 Well, hello hackers. I was going and the last one no other track at the same time than me So I'm feeling the pressure and after this talk you have the party So I try to be as fast as I can so first of all, thank you for for having the time to be here and This is the first the third time that I'm speaking here in in this conference Which is very good because that means that I'm still the chief data officer of telephonica I've been there the chief data officer of telephonica for the almost four years and Today I'm going to to talk about things that we are doing in in that area. So Three years ago when I was here at first time I was spinning what is what was our vision about data privacy and and the and the usage of Big data technologies inside telephonica last year I was talking about how we were thinking about to create our digital assistant aura And put it into our system and today I would like just to refresh What been doing the last year and sharing some thoughts about the the world in which we are getting into with Artificial intelligence and cognitive services because the last three years has been amazing. We've been in a period in which artificial intelligence is Getting to the human priority in a lot of aspects and that means that artificial intelligence is Doing things better than the human in things that normally we attribute only to two humans So in this talk, I'm going to talk about all this stuff and in the end I'm going to explain what are the living apps, which is something that we created on top of our strategy So this is the the first slide. This is This is the slide that I've been using the last in the last conferences that I deliver here And it's the slide that I received to to work as a chief data officer in telephonica the first day someone Gave me this slide and said Do it and as you can see it's very clear what I need to do into what I needed to do into telephonica Well, it's a it's not an architecture. It's just an explanation. It's just a Graph to explain what wanted to be done into telephonica and the idea is quite simple We have infrastructure into telephonica. We have the network. We have the it We have the the products and services devices the network that the IOT and sensor, etc We needed to transform telephonica with new capabilities. We needed people to and that understand The data so we started to create groups with data analyst Then we try to automate systems and to create automatic tools democratizing the access to the Insights that can be generated through big data machine learning algorithms. So Once that you have that the idea is that into telephonica We will be able to do things internally to optimize our services, which is the the first one on the top left Then creating new services that we can provide to our customers then give to customers the capability of Managing everything which is related to telephonica in an easy way, which is aura the idea in the center is aura and Right part is how to create technology for institution Companies, etc, which is Luca is the big data and artificial intelligence into telephonica And the last part is how to open this platform this for platform to third-party and create an Ecosystem which is what we are going to talk today. So in a nutshell If you've been listening at any time, what is the for platform in telephonica? At the beginning nobody knew what the hell was the for platform So I needed to explain to the executive in telephonica in an easy way What is the for platform? So I create something very very very complex to explain to them And I created this video Which is only two minutes and is focused to explain to top executive What is the for platform into telephonica as you know? Telephonica possesses a wide range of technologies services and applications Our three main platforms are different from country to country and this involves a great complexity Each of these platforms is generating data But each one uses its own data models and manages the data in a different way It's as if each one is speaking its own language and this doesn't seem very Efficient for a data-centric company does it with the digitalization work that we've done on these three platforms in the recent years We've achieved an ecosystem of very robust platforms that offer excellent digital services But this isn't enough our clients are asking for even more and also achieve the speed and agility that our clients are demanding It's important that all our technology within telephonica speaks the same data language by this We mean that each piece of data be it a client's name or fiscal identifier an antenna signal or the position of a mobile device Connected to our network is stored and processed in exactly the same way in every part of the group to do this We've been working with big data teams to build the fourth platform a layer that allows the whole group to speak the same language Now I'm not going to go into detail about the fourth platform But I do want to talk about two of its fundamental parts Firstly the unified reference model the URM This is the data model with which we store exactly the same information in exactly the same way around the world But as well as agreeing on how we store the data We also have to decide how we use it and the common model for working with this data is what we call the fourth platforms API When we incorporate these pieces the URM and the API into all our services products and applications We'll all be speaking the same data language But what impact does this have? Having built this common layer We're now capable to build on top of it technology and applications that work globally and by working globally We mean that you can build it once and then they work everywhere across the telephonic a group You don't have to reinvent the wheel each time which gives us agility At an internal level We can develop things that allow us to generate efficiencies for example Analytical models that allows to better manage our network and its deployment or a churn prediction system And for our clients we're able to build things that revolutionize the way we interact with them For example services such as our new global app and smart Wi-Fi Or even something as innovative as an artificial intelligence In summary in order to offer our clients the agility they expect from us at a global level and to truly be a data-centric company We need to speak the same language of data across telephonica In order to achieve this each new product service or app that we make in the future has to be built on the same Common telephonic a language that's to say that incorporates the full platforms URM and API This is what allows us to build technology solutions just once and use them globally Solutions that generate internal efficiencies and revolutionize the way we interact with our clients such as our and our new global app What this is the which is the video it was easy to understand That is a in a nutshell what we we've been working on the last three years We started with that idea and today the four platform is a reality in in 10 countries We have more than 178 million mobile lines With normalized data we have more than 70 solutions based on the four platform One of them is aura and the rest of of them are a smart pricing a network planner and our global app a smart Wi-Fi and secure network antifies phishing solution, etc. And more than 55 55 API is families to to do things and the idea is quite simple We create the global technology and it's connected to the four platform API and in each country if they have different systems It doesn't matter. They need to connect their systems to the four platform Gateway so for us is completely transparent that allows us to create Technology faster than we were doing before and one of the technologies that we created this aura aura is launched in nine countries And in in every country is connected to different Channels for instance in Brazil is connected to the contact center is connected to what's up? It's connected to the global app in other countries like Spain is connected to Movistar home Which is how a smart display a slash phone that we are connecting in to TV We have our own social networks in website, etc. So today our eyes are starting to to grow in in all the country Quickly because people love to use artificial intelligence. So when we are starting to to think about Using natural language to as a common interface to all our technology It was a little complex because at that time cognitive services were not so so ready to be using in corporates and We needed to do you to take a lot of decisions? So we had decisions about how to create an artificial intelligence And we needed to take decision like the gender if we wanted to have a physical appearance or not Uh, if it should be useful Rational or fun one of the things that people ask the most is can aura Tell a joke to anyone and we used to say look for a friend So aura is for helping you to do things in into telephonic And that was the idea of creating aura and we started to use aura before the cognitive Intelligence was completely ready to take that decision the cognitive services that we were using and one of the things that it was Challenging for us is that today is easy to use cognitive Services to to recognize The language to recognize and intend but when you look for small details things are not very well so one simple technology like war which is widely used in everywhere If you use microsoft war in spanish, there are more than 7500 words That are spanish words that are in the dictionary, but war Recognize it recognize them as a mistake, which is very bad and in in google We have Exactly the same problem. So because of that we started to work with the real academia de la lengua in spain and also with google microsoft amazon Facebook and twitter to create a project that we it's called leia and the idea is how to speed up The usage of spanish in all cognitive services that we are using today in our technologies. So We have this problem, but when you started if you start to work with cognitive services There are a lot of problems that today we need to face So another one is the sentiment analysis. It's very easy to To use a sentiment analysis to recognize if someone that has that that is interacting with your technology It's angry. It's upset. It's sad. It's happy Whatever, but the truth is that the technology in spanish is not really at all So in this example, which is a cognitive services to recognize the sentiment analysis You can see just exactly the same sentence But with a comma and it's not a comma a very big A comma of very big importance is just a comma to create a pose Is changing completely the sentiment in one side Parallel a kello no era un problema difícil era un hacker is positive and the other and on the right just Removing a coma which is just for a pose. It's not changing the meaning at all The sentiment analysis is completely different. It's negative So the technology is still not really for for spanning of course We have a lot of bias in all the technologies in all the cognitive services that we are using One of them is quite simple. It's gender bias in In google translation and you can use some expression like it's a scientific It's a nurse. It's an engineer. It's a secretary and And the the model that they are using for the translation service, which is a And a cognitive service is selecting the gender because of the bias So it's a scientific. It's a male. It's a nurse. It's a female. It's an engineer It's male. It's a secretary. It's a female which is not good that you Are planning to use this technology in a regular base application Because of that and I would like to explain this always because I'm putting an example with google Google is working hard in trying to remove that because in the end artificial intelligence learning from the data And we have the bias in the data. So we need to to understand when An artificial intelligence have been learning with bias How created this technology that I really like it, which is testing with concept activation vectors and ideas Understand what are the vectors that triggers one decision taken by a By an artificial intelligence and indexes these samples is are very clear in the left side Shibra Stripes horse and savannah on the right side. A doctor is white coat a state of scopes and male So in the right side we we can discover that one artificial intelligence that they use an artificial vision as a cognitive Services is learning with bias and we need to take care of these bias If we want to create technology in a professional way so Another thing that scares me a little is that we are starting to To use cognitive services to recognize people to recognize everybody. This is quite quite interesting This is the natural language recognition for For google also and the idea is that you can type Any expression and it recognizes Not only the sentiment analysis, etc But also who are in the sentence So in this example, I I wrote Sentence which says temal also was choked by Chuck Norris because we did a a picture together one year ago and Putting that expression Google is able to recognize that temal also is me which is in the wikipedia in spanish and Chuck Norris is Mr. Chuck Norris and is in the wikipedia in the in the states So it recognized you very well just by the name just by the name But if you go to artificial vision the thing is a little worse for instant Today we have artificial vision which is recognizing almost anyone Quickly right now is only available for celebrities. Of course, it's me on it This is an example of one picture Here we have a mexican actor. It's me and it recognized me with 99 percent of accuracy even without the beanie hat And paco prieto with 94.2 percent, but pacolo baton is All fashion is not it's not a celebrity for the artificial intelligent today And it's quite nice because you can Try different tricks. We exchange things with kevin mechnick, which is the most facial hackers on the On earth I let my beanie hat and I was using His glasses and the artificial intelligence also recognized very well And this is scary because this is today, but in three years probably It will recognize everyone around the world Because the artificial intelligent today has available data of your faces everywhere Facebook Instagram What's up? What's up is is encrypt the problem with this is that today we we've been developing also a high resolution cameras and I really love it You can go online and see a demo This is an example here. This is a picture one single picture From a high-res camera. It's a you can see the city here and you can say, okay. I want to zoom for instance here go Go go go Go go go And if everything goes well and I have internet connection you can see the picture of people That are All around So in the future with one single picture one high-res camera all of you are going to be recognized everywhere Everywhere, which is something that probably we need to decide as a society if we want this or not But the truth is that technology is also ready and it's not perfect as I said before in the cognitive services I was showing you some examples in which Cognitive Intelligence is not working perfectly But I have one more that I really like it which is Which is the next one the idea with with the The cognitive services is that you can train a cognitive services I can train my own cognitive services and depending on the data It will learn on a different way So every artificial intelligence that we are creating is based in the training of the cognitive services that we are putting into So this is an example with Alice Cooper. Alice Cooper is on the right You sure The artificial intelligence is telling you which one is is Alice Cooper and which one is me because you can see the rectangle The idea is that I publish this picture In with and one of my friends Text me through whatsapp with a joke that I've been supporting the last 10 years That is that I look like George Clooney Do you think that I look like George Clooney at some point? Well, I was upset so I decided to test it with artificial intelligence with cognitive Services, so there is a cognitive services in Microsoft Azure in which you can upload two pictures and the artificial intelligence tells you if People on both pictures are the same or not quite simple, so I Upload that picture with one George Clooney picture And in verification it says the two faces belong to different people Which is normal. We are different people, but confidence is only 44 percent and If you rise to 50 Is the same person So I tested with Richard Pryor on the right and as you can see we are not the same Same person So I decided to look for another picture trying to look like George Clooney So I look for one picture that I That I have from three years ago when I was in four years ago when I was in the executive committee of telephonica We needed to do a photoshop session And they did it two different pictures one with the bini hat and another one with Without the bini hat they needed to decide if my picture in telephonica should be with the bini hat or without the bini hat In the end we did a data driven decision So we asked everyone and in the end they decided to use the bini hat picture If you go to telephonica xcom website, you can see my picture with the bini hat But there is another picture that is also available to download Which is without the bini hat. So in that picture, which is that As you can see the bini hat changed completely your Appearance in this picture using that picture of George Clooney The artificial intelligence said the two phases belong to the same person Which is bad for me if at some point George Clooney commit a crime That is captured by a high resolution camera and use an artificial intelligence Which had been trained with this data because it's as sure So I thought okay, it doesn't make sense because I've been using the artificial intelligence For celebrities, which is also from Microsoft, but the training is completely different. So if we if You upload both pictures To the other artificial intelligence to the other cognitive services Which is training with a different set of data They recognize George Clooney and it recognized me on a different way. So we are facing things. We are facing Issues with artificial intelligence and especially with cognitive services that we need to think before use it Wildlife and of course Artificial intelligence is recognizing you very well. So today is very easy to create a fake person a fake artificial intelligence, which is Which is going to look like you which is going to act like you which is going to speak like you But it's not you we did that training with an artificial intelligence using Technology, which is an artificial intelligence that recognized me and another artificial intelligence trying to to generate pictures of me Which are which is not me So on the right on the left side is Enrique, which is one of my colleague And he is using artificial vision from the camera Which is recognizing the gestures all the things and on the right side is an artificial intelligence Very creating me in real time to have a video conference to to do an Skype to ask for increase of salary to do whatever and of course This could be this could be very dangerous because of course you can pretend to be a 16 years girl or you can pretend to be 14 years boy or whatever or you can pretend to be the CEO of your company and or there are Wide transfer to the bank. So this technology has been used for crimes, but This this have been on one year ago. We did this one year ago in a computer in a laptop and Today have probably you have been You've been watching a lot of videos with deep fakes or whatever the technology is almost perfect And the idea is a cognitive intelligent that recognizes you very well and another cognitive intelligent trying to To bypass their The the previous one. So every time this technology is saying it's not shema. It's not shema. It's not shema until It learns how to create a real shema that fakes this This previous technology so they are training themselves to create that that new reality But the truth is that we face a lot of Issues, but people love to use cognitive services people love to use cognitive Artificial intelligence to interact so last year the trends in 2018 people love video more than 50 50% of people use chatbots chatbots to interact with customers a day 40 percent of Of internet had been using on on tv So everything had been changing and probably you if you go to a taxi And you get into and ask for an address probably the taxi driver is going to use the mic To talk to the gps to introduce the the the address of your destination because people love it so We started to work how to democratize this and how to Use it both sides for internal purposes and to telephonica and for our customers So the first thing that we did it is less to create and a quite and A quite service to create in real time artificial intelligence chatbots with With facts, etc. So we created this which is look at official intelligence Wizard the idea is that it's a website It's a wizard you go type the questions type the answers type The different neuter answers that you want to to add and once that you select the icon in in my example is Isabini hat you just publish it you and you receive a cure a cure code and HTML code that you can Invented in your website. So this is one technology that we created and I explain in this because I put it in production in my In my block Which is here So this is my block My personal block. So here on the bottom right You can see a bini hat. Can you see? And you can next things to my chemo bot. Whatever for instance, you can say Yes, I'm a hacker and quick for it And well in this case the answer is a short video because you can put videos pictures texts, whatever you think is a A quick technology to generate artificial intelligence For instance, we use it with the cure code to put it in in shops or in posters or whatever you have a movie You can put the cure code or whatever Whoever with an smartphone can type things one I have a list of things that people ask to my bot And you don't want to know it And one of the things of course one of the things that they ask is do you have heard under your bini hat And what is a hacker? What is your salary? Whatever so quite interesting but On the other hand, we wanted to use it internally so I was in telephonica and I wanted to use artificial intelligence In all our products and services and one of the places in which people Use the most Voice is at home because if you are in a train or you're in a taxi or you are in a starbucks. You don't want to be Managing technology by voice You want to be some privacy you are not going to to ask some things to publicly on on a public environment But at home is very common the usage of voice So we wanted to join that trend we wanted to put Natural language technology with cognitive services In our technology and in spain telephonica is the first technology provider at home. We have 6.5 million of of Internet connection we have 4.5 a million of houses with fiber wi-fi Pay TV we have wi-fi rotors. We have pay TV over the top with movistar blues little we have Movistar home, which is our smart screen with Slash phone we have a lot of devices, but not all only devices, but also services. We have carrier billing We have cyber security. We have And home insights et cetera. So we decided to join both things and create a And more innovative Portfolio of services at home and this is something that telephonica had been doing the all the all the history in the in 1978 we were just commercializing wireline phones, but then we moved to Internet service provider in 1995 we created infobia, which is one service that democratize Internet access in this country at that time Spain was the first country in europe With internet connections people connected to internet Thanks to infobia. Spain was number one at that time in 1998 and 1999 we developed The adcl service with one rotor with wi-fi That wi-fi service today is the most Spread wi-fi service in in spain in 2003 pay TV and we created fusion in those in 2014 We bought we acquired movistar plus canal plus and we created this platform of pay OTV content and in 2018 we started to create new services at home like aura movistar home Connection seura, which is the security from the router that you can enjoy for free if you are movistar customer one question hands up, which is not a telephonica customer Please abandon the room immediately and in 2009 actually Two weeks ago we launched a movistar living apps and the idea of these living apps was quite simple. We have a lot of devices. We have um Set of box that controls tv. We have the wi-fi router. We have the movistar Home, we have the wire line phone. We have the remote. We have iot things Etc and it's like more or less like a computer. We have the screen. We have the operating system, which is aura connected to movistar plus. We have uh Interfaces for voice with the mic with the mouse, which is the remote with With arrows to to move through the Through the option. We have apps connected to the system. We have internet connection. We have content. We have It's like a computer So we started to put all the all the pieces together and create something that we call Internally home as a computer To create an architecture that joins all things that we have in At the home and once that we have That created is like a a computer probably you remember when Years ago the amstrad the spectrum you turn on the computer And that's it if you wanted to run anything you just say run The name of the program And the program takes control of the interface the keyboard the mouse the screen etc So that was the idea that we have with the living apps So how create how are going to create a platform to run digital experiences at home using cognitive Cognitive services and artificial intelligence and we did this so the idea is that we have right now this platform Which is connected in in to the four platform if you remember the first slide in which I was talking about the Open apis for third parties to create technology on the four platform So the idea is that we have connected all devices that our customers have are the houses To the four platform we collect data. We have apis to do things into the devices We control tv. We control the wiper rotor. We control the wire line phone We control the services that we provide at home and we can integrate third party We've been integrating on this platform several companies so far for instance We integrated netflix, which is integrated in our set of bugs. We integrated twitter, which is integrated in movistar home. We integrated In veria, europea and atletico de madrid as living apps. We developed our own living apps to run experiences At at home to control all services and we are working with all of them today And speeding up to announce a lot of them in the next mobile world congress But the idea is quite simple. We have the four platform. We have the devices the devices are connected to the four platform There is a common interface which is aura aura is a cloud-based cognitive intelligent Which is created in in microsoft ashur. We are leveraging of Microsoft ashur cognitive services So the customer interact to aura using different devices for instance the voice The voice remote or remote with mick which is going to be launched this month or one app with aura You can talk to or The all-fashion remote with the all-fashion remote you go to tv and there is an option which says apps and you can You are talking to aura through a different interface and of course you have movistar home. You have touch and mick or whatever a webcam Also, so once that you connect devices you have the interaction to the to the four platform If at some point one of the customers says, okay aura open Whatever application Then the app takes control of the home as a computer that platform and it can create different experiences So this is the look like if you are a movistar plus Customer right now you have the app section in which you can run the apps and This is an example that we created with the Europa the idea is that we don't want to migrate The mobile app or the website of the company to the living room We just want to create experiences that deserves to be enjoyed in the living room So in this case, they are they created we created together But the idea is coming from them is an experience in which someone at some point is having a discussion with With their partner and say, okay long time ago. We're not going on a trip and you say, okay aura Pause Netflix and then you pause the movie and you say Okay, now open in Europa and then you just ask for a trip randomly. So this is the experience Okay, now Okay, now I want to fly in December a weekend for less than 200 euros Okay, now I want to see the offer Okay, now I want to call to hire Well, that's the idea just different experiences for third party today You can have extra content for from atletico de madrid for instance in the living app before the match After the match in the middle of the week interviews, whatever you can do similar things to with iberia We are working with a lot of companies, but believe me doing this in atletico. It wasn't easy Today looks very easy think that we did our plan very in a staff three years ago with the fourth platform Normalizing data and creating the apis and once that you have the data and you have the apis The apis you can do a lot of things. This is the atletico de madrid. This is another example That's the idea just using artificial intelligence in the core services of a company like telefonica And take the most of the cognitive services taking care of the issues that we are facing you need to analyze Carefully what is going to do your artificial intelligence and how every single cognitive services you are using Works and create the best experiences for your customers. And that's all. Thank you very much for your time