 I would like to quote the famous saying by the ex-CEO of Apple, Steve Jobs. He says, you need to start with the customer experience and then work backwards toward the technology, not the other way around, right? So you find what the problem is with the customer. And then you work towards the technology that can solve it, right? Not like you have a technology and then you find a use case to implement the technology in that. So this is an interesting fact. The market analysis says that acquiring a new customer is like five times expensive, more expensive than just maintaining your current, your customers, okay? So that actually shows us how important is it for us to maintain our existing customers and then you give them what they want, the customer service you provide and then you keep them with yourselves. Not, you don't make them go for your competitors. So about Verizon. So Verizon is basically one of the largest telecom providers in US. So we provide personal and residential network services. Personal as in, we provide networks for your mobile phones and then residential as in we provide home internets and then your internet TV and then your home phones. So basically this is the number of active users we have as of now. 146 million users who are like having various kind of services that we have available and out of these we have four lakh people who are actively using mobile apps with us to maintain their accounts like paying their bills, now switching their mobile plans or getting new services, stuff like that. So that is a pretty big number for us. And seeing the 146 million user base, we really get a lot and lot of customer calls every day. And most of these calls are very minor queries regarding, let's say, I just wanted to activate my SIM card, I'm not able to find it. So there are some findability issues in the website or we don't place it right there where they are looking for, right? And then some other issues would be like Verizon has launched a new plan, data plan and then I'm not able to figure out how to do that. So these are very minor issues that doesn't need a customer service call. Why because every customer service call that we receive costs Verizon around $15. So I mean a lot of customer call citizens per day, we receive around three lakh calls and every call would actually cost us around $15 because that is at all free. So doing some mathematics that is so much of money for us. So every day Verizon spends around $4.5 million just for your answering your calls and your queries through mobile phones. So this is what we wanted to do actually. So we wanted to increase the customer satisfaction and decrease the customer call volume. So this is our intent and the whole exercise that we wanted to bring it out. So as a part of this, we cannot compromise on 24 bar seven customer service that we are providing now. Why because internet is so mandatory in US because people would have opted for a work from home or they would have been watching their favorite cereals with their internet TVs and then if your modem or your router goes off, then our credibility or reliability is gone totally. So we cannot compromise on this. But we still wanted to maintain our customer satisfaction and decrease the call volume. So that was really a challenging task for us. So that is when we found out this would be a suitable technology to adapt and implement in our system so that we would be able to actually solve these two problems parallely. There are lots of flavors in AI. So basically the flavor that the first one is the narrow AI or the weak AI that the industry is following. Which is basically working with the request and response manner. So you get a request and you actually have a response for that that is already there and scripted and then you actually show it to the customer. So that is the first flavor. The second one being strong AI which has a consciousness and mind, which actually has a little bit of thinking that is not available in any of our industries or in the science that is going on. Followed by the third one is the general AI. General AI basically is as intelligent as being a human being. Which is very far from the future now. And then the last one is the super intelligence, which would actually surpass your human brains. So now what we have adapted is the weak AI, the first one where we get a request from the user and then we give a response for them. One part of this weak AI are the NLP, which is the natural language processing, which actually helps us to understand what the user is typing or what the user is trying to search for. And then we process their language into a machine learning and then we give out the information. Using these two technologies, we implemented these text-based chatbots. This was one of the things that actually fantasized us because the market value of chatbots are around 700 million as of 2016. And we also had a fact that 80% of the businesses are going to have chatbots by 2020 for their business purposes. So all these facts merged in together so that we were really confident about solving our problems with the text-based chatbots. So we implemented these text-based chatbots both in our mobile app and in our website. So this is how it looks like in the mobile app. So we had a search icon with a plus, saying that it's going to be an enhanced search for you. And then this is a chatbot that will be available in our desktop websites. So whenever the user opens the search, it actually opens a chatbot. And then the user can search whatever they want, they can converse with the chatbot and it'll actually give you information that you're looking at. So I'll be showing you examples about how it looks like. So basically, we get around 15,000 in-app searches every day from our users. And then 45% of these searches are basically goal oriented. Let's say if a user comes to our app and searches for something like I wanted to trade in my old iPhone 6, right? So that is a potential business for us which we cannot lose. So if we give him the right result, he would actually trade in his old phone for a new one with us. So and then we get some revenue out of it. So we cannot compromise on these kind of searches. And 45% of the search analysis that we had done, we had researched on the searches that the customers had done previously. We're all goal oriented and they were made to get a decision out of it. So we finally told our customers not to search anymore but to converse with us. So the difference of searching and conversing is that when you search, you just use keywords. But when you converse, your context is very specific. So we get a lot of information about the intent and the context of how the user is trying to approach to us and how the user is going to get benefited out of it. So basically, let's say the user is trying to search for, so if the user is trying to search for his data balance, he just need to say, how much data do I have? And then the chatbot will actually take him to the module where his data balances are shown. And if you wanted to change your plan, you just say, so when you have to change a plan, you just need to say what plans can I change? And then it will take you to the most recommended plan for your kind of usage. And if you wanted to buy a Galaxy S, you just type it and then it takes you to your product description page. So basically, we are getting rid of a lot of search results, a lot of links that the user needs to go through. We are taking them directly to the context of what they are trying to search for. And these bots, they don't only work on chat, they are also working as an intelligent agent that actually will be like trying to find out what you're going to search and then they'll give you the right information up front. So whenever you open your app, you'll be getting the information that you're already looking for. Let's say you're running out of data, it obviously shows you how to get more data or how to get ways to get more data or extend your data limits. So that you don't run an overage costs for your bills. And then if your bill is ready, we obviously show it up front. So when you open our app, your bill is already ready and it's ready for your payment. So this gets intelligent as you are like using it more and more. And then the information on the front page of your app will start showing you more and more contextual information as per your usage. So when we wrote out this, there was a tremendous response. So one of our users, Angela Smith, from in Google Play, she has mentioned that nine out of ten times, she gets the right answer at the right time when she opens the app. And then she has been extremely helpful for the app with the app. And then she loves the app and she feels great about the app. So now that we have kind of found out a way to help people. But that is a very small amount for using mobile apps. But our user base of 146 million people is really vast and we really wanted to cover the rest of the people too. So when we again dig deep into the user patterns and how people use the mobile phones, there was an interesting thing that every individual in the US, they'll be using around two hours and 25 minutes on average to spend on mobile apps on their mobile. So and then complimenting to that, six out of ten apps they use frequently are for messaging. So and then one more fact that was complimenting was 50.6% of people were actually feeling comfortable to talk to a business over a messaging app than in a phone call. So all these facts were really merging to a center point where we wanted to find something, some way where we can reach the people using these messaging apps. And then we found out that Messenger has already had a chat bot and then they have a huge user base of 1.7 million as of April 2017, which actually would help us to merge the gap between our user base and reaching out to the users using the mobile app. So this, we are also rolling out the chat bots in Messenger app. So anytime you open your Messenger app, you will be getting notifications of if your bill is ready, if your data is out, and then you can chat with them, you can change your plans, you can trade in your old devices and what not. So all these things, implementing chat bots in all these three stages, we actually improved our customer experience to a greater extent and reduced our findability issues too. So following with this, the next step for us would be like working on the voice church. So the fact says that by 2020, 80% of your searches is going to be through voice. So we are trying to merge with Siri, I just want to update my plan. It actually would talk to our app and then give you the right information to upgrade your plan. And then same goes for Alexa. So Alexa, what is the trading value of my iPhone? It'll actually talk to our app and say, hey, your trading value for your iPhone is going to be $100, do you want to trade in? And then you go with that. So the next way of reaching people faster and easier is the voice searches that we are trying to work on now. And I would like to end with the code by Jeff Bezos, CEO of Amazon. He says, what's dangerous? It's not a wall. So as a telecom company, we are trying hard to transform from a telecommunication-based network provider into a technology company. Our recent acquisitions, if you would know, are Yahoo and telematics who are into searches and intelligent cars and stuff like that. So our transformation is hard and real, so we are all working on it. So I think this example and this use case would help you all to transform more into our businesses and as a designer. Thank you for this wonderful opportunity, yep. Any questions, suggestions for us? We have time for questions, one or two we can take care of. So I wanted to ask you, you had these benchmarks in the beginning and then in the end you said you improved customer satisfaction, decrease number of searches for issues. So can you share with us like the degree of improvements? I guess you measure them somehow like, was it a 200% improvement or like? Yeah, so the customer satisfaction index was considerably improved. Like let's say we had around 70% of people satisfied and then when we implemented all these in all these various platforms, it was around like 10 to 15% improvement. Because the app and the desktop users were a little bit low. And then when we had the chatbots in Messenger, that was tremendous. We were able to reach a lot of audience. So that also increased our credibility and reliability on our brand value mode. Okay, so we started out with data analysis of all the chats that I mean the searches that the user has done. So from the website and the app, so we used to take key words of it and then we used to categorize how the user is actually searching for what information, how he navigates and finds the information. And what and how much time he takes it. Let's say he wanted to trade in his old phone, he just goes and types in trade in. And then he has a long list of search results from our website, which has a lot of articles and then he has a lot of videos to watch. And then we're not even sure if he has already tried in it. And that was a potential search that was missed. So by analyzing all these keywords, we found there is a trend that the user searches and the most searched and least searched. And then we had chatbot scripts written for the most searched keywords. Let's say trade in my iPhone and then we obviously know their account information, what iPhone is linked. They don't even know, they're not even necessary to specify the model number or the color of the storage, we already know that. So you just say trade in my phone, we know that it's an iPhone 6, 32GB, we give them the right answer by the chatbot. So that is how we built our chatbots.