So from one of the things that we've learned from working with thousands and thousands of customers doing this, is the biggest difference that we see in people who are successful, and people who are not in terms of machine learning and AI, it's about focus and it's about speed. I wanted to just show you this, it's is based on two of our leadership principles at Amazon. One is called "customer obsession." It's really about, thinking about the solution, and working backwards from the customer, not thinking about the technology, and working backwards to the solution, but thinking about the customer working to the solution. Part of the way that we teach customers to do this, we just talked about training in our "yes and" [an improv] exercise. Training is the number one question that we get, number one question. Part of what we do is before we get into the technology, we ask companies to write a press release. Tell us what you're trying to do from your customer's perspective. Short headline, one page, what is the problem you're trying to solve? What is the solution that you propose? What would customers say about that particular solution? Would that be something that they would like or not like? The second thing we have them do, as part of this training, is to write a frequently asked questions [FAQ]. Now, this may seem easy because, when I first wrote my first set of questions, I thought, "Well I'll just write my own, but they've got to be really tough." It's got to be what would your customers say if they were sitting and having lunch with you, that they loved about the solution, or if you're having cocktails, what would the customer complain about this particular solution? You need to think about what are some of the toughest questions that you're going to get? Then you start defining what you want the user interaction, and the user experience to be. Always start with the customer in mind. What do you want to have the customer experience as they're going through the design of AI? The reason this really matters, in particular, for artificial intelligence, is that there are so many companies that come to us, and they say, "We want to do AI," and we say, "Okay, well, what problem you're trying to solve?" They're like, "Pick one. We want to do AI." Anybody had that? "My boss read about it in a airplane magazine, we want to do AI." That's not the right place to start. The right place to start is: what is the voice of your customer telling you that they need? And then using AI to help solve that challenge or that problem. I do know because I had a round table with a group of entrepreneurs yesterday. Any entrepreneurs in the room? They told me that if they put AI anywhere in their company, that they get a higher valuation. I think that's interesting, but that's not sustainable, that's not long-term. You really need to think about what that customer problem looks like. I wanted to talk about one entrepreneur in Silicon Valley, who invented something, thinking about the technology first, and what he created is a AI powered salt shaker. Sounds cool, right? An AI powered salt shaker. You actually use your app on your iPhone, if you want to dispense salt, so you don't just pick up the salt shaker, you first have to plug it into your app, and you type in, "I want a quarter teaspoon," or "an eighth of a teaspoon." But amazingly, that salt shaker, will remember your patterns. It will learn, and say, " Wow, in the morning Sandy wants a quarter teaspoon, and in the evening, she wants an eighth of a teaspoon." But wait, there's more using serverless and container technology as well, they also connected it to Alexa, so it also has lights in the salt shaker. This blue, will turn to red, and pink, and it will show you a light show, that again learns your mood. Do you want it fast at night, and slow in the morning? Now this uses every buzzword I can think of. It's got IoT sensors in it. It's got AI, machine learning, it's even got connection to Alexa. Do you guys think that this is working backwards from the customer? How many of you would use an app to dispense your salt? This company did get funding, but did not make it. They were really cool with AI, but they did not really drive what they needed to drive. As opposed to a company like Lionbridge. Lionbridge is the world's largest company that translates languages, and they localize content. They saw a problem with the cost of doing that manually, as well as the time, that you would spend in translating manually. What they came in with is that was their challenge, and they use machine learning and machine translation, in order to solve that particular customer problem. It reduced their costs by 20 percent, and reduced the time that they have to translate and localize by 20 percent. Do you guys see the difference of the two examples? One is tech-oriented, one is customer oriented.
