So are you feeling like a chef? An AI chef? Feels pretty good. You've been cooking, you've been organizing, it looks pretty good. You're plating things and realize, well, what if I just ordered out? What if I want a frozen dinner that's not bad? Well, right! You want to borrow someone else's work through a plugin of some form, extend what you're doing, and make you even more amazing. So let's reach into the freezer and try out a kind of frozen dinner, a design thinking plugin. You figure out the SWOT stuff, let's go grab another one, work with it, and keep cooking so that you become the best AI chef ever. Let's do that. So we're gonna tap into the design thinking plugin, get excited. Design thinking is the five-step process, empathize, define, ideate, prototype, test. We'll get into it. We're going to dig into a plugin to take advantage of this incredible plugin that I hate to say this so boastfully, but I made it and I was excited when I made it. But again, if you don't like it, you can change it because it's in the files. So design thinking, it's automated, it's pretty amazing. It's a plugin and the plugins are important because as you take your AI capabilities and and make them into transportable plugins packaged so that someone else can use them, you'll be able to use them in all kinds of places. Have you tried out the chat GPT plugins? That's pretty cool. And at Microsoft build, you heard the EVP of AI, Kevin Scott talk about more ways to use AI plugins across the Microsoft universe. And just in general, it's a great way to share recipes. Okay, let's do a quick inventory. Inventory. One, you know how to make a kernel. Fantastic. Two, you can make semantic and native functions. You can do a lot with these. And number three, you're able to use the business thinking, log in, and you're able to process SWATs in ways you could never imagine. Well, I'm not sure if you imagined them before you took this course, but I'm kind of bewildered at how amazing this stuff is. And now we're going to do design thinking plug-in stuff. And here you are. Just a reminder, here we are. Okay. So first off, what I want to do every time we want to get a kernel ready. Let's get a kernel. Get a kernel ready. I like work down so much. Okay, and I'm gonna bring in the boilerplate that you already know. Let's get that kernel ready. Let's get the message we like. A kernel is now ready. Very good. And next up, what we're gonna do is we're going to do what is important, which is to start from customer feedback and do design thinking. So let's start back with the customer, shall we? How do we do that? Well, first off, we wanna bring all those slots back into view just because they're useful, bringing in the slot questions and the responses. And what I'm gonna do next is gather a bunch of customer feedback. Let's get that, shall we? Let's get some customer comments. Okay, so this is, again, this is our business snapshot as questions and answers. This is the customer comments, 10 comments that are sometimes positive, sometimes negative, that's how business works, right? And what I wanna do now is access the design thinking plugin, which now you know is a folder somewhere. Let's sort of get our bearings, shall we? It is a folder somewhere. And let me show you where it is. Right? So there's a plugins SK, there's design thinking, there's a define and empathize SK prompt.txt. What do they look like? Go ahead and open your file browser. And if you want to look at the empathize prompt, it looks something like this. It is following our anonymized comments from customers, takes an input, and then convert to JSON, the list of sentiments. These models are very good at calculating computing sentiments. So they're good at that. That's the Empathize plugin, does some magic. And it's gonna take these customer comments. It's not gonna take these swats, we're gonna use them later. And we're going to run this input with that design thinking plugin. We're gonna access the plugin directory. And then what we're gonna do is we're going to do our favorite thing of importing the design thinking plugin. and we're going to run the kernel with the empathize function. And we're gonna take the customer comments over here, all over here. And then we're going to have it run the empathy analysis from this escape prompt file inside this place in the file directory. And note that if I change this and save it, it changes the prompt. Do it yourself. And you should do it. Let me show you it running so we can look at the food and pick it out around on our dish. So what did it do? It categorized different types of complaints about seat condition, praise for the garlic pizza. Someone likes it. Frustration because always a new person is serving them, doesn't know the layout of the restaurant. It's like a neutral response, like why aren't there calzones? And lastly, there's no online ordering. The pizza shop owner hasn't done any digital transformation yet. So what do we have here? We have a design thinking, instant empathize, magical AI, generative moment that took in these comments and it generated a sentiment map. So now that we have that, you want to remember what is design thinking. And design thinking is a simple set of steps. There are five steps. Most people argue that there's no steps in design thinking, that it's a set of activities you can sort of do in any order, really. But we did empathize, and we used large language model AI to summarize the feedback. Now once you have this feedback, you can then convert it into input to defining the problem. You know, you can never solve a problem unless you understand it well, as you define it. So let's rerun the plugin with the original information. It's calculating that. Okay. These are the responses. And now I'm going to send it into the define plugin. Here we go. Okay. So, okay. My result equals, wait, I can't help but love typing because that way I can feel it with you. So just bear with me. I'm going to use the empathize plugin. I'm going to feed that output into plugin define, okay? And then I am going to give as an input string, I'm going to give the feedback, customer comments, okay? And then I am going to output them. Let's get a fancy plated statement there. So what's going to happen is now I'm not just calling the empathize component of design thinking. I'm calling empathize and define it's going to do this and chain them together the way the kernel processes things sequentially and what happens is. Let's run that. It's going to take the empathy from the customer feedback it's going to feed it into define which is going to hypothesize what kind of problems exist so. So first off, great garlic pizza, okay, and well, great chef probably, and great ingredients, high turnover rate, insufficient training, absence of calzones, well, maybe we need to sort of rethink what we serve the customers. And so this example of taking design thinking to number one, take the feedback, and then define the problem. But design thinking is actually about innovation, and innovation is about making new kinds of ideas. And so you might take this knowledge you gathered and push it into an ideation engine to ask questions about what you could make as a solution. And then you want to prototype it and you also want to test it with real customers. This is empathize, define, ideate, prototype, test, five different components of design thinking. Well, you can do it with the design thinking plugin as well. Well, let's show you it running with four phases, all chained together. We're gonna empathize. We're going to define the problem. We're gonna ideate solutions and we're going to suggest prototyped, you know, some of these people paper prototypes of things. We're gonna prototype things out of paper as a next step. We're gonna stream this all in into a kernel. And what happens is something, I think once you do this kind of work, this kind of business strategy work within large language models, I don't know about you, but I am bewildered. And I'm also excited because a lot of bad ideas can get thrown out quickly. You know this word hallucination sounds pretty bad, but it's actually what humans do. They make mistakes. So when you're playing with business thinking, design thinking, these two plugins, you'll discover that, hey, it actually gets it wrong sometimes because often these kinds of questions don't have a single correct answer anyways. So this is what's suggested. So to improve seating comfort and cleanliness, create a paper prototype of the restaurant seating area. Well, that's how to use paper, right? Offer discounts, make paper coupons, train servers, make a simple training manual, cheat sheet out of paper, make a loyalty card with a punch cards. This is like 10 ideas that I think if you were to go out and pay for them, you could pay for them, but here you got them essentially close to free. I mean, tokens do cost money, but it's kind of amazing. Now you might say, hold it, John, you're getting off the hook because you didn't let me test the idea. Well, it turns out that more and more services are emerging where you could basically give a large language model a role like a 40 year old person with a certain situation and suggest how they would respond to this prototype. All right, let's do some of that, shall we? So let's make a prompt. We make a prompt to test the output. Let's say I wanna be a 40 year old man who has just finished his shift at work and comes into the bar. They are in a bad mood. Right. Close that. And they are given an experience like input. We're gonna give the experience and then summarize their possible reactions to this experience. Okay, so now we're gonna call this a test function. We're gonna do this inline just for convenience here. We're gonna add the configuration information and let's take one of the examples above. Let's choose one of these things here. Let's say someone walks in. And they find a simple loyalty. So this is one thing we could do. Let's pretend we've made a loyalty program. So a simple loyalty card that includes details, other importance, memoirs, is given to every person visiting the bar. So that's a potential paper prototype. Let's now test this. Let's run this prompt through the kernel and let's print out the results. So we're basically taking a prototype idea and we're going to test it on a synthetic person basically. And of course, there's like a sort of a special message from OpenAI that I can't predict it, but I did ask in the prompt, please go ahead. So however, some possible reactions could be, I might feel indifferent because I don't care about your loyalty card, or I may feel annoyed because I just go into the bar, or I might be interested because I might wanna be someone who takes advantage of being a regular customer. So these are really useful responses that you can imagine applying to any of the prototyped ideas and simple example of how to use this design thinking plugin, it's different functions, and just really helping to test ideas around the small business owner, pizza shop owner, construction company, you name it. So let's remember what's happened so far in this lesson, which is, I want to keep saying this over and over, which is there's something we haven't done yet. What is it? Well, we're showing off the completion engine. The completion engine is generating things. They're pretty cool. We haven't done any semantic similarity. We're going to do that next and just wait for how that feels. That's a different way to do things. Let's go into it.