Get Microsoft AI Certified with this 30 Day Challenge

I love the lab exercises in this, uh, challenge that Microsoft has put together. They're really good. They show you some realistic examples of how AI is being used in the real world. Um, so I'm going to encourage you that even if you go faster than maybe you want in order to hit the deadline, which is next. It's a week from today. You have 7 days to complete the challenge. There are 16 modules, but you'll see in just a second how quick, um, those modules can go if you let it. As with all things, you can burn through a book, right? Or a workshop, um, really quickly and not really gain much. So I will encourage you to go back. All of this content is free. Finding it is hard. So now you'll have a link to go and go through that content. But you will want to make sure that— oh, I did it in about 3 hours a day. Thanks for that question. So how many hours a day did it take for finishing it in 3 days? And I might have finished it more quickly. I had someone finish it in a day, right? So it's not like— I don't even think it's 8 hours total. So you'll see people in the certificate. Oh, I'm going to show you right now, actually. Are you able? Oh yes, thank you. So I'm going to send you 2 links. This first link that I'll put in the chat actually is the link to get reminders when we do challenges. I'm hyperscaler agnostic, though I love Microsoft, which is why I'm doing it first. I also will be doing challenges for Google and Amazon through the rest of the year. I started a nonprofit and have gotten grant funding to do just this work, and it's super exciting. And I just love the idea that more, more people can get exposure to the kinds of solutions we can build with this stuff. So first The first one, obviously you're starting to see a pattern that I can't easily type and talk at the same time. So inclusioninspires.com/ai-challenge. They will all be at the same address. Once you sign up, you get access to all of them that we'll do throughout the year. And like I said, it's about every 6 to 8 weeks that I get them launched for. We have 4 more, I think, on Microsoft. There's 3 on Amazon and 2 on Google. So it's exciting times. Okay, so that first link is to sign up so you get notified of the recordings that go out. I did a 30— I have 31. They're less than an hour, 40-minute recordings today. I'll try to keep you at about 40 minutes. And my goal is then you have an hour to take the first, the next 20 minutes, and I'll stay around. So if you need to chat with me, you can, but just go through that first unit just so you can like literally start those engines rolling. So you have an hour here. I'm gonna spend the next 40 minutes going through the challenge, showing you exactly what to do. It's— I helped write some of this content, specifically the computer vision stuff. I'm very proud of it, but I think you'll find it. I mean, it's just, it's really intuitive. And if you're into AI, you'll find it's kind of, it's pretty exciting. So So sign up for that challenge. That will then give you an email, but I'm going to give you the link right now to the Cloud Challenge. This is where all 5 of the challenges that Miami-Dade is hosting for us. Miami-Dade is the college that Microsoft granted these certification vouchers to. It doesn't matter. You don't have to go there. There's— I mean, they just pick an institution to facilitate the transfer of those vouchers. And obviously Miami-Dade, actually, just to give them a shout out, they are building the first applied AI Associates, Bachelor's, and hopefully Master's program. And applied AI is literally what this is about. So it's less about becoming a data scientist and learning to write algorithms from scratch, though this challenge will dabble tiny bit in there. It's more about leveraging existing artificial intelligence models, customizing them for your use, and then building unique assets like unique models from things that have already existed, right? It's kind of like leveraging existing bass lines in music. Very, very similar for those of you music people out there. You know who you are. So we have several available challenges. You'll hear, you'll see here. Did I paste it in? Yes, I did. Okay. So if you go to that second link, you'll see this is the Miami-Dade site. However, every time we run this challenge, it might be Miami-Dade. They are huge fans and friends with Microsoft. So, well, chances are we'll run this multiple times. But it also could be another institution. I've done it probably with 7 universities so far. It doesn't matter. But you'll go in, you'll register. I picked, of course, the AI Fundamentals. There's tons of other ones in here that are pretty cool. Power Platform, which is our new low-code, no-code on Microsoft. Data Fundamentals, Cloud Fundamentals. Some of these you might find you wish you had taken when you get into some of the nitty-gritty of these these modules that we'll talk about. But there are no prerequisites for the curriculum that, like, literally, it's gonna be kind of like me giving you tracing paper, right? As an artist, these will be a bit of like tracing paper over something that's complex. You might not understand all of the artistic technique that went into it, but you'll be able to trace it and see what that outcome looks like. Build a little muscle memory. And remember, the goal is you'll understand the fundamentals, and that 100% you will get, even if you don't know every, you know, detail of how each one of these models that you're gonna be using is made. So I'm gonna go ahead and click on that link. That'll take you to the challenge page. It'll tell you how many days left we have— 7 days, 4 hours, 48 minutes, and 40 seconds. I have, like I said, have gone through this multiple times. You will have to create an account at Microsoft. So if you don't have one already, you'll have to create one. You will also have to sign up for the free Azure account, which it— the, the modules will walk you through creating this account, give you the links to do it. I'm gonna show you what it looks like so you know you're in the right place. You also have unlimited access to me for the next 7 days to get this done, right? So if you get stuck, you can email me in the challenge email that you'll get. It has all my contact information, but I am like, this is 7 days, let's do this. You could totally do it. Um, we've got 68 people total registered for this challenge, and the only people that have promoted it is me. So, so it'd be cool if we had, I don't know, 30% like actually finish it. Um, and I said this at the beginning, I did kind of shoot myself in the foot because I admitted that only 10% of people finish what they start. Some of you know this statistic just being in the world, but only 10% finish things they start. You have started this. I encourage you to finish it, but know that 90% of the world kind of reads the first chapter and stops reading, right? And even if they want to finish, like, it has nothing to do about intention or boredom. Like, it's just cognitively, it's hard to keep our human attention, um, on something. So that's why this 7 days might be better than those who started with me, you know, 21 days ago, um, because 21 days is a long time. 30 days is a long time, but 7 you might be able to just knock this out. I encourage you to. If you scroll down, you'll see these are all of the people who finished the challenge, which means— and Sandeep actually sent me a message, he's like, I got my voucher, thank you. Um, some of you are not from the United States, uh, which is totally fine. Um, but it will ask you for an email address or a physical address. So if you are not from the US, you do need a US physical address send me a direct message and I will send you the address to use. And I have that relationship with Microsoft. You will not get in trouble. But they did ask me to so they could identify those. And that's just— it's like a logistical— the way they coded it was wrong. Obviously it wasn't— it wasn't inclusive of our worldly audience. I don't know why. It seems kind of strange. But anyway, so that will happen. And I will, like I said, I'll just send you that address. I won't give it to you here because I don't want you to use it if you don't need it. They do wanna send you swag, so you wanna get that and give your own address for it. But lots of people in the challenge. I'm still on the first page, so that's exciting. It doesn't update because I went through 3 modules today trying to like, you know, show you all like, look, I'm on 5. It didn't. So it didn't. But I did go through 3 modules just like an hour ago. It takes a couple hours to update, so don't be like refresh, refresh, refresh if you're into the leaderboard. All right. So once you're there, it's going to tell you instead of continue learning, yours is going to say join the challenge. And once you join the challenge, which means you have to create your account and it'll get you in the challenge, it's going to take you to the curriculum and there's 16 modules. I'm going to scroll through them faster than you can read it. It's super cool stuff. I want to walk you through what you're going to be learning to hopefully motivate you to finish. The first is very simple, high-level overview. This is why you'll start to see it, why you're gonna be able to go through this pretty quickly, right? It's like 3-minute snackable videos, right? Like click, click, click. It's very simple, especially the early stuff. These are quick and intentionally so because this is of course Module 1. As I'm gonna go back. In module 2, it's like, all right, let's get this stuff started. It actually is going to encourage you to open up machine Azure Machine Learning, and I'll show you what that looks like. The first thing it says is if you wanna use machine learning on Azure, you have to know what the cloud is. It's not gonna teach you about the cloud, but hopefully you have some context to what that means. But if not, that's okay. It's gonna walk you through. And you'll see here as I take you into Azure, it's gonna ask you to create a machine learning compute cluster, which really is literally, if I had like one of those big boxes we used to have at our desk, the desktop machines, like I have one, it's just not in this room, but it's literally gonna hand you a couple of big machines that have, you know, hardware in them, network cards, all the things you need. Here's the beauty of it. If I wanted to do what we're about to do, I'd have to know a lot about hardware, networking, server configuration, like all this nonsense that has nothing to do with the actual AI, right? Just to get started. So all of that is peeled away. We don't have to do any of that, but we do have to actually create the resources. You will not be charged for this compute in the context of this challenge. However, you will get charged if you leave these resources running. So you can create them for free. But your job at the end, do not leave them running. Every good cloud architect has had the moment where they're charged like $1,000 for stuff they forgot to shut down. Don't let that happen to you, right? And it'll tell you, uh, in the challenge exercise, please go back and shut down the stuff you haven't. You'll notice here, I just created this 2 minutes ago cuz I wanted you to see, because when I finished my challenge exercises, I was like, delete, delete, delete. Here's the cool thing. You will need those resources every time you do the next exercise. But here's what I do. Bonus points for, or maybe gold stars for those who choose to do the same. I just went back and recreated those assets, right? Building the muscle memory. So it'll say I need a compute cluster with a certain type of physical machine with a certain amount of, uh, instances available with a certain type of security. So I make all those selections. The first couple times I kind of have to look at the instructions, but the third and fourth time I did it, I didn't have to look at the instructions anymore. And now you can claim that, like, you, you know how to configure a compute cluster for machine learning, and that's a pretty unique skill in the world right now. So just by really repetitiously doing some of these exercises, you're gonna gain some pretty useful skills. In the industry. But that is, it's called AI infrastructure. It's kind of this hidden skill. Most data scientists and data engineers don't know anything about what it takes to run what you're doing, but I'd encourage you to take a look around. So this is in Azure. I create the hard, like the hardware that all my cloud software is gonna run. And now I go over to what's called the Machine Learning Studio. And so I'm not gonna I don't wanna get you too confused, but this Machine Learning Studio allows you to use what's called a WYSIWYG environment. It's basically what you see is what you get, right? You have the ability to click and drag components onto a palette or a freeform surface. You connect them together and it builds a machine learning, what we call a pipeline, where you inter— insert data at the top and a prediction or a probability comes out at the bottom. It's a bit magical. And like I said, if you do it as fast as I did it the first time, you won't completely understand what you did, but it'll seem kind of like magic that you were able to. For example, I think one of the exercises we use is auto pricing, right? If I put in, you know, the make, the model, the year of a used car, it's going to, based on pattern match data, it's gonna do a regression analysis and gimme a probability of how much that car's probably going to sell for. It's in my mind, I'm like, wow, that's pretty cool that I was able to do that and not write any code. So that is the, the world we're moving into is how do we build solutions to, let's say, a used car salesman's problem of figuring out what to price things at in a data-driven way that's inclusive of geography, location, demographic, right? How do I make sure that I'm not accidentally pricing things differently because I'm talking to a different demographic of a person or I'm talking to a different zip code. Do you know that that happens? Like inherently human salespeople will change the price that they offer based on where they think you live or the kind of car you drive or the purse you have. It's crazy. So AI will do the exact same thing if we're not intentional about it. So one of the things we'll encourage you to do as you go through the exercises is look at the data. And then we're gonna ask, ask you to actually fix the data because humans are, you know, crazy, terrible, biased people. They're actually wonderful, but they are biased. And you are going to want to detect that bias before you build any AI and actually complement little pieces of data that complement that dataset to make it less biased. We call it mitigation, and it's a human. Who does that, right? The AI doesn't know how to mitigate itself. It's why the human in the loop is so important. All right, so going back to, so this is the machine learning studio. You'll do 5 or 6 exercises here. Let me go back and just walk you through the, oops, I, I think I passed it. So let me go. There we go. I'm gonna go back in to that content. So the getting started is easy, takes 34 minutes. The first lab exercise takes about 45 minutes if you read. If you, you know, it's like I did it in 20 minutes, I think, in half that time. So just it depends on how fast you read. It depends on how long it takes you to look at something and it tells you what to do and then do it in the console. The other thing that I'd say 90% of the people that have completed the challenge so far did is they read the steps and they looked at the screen captures and went, okay, I kind of get what's going on here. Because remember, this is a fundamentals test. It's not going to test you on what was the thing you clicked on on the second page of the third thing, right? Like, the benefit of the lab exercise is to reinforce the fundamentals. So it is possible for you to get the gist by just kind of walking through the narrative, um, but I'd encourage you to play around if you've got the time. Like, play around with this. It's not going to take very long. So one of the things I do want to share with you, if you're not familiar with AI, is that there's lots of different types of AI. We are focused on the fundamentals, which means that we will primarily be using supervised learning. And in supervised learning, there's a couple different types of models. There's regression models, which are the typical models that say, you know, I am 95% sure that that's a cat and not a dog. Right? Classification is often something that, we see used as well. And then the other type, uh, time series forecasting. So in other words, like for example, if I wanted to know, um, at a certain season when I, when I should sell a used car, right? I wanna look at the data and figure out what specific time series I should actually do that activity. And there's lots of different use cases for this. So one of the best, um, parts of this activity is that we're gonna give you real use cases. One's in automotive engineering, one is in healthcare, right? So we're gonna give you real tangible use cases for how to use these automated machine learning processes to solve real-world problems. I get like multiple startup ideas every time I go through this material. I'm like, oh my gosh, there should be a startup for that. And, and it's also something that if you know startups, you might wonder if they're using these models because they so closely align with the services these no-code models offer. So we're gonna talk about building regression models, building classification models. These are all different types of supervised learning, which just to give you a slightly more deeper clarification on that, what that means is that you're going to create datasets and you're gonna split the dataset into two pieces. You're gonna use half your data to train the model on what good looks like. Oh, I'm not very familiar. Say, this is what a properly priced car is, or this is what a dog looks like, this is what a cat looks like, right? So it will provide very specific, you'll provide very specific examples that the model's gonna use to make a prediction. It's probabilistic. And then you're going to have another set of data that you're gonna use to test if those predictions are actually working before you move into production. So you'll notice, you'll hear the terms training dataset, right, as we go through this. And that's just to indicate, or it'll ask you, I think one of the questions is around, you know, when will you need two datasets, right? When will you need a training dataset and an inference dataset? And of course, the answer to that will be in these supervised learning models or regression or classification models. All right. So, so those types of models are the most common and easily accessible value that businesses can, can attain from AI. And these are probabilistic models, right? So sometimes people get confused about what is AI and what isn't. And what it always comes down to is deterministic, meaning I often hear people talk about decision trees. I used to write code that looked a lot like AI, right? Where I would so clearly know what the customer was gonna do, I could choreograph exactly what they were gonna do step by step by step in code. And it would look to the user like, how did you know I was gonna do that? Right? Like it would look magical to the user, but it was still code. It was still very specifically, if you do this, then do this, then this, then this, switch cases, if, if else, right? Very conditional decision tree, even pattern matching logic. But it was all programmatically defined. There was no presumptions. In machine learning, we move to a probabilistic model, which is where these supervised learning models, their whole job is to give us a percentage back, right? A confidence number. And so if a model gives me a confidence number of 68% or 22% or 99%, that number doesn't mean anything until a human like myself determines a threshold for a decision. So I have to go in now and say, okay, if you tell me 70% or better, make a decision for yes. If you tell me 70% or lower, say no. But I, as the human, have to write that code to determine when those thresholds are engaged. That's the difference with AI, right? One, it's deterministic. It's already predetermined, written out. You— that decision path has already been laid out. Now we're gonna let the customer make specific choices and then we're gonna predict their next choice and decide whether or not our model made a strong enough prediction to take action. And you can understand in organizations like, or in industries like healthcare, how dangerous, right? That could be if we get those thresholds wrong. Or in retail, how maybe loose those thresholds can be. So it's very industry-specific, which is why it's cool that each one of these will go through, um, and talk to you about different industry examples. All right. So once we get through that, the purpose of the first 4 labs are to walk you through what the different types of the most common high-value machine learning models are, which are the regression, classification, um, and clustering models. Then we get into the fun stuff, which are applications of those baseline models. One application is computer vision. I use this model all, well, all the time, but it's, it's computer vision is the ability for a computer to predict what it sees on the screen. You often see this in like bounding boxes being placed on a screen or, you know, if you ever watch autonomous driving footage, it's got all these boxes and percentages all over the screen saying, I think that's a dog. I'm 20% sure. I think that's a human. I'm 90% sure. It's all of this analysis that's happening in real time. Computer vision provides that. Now, computer vision is trained on a baseline model and a baseline set of data, which means that it won't know specifically how to solve any specific problem other than, is that a tree or a cat? Right? Like, it'll be able to make these general, references. If you want to create a computer vision model that you own the IP for, It's customized to your use. Let's say I'm a produce grower and I wanna be able to tell bad produce from good produce, or I wanna be able to detect to the day when produce will go bad. I'm gonna wanna customize that baseline model. And that's what you could see here, the custom vision services. This is one of my favorite exercises. It's so fun. I went in and I chose to use, uh, Chip and Dale, the Rescue Rangers. I don't know if you know, but a movie just came out about them. Anyway, I went and created, uploaded 21 images of Chip and Dale, and trained the model to tell the difference between the two so that I could pass it an image of Chip or Dale and it would be able to say, oh, based on what you told me, it's Chip. And it's just fun. Like, it's fun to be able to create software that makes decisions on my behalf. And it's a way to really, like I said, just implant some of these fundamentals. Object detection is another function of computer vision. So detecting objects, basically it says if you're looking at surveillance footage, I wanna get notified when there's someone on the premises that shouldn't be there, or when there's someone somewhere during a timeframe that shouldn't be there. So that object detection is, again, I, a computer vision model can detect any object, but it's my job as a human to actually programmatically say when this type of object is seen, or if anything is seen at all, I wanna know about it. And an alert can be created, or any code could run, right? An alarm could be set. Tons of things could happen. But that object detection is really important. This was just in the news, computer vision related to facial recognition. Microsoft is actually pulling out of facial recognition. But just so you know, Amazon, IBM and Facebook all had production face recognition disasters that caused them to change how they went to production. So this is not new. Microsoft's super valiant in going proactively, right? Not waiting for something bad to happen to do it. I, I like it. I like the idea that they recognize the flaws in a facial recognition model that's not properly trained. But I do hope, you know, that You can do facial recognition, right? It's just expensive and requires lots and lots of data, and that data is getting more expensive by the minute. So companies are just not interested. It's not financially feasible for them to invest in software that works. So that's why. But it's still cool. So we teach you that. Oh gosh. Okay. Oh sure. I know. I don't wanna upgrade. I don't think. So we detect Let's see. Oh yeah. So facial recognition. The next one is computer vision, but this is OCR. So optical character recognition, I think it's called. Doesn't really matter. It's like HTML. Who knows that anyway? Oh, did I get it right? Okay, good. It's one of those things, right? You're like, wait, brain. Okay. But it's really neat because we now not only have— oh, thank you. All right. Ah, Andrea, nice. So all of these specifically OCR models, they started off with being type, like, you know, text on a screen that you could read. But now they've gotten very fancy. We've got two handwriting models. It can even recognize, dare I say, like medical professional handwriting. You should test it. It's pretty impressive. So I've had like, We did a government project where people had forms that we wanted to ingest digitally, but because they were in an office, they'd slap on Post-its and write on the forms and then also ingest that. And so we had to figure out a way to be able to search for content that was on the Post-it on a form. Crazy. But it's now entirely possible. And like I said, the only skillset we needed was a developer. We didn't need a data scientist, right? We just needed this model and a really patient developer to curate all of that. Pretty cool. Yeah, exactly. They get taught that, I think, Emily, to write in like a way that's not legible. I feel like it's job security or something. So the last couple are really useful use cases, and this is why I encourage you to kind of go through the whole thing, even if at first you just go through the content and then go back and do the labs. But one of them is Forms Recognizer. This gives us 100%, 100% accuracy on digitized forms. So I don't think I have a form. Yeah, like this is like a government form, right? It's the exact same form for the entire state. Everyone uses the same one. It ingests it, digitizes it. You can also pass in— I don't know if you know this, but if you Google driver's licenses, a bunch of fake driver's licenses will come up, images.. And you can grab those images and pass them through Forms Recognizer, and it will extract all the data off of those driver's licenses and put them in code. But what happens now? The benefit is, is like, if I just give a scanner to a dentist office, I never have to fill out the forms they give me. They could just digitize everything, right? Extract it from my driver's license and my insurance and automatically fill out these horrible packs of paper that I have to fill out for all my children. Every time we get a new dentist. Alright, so that is a really cool one to play around with. It also again is a low-code implementation. Then we get into the ones— some of you know me from before, I'm an Alexa person. So I grew up on Alexa. That's my AI journey started. I love language, natural language specifically, and voice. So you can see here, you'll have a bunch of services around language. And it's not just spoken language, it's written language, and what you'll find is actually they are the same for the most part, right? There's just one additional model that's used to turn text into speech, but almost everything is happening in text. So I will encourage you to go through these because it'll give you a way of manipulating and more importantly exposing anything you're doing in technology as text that can then be human readable and writable in a natural language format, which means your customers can just talk normally to their, your systems, whether writing it in or speaking it and invoke, you know, invoke complex functionality. It's pretty cool. So we, oh, here's another one. So recognizing and speech synthesis. So recognizing speech is cool. I use it on Alexa all the time. She still hurt me. Anyway, I use it on that device all the time because I have one in every single room. And when I go shopping, it'll tell me who said what to Alexa on the thing, which is kind of nice because I'm like, oh, gummy bears. No, no, Mommy didn't ask for those, right? So I could be like, kid number 4 asked for those. No, we're not doing that. You just got back from the dentist. No. So it's kind of a neat feature. And then also though, voice synthesis, being able to create, which you'll get a chance to do if you choose to create your own voice. So you will upload a snapshot of a bunch of audio data of your own, and you'll be able to create your own synthesized voice of yourself. Yes, it can be Black Mirror-y. I know it could sound creepy, but it's kind of cool. So I encourage you to do it. I created the voice of the um, who— he was like the head of the CIA in the '60s. His name was J. Edgar Hoover. I recreated his voice, and it was very creepy because he doesn't live anymore, and I was able to have him say things that he never said. Um, so it's, it's, it's an interesting time, but there's real practical reasons for voice synthesis. Val Kilmer uses voice synthesis today to replicate his own voice because he lost his voice in a degenerative disease that he has. So it's interesting times, especially in healthcare. Yes, Bridget, absolutely. I know it's scary and cool though. Scary and cool. That's like my new verb, scary cool. So at the end, we will do my favorite activity, which is build a bot in minutes. Chatbots, I know you probably know this, got a bad rap, right? So they got a bad rap. Over the last couple years, but Microsoft built a framework that allows you to ingest any PDF into the knowledge base of a bot and publish it. So within 5 minutes, you can take anything that's on a website and turn it into a conversational app. And you do that with this, with this module. So I encourage you to get to the end and, and play around with some of these cuz there's, it just opens up a whole world of opportunity. And those are the types of things, like I said, I get to these and I'm like, oh my gosh, I should have a— I should build a business around this. So if you're an entrepreneur, you're gonna like love this stuff. If you're an intrapreneur or working inside a company, you'll— you could potentially do some magic where people have struggled, or you're like, you know, if you're got a feature you always wanted to release, you wanted to add translation, which now is like 2 method calls away, right? One, to detect the language you wanna translate to, and two, the actual live real-time translation. I mean, this stuff is— it's the best-kept secret, which is why I'm doing this challenge to hopefully introduce more of us to it. So there's a bunch of stuff in here. And like I said, I do— of course, I have 40 minutes. This is intentional. I could, of course, upgrade, but I'm not because then I'll talk to you guys for 2 hours. So I'm trying to be good. But you have 20 minutes now on your schedule carved out I'd encourage you to go ahead and get started, right? Go in, get at least the first Getting Started on Azure AI. Do the first module. It's 30 minutes. It'll take you less than that if you go through it all in one— at one time. It's like 3-minute videos. And then if you're super adventurous, go and do the automated machine learning. None of this requires extensive amounts of coding. None of this requires you to be a data scientist or statistician. All those skill sets will be helpful, but you will learn a lot about some of the tools the biggest companies in the world are using right now, and it will give you a leg up. And more importantly, you will qualify for a certification that you'll be more than qualified to do and pass, and you'll have a chance to, you know, get out there and hopefully build some good AI, uh, cuz there's, you know, some pretty dangerous use cases out there. So we need more of us to go out there and change the world. So I'm gonna end with that. But what I am going to do, you'll see this show up in your inbox momentarily. I'll get the recording, but I'm also going to send you my number on WhatsApp. And if you are up, if you have a question, just WhatsApp me anytime. I carry my phone with me. It's 7 days. It's not indefinitely, although if you call me, you know, if you WhatsApp me like 2 weeks from now, I'll probably respond because you're all my friends. But, but Specifically to this challenge, I am here for you anytime. If I'm sleeping, I promise as soon as I wake up I'll get to it. I'm very passionate about your success on this project. If you can do it, like, do it. Like, this is the time. So I hope you enjoyed it. Thank you so much for coming. Like I said, you'll get an email from me in, I don't know, 10 minutes. I don't know how long it'll take to download this thing, but I hope you found it helpful. Thank you so much. It's so good to see all my friends. Not all of you showed up on video, but that's okay. Andrea, thank you so much. Emily, thanks for chiming in. Everyone was awesome. Thank you. Thank you. Thank you. No code, right? I'm telling you, this stuff is amazing. Ah, thank you. This is so cool. So you guys are awesome. All right, I'm gonna let you all go. Thank you. Look for that email. Ping me directly if you have trouble. I am here for you. Go do it. Go forth and prosper. Okay, wait. Okay, bye.