Selling in the age of the customer

Thank you so much and thank you for having me and for bringing up expectations. So be prepared to be disappointed. My chatbot broke down last night, but it will be up tomorrow. So, yeah. I have about 20 minutes to give you a brief tour of some of the stuff that we are working on, some of the stuff that we see big brands are working with, with regards to artificial intelligence. I think it's... Is someone running? Cool. So this is the formula that I'll be working by. I'll try to briefly explain you why I think artificial intelligence, specifically chatbots, is interesting and important. I'll give you a brief look under the hood of how it works and why it works. And give you two or three applications of artificial intelligence, more specifically chatbots, not chatboxes, as we just heard about before. But first of all, let's look into the mind of the consumer. It's already been mentioned twice today, so obviously loyalty is really important. And the reason I think chatbots is so interesting is because it sort of fulfills... The promise of through a conversation it will actually learn more about you, and in that way it will create an emotional bond so that you'll become more loyal. Loyal clients, customers, whatever you want to call them, they have a higher lifetime value. So that is the reason why, at least part of the reason why it's interesting to start digging into artificial intelligence. But if I was to say four reasons where... this fits in or where it can actually help us, it's customer centricity, which is sort of the theme today. The second we heard Lars talk about, content is king, but I believe that context is God. I'll give you a brief example of that. Cognitive overhead. I read a recent study saying that we can only accommodate around 100 pieces of information or decision each day. And I think most of us feel an overload, or at least we have to use our brain to process it. We have to process a lot of information and to make decisions all the time. And I do believe that chatbots and artificial intelligence can help us here. And then finally, I don't know whether I would say famous, but at least it's the problems for people developing apps that they're not being used. We have too many apps. We're not even using them on a daily basis. We're not even using them on a weekly basis. So I think that's where also the chatbots come in and might overtake some of these apps. And first of all, customer centricity has been mentioned a few times. But we can see on a 10,000 feet level that companies are investing in the functionality and technology that is focused on understanding what the clients are doing and understanding what else can we do. And not just figuring out, okay, so is this customer or potential customer asking for this product or this product? We need to understand what is their effort. We need to figure out how can we create a product that actually turns them into an awesome person who can do rad shit, whether that is building a wind turbine or whether that is actually feeding your child. That doesn't matter. We need to look beyond the products. And the chatbots and the artificial intelligence will gather that data through a conversation so that we will have an idea of how can we actually bring them this product that will one-up them in a way that will make them feel like they're a part of the product. So it's to say in the gaming world. Five years ago, I started a company called OnSilo. It builds semantic search engines. They're actually located just a few hundred meters from here. I think it's that way on Navitas. And the core idea of OnSilo was to beat Google in search because search really is broken. Today, it's a game of who pays the most to get the clicks. It only looks at specific words, whether they are featured in an article or on a product page. And then it looks at the results. And it will bring that up to the front plus a lot of other ways that you can actually influence the algorithm. But the great thing about Google is that it will give you exactly what you want. But the bad thing about Google is that it will give you exactly what you want. So you will never come outside of that, the way that you describe your situation. And I do believe that chatbots in a way could help through a conversation actually start to understand what else is there. And that might solve your problem. Other than just this one product that you might be asking for. And the much talked about traditional customer journey. I don't know whether you can see yourself in this. There are obviously a lot of touchpoints today. And you have to bring all that information and all these touchpoints into one sort of seamless journey. And I do believe that by using chatbots and by introducing artificial intelligence, you might actually streamline this so that you will only have one interface, one communication, one conversation. That will actually help answer all these questions that you have. And that today you might go to a forum to look for reviews or ideas on how to use a specific product, etc. We do believe that products such as Amazon Echo, I think Simon just briefly talked about it in the beginning. That this will be the place where you want to integrate. This is the new algorithm. This is the new Google. Because if Amazon Echo will be in your kitchen or at your office and you can ask it to buy stuff, that's the algorithm you want to influence. It's not Google anymore. And the reason this is interesting is because it will understand the context. Because you will ask it other things. It will know what you want to listen on your radio. It will know what you tweeted. It will know that you forgot to buy flowers for Mother's Day. Such as I bought a flower. I forgot to do. And we actually pulled out the artificial intelligence of the Amazon Echo and used it in a smart mirror in our toilet in our office just across the street. So now you can have a conversation with Alexa as she's called at the toilet. And you might actually perform some transactions there. You might buy flowers. You might see what's next on your calendar. Whatever. So a streamlining of the customer journey. And finally the app fatigue. I think I have about 200 apps in my smartphone. And I think I use about 7 or 8 of them on a constant basis. The rest of them I completely forgot that they were there. And that's probably what most app developers are facing today. That even though it solves a problem, it's just forgotten in the masses of apps that you have available. And what we see that bots might do, at least some of the big messaging services, they will take all that functionality and build it into the messaging platform. Because if you have Google that have 800 million people using the messaging platform, why not build in the currency converter? Why not build in the webshop? Why not build all these other kinds of functionality that you have in apps? So that's the big trend at the moment. Okay. So let's move a bit closer to the smartphone at least first. Take a look under the hood. First of all, as I promised to disappoint you. By not having a demo. But I think in general when we talk to clients about using artificial intelligence, building a chatbot, they are impressed with the output. But they are completely disappointed when they learn what is actually behind. And first of all, I'm not a coder. Not at all. But I built a chatbot that worked yesterday. And it's, I mean, it's not basic. And that's probably the reason why it's not working. And that combined with my lack of skills within Google. And coding. But algorithms, they're really great at transactions. They're really great at doing math, statistics, the boring stuff. We people, we find personalities, emotion, all that kind of stuff really interesting. But the algorithms are slowly moving up and displaying a personality. I'll give you some examples later. You can say there are an evolutionary process going on right now. We have four different... Steps. At the top you have bot-aided humans. B-O-T, that is. That will have a conversation with a bot and then ask a question that some user posted to them. So they will go to an artificial intelligence and say, I have this question. And it will bring this service, it could be search personnel, an answer. And then they will forward it to the client. Secondly, you have the human-aided bots. And that's quite a big thing. I think that's more or less in general the stage we're at now. You have built a chatbot that will answer questions. But there will be a human curator placed in between. So they will look at the answers and say, okay, it makes sense when you look at the question that was asked. And then it will forward it to the human. This is also being implemented at the moment where you will have a request and the task will be redistributed either to a human or to a bot. Depending on what is asked for. So when we know now the sort of the frame of what the bots and the artificial intelligence can do, we know when a question is suitable to send to a bot or whether it should be asked by humans. You actually have a way to reroute questions. It might also be rerouted to bot-aided humans, etc. And then we have the final layer. There are no humans involved. This is just an algorithm, an artificial intelligence that keeps getting better the more people ask. So that's the final evolutionary step. So what's in it? What's under the hood? First off, you have natural language processing. When we built the OnSilo engine that's implemented at a few big companies around the world, we took the Stanford path. It's getting a bit technical now. But they machine read every article ever posted in all newspapers in the US from 1850 to 1860. And the US from 1850 until today to understand how people actually explain themselves. How do you explain a phenomenon? What are the words used in which and what is the syntax? What's the meaning of it? Secondly, you have machine learning. This is where you don't tell the engine to do something. The engine or the machine will actually build up learning algorithms so that it will improve without me having to tell the machine. When you get this kind of request, you need to answer this. It will slowly start to build that intelligence itself. Then you have answers generation. That's the part where you mix and match questions and answers. You will look at, well, this person has been asking questions about this for a long time. Now I need to provide a new answer because obviously the answers I gave before weren't the right ones. And then at the end you have APIs that connect to messaging platforms. It could be that you want to build it into your web page. It could be wherever you want to build in a bot. It's more or less possible today. And there's a great bot rush going on right now. I just heard that some of the big venture capital firms, they sort of gathered up that around 99.9% of all the investments they made during 2015 had small degree or large degree of artificial intelligence built into the product. So if you're looking for funding from Silicon Valley, you have to at least use the bot's word artificial intelligence. But you can see that it's going in all kinds of directions. You have those that are platform agnostic, meaning that you can integrate them anywhere, as I just briefly mentioned before. Or that it's an old platform such as Siri. I actually built on silent with Matt Sruide, a Danish guy who helped build Siri a long time ago. I think it's eight, ten years ago. Then you have the generic intelligence, which means that you can answer it anything. Actually, the bot that I built would answer questions specifically on innovation, but it might also tell you the time of day or that IWDK was going on at the moment. And then at the other end, you would have, for instance, an ABA chatbot that only asks or answers questions related to that specific subject. So we see them moving in all kinds of directions, some that might fit into all kinds of application areas. But it's not the end. It's not new. I think this is one of the first mentions of a bot. This is a shop I built by German students some, let me just cut to that, 18 years ago. And it was kind of basic. You would put in a request for something and it would look online, as people called it back then, to figure out where is the best place to buy this, calculating the time it took to ship it to you, the price, and other factors. And it would give you a reply. And you could go to that website. Obviously, what we're seeing today is that you would build in the transaction into the messaging. You wouldn't just be given a link to a web shop or wherever. It would actually, if you give it your credit card details, it will provide you with a one click shop. So it's not new. But let me just give you a few examples. I think this is a very basic example of how you can use common artificial intelligence to have a conversation with a client to help them find what they're looking for. H&M, they built a, you could call it a decision tree. You would start out by saying, are you looking for men or women's clothing? So that's A or B. And the next level might be, are you looking for pants or shirts or hats or whatever you're looking for? So that's so really, really basic. But what they've seen, and I've seen some data on it, is that people prefer this way more than going to the website and doing exactly the same thing. It's completely the same journey that you take. You go through this digital application. But people prefer this because they're used to actually chatting and using different messaging platforms. So that's one of the main reasons this actually works. I haven't seen figures on whether it provides better sales figures, but it needs to be seen. But I think it has a great potential. Another area that this can do is that it will integrate into platforms such as Slack. This is KIPP. This is a group. It's a group buying bots, if you can call it that. But people actually vote for different items that they want to buy. But more people have to decide which items to buy. So you'll have a conversation with the bot and you'll say, okay, I have these questions about the different alternatives. And it would aggregate this and provide you with two or three examples of something that solves your problem. And this kind of intelligence will be built into all these platforms. And it's already on two of the larger messaging platforms. But it will be integrated into all those that you might use. So in general, we see that brands are slowly discarding their focus on apps. I'm sorry to say that if there are any app developers in here. It's going very slowly. But they will be moving into the messaging platforms, integrating into the app stores of the messaging platforms. Because there's 800 million people using Messenger. And there's like, I can't even remember the figures for Snapchat and all these other, Kik, Telegram, etc. There's a large user base. So probably most of your users are already in there. And if you can use or if you can embed the transactions that you actually had an app for before, it makes a whole lot more sense than being an app on the fourth page on an iPhone. And in Denmark, I didn't know that there was examples here that you had. I thought you said chatbot. It was a chatbox. But these are, there's a bar actually in Aarhus here where all the chatbots that were laid off, they meet once a week. There's Eva from SAS. She was giving out actually complete nonsense. So she was cut off. There's Anna from IKEA. She's just been taken down for maintenance. But she's been down for maintenance for two years. So she's probably not coming back. There's Emma asking questions. She was supposed to give answers, but she was actually asking questions instead. And then you have Knud from Måns Kommune. That's a sad story. I won't get into it. But it just means that there have been attempts at it today. But I don't think that they had the technology available to become successes. And obviously Microsoft had to give it a go as well. I don't know whether you heard this story. But... You have people who have been working with artificial intelligence for 15 or 20 years. They know how to game a simple algorithm. So this is what happens if you start feeding it words. And it will start structuring and thinking, okay, this is what the user wants to hear. And it will provide that answer. So obviously there are some pitfalls still. And they have to take it down immediately. Which was actually kind of also what happened to me. But another story, I think, if... No, I mean this is serious. If the alternative is talking to a TV marketer, we all hate that, or listen to corporate communication as this guy. I think he's one of the bad examples. Then I would rather talk to a chatbot. I would much rather talk to a chatbot that would ask me questions. And then slowly start to understand what my problem was. And figure out, well, Anders uses these kind of words. He might actually swear a lot. So I'll actually... Down the road I'll start swearing so that we'll build a bond based on that. I would much rather have that than this guy, Torben Sandler. But we all know that. So to the bot that I built. It was working last night. But then at some point it was... Me seriously started using my own Twitter account, my private Twitter account, to send out just complete nonsense to people who had mentioned me in a tweet for the last six months. I have no idea why. So I have some red eyes because I stayed up late. So right now I've given the phone to my son. He's about two years old. And the reason I think that is a funny analogy is because that's sort of the level, cognitively speaking, that these chatbots are at today. So I might as well today use my son. But I don't know who's going to win the race. At least it's... I paid $9 to have it hosted at chatbots. I built it on the Pandora bot. I took a... I took a basis content package, modified it with answers that fitted the context that I was looking for. And we uploaded it through... On Twitter. And it was actually making sense for half a day. And then it just blew up. So right now it's way cheaper and easier to just ask August. Because I think two or three out of ten questions he will actually make sense. And that's better than the bot. So to sum up. I think my time is more or less up. Okay. So to sum it up. We are building bots at Innovation Lab. Sputnik 5 is a micro company that I'm working in. But we're working more conceptually to help clients figure out where does this make sense to try it out. And as we've already heard from the speakers before, you need to build, learn, and adjust. As this is also a great example of. You wouldn't take a bot and just throw it out to if you have like 100,000 followers on Twitter. And then just cross your fingers and see. Ah, we hope that makes sense. So definitely we see brands moving into this. Seeing that, okay, people are not really using our apps. We might as well build functionality, transactions, whatever we want to do into the messaging platforms that are already out there. The second part is that it doesn't have to be all text. It could be graphics, animations, whatever kind of file format you want to build into it. That's actually possible today. So it's not flat 2D text chatting. And on a general level, we see with bots not dying out, that's going to be. Sorry, with apps losing their momentum. It's going to be a re-bundling of functionality. Because now today, I don't know whether I'm an average user. But I have about 200 apps. And that's just too much for me to oversee. I can't find them. I forget that I have apps that I can actually perform stuff. But I'm on two or three social media platforms. And I have a few messaging apps that I use. Why not re-bundle functionality and put it in there? So I'll just ask, send my mom some flowers instead of having to find the app from a florist. And people are already familiar with messaging. So that makes the onboarding easier. And the final two parts is that you want to design for the context. It really is God if content is king. And make it convenient. And the way that we try to do this, at least conceptually, is to figure out can we find some uncertainties in new markets for a company. So don't start with your core user base and just put a chatbot out there. Because they will just wreak havoc on the relationship that you build. So thank you for your time.