 getting my slides up. Why don't we talk about why we are here? Why we are here in this room? Why we are not in the caves? Any guesses? Well, because over a period of time we have acquired intelligence. We have created an ability to retain information and learn from it and pass it on to others. And that has essentially helped us human race from caves to a nice office like this at Microsoft. And I think this is the fundamental thing, intelligence that essentially separate us from the rest of the species. So today we are going to talk about artificial intelligence and its usage in the retail industry, specifically how it is going to have impact on the e-commerce and the stores that we used to. So first of all, what is intelligence? Intelligence is essentially our ability to infer from the information. Whenever we have any information, whenever we are looking around, we are essentially getting information from various sources. That information could be how much traffic is there on the signal. Is it good enough for us to cross the road or not? What time it is and if I need to catch a train in the MRT, do I have enough time or not? So we have different information and we use this information to make some decision. And often the process of converting that information into a tool that helps us in taking any decision is the knowledge. And that is what is defined as intelligence. But what is artificial intelligence? How do we define artificial intelligence? So artificial intelligence is not a new field. Even in the 1950s, when the computers were coming up, many people thought that the robots will rule the world. And the artificial intelligence was defined as a computer program that has ability to mimic the entire process in which a human captures the information and infer a decision. And that could be for everything. So in this very simplistic context, when I look at a phase that I am familiar with, the moment I look at it, I know what their name is, if my memory is good enough. Can we do the same thing with computers? Probably not. Probably yes. So just to give you an example of how it looks like in real life, if you have kids and if you were to teach them anything, how would you give them that information? You would repeat that information in different context, in different settings until they are able to recollect that information and create their own judgment. So for example, you would probably show them different phases of cat. And hopefully in future, when they look at an animal that is very much similar to a cat, they would be able to infer that based on whatever I know so far, it looks like it is a cat. So we teach them with the repetitive information until we find a point wherein they are able to infer based on whatever they know that what this object is. But for very long time, we could not do that with computers. Computers were really good at doing analytical stuff. So if you tell them what is 2 plus 2, they would happily tell that 4 and they would do it thousands of time. But if you show them two phases and ask them this question, whether you are looking at a male or a female or a transgender, they would not have any idea. If you show them a picture of a lion or a cheetah or a cat, they would not know whether they are looking at an animal or a flower. So scientists at Google in 2012, they trained their computers to identify the images of cat because cat as you would probably know is probably the most shared video animal. So there are tons and tons of videos in YouTube about cats. So YouTube, in the Google engineers, they fed entire YouTube content related to cats to their supercomputers with an idea of creating a blueprint of what a cat might look like. That computer went over images of cats and soon after a long time, they started forming an image of how a cat might look like. What are their facial characteristics? And after they did it for hours and days and months, they thought that now our computer is smart enough to draw a picture of a cat. And that in essence was something that kickstarted the interest of the industry in artificial intelligence. But why it has become relevant again? As I mentioned that the artificial intelligence was first mentioned in 1950s, but what has happened that has made it so popular now? Why organizations like Amazon, Microsoft, Google and plenty of other startups are working in this area? Well, there are a couple of reasons. The first reason is that technically now it is possible to process a huge set of data. So for a long time, our entire computer architecture was based on the CPU, that is central processing unit. And because of the limitation of CPU, it could not process the huge amount of data that is needed for an AI based application. Then came the GPUs, graphical processing unit. And that essentially powered the AI revolution because with GPU, it became possible for people to analyze tons of data. So even on a single chip, there was ability to have thousands of thread concurrently working, analyzing the data and making a meaning out of it. And now using this, engineers could build very interesting things such as these beautiful products that we see on the screen, Siri, Amazon's ECO or the Google Home. They are excellent product. And they are going to redefine the way human interaction works with the machine. They have become both so good with NLP, their natural language processing that the moment you speak, they would be able to identify what you are speaking. And with all the other information that they may have about you, they would be able to identify what your context is and how they can utilize that context to serve you the information that would be most relevant for you. So let's look at some of the domains in which artificial intelligence is being used. So the first one is fraud detection. So we have plenty of financial transaction happening in financial industry, happening in banking industry, happening in the trades. How many of those transactions are fraudulent? And how do you identify? Well, apparently, all the transactions that are fraudulent follow a specific set of characteristics. And using AI, if we give them a bunch of data, like millions and billions and trillions of data, and we flag that these are the fraudulent transaction, the computers have now become smart enough to identify that what is it that is common in these fraudulent transactions? How can I make a hypothesis that whether when I see a transaction like this, is it going to be a fraudulent transaction or not? And then they can test this hypothesis on the next set of transaction and see whether that is correct or not. Similarly in the medicines, we know that how important analyzing things like DNS are. But they are very resource intensive, finding cure for things like cancer. You have to do permutation on like thousands of different variables that may impact whether a simple lump is a cancerous or not. So, doing these kind of analysis require people to analyze huge set of data and systems powered by JPU can allow that. Industries like legal industries, where there are tons of cases, historical cases that become the basis for the discussion and the basis for the judgment for the legal industry. Artificial intelligence can help that. So, there are all the important aspect of our life are now being impacted by artificial intelligence. And I think retail is going to be the next one. Now, why I think retail would be impacted by artificial intelligence? The first and foremost, in order for us to create any hypothesis, we need some data points. We need some data. Unless we have some data, we will not be able to create a hypothesis. And in detail, there is no dearth of data. There is data everywhere in the retail. So, for example, when somebody is walking in the store, retail stores would have data about that. In what aisle they are going, what is it that they are purchasing? How much time they are spending on the aisle? How often they are going to the aisle? What is it that they are purchasing? What is their pattern of what they are purchasing? Is there any seasonality element in what they are purchasing? So, there are plenty of elements related to the retail data. And on top of that, they have historical data of a person, historical data of how that store may have performed in a specific season. Historic data about competitors, how they have done in the retail sector in a specific season for a specific type of cloth or for a specific type of product that they are selling. Then on top of that, our life has become public more or less with the mainstream usage of platforms like Facebook and Twitter, where people are sharing every little information about their life. People can kind of get information on the trend that looking at this data, what is it that we can derive that can help me restock my retail store in such a way that I am not over stocking. And I am stocking with the stuff that I know is going to sell because I have ability to analyze the trend based on the data that is present on Twitter. We can get a better grasp on the demographics of our user. It is no longer sufficient for people to identify just the gender and the age bracket for their demographics. A dataset powered by artificial intelligence will tell them very specifically what kind of people are going to buy your services, what kind of profile they would have, what kind of interest that they would have, what kind of communities that they go in. It will give a much richer dataset and not only that because it is powered by AI, it would be able to experiment, learn from it and adjust if needed. So for whatever reason, let's say if somebody is creating a hypothesis that people with interest in PHP would be able to buy this particular book. They would run this experiment for that people, for that group of people. They would then tie in with the conversion rate. If the conversion rate is looking healthy, they will keep on running that experiment. If not, they would probably change some parameters and see if that works. And that would happen automatically because they would be powered by AI. And similarly with the advent of facial recognition, they would pretty soon have cameras pretty much on every aisle of your retail stores. And using that information, as you are walking in, they would be able to identify who you are. And they would be able to tie in that information with your identity and make sense out of what could be the purpose of you in this store. If they can identify that purpose of your visit to this store is to, let's say by a pair of trousers, they can potentially give you a specific coupon code just for you. And that is possible. So let's tie everything in and see how it might look like in future. So let's say you are searching for a specific product online. You're searching for a pair of trousers. And while you are searching the pair of the trousers, your information would give, your store would, your, your rabbit store would know who you are, where you live. And based on that they would give you probably information on which one is the closest store where you can find this pair of trousers. So they are already using that information to give you a personalized, a very unique experience that is just created for you. And irrespective of the technologies that we use, whether it is Magento or Basepoke, we need to make sure that when we are creating our stores, when we are creating our experience for the users, we are keeping that in mind because the norms is not going to be that user is going to be able to just perform a search. They would expect that all the other stores are offering contextual search. They would expect that user are, that other stores are using their personal data to give them a unique experience that is making their, their life easier. They would expect that from every store because that would be the new norm. Now let's take that journey forward. But assume that a user has just searched for a pair of trousers, but they have not completed the purchase for whatever reason. Now they go to a store. The store would already know that this particular user, as you walk in, they would identify with your face that you, this is the user that has performed a search on the website for this particular product. They would correlate that information and know that you have an interest in trousers, but you did not make any purchase on the website. And now in, you are in the store. So there is a possibility that you might buy this, buy this trouser. Then on every aisle, they have this camera. They would know what is your location in this store. They would identify that now you are in an aisle where you can potentially look for the store. You take out those trousers, you have a look at them, you try and make up your mind. But for whatever reason, you do not buy that product, you put it back. And at that stage, they would probably know that you have not taken it to your cart using the sensors that they might have to detect your activities that what is it that you are doing in that aisle. And as you walk out of the aisle, they would know that you have not taken your product. And imagine at that stage, if you have your, their mobile app installed on your mobile phone, you get a message saying that, hey, we are running a 30 percent off on these trousers. Would you be interested? Chances are that you would buy. And the chances are that if you are running your e-commerce store, you would want to replicate that behavior for your users. Because that is going to be the norm. And if we are not following the norm, then we would probably be left behind. So AI would open up lots of opportunity because of the kind of data that it would have, because of the potential it would have. And this is just one scenario. But then there are other things such as chat boards. People will not expect that you have a dedicated support line where you are put on hold to be able to speak with a customer. You would have automated chat board who would know who you are. And depending on the information that you need, you would get your results instantly. Because often, whenever you call a call center in the context of e-commerce, the reason of the call would be, where is my order? Now all these things can be handled very easily with the help of things like chat board. They can be handled very easily without you typing anything. If you have a device like Amazon ECO or Cortana or Google or Siri at home, you can just ask that question, hey, I am expecting this order from Amazon, where is it? And they will tell you that this is where your order is. So the things around us in the e-commerce ecosystem are changing rapidly because of the artificial intelligence. And there are already many players who are working in this field. There are many players that are working in different aspects of the e-commerce ecosystem. It will have impact on the manufacturing because G has a software called better software. And it can get integrated with any manufacturing unit. And depending on the thousands of data point that you can collect from your manufacturing process, it will tell you whether the product is defective or not. What would be the impact on the e-commerce? You would probably have less faulty products. As a result, your margin would probably improve. And you would have to plan in such a way that you would expect less returns. You would probably need fewer people on the resourcing side who handle your support calls and things like that. You have voice control assistant. There will be more intelligent product recommendation tool. Your product selection would take care of the season, would take care of the trend. And as a result, you would probably be able to stock with the stuff that you think that you know for sure that is going to sell. You would have ability to maybe provide contextual search, visual search. So you don't even need to type. You can just, within your e-commerce context, you should be able to say that, hey, I'm looking for a blue pair of jeans and my size is 32. And that should be good enough for them. Better still, you should be able to take a picture and say that I'm looking for a product like this. And the computer vision would become smart enough to gather all the information that needs to be gathered from that picture and perform that search for you. And the moment we have few big players providing this kind of experience, the benchmark would change. Then it does not matter whether your e-commerce store is a small or big or medium. The benchmark would change and people would expect same things from every single e-commerce store. We will have changes in the physical retail store. So there is a product called PAPR. And that PAPR is a robot. It's a physical robot. And SoftBank has invested 200 million in installing those physical robots in different banks across Japan. And these robots, they know who the customer is when the customer walks in. They can offer the same kind of services that a cashier gives them. Imagine in future's retail store, instead of your traditional POS system, you would have robots like PAPR walking around in the retail store. How do you, as a e-commerce store, kind of integrate with that? How do you work in that ecosystem? These are the challenges that we would face in the future and it would be important for us to think about it now. We would have technologies like augmented reality. I'm sure most of you may have played the Pokemon Go game. And it was probably one of the best example of how you can mix physical life with your digital life. And the things like this will become more commonplace in future. When you, let's say, go to a store and using your augmented reality app or the e-commerce app, when you put your camera in front of a product and imagine if it is a food product and you get all the information about the origin of the content that has gone into making that product, calorie information or any other information that may be relevant in the context of that product, things like this will become commonplace. And we would need to provide the same experience because that would become the new benchmark. You may have heard about the experiment that Amazon did in Seattle with Amazon Go Retail Store. So if you have not heard of it, Amazon started on employees and there was no checkout desk. There was no employee in that store. There was no security guard in that store. If you work for Amazon, you would just walk in. And as you walk in, they would know who you are. They would automatically assign your identity with your Amazon Prime account. And when you are picking up any stuff from the shelf, it would know what you are picking up because every shelf is controlled by the sensor. It would know what you have picked up. And when you walk in, walk out, it would just detect that money from your checkout because with facial recognition, with sensors, they know exactly how much time you have spent, what you have picked up and what is the cost of that. The experience like this will become normal in future, I suppose. Changes like this are going to impact everything in the retail sector. And I think a platform like Magento would be benefited a lot with the changes in AI. Why I think that would be the case? The first Magento has excellent open architecture. And by having an open architecture, it can expose the entire product catalog and the sales data to any system that can be powered by AI. So if as a web store, if you are running a Magento store and if you want to utilize the power of AI, you can very easily integrate it with other AI libraries that can give you that analysis by just exposing the data that is already there in your Magento platform. It is an open architecture, so it can consume data from any web services. Most of these libraries that are powering this AI are now exposed as web services. And Magento, as you may already know, can be integrated very easily with any external web services and provide the data that it needs. Magento can provide deep integration with every aspect of e-commerce ecosystem, whether it is supply chain, to POS, to logistic and delivery. And because of this integration, right now that integration may be with POS, in future this integration may be with the robots like Pepper. In future that integration may be added with the facial recognition, the checkout may be related with facial recognition. It has got excellent APIs and integration with different APIs, makes it extremely easy for any Magento based application to take or utilize the power of AI. And on top of that, there is an excellent ecosystem of the marketplace. And many people have already started building this AI based extension that we can just use, just enable them in our web store and start using them. So there are for visual recommendation, there is an extension called view.ai. You can just take a picture and view.ai would suggest related or related products that look similar to the image that you have taken. We have search spring as an extension that provides contextual search. It would understand based on the customer data that it may have, that what is it that you are searching. So if you are selling books and if you are selling coffee and if you search for Java, it would know what is your customer profile and what kind of likelihood you have to purchase a coffee bean or a book. It would be able to do that. And I think as a solution provider it is on us to basically care about these things because our retail stores or our e-commerce stores they may or may not know about the implications of the AI. So I think we need to work as an educator in the retail sector specifically and give them the opportunity to embrace these changes that are there because of artificial intelligence. We need to think about the challenges and the changes that they would have to go through and provide support that Magento community or the Magento ecosystem provide. So with that I would like to ask you this question that what is it that we can do to help them. Anything? Well if there is nothing that we can think of right now I'm sure in future we will have plenty of things that will be able to contribute because artificial intelligence is here to stay. Let's have a conversation about it. Thank you for your time and attention guys. Happy to take any questions that you may have or we can talk about it afterwards. Thank you.