. So just a reminder to ask the most common question that we always receive is I will be sending a follow-up email by end of day Monday with links to the slides and links to the recording of the session. And feel free to type in any questions you have in the bottom right-hand corner in the Q&A section. Everyone's kind of quiet today. I think the holidays are upon us. And then look forward to the series next year, Adrienne, where you will be continuing it on the second Thursday of every month. And so as you mentioned, there are modern AI and cognitive computing, boundaries, and opportunities. And likewise, you'll be speaking at Smart Data, our Smart Data conference in January in San Francisco. Yes. Very excited about that. What are you talking about there? You know what? I have, as I recall, I've got two tutorials and a couple of sessions. So I don't actually have all that at hand. I know I'm doing a panel. I'm doing a couple of tutorials. There'll be something on AI in the future of work. There's definitely something on natural language processing. Okay. And give me the URL. I will, yes. We can all see what I'm doing. I will indeed. And there's a question here about a discount. And yes, indeed, we do have a discount, especially for our webinar attendees. So if you use, let me just pull it up here and make sure it's not plural. Use the webinar as a discount code. We will receive additional money off. I'm very excited about that. And then we do have a question coming in here, Adrienne. Could you explain what you meant by modern AI and how it is different from conventional AI? Sure. One of the problems that we have in this field in particular is people throw around terms and everybody nods like they know what we're talking about. For me specifically, there are three things when I talk about modern AI. It's all of the technologies for understanding, learning, reasoning, and planning, which are kind of in the core of that diagram. But what makes it different from the way we approached it in the past is the availability of two things. One is big data and the other is machine learning. And it's not to say that there wasn't a lot of data before or that there wasn't experimentation in big learning. But I noted earlier in the year that when I was getting my start in AI and writing my very first natural language processing app, I don't want to say when that was. The way we approached things was largely rule-based. It was largely statistically-based. It was, in some cases, an attempt to mimic, to do biomimicry for the way we process information. Machine learning, the way it's done today, which is based on a model of neural networks and deep learning, which is a multi-level neural network approach, was really not invoke at all. And in fact, the first dozen AI textbooks that I dug out of the basement, not one of them mentioned machine learning. And now it's almost synonymous. It shouldn't be synonymous. It's an approach that's very useful. But if I put it all together, so to me, modern AI is, all the AI topics, from vision to speech to reasoning, plus machine learning and big data. So it's those three things. Perfect. And that brings us right to the top of the hour. Adrian, thank you so much for this presentation and for the series this year. I really look forward to next year. We've got some great topics like you've got posted right there on the screen. And thanks to all of our attendees for being so engaged in everything we do. We just certainly appreciate it and look forward to hopefully you will join us next January for modern AI and cognitive computing and likewise at the Smart Data Conference, where you can meet Adrian in person. So I hope everyone has a great day and thank you very much. Thanks, Adrian. Thank you. Take care, folks. Cheers. Happy holidays.