 Hello, everyone, and welcome to our next EDW session called Data Strategy, Creating a Breakthrough Inside Organization in a Data-Rage World, which will be presented by Jason Perkins, the Head of Data and Analytics Architecture at BT. Just a note, due to a conflict, this session has been pre-recorded. All audience members are muted during these sessions. Though the speaker is unable to attend, please do feel free to submit questions in the Q&A window on the right of the screen to be reviewed at a later time. Please note that there is a link formed at the bottom of the page for this session survey. This is where you can submit session feedback, and we encourage you to do so. So let's begin our presentation now. Thank you and welcome. Hi there, and welcome to Enterprise Data World. My name is Jason Perkins, and I'm the Head of Data and Analytics at BT, the largest fixed mobile digital communication provider in the UK. Here today to talk to you about creating a breakthrough insight organization in a data-rich world. No matter where we are today, whether we're at home, whether we're in the office, whether we're at the shops, whether we're out with friends, whether we're out in the great countryside, out seeing our world, we're connected everywhere we are today. And really, the pandemic has brought that home, as the importance of being connected and why it's fundamental to the way we live our lives today. I see that in my teenage boys. It is up there in the massive hierarchy of needs next to water, connectivity. It is all important, whether it's fixed or it's mobile or it's Wi-Fi, that they want to be connected all the time. This is a trend that's just continuing to exponentially grow. Today on average, 10 devices roughly in the average home in the UK and Europe. But we know we're predicting over the next five years that that's going to grow to 200 devices in the average home, a real massive growth. And that trend will just continue as increasingly more and more devices become smarter and connected. And really what that brings is a wealth of data to us to help us actually understand what's happening in this world, a real digitization of the real world. And that will help us deliver brilliant experiences for our customers and our colleagues. However, that does require that we think somewhat differently about how we treat that data in this data rich world. And that's what I'm going to talk to you more about today. So in order to really unlock the insight from all this data and all these connected devices, we really need to connect the dots. What you can see in this diagram is we're seeing an explosion in the number of channels at the top there. So you've got the physical channels, so your retail stores, your account managers, and the pop-up areas, your traditional brick-and-mortar stores. You've got then also your online stores. You can see your websites, your affiliate, your partners. And increasingly seeing also through social as well. So your Facebook, your LinkedIn, just a richness in terms of the number of different channels you can actually engage with organizations on today. And fundamentally, we want to connect across all there. So we're able to provide relevant conversations with your customers, relevant experiences, and across those better channels. And then we've got in the middle of the year, the different types. So you've got your business-to-consumer type relationships, or whether that's an individual or a family or a household, into your B2B market, whether that's a man-in-a-van or a mom-on-pop store, through to your corporate, your public services, so your big multinational banks and other organizations, your consumer package goods, international organizations, through to your wholesalers, and then up into your multinational companies. So we operate in over 180 countries around the world trying to connect for our major customers a seamless experience. And then we've got the different connections we have as well. So what we're seeing, a real transformation in the type of traffic that we're bringing over these connected networks these days. I remember back in the day when it was all about counting minutes and it was about voice. And the more people were speaking on the phone, then experience was great, revenue grew. It was quite simple times. It didn't seem that at the time, but today where you've got a wealth of different types of data, data makes up over 85% of the traffic on the networks these days, and that percentage keeps growing year on year. And what we're seeing there is you've got your media. So lots of people consuming on-demand TV and series binge watching, being one of the great change in the TV habits during my lifetime. As well as the connected, the mobility, so you're connected anywhere, whether the gaming, whether it's mobile or again more households, through to the different connected devices we're seeing. So the great growth, the coming growth of internet things, so whether that's connected your drones or your connected toothbrush, your connected multimedia systems, I've got smart TV, we've seen massive growth in those sort of areas. And then also into the business market. So we're seeing your smart buildings, smart data centers, smart shopping malls. We're seeing digitization of pretty much everything. And what we're trying to do here is connect all these different devices and these channels and these customers to really understand what's going on here, trying to create that digital twin of the physical world so we can really optimize it and actually provide experiences to our customers and to our colleagues that really build on that. And that's one of the real challenges when we think about scale, we've talked about the growth in the number of connected devices. We've also got massive growth in the amount of data being carried over these networks. So time seven growth over the next two to three years, that's what we're predicting across fixed mobile and Wi-Fi. And we can take some, what we're looking for is inspiration around how, what sort of patterns work? And then we can actually look to the real world. How is the real world managing these complex adaptive systems? So we can look at nature, so you look at your organic ecosystems. So the animal kingdom, whether it's nature, with the great forests of the world, there's real patterns there around how the world's managed to scale that, as well as cities when the great innovators of the innovations of the last 200 years, how people have come together and produced more value by coming into the great cities. We've also got the social platforms of the last 20 years where that's really creating new digital connections with people. So people can have bigger networks of people they connect to than they could in the physical world. And then you've got your organizations, how organized that continues to evolve over the last 200 years. Organizations, again, how do they scale organizations? And as well as the communication industry, how do we manage complex networks and manage that growth, both the growth and the scale? And why is that important? Well, what we see is there's certain laws and those laws have certain properties and those properties can be quite useful if we think about them in a data rich world. So if you look at the physical world, you look at cities, what we can see is that wealth creation, crime innovation actually superscale in cities. That's why cities have been a great growth engine for us. So the more people have got, the more wealth they create, so it's not a linear scale of one for one, it's actually above one. Wealth creation, innovation truly scales in these mega cities. And that's why we're seeing cities continue to grow. Obviously there's some downsides to that. We also need to think about like crime that tends to also superscale. And clearly that's not a desirable property. Pollution was also another thing you need to look at. So we need to look at these and think, what can we do to manage these scales? When we look at nature, what we see is actually animals, we share a lot of common property. So roughly in the animal world, 1.5 billion heartbeats in a lifetime of an animal, doesn't matter whether you're a whale or you're a shrew, it's roughly even. The bigger the animal, the slower the heartbeats, typically have a longer lifespan, smaller animals typically have much shorter lifespan, but their heartbeats much, much quicker. There's clearly some universal laws that are regulating this. Humans probably being, we're the one lucky species managed to defy in the last 100 years, this sort of rule, we've managed to extend our life beyond what was true for the previous thousands of years. Also we're seeing in social networks how news can scale, especially if there's a lot of talk about fake news at the moment. And how that scales, again, we can see that super scales often, especially the more tantalising it is, tends to get around. Again, we can look at that, not that we want to promote that fake news, but it's more interesting than the pattern and how is that actually scaling? How is that spreading across a network? And then we also look at urbanisation, the captures it grows. So what we see here is, yeah, these are where we're scaling more linearly. So what we see is actually a much more, actually sub-scale is actually lower than linear. So that's a, instead of one for one, it's below one. So every one person you add, actually, it gets more efficient. So we're more energy efficient, for example, in the big cities than we are if we're actually out, spread out more across an area. Now there's some efficiency that comes in. So that's the columns of scale type rules. Again, very interesting when we think about data networks. How do we unlock all that connected data in our organisations? And there's some great writers on this, Philip Teplow, I can certainly call him out. Also, Jeffrey West has written a brilliant book on scale. And they talk about, as well as some greats of power of networks and other great book around this sort of topic. And what we see here is there's real universal laws. And I've called out three of the big ones here. One is, in terms of these complex data systems, you need to think about how you space for, how do you cover the entire ecosystem? That's the fundamental, we'll talk about what that means for data analytics in a second. We also need to think about terminal units. So typically what we see is a largely identical. So whether that's buildings, which have similar plug sockets all over the building, how do you scale that? By having it identical, as well as in nature. If you look at capillaries, largely similar across all animals. And then also what we see in these large adaptive systems is they're self-optimising. They evolve, they adapt, some important principles there for us to consider. So let's just think about that from, what does that mean for our data and analytics organisation and our architecture and our strategy? Well, when we think about space building, that's what we truly mean by democratising. We need to fill the whole of the organisation with data and insight. No, and that's from the C suite all the way to the front line and beyond. In these extended organisations, ecosystems, what we're seeing is partners, out in your partners, out into your customers, you need to really unlock it. So it's not about decentralisation or centralisation, it's about redistributing that across the whole organisation. And what's some other patterns to do that? Well, this terminal units idea is quite powerful because what we can think about there is actually, if we have common ways of doing this, then it's much easier to democratise this across an organisation. So common data platforms and common services that people can engage with, whether that's APIs, whether that's AI algorithms, whether that's data visualisation tooling, whether that's data engineering tooling. Common ways of doing that allows us much easier to connect those pipes, to connect that insight across an organisation and really give you the cumulative effects. That's where you're building up the cumulative effects and these network effects of connecting all these things together. So one-on-one doesn't equal two, it equals three, four, five. And then also we're thinking, how do we optimise this system that we're producing with all this data in it? Now, how do we do that? Well, self-optimising, what we need to, it's about experimentation here, this is what we want to take and a test-to-learn approach so we can actually evolve the system based on insights. So we'll talk a bit more about that later, but yeah, generally we want to drive lots of experimentation across the whole organisation as well. So lots of experiments and test and learning. Not everything will work. We need to test and learn as that stuff, as we prove that actually there's evidence that this is heading the right direction. I'm an architect so we have to cover a bit of architecture so let me go into that now. So when we talk about data strategy, there's a number of pillars to data strategy. And the one I'm going to focus on today is really the middle pillars here around architecture as an architect and the roadmap. I will also touch on some of the other pillars, but in 40 minutes I'd rather dive into one of the pillars than give a broad brush view across the different pillars. But what I guess I'm saying here is architecture is fundamental to a data strategy, but there are other pillars you need to consider to really truly be, if you want to build that data-rich organisation that we talked about at the start there, you need to think about your vision and principles so that everyone's on the same page. You need to think about what the business outcomes you want to deliver across the whole organisation. You need to think about how you manage your assets, those shared assets across the organisation. What are the most valuable assets in your organisation and how are you going to unlock them across your organisation? How are you going to unlock the actual skills? How are you going to increase and make data part of everyone's job as part of your data culture? And then also think about the ways you work. How are you governing it? How should you operate in what your funding approaches are to this? So all important things. And as we said, self-optimising, your strategy needs to be tested and adjusted. It's not a one-and-done thing. It's very much an evolutionary thing. And you should prove that through these lighthouse cases that actually evolve as you go through your journey. So here's a data analysis reference architecture. So in terms of the data reference architecture, then I like to think of there's two major steps. One is you have to provide the right foundations. So we'll start at the bottom. We need to build the right foundations. So we'll start at the bottom. Data marketplace, if you're truly going to be an insight-driven company, you need to manage and govern and assure your data assets. So you need to know what data assets you've got. You need to manage it as an asset. Manage it like you manage your capital. Manage it like you manage your people or your assets and your other assets, your physical assets. That's the way we need to manage data. And hence, that's why we've got the pillar on the right in the gray here, enterprise information management. These are all the common capabilities you need. And this is fundamental to the whole strategy. You need to know what data you've got, what's it being used for, where it's being used. Now, these are the fundamentals. Also, you need to democratize this information. It isn't a centralization. It needs to be adaptive. It needs to be democratized so that people can understand. So this is where the marketplace idea comes in. We need to treat it like a marketplace so people can get the wisdom of crowds. We can crowdsource this information so that people can find, they can understand, and they can collaborate around our data. So that's the fundamental first pillar in our enterprise data foundation. The second one is really we need a data fabric. So we need to be able to unlock and mobilize our data. So, and that requires a rich array. We focus on an event-based architecture because what we see in this growth in connectivity is a massive growth. The exponential growth is an event-based data. So that's those connected devices. It's data, and that data needs to be streamed. The sort of use cases we're seeing is a lot of low-latency use cases where people need rich data in the moment. That doesn't mean you still won't have some data that moves at batch, especially where you want to group and you want to aggregate your data and you want to combine your data. Those will also be interesting patterns as well as virtualizing your data as well. So, you know, not always collecting it all to the center and a streaming at all. As well as the other big trend we've seen is the citizen data and the citizen data integrator. So what we're seeing is we still need the expert data engineers, but we also need to grow as part of that. We need to give the data preparation data wrangling tools to the business so they can actually go off and do some experimentation. And the final pillar in our data foundation is really that agile data platform sitting at the heart of our architecture and essential here. This is, you know, what we're seeing, you know, what we're doing is moving our center of gravity we should talk more about in a minute, to the cloud. That's really to unlock and enable us to unify that data. So what we're seeing is not only connectivity but convergence. People want to be able to see across. I want to be able to see the customer's journey across these different connected devices, across these different channels so we can respond better to their needs. And hence why the agile data platform is important. And this needs to, you know, it needs to be able to cope with self-service. And one of the things about self-service is pretty unpredictable. So elastic capability is fundamental to that. So you can respond to those elastic demands of the business. You know, as you democratize the querying of the data, the ability to do the analysis, you need to have an elastic platform sitting behind there that can respond to those needs. And then we've got the other, then, is how do you grow? How do you grow the insight and analytics consumption on top of there? So here we've got three main strategies. I'll start at the top. One is promoting self-service. So this is really giving analysis freedom out across the business to your analysts and your, to your analysts and your consumers so that they can ask new and better questions around the data in the moment, in those meetings. So moving away from stale curated data to, you know, which has been polished, you know, over a week before you go to the meeting, actually being able to, you know, navigate and interrogate that in the meeting. So you can actually, you know, so people don't have to go away and find out what really happened. You know, that enables a much better business conversation around what's actually happening in the business. Number two is we want to provide an app store. So in terms of these insight applications, we really want to be able to share and collaborate around that. So we think it's connected, the connected world and the conversion world. We want to be able to, so, you know, what's happening around mobile experience? What's happening around the, you know, on TV, TV viewing? What's the latest hot contents and programs and series? You know, these will be on an app store that'll be shared. Now, analytical content in the apps, this will be a mix of different types of content. So it could be a dashboard, it could be a report, could be APIs, could be algorithms. We really want to enable that app store that allows you to collaborate. And people can like and they can promote and you know, and you'll have, you know, you know, we want to create a social collaboration experience alongside. So this is actually complimenting the data marketplace I talked about earlier, really. These are the insight applications that are based on top of that data marketplace. We see these as really two sides of the same coin. And finally, what we have there is the cognitive services. You know, a very fast-changing area here, cognitive. You know, we call this cognitive because, you know, part of my responsibilities is around data and analytics. And I see AI as something that also touches on digital. You know, so cognitive is really for me as the brain. You know, you're trying to match or improve on human decision-making, you know, so the brain. But, you know, AI can also cover the other senses or like your eyes, your hands, your smell, you know, robotics, all that sort of stuff. That's not a core focus for me. It's, I mean, one of the other architecture main. So I support that by providing the brains, but not necessarily all of the part of it. So really this is about the machine learning, deep learning, data science capabilities as well as decision management. So here we see, I'm going to talk more about later, but what we see is an explosion in the opportunities to augment and automate a lot of these decisions. So, you know, as we move into this digital world, a lot more decisions need to be made in the moment. And, you know, it's just not possible for humans to do that in the moment. However, they are helping augment. They're the ones who are in control of this cognitive services. So they're the ones who are setting the business rules through self-service decision management and also plugging in the algorithms. So, you know, plugging in the algorithms and actually enabling that experimentation and that competition between those algorithms as part of these frameworks. And roughly, you know, we've grown to over a million decisions a day today, automated and augmented decisions a day. And I see, we see that growth continue to grow out to 5 million and 50 million by 2025 as this connectivity explosion continues. And as we, you know, 5G and self-healing networks and ultra-fast self-healing networks, there's a lot more automation augmentation to come. Okay, on to the next slide. So let's dive a bit deeper into that cloud, the agile data platform that we talked about, you know. So what is that? Well, what we're trying to do here is really transform the speed of tying to market, tying to data and tying to insight. So we're trying to reduce that time. It takes the business to get access to data and then that you're able to produce insight from that data. We're also trying to enable the ability to experiment. And what we see today is, you know, an enormous amount of time in the experimentation is really focused on, you know, finding the data and doing some of that data prep and data wrangling. We really want to focus, you know, so we want to make the data more easily available so they can focus more of their time on producing the insight, really. And also make it easier for people to experiment, you know, really, you know, back to that space-spilling idea and then experimentation across the organization. You know, and a common platform is fundamental to that. You know, having silos, you know, having mobile over here fixed over there, TV over there, you know, those are just barriers to providing that enterprise insight. You know, so this center of gravity, also what we're doing here is we're actually trying to provide a common data platform. So what we saw in the previous generation was, you know, taking us, you know, your data warehouse, you know, complimentary capabilities, your data warehouse provides you the curated consensus around the business, you know, your complex schema converse SQL, then your deep, your deep data, your event data and your big data lake. But what we ended up doing off on the business was they moved the small data to the big data to augment it to enrich the big data, or they were taking aggregated data from big data over to the data warehouse to augment, to provide more signals around what was happening to the big data, you know, so bringing this into a common platform, you know, but recognizing it's a common platform doesn't mean it's the same storage technology, but, you know, that will bring it closer together. So common capabilities here. So you still might have, you know, a columnar database to do your data warehousing, and you might have an object store to do your big data type analysis, but at least it's closer together and there's less friction from having a common technical platform and services that you can provide in there. So common data fabric, for example, common data preparation tools. And also what we see here, what we're really promoting is this data product idea. So what we want to do is, you know, again, that distribution, this space filling idea, we want to really break down and think about data as products. So we've got our, you know, and we'll talk a bit more about this later, but, you know, fundamentally thinking about our data as a product, so, you know, not thinking of it as, you know, in the big data schema on read container idea or the, you know, sort of data warehouse model if, which has a single centralized team looking after it. We're trying to, you know, really distribute this, you know, I don't know what that means, but key thing to see about bringing it into the, into the context, you know, that's where the ML innovation's happening at the moment. You can get access to the latest AI in ML and it's moving at a pace. So, you know, you know, so that, you know, we see some real advantages in moving that closer to where all that innovation's happening. Anyway, okay, so, so I touched on the product thinking, you know, so, and this is really, you know, leveraging those ideas that have come out of the retail world where, you know, how they think deeply about how they manage their products, you know, and we're trying to apply that to data, you know, and we're leveraging some of the work, you know, 3D and the data mesh architecture, you know, these are, you know, this is where we're getting the inspiration from for these ideas. So, you know, source data products, you know, these are, you know, largely mirrors of our source systems with some, you know, where we've actually done some work to make it a bit easier to consume, you know, so we do do a bit of work, it's not an exact mirror, we do do some work to align it with our enterprise data model, just to make it a bit easier to consume, but these are roughly map one to one to our master systems, our key operational systems. And then what we do is we have on top of there the consumption data products, which are owned by the business, and those are really focused around a set of use cases, you know, and again, it's about thinking about, it's almost thinking about it like a serial box where you write on, you know, this is what this data product's gonna do, and we actually do some portfolio management to ensure that, you know, we understand how many boxes serial we have and how many boxes of, you know, dairy products we have and, you know, figure it as a store type idea. You know, the really fundamental, is thinking of data domain as the first cast concern, is the, you know, fundamentally to this, is the standards, you know, the data standards that we apply. So we do, you know, with being a product owner, comes a lot of responsibility to quote Spiderman. So, you know, you have to really, you know, what we have is we have a set of standards that in order for people, you know, so, you know, data products go through a life cycle, so alpha, beta are generally available and the standards they have to pass to be able to qualify for generally available data products. You know, and that, you know, those are stuff like, you know, they have to be cataloged in our data catalog. They have to have a defined interface that they provide to consumers. They have to have an owner, you know, to create a data owner who's responsible for that data product. They have to set their SLAs or LAs around what the service levels are for that data product, how fresh the data is, data quality checks alongside that data product. You know, and those are fundamental, you know, those are fundamental to enabling you to join those, bring those data products together, which is, you know, that's where we see a lot of the value is, you know, it's not about having data products. Data products should be able to provide some value in its own right, but it's really combining those data products. There's also a lot of value in doing that as well. I mean, and that's, you know, that's the fundamental idea behind data warehousing, which we still want to leverage and use. So we also got a lot of trends in our, you know, about those insight services, you know, we tend to break those down into the three main personas that we focus on. But, you know, the fundamental idea here is again, going back to that space filling idea that we really want to grow that triangle we have at the top there, grow the number of users and grow the number of users in each of these areas. So, you know, consumers, we can really see, you know, it's really about expanding insight consumption across the organization. So how do we get more insight and spread it from that C level out to the front lines? You know, and we're seeing, you know, a lot of innovation, you know, a lot more interactive capability. We're seeing a lot of capability to move insight across those different personas. So analysts can share more easily with consumers as well as data scientists can share across there, you know, that's important. But also new interfaces. So, you know, interfaces like search and conversational interfaces, you know, with insight bots, for example, you know, that help our users different ways from to talk and get access to insight. And then we've got our self-service analysts. We're really trying to speed up, enable them to speed up insight identification. So, you know, again, a lot of innovation we're seeing in the capabilities around that space, you know, with a broader range of visualization, a broader range of recommendations around that, natural language generation. So you've got some words to go with the visuals you're seeing. So to help understand the picture you're seeing, as well as more collaborative capabilities, you know, really around, you know, around those reports and it's much easier to share and have a social collaboration with other users around that. And then finally, we've got our data scientists. You know, we're really looking to scale that out. So that's really from the citizen data scientists up to the expert data scientists. So, you know, massive, massive innovation in those area. You know, and we'll talk a bit more about this in the next slide. So, you know, here where the other two sectors tend to be more mass markets, so we're looking at more common tools in this area then it tends to be much more smaller community of much more experts. So we've got more flexibility around bring your own tools because there's not, you know, we want to unlock that value in this expert community. Where typically in the other two, it's much more about connecting and distributing the insight across the different communities. And also what I want to talk to you about is, well, you know, so, you know, AI and cognitive strategies, you know, a big opportunity for every industry, to be honest. But, you know, we're thinking quite deeply about how do we approach AI strategy? You know, AI is everywhere. You know, I say, you know, there's not a division or a unit that can't benefit from smarter decision-making. And what we're thinking about is, you know, where do you buy? Where do you build your AI? Where is the differentiation? What is commodity for your business and what is differentiation for your business? Where do you want to own the learning? Where do you want to leverage learning from the rest of your industry or cross industries? You know, and where's the, you know, how do you balance the value versus that competitive differentiation? You know, and those, and that affects your AI strategy, you know, because if you're thinking of talent analytics, maybe you buy that. But if you're a sports-based company, that's probably a differentiation for you. So maybe you want to build your AI around there because, you know, it's some unique learning and you want to own that learning. You don't want to be given that learning out so they can be sold to other companies. We're also seeing self-service. So those, you know, we need to grow. You know, we've got around 100 data scientists, but that's, you know, that's less than 1% of the organized 100,000 organization. But we want to actually spread some of that data. You know, data science AI is a spectrum. You know, everything's an expert, a really complex, you know, it's big, hairy, gnarly AI data science problems. Well, that's where the, you know, those are the stuff we'll target are the experts. But there's other smaller and medium-sized problems that we are looking at tooling, where the tooling can help in those, in those areas. Experimentation, you know, is fundamental, not just to AI, but also to the analytics as well. But we're really looking at how do we industrialize that experimentation? How do we provide easy ramps from going from experimentation to production? How do we allow ourselves to be able to go in there and actually test out new AI? You know, what one thing I'm confident of is over the next five years, there'll be a huge amount more innovation, you know, so being able to adapt and bring on new algorithms and new capabilities is going to be important. So, you know, those are some of the fundamentals. Also, we want to enable, you know, the experimentation, so being able to share what experiments we've had, whether they're successful or not, so that actually, you know, if you're going to be a learning organization and that information, that knowledge needs to be systematized so that everyone has access to it, you know. So if I do an experiment, other people can know if my experiment was, you know, and actually learned from me, not have to learn it themselves. Ethics and responsibilities clearly important, you know, we're thinking deeply about that. You know, we're looking at how, you know, there's a lot of work going on around what some of the ethics policies, as well as, you know, some of the tooling, how we're tooling can help. So how can you de-buy your data? How can you make sure your, the outcomes are responsible, are transparent? You know, so some interesting things happening in there, you must key another key pillar, as well as then, you know, the broader society. So how do we use this connected world to actually benefit the whole of society? And how do we actually use it as well as part of convergence? So, you know, how do we connect it with these other technologies that are exponential technologies are coming on that can really make a difference in society? So let's bring it all back together now. I've talked across a number of those topics. I thought I'd throw some use cases at you just to give you some examples. So one of them where we've had some great success is NUSAN's calls. You know, virtually everyone gets NUSAN's calls these days. You know, 75% of our customers receive at least one and probably more than one NUSAN's call per month. And what we've done is taken over two billion records and use that to develop algorithms that will allow us to filter out those calls and take away those NUSAN's calls away. Now these are unknown, you know, unknown callers to our customers. If it's a friend or family, we've got another product that takes care of those type of NUSAN's calls. But this is really those mass caller type scenarios where they're calling lots of people. Now, this is an adversarial AI. So we've created this compound index but we have to continue to evolve that. So what we find is like fraud, this is where someone, you know, you come up with a defensive algorithm. Someone else tries to find a new way to get around your algorithm. So you can keep constantly evolving your algorithm to keep making sure that your net is capturing as many of these as possible. We're also looking at techniques to actually discourage people from doing that as well. So, you know, there's some, so for example, we always terminate the call. So we make sure we charge the offender, which is quite a clever technique, but we don't terminate it with our customers. So that's useful, you know, other things to think about as well as just the analytics is the actually what you do with that analytics is really important to drive the right behavior back to the front line. Also, we've been using it to understand, you know, our cloud idea to actually understand customer service. So in our mobile, better understand our customers by bringing the different data together, both our online and our offline data. So really unifying that data so we can understand better our customers, understand how they're interacting with us across the different channels. Which channels do they want to be contacted on? When do they want to be contacted? And really also what the cloud allows us to do is we can elastically expand that. We can test five times quicker at that five times quicker and we can also test it a lot cheaper. And we can get also a better results, you know, by testing different algorithms, more permutations. You know, what that means is that, you know, we can do more. We can do more and we can get a better results. We can get more and we can do it at a lower cost. You know, so we've had some fantastic results in actually doing that, bringing our data together and really understanding how to make sure we've got relevant marketing that to our customers. So, you know, what are some takeaways really from building this? Well, I'd, you know, think about how you're going to scale this. This is one of the big challenges for organizations where you've got so much wealth of data coming towards you. You know, one is, you know, we need to treat it as a core foundation data in your organization. That's a fundamental, you know, it can't be something that's done on the side or an output at your organization. It needs to be fundamentally thought about alongside your people, your data and your core services. You need to really think about that data. You know, the other key one is really thinking, you know, human centers approach, you know, how do we enable it? You know, how do we enable that insight driven mindset for our customers and our colleagues? You know, it can't, you know, so how do we, and how do we spread that across the whole organization? You know, that space spilling idea is fundamental to me, you know, in terms of democratizing across, you know, and that really requires not only invest in the architecture, your strategy, but also your people. And that's both, you know, your colleagues, your customers and your partners. You know, that's a fundamental shift I see that's going to have to happen. And then finally, the other big idea I've brought is actually thinking about, you know, that terminal endpoint idea again, you know, so thinking about how do you actually enable some common services, modular common services at that? Because you need that flexibility. There's a lot of changes coming. You need to be able to swap this out and try out new stuff. But those modular common services, does that enable you to actually, to spread that out? You know, if everyone had different electrical sockets or you all had different capillaries, then it, you know, we would not be as successful today with our great cities or, you know, the animal kingdom, if we'd have gone with different types of fittings there. You know, so those are fundamental ideas I think we can take from, you know, from nature and from, you know, other industries, you know, where that stuff's really worked. And probably finally, obviously the self-optimized, so think about how you experiment and how you test and learn with these ideas to see what's working and where you can really, you know, automate that and build that into your code. You know, so we have a dashboard that shows me the architecture health of our ecosystem. So I'm learning about what is working, what's being used, where they're finding value and where there's friction in our system so we can continue to evolve our architecture so it's really delivering what both our colleagues and our customers want. Well, thank you for listening to me today. I hope you've enjoyed the presentation. We'll now move on to Q&A. Thank you. All right, just as a reminder, though the speaker is unable to attend, please do feel to submit questions in the Q&A window to the right of the screen and they will be reviewed at a later time. Thank you. And it looks like that completes the session. Thank you, Jason, for this great presentation and thanks to our attendees for tuning in. Please complete the EDW conference session survey. They're located at the bottom of this page. The next session will start in a few minutes. Thank you.