Day 3: AI Challenge - Building Your First AI Employee

Welcome, welcome. Hello, hello. Welcome in. It's amazing. Admit all. Come in. Welcome. Hi, Jill. Hello, Anna Christina. Hello, Lauren. What's up? Hello. Good to see you from just across the way. I— if you're in Miami these days. Yeah. Hi. We have a couple Jills. Um, hello, Firefly Notetaker. Welcome in. Nitin, Patricia. I see you all. Hello. Hello, Sinan. Jill, good to see you representing Um, from the Golden AI Academy. You'll have to tell all your friends because some of them couldn't find their way here. And I was like, I can only lead you to water. It's funny because you have to register each time, or at least— yeah, yeah, you shouldn't have to register. Did everyone have to register every time? Oh yeah, it kind of forces you to do that. Zoom registration. That's annoying. Okay, good to know, good to know. Maybe I'll fix that for tomorrow so you don't have to register. I think it's the same link, so that's a, a Zoom feature I was not aware of. All right, welcome Reza, good to see you. Sybil, nice to see you all. Welcome, welcome. All right, today I fear may be the tipping point of overwhelm if you're not there Just fair warning, I'm try— I try— I went through this multiple times. I'm like, how do I make this better? But it's the reality of the day, right? Like, we have so many tools are being introduced to us. So here is my gift to you today, is that I am going to give you a rubric, a framework for analyzing these tools. I'm going to give you a couple tools that'll allow you to test things all at once. Uh, we call them tool aggregators. But this, these first 3 days are the hardest part of agentic development in order to make sure that you have an agent that actually works and serves everybody. Um, let me go in. So are you, uh, Sky, you're able to like admit people? Okay, great. Thank you. Um, welcome Marisol, Liz. Good to see you all. I hope you've enjoyed it so far. So just to recap, we have to kind of jump right in because we got lots of stuff going on today. Um, so first and foremost, I will go grab the deck, uh, for you. And let's see here. All right, I don't know, I don't know. I'mma keep it, I'mma keep it. Um, I'm gonna grab the deck for you while we're getting started here so you all can follow along on screen or on your own. I need to hit the share link and Copy link. And don't forget all of these notes. Maybe I'll give you a quick preview of that as well. But all of this is sitting inside the school community as well as— oh, I forgot someone asked me for the transcripts. Maybe, Sky, you can take down the note. I just want to grab the text file from Zoom. I opened Zoom a couple times today but never ended up— got distracted before I got there. So if we can make sure to just grab that text file and throw it into the resources so that way people can throw the— everything that was said. You can grab the transcript from YouTube by like going to the transcript and highlighting it, but let's not even make that a requirement. But it is available right now if you want to do that, but you don't have to. So we'll add that in. Awesome. I did grab this link, so let me go and throw it in the chat to get us started. Welcome everyone. Natasha and Natalie, Nazar. I'm going just for the ends right now, so welcome. All right, here we go. Uh, okay, chat, welcome in everyone. Let's see, uh, I know, sorry about that. So sorry you had to register again, like day 3. We're friends now, we should not have to do that. So I'll try and undo that by tomorrow. It might be a Zoom thing that I somehow checked a box I can't uncheck, but I'm gonna try. I'm gonna try. What's up, MV graduates? That's right, so we have AI Mastery. I didn't know you just graduated. I would have made a big deal about that on Monday. Um, so congratulations. It is no small feat. Uh, many, uh, will be following in your footsteps, uh, come March and starting the cycle all over again. Uh, hopefully you found a lot of value. I got an opportunity to teach in the Mindvalley Summit, but then also in the, um, in the Mastery Program, but they only gave me one tiny little moment at the very, very beginning. And then I was like, that's it. Like, do I not get to talk to them again? So now I know, I didn't know. Uh, so this time around I'm gonna be like, giving just a little bit more guidance. Cuz as you go through, if I was first, I would've, you know, prepared you for what was coming. Uh, but congratulations to everyone who graduated from that Mastery program. It is no joke. It's like a master's in, uh, AI for non-technical people, right? For founders and business owners. And so, uh, really, really challenging program. One of the most challenging, which is the promise, right? Like that's why we did it. But, uh, but really challenging. But hopefully, yeah, congratulations to all of you who went through it. Yes, exactly. So the word got back to, uh, the Mindvalley team and they were like, apparently students liked you, so you get to come back. And not everybody gets an invite back. So I'm really grateful for that. Um, but my hope this week is not to give you a master's, um, but to get you functional, right? I want you to get value immediately. And here's what I found. This is true about me. I'll just, start with this. I— oh, did I hit copy and paste? Maybe. Let me just make sure. Yeah, I get— er, squirrel. There it is. That's the deck for today. So at Amazon, when I was invited to join Amazon Alexa, you know, like over a decade ago, I knew similar to all of you, right? When I started learning about AI, I didn't know anything about it. Now granted, this was 2014, so it was a long time ago, but I didn't know any. All I knew is number one, I was ambitious. I had a goal. My goal was to create a user interface that my son who has Down syndrome could use to interact with the world. That was like my laser-focused vision and help moms and caregivers like me just survive. Like, how do we, you know, find our own sense? And this is why Mindvalley people and I, like, we're family because I was building back then, you know, mindfulness and kindness and positive prayer, daily affirmation, all these capabilities. Into a platform that the engineering team around me was like, psh, who needs mindfulness? Like, why are you building that? Nobody wants that. They were all where some of you are now, like they need engineering help and they need, like, they were just in their own world. And I was the sole representation of maybe it's gonna be more than just a kitchen device that gives recipes and workout apps. Maybe it's gonna help people and. Uh, when I started building these system, these applications, I didn't know anything. So all I could do was look at people who were just a little bit ahead of me, copy what they were doing. And here was my plan. So back then we had, we have GitHub, you, we have GitHub today. I would still encourage you to use it, but I went into GitHub and I looked for people who were building applications, not the ones I wanted, any apps on Alexa. And I would copy, I would clone the repositories. And then I would rebuild them. And what that means is like I would literally code every, uh, file, every line of code I would rewrite. Now when I started, I had zero clue what I was doing. I was just like rote memorization, like, okay, A-A-L-T-A-L-T. Oh no, that's a button. Okay. Let me, like, I didn't know anything, but what ended up happening very, very quickly is as I started building other Solutions, which is what I'm gonna share with you today. As I started using other tools and seeing other examples, I started gaining a little bit of confidence because as I built more things, I was like, oh, I remember this was in this other app. And oh, I've seen other people do this. And so I started almost like, if you think about it like music, it was like learning bass lines in music, right? Like by themselves it wasn't helpful to me, but as I began collecting all of these little applications that people were building, which you're all sharing, I mean, we just saw Bill was super nice and shared what he did yesterday. And it was really cool. Like, I was like, this is, this is awesome. As you collect those experiences, you now can go, oh, I remember a friend built this app. Maybe I could use that for this part. And maybe I could use this LLM agent thing from Lovable for this part. You start assembling all of these baselines. Like, nothing is novel, right, in artificial intelligence. There are maybe, I don't know, 10% of use cases in artificial intelligence have never ever been thought of before. Maybe there'll be more in the future, but most of it is predictive. Most of it is looking at how we've done things in the past and coming up with new ways of doing that. So in the sense that it's looking at existing patterns and we're either improving or optimizing those patterns, and that's what I learned is the most powerful skill you have is How quickly can you collect all of these experiences and then use those to create something new for your audience, new for your customer, or new for yourself? So that is my hope, is that after today, again, I, for those of you who are just joining, I am a little worried I'm gonna overwhelm you. I did collect, like, there are thousands of tools. I did create a short list, but it's still a list and What I did at Amazon is I built a hundred— I went and built thing— I built an Alexa application a hundred different times, a hundred different ways. As a result of that, I became very good at building whatever I wanted cuz I had experience and I had just this wide range of experience. And I did that in a very short period of time, 6 months. And so I just kept building and building and building. I didn't use half of them. One of them, fun fact, for those of you who are NFL people, uh, in the US we have football. It's not happening right now, but, uh, there was a football player, Marshawn Lynch. Some of you might know him. He's still very popular. He's doing like DraftKings commercials and all sorts of stuff. He was in a movie, Love Hurts. Like, good for him. Um, but at the time, this is back in the day, I got, uh, voluntold to go help Marshawn Lynch build an app for his store. And so I got experience. He had built a store called Beast Mode in Oakland, and he wanted to make it a smart store. So when you walk in, you'd say, open Beast Mode, and the blinds would go up and the lights would turn on and the disco ball would go around. Um, and the bar would open. Um, and so, but I had never done that before when that had happened, but I figured it out when I was there. But I was good at figuring things out. That's why I'm so passionate about AI solution architecture is that it's a little bit about knowing the tech and having that background. It's a lot more about can you see a problem and envision a solution? Or are you the person who sees a problem and goes, uh, I don't know what we're gonna do here. I don't think there's any way out of this. This is a disaster. Like, I never— every problem I see, I kind of see through the lens of enthusiasm, excitement, positivity, and opportunity. And that is hopefully what you'll see when I go through all of these tools is like, Wow, there's a cool opportunity to look through 8 to 10 tools, go through a rubric of measuring how those tools, um, match up to each other. And there are 3 things I'm gonna ask you to focus on, but there are about 10 different lovers that you can use. Um, and then from that, pick your favorite. Now that's all I'm gonna say about that right now before I get too far ahead of myself, cuz I have very, I have Noelle ways of doing things and I'm gonna share that with you. Because it's working for me. But I'm also going to encourage you to use the rubric I use to develop my habits, right? So that way, the rubric through your lens, through your eyes, through your lived experience, you might land in a totally different tool. I know there are people on this call like Frank and Bill and, you know, even Padmini, right, that come from an engineering background, in which case you might want a more engineering-focused tool. You might want a tool that allows you to put your fingers inside the code and mess around with stuff and maybe even break things. And when things break, you're like, yay, how fun is this? Right? Not all of us are gonna have that experience. Like some of us don't want it to break at all. So I'm gonna go through the entire arc from, uh, no-code all the way to complex code and share with you the trade-offs between each one. So, uh, yes, exactly. And, and, you know, as Natasha, as you like in the engineering world, like we, we can choose some problems we wanna know a lot about, some maybe we want to abstract ourselves from. I certainly have to be very selective, otherwise I end up in the weeds on everything and I don't have time to do the most valuable things that my business needs me to do cuz I'm heads down. Now I will tell you, those of you who have joined VIP, Of course, last night I'm, I have PDFs that I created for the, the week and I was like, or I could build something. So of course I built a, basically instead of giving you, you know, it's kind of old school. It is what I was thinking. I'm like, I'm gonna give you a PDF with lines on it, which I've done so far and they're fine. Um, but you're gonna fill it out. And I was like, but now we're doing, we're measuring applications. And wouldn't it be better if I'm doing some kind of rubric measurement that at the end of me filling all that out, I actually get something like a diagram or visualization or some aggregate report? Like, we could do that now, and it took like 20 minutes. So the benefit of how I work with you all is that I built it in Lovable, but it is editable. So you get to go in and grab that same Lovable and build on top of it. People like Renee here, Clayton, she's one of our AI solution architects in the AI Leadership Institute, extremely successful consultancy. And she, you know, she was one of the first that took my very, you know, small version of an AI readiness assessment. And she, like, took that as an idea, like, ooh, that's a good idea. And then kind of created this real— and I don't know, Renee, if you're here, you could definitely throw it in the chat if you want people to see it. But, like, made it better. And that's literally the beauty of a concept I'll introduce you to if you're not familiar of open source, is that I don't believe in scarcity. It's a mindset that I have. I believe there are so many companies that desperately need our help. Some of you are here to help yourselves, right? Which is good. Please do, cuz you helping your business do a better job means you'll be able to serve the people, right, that you're trying to serve better. But some of you are actually going to get the bug of using these tools and be like, I could help other people use these tools and I want you to. So everything that I do here, that's why I share the deck. That's why I will give you the lovable prompt. It's why I try to be as, um, as transparent as possible so that you can see a lot of this is just being brave enough to try. And that is my goal for the rest of this, uh, week is to get you in the habit of trying something new. So today I'm gonna show you a bunch of tools. Tomorrow, um, we're going to focus on a single tool that has 6 agents in it, and I'm going to walk through all 6 agents, and I'm going to encourage you to build those 6 agents with me. I'm going to give you— I don't own the platform, so I'm going to give you a QR, a link and a QR code for the platform, um, and, uh, what is it called, a promo code, right? So you get like— it's pretty cheap, and then it's a 70% off code for a pretty cheap thing. So I don't know exactly how much it is. I It's like $10 maybe, I don't know. Um, but we're gonna do that tomorrow just so we have a common ground for building something that's functional and every one of us can use. Um, and then on Friday, most importantly, as I push you out of the nest, I wanna make sure you understand like everything we will have done up until like including tomorrow will basically be we baby tiger mode, right? We will have built small, a small little baby tiger. That has the ability to do a lot for us. But what are the things that you can do to protect yourself from that baby tiger getting big and hurting you, your business, your clients, your employees? That is the focus for Friday. And, and it's fun, I think, uh, but there's a lot of cool stuff happening in, in the industry, um, that gives us— oopsie— uh, that gives us an opportunity to, um, change our, you know, basically set ourselves up for long-term strategic success. And that's why we're doing the work we're doing today. Um, so the homework today, I will start, I'm just gonna like lay it on you so you can mentally process it through the rest of the call, is to take, if you are like me, take all the tools I share and run this rubric and measure how through your lens of experience you might use one or the other. The bare minimum is 3. Pick 3 of these tools and test them. You now have a bunch of words that you've written down. You've got the guardrails, you've got an, a system prompt. I'm gonna ask you to go into these tools and be like, hey, how could I build this system? And I, like I said, I'm gonna give you a set of questions that you're gonna ask along the way. Oh, oh good. Hi Renee. Oh, were you talking to me? I thought you were answering a phone call. I didn't know, so I muted you, but you can unmute now. If you want to. Hello. Yeah, I know. I was just talking about— I didn't know, so I was protecting you from yourself. I didn't know. Oh, no, you're good. No, I just, I was trying to wait for you to take a breath, but you didn't. I did. I do. I breathe in a very— you just couldn't hear it. But yeah, thank you. Yeah. But yeah, feel free. Noelle did, which is beautiful. I put it in Lovable, and then I made— created in— also recreated it in Anti-Gravity, and then I recreated it in Manus. Both. And then it's been live on my, my website for my clients. And now I've actually redone it and I'm still tweaking it. But it was a lot of fun to see what Noelle did and then just to make it mine. Yes. And that's like, I think the goal, right, is see as many different examples of what people have done and use that as a starting point. Like, because, you know, I did it for a very specific reason, which was to teach. Like, I was using it as a vehicle to help people who are learning to be solution architects. Realize, hey, there are new ways to assess knowledge, to get information from your clients. But now that becomes a new baseline. You don't have to start with, what will I do? You can start with, oh, Noelle did this cool thing. What if I added new functionality to that? What if I created a better experience? What if I added color? You know, like there's so much more you can do when we all kind of share our ideas together. And so some of you, will want to play very close to the vest, and that's totally fine, totally up to you. I just— that's my ethos, is more to share as much about what I'm doing. And I hope— I mean, on Alexa, every single skill I built on Alexa, I made open on, um, on GitHub, meaning anyone could go copy my code and publish their own skill. They didn't even have to change it. As a matter of fact, two people built a mindfulness app that was exactly my code. They didn't change it at all. And it just gave me a sense of like bias is real, cuz it was the exact same code, same everything. Um, but for whatever reason, those versions ranked higher in the list and I was like, why? Anyway, and that took me into a, like when I first started working in AI, I wasn't, I didn't even know what responsible AI was. I didn't know that that was a thing. I didn't know that models could make bad choices. I didn't know that they could be pre-trained with data that would help them make bad choices. Like, I didn't know any of that. It happened organically as I started seeing like weird behavior in AI systems. And so now, needless to say, I'm very, very focused on that. All right, let's see here. Do, do, do, do. Yes. So thank you for sharing that. Yeah, I think it's great. And of course, we have lots of people who share that enthusiasm of sharing and growing together. So I'm super excited that you all have found us and that you're here. Now let's jump right into the deck. Some of you have seen it already. Oopsie. Let me, go back to the beginning before I share it. I think that's right. All right, cool. Now I'm going to keep my list on— oh no, I should hide it because some of you won't be able to concentrate. I wouldn't be able to. So I'm going to put this in presentation mode so we can go through it. But today is all about tool selection. So we've done the hard work of designing the AI system. We've said, okay, here's the agent I want, here's the job I want it to do, here's the data it needs to do it, here are the guardrails, here's the type of decisions it's gonna make, here are good examples of how that decision's gonna happen. Here are some maybe negative ex, uh, examples of what I don't want to happen. So we framed out the, the description of the behavior of this agent. Now you have lots of options for where to build it. By tomorrow, I'm going to— because some of you are going to be like, ah, there's so many tools. I want to, just as a service, make sure you realize that there are very, very good tools on the market. But I also want to caution you that in tool selection, my immediate bias is to organizations that have been alive for more than a decade. So that is my immediate— and that's because I come from the enterprise. Right? So I come from enterprise, so my, my mind, you know, the voice in my head says, you know, be careful investing in a company that could get acquired, could get squashed, right? I call it Death Star acquisitions. Like if I give a company my data, I wanna know the CEO. I wanna be able to see their directors and their board of directors and their executive leadership. I like to see, you know, some women and people of color up there. Um, I have some very, you know, I wanna go and, and see who is behind, right? The, the tool that I'm number one investing in, but I'm also in today's day and age giving a bunch of data to. That is not true about these tools that I'm about to share with you because there is a bit of limitation when you go bigger, right? When you start working with Microsoft, Amazon, Google, NVIDIA, they give you tools. However, the learning curve is steeper. And the cost structure is steeper. So it, it's very, it's a big struggle for me. So what I'm gonna share with you today is the enterprise rubric I use to evaluate smaller companies, and I hope that will be helpful to you as we go through today. Now, we have Monday defined the role we were going to go after. Some of you chose social media, email managers, um, lead generation tools. I am going to focus— we have 6 agents that starting tomorrow we're going to build. So I, uh, in one of the promises I made, uh, when I started this challenge is that I wanted you by Friday to build 6 agents. We would do it in 5 days and it would give you 40 hours back in a month. I think you'll easily do that. My agents right now, I think I just hit 300 hours, uh, since February 1st, you'll see that data again. I use it every single day, and I'm gonna share with you exactly what I do. That is a very specific toolset that I picked to solve very specific problems in my business, which is email management, calendaring, social media consistency, um, long, long-form blog posts for Substack and my LinkedIn newsletter, um, uh, a lead generation tool. A legal assistant to help me with NDAs and, you know, um, proposal, uh, requirements. Um, I all the time people are like, hey, would you mind consulting on this project? And then they send me a proposal and I'm like, I don't got time for that. Um, but with the help of my assistant, I can throw that now into my legal assistant and it can give me some ideas about like what to be worried about, what phrasing to change. I can even ask it to write me the email that'll help protect me, right? In, in, in a very short timeframe. And then finally, of course, just a receptionist somewhere where when I go— now, granted, I, my main mechanism for getting in front of people, like most of you know, cuz this is how we met, is to do public speaking, whether that's virtually, um, in a webinar type environment, um, or whether that's in person. Now, fun fact, I did not know this yesterday, But this is super cool. So one of the tools I'm gonna share with you today is Make. And remember yesterday I was like, I can't remember the client, right? Make is the client that I met with that I had to leave y'all for. So make.com is hiring me to come. Now granted, it's a virtual session, but it's recorded onsite in New York. This is, I'm not, I'm not gonna be too like spiritual or religious, but I'm like, this is when you know things are working in your favor. All right, because it's a virtual event, but they required me in person and it's in New York. So I get to go see Hamilton on my birthday. Like what? You can't make this stuff up. So, so that's what, so it's make.com. So isn't that, that's pretty cool. So they have a, uh, and I'll encourage you to go check it out if you want. It's called Face Value. Um, is their lead, uh, thought leadership, like webinar system service. They've interviewed Kassie Kay, they've interviewed a bunch of people, uh, Claude Silver, who's the, um, Chief Heart Officer at Vayner, uh, Gary Vaynerchuk's company, VaynerMedia. So lots of really cool people. I'm really excited to be on that stage. But the reason I mention that to you is that when I go through and I start thinking about like the agents I'm gonna share with you, I have stages that I'm on. I have content that I'm creating. I have a social media brand I'm trying to build, right? I have customers that I'm taking care of. So many of you kind of fall into this bucket, or at least who you might be 3 months from now or 6 months from now. So I felt it would be valuable for me to share with you the exact 6 agents that I'm using on the platform that I'm using and all the questions I ask that platform, which the CEO— I just had another meeting with them on Monday. What, you know, they're like, okay, Noelle Russell, how else can I help you with anything else? Do you have any other questions? But at least I've done that homework right for you. But I'm also now gonna walk you through those exact questions that I ask. And yes, sometimes the platforms are a little irritated, but I will tell you, many times these platforms create FAQs on this exact question set that I've asked them. So I feel like I'm, I'm doing a service to the world. So that's what we're gonna do today is tee us up for analyzing the tool sets that you want. I'm gonna show you the top 8 agentic platforms. Again, based on a rubric that I've developed. And then in VIP, I'm gonna give you a, a just a quick tool that you can use to get that assessment. I'm gonna also give you the edit capability on that lovable prompt so that you can go in and create whatever visualizations or reporting that you choose to do. It's just a start. It's just a start. So I'm super excited for this. But how cool that it's Make.com. Like, we were just talking about them yesterday, and then I was like, Who is Solanas? And it turned out to be Make. That's crazy. All right. So today it's not just, it's really not just about the tool though. I will be demonstrating tools. It's about building an operating system. And this is something I learned at Amazon a long time ago. There was a, uh, one of our head vice presidents, his specialty. And, um, I wish I could remember his name. I'll, I'll have to think of it. He's connected with me on LinkedIn, but after his time at Alexa, which was very hard, he like went dark. So I haven't seen him since, uh, because that's what happens, right? Like you work really, really hard, you launch a really big thing. He had a lot of stock, uh, involved in that and he's now enjoying his life, I hope. Uh, but he was hyper-focused on operational excellence and what that meant was we never built a thing to satisfy like a feature request. We were always thinking way ahead, like, how does this feature fit into the operating system of an organization? Or how does, like, when we were building Alexa, we were trying to build in, uh, two types of Alexas. One, how does this fit into the operating system of a family? Meaning, uh, meal planning and chore charts. And I don't know if you know this, but all of these things are built into Alexa. We didn't do a good job of telling people that it was there, but it was like a, an ecosystem of capability to allow you to just run your family better. And then we did the same thing, it was called Alexa for Business. We did the same thing for business and we attached customer service and onboarding and all of these processes. As a matter of fact, one of my favorite projects we did, uh, was we, I don't have a version of it. Oh, maybe I do somewhere in here. Oh yeah, I do. What? These exact devices, right? You've probably seen these before, these little dots. So What we would do is we'd put on this device an onboarding application. So when a new— and I definitely probably— oh yeah, I do. What? When a new employee would come on board, they would go to their desk. I'm not going to untie this, but you know, these are like headphones, good old-fashioned headphones. Some of us still use them, not Bluetooth. But you'll notice on these devices is an auxiliary out, right, for headphones. So you'd put the headphones in and you would say hello and the onboarding process would kick off. And it would be a conversation. It would be like, great, welcome to the company. And it would do like a podcast style approach of tell kind of what you would do today with like NotebookLM, but we built it into Alexa. So Alexa's like, here's your company, here's your hiring manager, here is your buddy, uh, here's what they're gonna do. Would you like to begin the first course? It would, it would like walk them through everything they needed for the first 90 days. So it did their ethics training, like everything. And all they had to do was put it in their ear and it would tell them. We also created the ability when they were talking to, and this is 10 years ago, when you'd talk to the device, it would change the screen, right? So there was an interaction between, because remember, even though I'm talking to this device, it's calling code so that code could change the visualization that they were looking at. I mean, just think like even today I would love that onboarding. Like even today I would love for us to build something that allows someone to have a voice conversation, and as I'm talking to them, it changes behavior on my screen. And that, like I said, it's available now. It's just, do we have people that are willing to build that? But that is kind of operating system level, right? I'm not just thinking, do I have a chatbot that'll coach someone? It's kind of like what we were talking about in VIP yesterday, right? Where we were saying like, you could build a chatbot that's a coach, or you could build an entire application, an operating system for your, you know, recruits or for your downline. And then of course have an LLM as part of that experience. Like, it is now time to think bigger. So we're focusing on one thing because it's better to build one than 100, but just kind of remember that our goal is to build an operating system for our family, our business, our team, not just a one-off agent that does one thing really well. It should be doing something really well, but in the context of When it gets its buddies, it's gonna massively, right, grow, escalate, scale, uh, the capability of the team, the family. Now, why do these tool decisions fail? So one of the many reasons AI fails in production or prior to launch is because the tool they picked for the problem isn't the right tool. Today, more than ever, the tool selection is a critical component of whether or not your AI ever gets out out of the gate. One, which maybe all of us are, have fallen victim to, is the cool stuff, right? The trend-based adoption strategy. The most organizations, when I walk in and I'm like, hey, so what, what AI are you using? They'll be like, well, so I got this LinkedIn thing, or I saw on hux.com, or, you know, like, whatever. I, heard about this tech, I saw it on the internet, uh, Instagram bubbled it up and I went and used it. So many of the great tools I use actually came to me this way. But using that as, um, an isolated, uh, only, like the only way that you choose your tool will set you up for failure because that trend is, is marketing at its best. Someone is hooking you and getting you to use something based on words that may or may not be true. And so what ultimately happens is that you end up adopting a technology, and then at the end of that adoption, you realize that the promises that were made in that trend, uh, trendy marketing, uh, conversation are not actually possible. And that's why these questions are so important. Pricing first. Many people, of course, are like, is it free? And I, some of you know this, right? I always ask my audiences when I'm on stage, I'm like, You know, what, what, what currency are you operating in? Like, what are you paying with if you are not paying with money and talking to an AI system? Like, what is the currency? And, and I kind of joke about it, like, it's your soul that you're, you're, you know, you're selling, but it, it's not necessarily your soul, but it's definitely not, it's not just your data, right? It's your data, it's your insights, it's your pattern of, you know, your predictive pattern of doing things. There's so much value in the way we communicate, that you're giving that away in exchange for a free tool. So the first thing I encourage you to do is not make a price-based decision on the technology. And I will say, for the most part, you get what you pay for. With a grain of salt, there are certain platforms that are just too expensive. They're too featureful for your organization. If you're a small company, I wouldn't, I wouldn't unnecessarily suggest you buy the highest level license in Salesforce. Salesforce that was built for enterprise companies with a million contacts. I wouldn't necessarily suggest you go buy Agentforce in Salesforce for a business and you're a solopreneur, right? I wouldn't— you could, but like, don't let price like, oh, they're expensive. They said they do everything. I'm just gonna buy them. It has to be— that's why I use the word, the term rubric. It has to be a collection of variables that you use in order to make the right decision. So the outcome of this is that you end up with what we call shadow AI inside your organization, especially if you have employees. People are just gonna start using AI to solve their problems 'cause they saw it on the internet. Instagram told them to, LinkedIn is full of, uh, suggestions in their inbox, and they'll just start using something. ChatGPT is now by far the de facto standard. If anyone says they're using AI and I ask them what it is, 80% of those people are like, ChatGPT. And most of them are like, and I named it Chad or whatever, right? Like they're, they've become very, very comfortable. So comfortable that they're anthropomorphizing in with a name. And that's fine. But likely they did not actually question, are they using the right tool for the right problem at the right time for that solution? And that's why agents are so powerful. Agents provide us the ability to avoid what happens when we just pick a model for no reason, right? I mentioned shadow AI, but many times the tools that we pick are not good at the problem. One thing I'll, I'll share with you about how AI systems work, machine learning, you know, at its finest, is that when you start using an AI model, when you start using it and you are asking it a question, it is already, in order for that model to exist, someone, typically a data scientist, will have picked an algorithm based on the behavior of that algorithm. Either it's gonna, you know, identify, uh, it's gonna classify the information right or wrong, uh, dog or cat, or it's going to find, it's gonna cluster, right? The information. Like it's, it's already an algorithm that's pre-designed to solve a very specific problem by the time you get there, right? So you're using a model that the problem's pre-designed. It's also been pre-trained. Right? We know with ChatGPT or any GPT-like model, right? The P in GPT stands for pre-trained, which means someone, a data scientist, has gone and selected the training that is best suited to solve that problem that you might not even know they solved, right? So now this model is pre-trained on a very specific problem that you didn't design. So every model is this way. Every model has an algorithm that was designed to solve a specific problem. In the case of LLMs, it's designed to do text prediction in human language. That's not a very specific pro— I mean, that's a specific problem, but not one that serves you. So when I, when I'm going through this exercise, what I want to make sure that I'm thinking is that every single model will require me to define the problem first. That's why what we did on Monday is so important. Now I'm only going to look at models that are designed for that type of problem. And they are, there are models that do this. If there aren't, every single model I will now be producing its behavior, meaning I'm not going to ask it questions unless I've already told the model through instructions or a system prompt the problem I wanna solve, right? How, and the data I want it to use to solve that problem. And of course, what good looks like, right? So every model you use, if you do not tell it the problem you're trying to solve, the data you wanna use, and what success looks like for that model, know that that has already happened. Someone has already done that to the model, but it's not your problem, your data, or your success. So that means you're already predisposed to this model not doing a good job for me, for you strategically. So that's, What we're gonna do now is everything we did yesterday was creating this system prompt, creating the instruction set that we want to use in order to make sure that the model does what we want for the purpose we've defined with the data we've provided. So that's the work we did on Monday and Tuesday. Today, now we're gonna pick a model that maps to that outcome. And most people don't do this, right? Most people are like, Or worse, they're like, oh my gosh, shiny button, ChatGPT, Opus. Like, I do this too though. I mean, it's very hard not to, right? You get a new ad, um, or someone tells you about a new model and you're like, oh my gosh, I'm going. Um, but, and it's fine to do that, but just do it through the lens of this, this measurement criteria. All right, so here are a few of the things that, and again, in VIP I'm gonna give you this framework. In the form of a tool. I will— this is also documented in the tech, the workbooks section. It's the text part of the school community. So you'll see that later today. But these are the ones that I wanted to highlight. There are 10 or 12 of them. These are the ones that I pulled out that I wanted to focus on. One, first and foremost, is it easy for you to use? Now, this is very subjective. If you are an engineer, ease of use is different. Than if you are a solopreneur running a consultancy, right? Like, your definition of success in each one of these categories comes first. For me, ease of use is how quickly can I get my result? I don't necessarily need to see the code, but I do want to understand transparently how the model's making its decisions, which I'll get to in a second. Speed to value. Again, I want it to be easy to use, but I also want it to get me my result extremely fast. Now, these are all what I would consider levers, right? So if you're kind of thinking about it like a Likert scale, on a scale of 1 to 5, you want to kind of go through each tool and be like, okay, it's about, you know, it's like lovable, super easy to use, speed to value pretty high. So I'm like sliding this little slider, uh, to a— in 1 to 5, 1 to 10, you can decide. Um, tool connectivity, this is huge, right? Tool connectivity. The ability for it to actually talk to other tools. Some models, some platforms are pre-connected, meaning that they have a set of tools. Make 2,000 connectors or something, right? N8n, hundreds of connectors, maybe thousands. I don't know exactly how many. Some of you might know, but lots, like so many you can't even— you won't be using them all. But the idea is like, is not just does it have 2,000 connectors, Does it have the connector you defined yesterday, right? Is my data is in SAP? Is there an SAP connector? If there isn't, I can't use this thing, or worse, I have to hand-build that connector. Now, these are challenges that, some of us come from big consulting companies. These are challenges we actually hoped for, right? When a client got to a point where they're like, ooh, no connectivity, I'd be like, gold, I get to build it from scratch for you, right? Custom development. But in the age of AI, that will slow you down. It will make it very difficult for you to get to production. Debugging. This is probably my favorite. I want to be able to know how decisions are made and step through that process. Another way of thinking about that is auditability. Can it be audited? So I'll give you an example. I think I showed this to you yesterday, but I created a platform over Christmas, because, you know, like Renee said, I don't really take a breath very often, but I'm— I love it. So that's the problem. So I build out, you know, this, this assurance platform. And in this assurance platform, it basically— its job is to use the rubric that I've defined to make sure that the AI systems in your company are safe on a bunch of different levels. And so I created a platform called ACP-1. It's a control plane for seeing all the AI. But let me show you exactly what it looks like. Oh, this is a little bit big. Let me make it a little smaller. Move it like this, and then I'll zoom in a little bit so you can see it. There you go. Oh well, it's maybe zoomed in too far, but you can see here, right? Like this toolset is to give me visibility to all the little baby tigers in my company, right? How many systems do I have? How many of them have gone through? Through what I call AI Assurance, which is 6 levels of— do you— actually, I'll go and show you what these, um, different options are. But like, let's— this one's broken, not intentionally, it's meant to demonstrate it. Oh, let me go to certification. So you could see here, these are all of the systems and their concerns. So this is the fraud engine, this little fraud AI system. These— this is the metadata I collected about it. But here is what makes it assured. Do you have a chain of command? Do you have a list? An executive owner, a business owner, a technical owner, my favorite, the kill switch owner. Do you have the person who's gonna pull the chain on the chain if they find out the model's hurting people? You need to know that person's cell phone number, right? Like we need to call them. Um, do you have an audit trail, right? And do you have policy documentation? Like, do you have a model card? I don't know if you know this. Some of you should, might already. But every model you use, it might not give you the source code, right? That vendor might not be like, here's the source code, look at it all you want. They might be like, no, that's our IP. But you can ask them for a model card that describes its behavior. And if you have that, everyone in the company, or at least the decision tree, should know what that model behavior is. So these are the types of things that as an executive, I want to know if you're adopting a Baby Tiger in my company, I wanna know, right? Like, is there a human in the loop? How does that operate? Who's the human? When do they get notified? I also wanna be able to download all of the evidence that this is working or not working in case someone, my customer comes to me and says I failed them. Or you might remember, right? Like that Chevrolet dealership that sold a truck for a dollar. I can assure you the CEO was like, can I see How this happened. And I can also assure you that there were not great answers to that question at that time. There are now, but, and, and they are in that dealership. It's not rampant across auto dealerships though, right? But you can kind of see the idea, like the more information we have, the better. And these are what we wanna do is ask questions before we invest in a tool to make sure that I can populate a list like this. And of course you all can look at this. It's public, it's open. As you know, that's how I roll, right? I want to make sure that you can see it. But this audit trail, right? I want to know every decision my model makes. Not everybody, right? Not every platform provides that level of debugging, traceability, explainability. Now, I will tell you, most tools do, but here's the least you can do. We actually call it in, in my framework, many of you who are in the school community, You have taken the AI Bot Builder program, and when you build a bot with me, I teach you the 5 Ps of prompt engineering. And one of those Ps is what we call persistence, which means it's up to you as the architect of an AI system to actually save the information that would create your audit trail. Like, the data's gonna come in cuz the person's gonna talk to your system. Do you save every interaction? Have you budgeted for storing every interaction? All of those things are things that I decide on a tool. Does it do this for me? Do I have to accommodate that? That's why debugging is so important to me. It's like when something goes wrong, when that baby tiger bites someone, like, how can I prove that my system was not at fault? Or if it was at fault, how do I find where that failure happened and fix it? And that often, that transparency is very, very challenging. Of course,, how easy is it for me to, I called it shareability, but also portability. I'm really, I grew up, uh, in technology in a time where we had this concept of vendor lock-in. Do y'all remember this? Some of you technologists out there might remember, but basically we would call you like an AWS shop or an IBM shop or a Microsoft shop. And what that meant is that you were on that platform and you could not use anything else because you were locked in. If you tried to connect to another system, it would break. Like it was very, very, brittle. You had to use everything from that one vendor. And of course vendors kind of liked it, right? Because they were like, you're mine and you can only ever— but we are now in a world where best of breed is better, right? That I'd rather have, we call it multi-cloud. I'd rather have a little bit coming from Microsoft, the best stuff that I know of coming from AWS, the best guardrail application coming from NVIDIA. I wanna be able to piece together all the things. That means that— that's why I like that symbol. It means that there has to be kind of composability, uh, as part of your tool. Again, maybe not. Maybe your tool set— maybe, maybe you wanna keep it simple and you don't wanna use multiple vendors, but this is just a lever that you're gonna identify. Is this important to you? Is it not important to you? But at least ask the question, cuz the worst thing you could do is buy a product that is gonna lock you in as a vendor, and the first thing you wanna do is connect to another tool. Um, how well does it support team collab? Um, I, that's one of the early questions that I ask is like, is there the ability for me to add members of my team? Some of these tools are built for solo, you know, for individuals, B2C we call it, right? Business to consumer. In which case they haven't really thought about teams because they're thinking you're just gonna use it yourself. I personally, everything I use, even if I am using it myself, I wanna think about how am I gonna scale? If I, if it doesn't have team collaboration capabilities, it means I am stuck doing it forever. So I always look for that, even if I don't foresee using a team in the, in the short term. And then cost transparency. Do you know how much it's going to cost you? The indicator for this is a calculator. Does the vendor provide a cost calculator before purchase? Doesn't mean that they can't give you this transparency, but it's a really good indicator that they've at least thought through it. When I ask this of organizations, I kind of look at their face 'cause I always like to have a call with them. And if I see like this look on their face, like, oh, it's a good idea. That's a concern for me. 'Cause I would prefer they have thought about cost. I'm, again, for you engineers in the house, we remember getting the surprise cloud bill, right? Like in the 2000s and maybe even 2010 where you start using cloud and you don't realize that it's charging by the minute, and then the next day you're like, $3,000? Like, wait a sec, where did that— and I've gotten cloud bills reversed, but we don't have to do that now. We— now I know, now I know that I can get surprised by a utilization or usage bill. And I don't want that. So I wanna know their transparency on cost well before I start using it. And of course, will they help me when I get stuck? I hate this phrase, but I'm gonna say it because it's how IBM, Red Hat, Microsoft, how they phrase it, Amazon. But you want to have a single throat to choke. It sounds terrible. But what that means, and, and that's like, you can look this up, they say that in their documentation. But what that means is you want to have a single line, and this is how Amazon organizes its support teams, is that there's a single-threaded owner. I prefer that phrase, but if I said that, you might not have understood what I was saying. So single-threaded owner, I want one person I can call who's gonna solve this problem. Some of you who are in our AI Leadership Institute community, you know this, I've been talking to OpenAI's support team, which is an AI system since January 6th, and I have been looped around to the beginning of that conversation 6 times. And I, it just happened again where they're like, I'm sorry, we don't understand the question you're asking. Can you start over? It's super frustrating. So I wanna know, am I gonna be talking to AI, which is fine, maybe it'll be good, but what's my re— how many of you have done that, right? When I get on the phone, if I'm, or if I'm in an email, I wanna be able to say, human representative, someone please call me. I just wanna talk to you. Is that even an option?. And I will tell you, in some of these big companies, Meta, um, OpenAI, talking to a human in a support capacity is very difficult if you're not paying an extremely large amount of money. So you should know that. It's okay, but you should know. All right, so that's our tools. Here are the, the scales. So as I, um, jump into a few demos, these are the scales that I am constantly measuring, which means, I don't know if you've ever heard that term that says, um, easy, fast, um, and customizable. Pick two. You can't have all three. You have to pick two, right? So in this case, you're going to have— like, here are three different measurements, and you won't— you'll be able to pick one side on each one. So if it's easy and fast, it'll likely have to be a generalist. If you want higher levels of control it'll likely be slower, but be more reliable, and you'll have the ability to make it more specialized. So these are all trade-offs that you'll have to be considering as you're building it. All the tools will have different trade-offs. This is why, as I mentioned before, agentic, uh, architecture is so powerful because if you build your agent small enough, one tool could be super easy, super fast, right? And general. And another, the next tool in the workflow could actually have more control, right? So you can architect, and this is why we, we are big fans of AI solution architectures. When we look at the how all these agents are gonna communicate, we can architect where Easy fits in, where Fast fits in, right? We can pick which parts of the workflow we use very specific agents. If your agents are too big, you're gonna have to make these decisions at a macro level. So it's really important, the smaller the better. You can add more functionality, but you might want to consider, maybe I want more control in this agent. So maybe I don't add it as a feature to my, you know, my fast agent. I create a smaller, more focused, more controlled agent, and now its latency is a little longer, but it's going against my data, and I have a higher level of trust with the answers that it's generating, right? So those trade-offs you're gonna have throughout, throughout the experience. And I just want you to be aware of them so that, you know, I mean, nothing comes free. So the first, I wanted to just give you an example of a couple of tools that we're gonna be using. This is my little scorecard, right? So you can see in this case, this is the Lovable app. You go in, I demonstrated it for you yesterday. I'm gonna show it to you again right now. Of course, ease of use, high. Speed, very high. You type in the words, it builds you a thing. Lovable, you could probably do slash any other vibe coding platform. Claude, there's a lot of different options. The flexibility is moderate because you would have to put it in GitHub. There's a separate connector tool, then you have to turn on the ability for— anyway, there's a bunch of steps that are required, so it's it's possible, but not super easy. And your ability to govern it is limited because you don't build the code. So you, it's hard. You'll have to add additional, um, layers of governance on top of the application that you build or design it with your words and burn more credits to add that in. What I like about other platforms like SageMaker on AWS or Bedrock or Microsoft's Copilot Studio. These are all great tools, but the, the benefit, and the benefit with that is that they have guardrail capabilities built in, right? This, when we say governance limited, it means you're going to have to do that work. So it's available and you can add it, but it, the burden of responsibility lies on you, not the platform. Now that's just one example. And again, I show this to you more to be just an example of what that scorecard can look like. And the process that you're gonna go through, right? So the first stage, we're gonna experiment. So I wanna share with you right now a tool aggregator that I use to experiment prompts on a bunch of platforms at the same time. There are lots of them. I use Poe 'cause Poe got me early. And so I have lots of experience in these, and they have— you could just look, there's thousands of models in here. Custom-built models, Character AI models, like just lots, lots and lots and lots. So what I would encourage you to do as you're building your solution is just see if somebody else has built something that's similar. May search for it, see if, and if it's similar, look at what they've done. See if there's anyone complaining about it. Make sure that you don't fall into the same holes, like leverage existing 3 and a half years of experience in the marketplace. Place. There's another one called— and if you know of one, please throw it in the chat for everyone because we collect those links and throw them in the classroom. Um, I think it's called Majestic. Uh, I don't know. Um, or Magic. Oh no, I'm afraid. I'm afraid. Be very afraid. Uh, let's see. High performance AI infrastructure. We'll see. We'll see what this is. Um, Ooh, submit resume. That's cool. Okay, no, that, that's kind of neat though. They're like, you can do nothing here but apply to work here. That's awesome. Ooh, look, they just got $100 million. Okay, so that's not the right thing. But there are other AI aggregator platforms. Now the benefit is that again, like I said, you'll pick 3. So I, these are the top ones that I use, right? Sonnet, Gemini, GPT-5.2. When you go in, you can basically make a request and then copy and paste the exact same prompt on 3 to 5 models. And what you'll see is, for whatever reason, bias, they don't give you the same answer. And that will be very informative to you, because as you go into these tools that ask you what model you want to use, it's nice to have tested out and given your own, gotten your own perspective on how well these models perform. The next thing, let me go into, is, so one, use a model aggregator to test out simple prompts so that you can see that some models are better than others. The next thing I always like to encourage you to do, and again, it's hard, it takes a little time, but whatever model you pick, whatever agent, agent platform you pick, and the model they give you, I encourage you to leverage this platform called Helm. I'll grab the link and put it in the chat. So this plat— oops, I don't have the chat. Hold on. Where is my chat? Where is it? There we go. Uh, almost found it. OK, good. Awesome. Oh, thank you. Thank you, Sky. So Sky threw it in the chat. So this is the leaderboard functionality. It actually sits inside Microsoft's, Microsoft Foundry, which is their engineering build tool. It sits in NVIDIA. They've also implemented it So this is not new. This is almost 7 years old now. There is, but what it allows you to do, and I'll just look at Helm Classic, it allows you to see where in certain metrics where these models sit. So Llama, uh, Mistral, Cohere, like you might not be familiar with these models, but when you go in and you use an AI systems, typically they're gonna tell you what model they've selected. And most importantly, if I go back to the leaderboard, There are very specific— look, calibration. How calibrated are the answers? How robust are the answers? How fair are the answers? Are they biased? Is there toxicity? You should at least know at a high level, is the model I'm using prone to be toxic? Like, look at this. Look at the difference between accuracy, 94%. We can definitely build a model that's accurate. Toxicity? 82%. I just, I mean, just like, that's interesting to me that any of these models, if you look, Cohere, Red Pajama, Bloom, Falcon, you know, like Babbage, of course, these older models, but like toxicity is pretty high. That's 18%. I don't know. Anyway, so it's worth knowing. So I give this to you again as a tool so that once you see the model that you've been given or you find out the model that a tool is using 'Cause most tools are model wrappers going in toxic. Oh, great question, Zara. So toxicity is like the tendency for a model to promote self-harm, sexism, or racism. Bias is different. Bias means that in the model, it's predisposed the data to think a certain way, and it might not be bad, but toxicity is sexism, racism, promotion of self-harm, Oh, that's a good question, Bethany. At what point do you need a separate computer for these systems? Most of these systems are running in the cloud. So the only time you need to have a separate system is if you decide to do what we call privatized model development or private LLMs. And in some cases, if you're working in a hospital or building for medtech or fintech regulated industries, they likely will want you to. Um, but yeah, I will tell you though that some of these systems also consume a lot of resources, meaning RAM, like random access memory on your computer. So depends on how old your system is. Um, I have a 2-year-old Mac. I— it has 2 GPUs and it crashes at least once a week because I don't have enough memory. So like Yeah, uh, so I would basically use the technology, find out, you know— oh, one of the things I would also use is just monitor the activity on your system. I got it for Christmas. I got a new Copilot, uh, Copilot PC. And the new PCs, this is— if you're going to buy a new PC, look for what's called an NPU, a neural processing unit. What this means is that it is going to deflect to a neural processing unit all of the real-time inferencing that you're doing in your browsers, it deflects, or inside tools that you've installed. So I recently installed Claude on my machine as an app on my computer. So now all of that inferencing happens on the resources on my machine, but with an NPU, which Copilot PCs have, and there's new commercials, there's lots of computers that have these now, but with that NPU, it deflects that real-time inferencing to a separate processing unit, which means your CPUs and GPUs don't get taxed. So that is a really— if you're going to be shopping and also just get in our community and ask, so many of us have gone through this process of buying new ones. So this is a good way to look at it. So in the last couple of minutes, I just want to take you through my favorite tools. I'm going to pin this back so I can do it. I'm going to do it in a specific order, and I want you to notice some of the behaviors. So the first thing, as I mentioned, Claude, you can install it. If you install it into— to look, notice I'm using a platform or a browser called Arc. That's how I can create like these easy ways of managing all my browser tabs. Um, if you like Arc, you can use it, but if you were not on Arc and you're on Chrome, it would say add to Chrome instead of, or in this case, remove from Chrome. Um, but this Arc is based on Chromium, which is the OS for Chrome. So that's why anything on Chrome can work in this browser as well. But this allows me to now get— and it's in beta, but it allows me to use Claude inherently in browser-based functions, which is pretty cool. So I thought I would share with you a little bit about what that looks like. So I installed it and it'll say here, for example, hey, you've got Gmail in your browser. This is just one of the challenges. Your inbox is full of meeting RSVPs, which it is. Can Claude archive these for you? So this is an agentic workflow that's going to connect to Gmail, go through my Gmail and extract things out. Now, in my mind, it's not fully agentic because it doesn't satisfy our one thing we did this week, which is, does it do a role for me? Does it help me finish a workflow? Not just do one part of it, which is, this is one thing, but I need all the things. So it's cool, but I test it out. So I give this to you as, as an example. One of the— oh, I wanted to go through a couple of the examples that they offered, right? It can create a shopping list. This is relatively new. I was on the beta for Google's AI shopping process. So now it can go and you could say, I'm on the hunt for new headphones, or I need a frame for a new picture. And it can go and find those things, put them in a cart or folder for you so you can actually evaluate. You can even ask it to buy on your behalf. And then the last example was building reservations for you. So again, these are all— this is what I call vending machine AI, right? Where we're asking AI to do a thing. What I want is to connect all these things together in workflows so that I operationalize the use case. So one of my, one of my favorites is Relay. I'll throw these links into the chat as we go. And I will, I have this tool list. So I'll also provide that to you in the notes of the class. But Relay, as you can see, the difference with Relay is that it's workflow-driven. So you'll see Competitive Analysis Assistant, Social Marketer, Reddit Agent, Personal Assistant. There's a distinct— me too, Javier— yeah, there's a distinct difference with Relay than these other ones that I'm going to show you. Then we have Zapier. Zapier is old-school business process automation. However, today they allow you to make them agentify, which means that you can build agents that call on these workflows. So I give this also to you as an option, and I would encourage you, you can get every answer to my questions that I'm presenting to you in the homework. You can get these answers from the documentation, but if you want, start for free. The nice thing about Zapier is when you start for free, it doesn't automatically trigger a pay cycle, right? It actually just stops working when you stop using it. And I appreciate that. Um, yes, exactly. Uh, okay, here we go. And then it's funny, my son's name's Jacob, so I was like, what's happening? Um, here's the next one. This is Crew AI, same vibe, uh, as Relay. Crew AI is also pretty fun. Um, it is again building— you'll notice this, I'll scroll down— it is a no-code platform. So these no-code platforms allow you to ask for what you want and it builds the sequence for you. It also has a marketplace full of connectors that you can use to connect your system. So you can see some of these questions that I asked. Oh yeah, they're, they're, I've been throwing them in, Amy. Let's see. Yes, I am throwing them in one by one. But yes, I will be re— oopsie. I will be relisting all of these in school as well. Um, that's why I throw them in here. Uh, oh, I see. Okay, hold on, let me throw in Crew AI real quick again. Um, and then we'll aggregate them all at the end too. All right. Oopsie, I grabbed the wrong one. You are— you already have that one anyway. Sky can also repost them too. Oh, we're getting to Make. Make is coming. Here's one. How many of you have used Gumloop? Does that sound familiar to any of you? So Gumloop was, uh, one of those that some— it was a go-to-market strategy that got my attention. Um, they are very popular in retail AI spaces. They've got strong relationships with Shopify, Walmart, Albert, um, Instacart, right? So really, really interesting workflows here, right? Love the name. Exactly. I will tell you, don't judge me, but a lot of my tool selection is like, I love that name. I vibe with that. That's cool. I like what you did there. And then I run my rubric, right? Do you have connectivity? Do you have tools, right? Those top 10 features, those are the questions. And then I evaluate the trade-offs between ease of use, complexity. Do I have access to the code and those types of things? But again, they have a visual builder. They also have an enhanced decision-making capability, which means you can add business logic into the interactions between each node. So, oh, thank you, May. I forgot to put it in there. So thank you for doing that. And I know we're going fast, but this is not meant to be the evaluation. It's just to show you, I'm giving you a few. I'm gonna encourage you to pick a few of these. I'm gonna publish all the links in the community. Just pick a few and go through the list of questions that I will be posting in the homework., option. Uh, close this out. Okay. N8N. Here we go. So N8N gives us the ability— how many— some of you have used N8N. This is what we were talking about yesterday in the VIP section where you build your workflow block by block. This is pretty old school. We've been doing this a long time. We called it business process execution language. It allows you to create blocks. It allows you to orchestrate and transform content between each block. Highly customizable, highly integratable, great place to start if you have a highly complex workflow that you are building. Thank you, May, you're amazing. I owe you a sweatshirt. Amazing. How many platforms? Uh, 3, Gabriella, 3. So 3 platforms. If you want, you could do them all because that's how I am. So I'm giving you more than 3 because some of these are going to make sense to some and not to others. Let's get to Make. Make is my new friend, right? They hired me to come do their webinar series, uh, in just a couple weeks. So I'm super excited. On March 17th, I will be there. So Make is an automation tool, very, very similar to N8N. The difference with Make is that Make came before, so they have a lot of more traditional connectors. Here's why this is a good thing: if you are inside a business that is old And I don't mean, you know, 4 months old. I mean like it's connected to systems that are not new, 10 years old, 20 years old. Make will likely have connectors to support that. N8n might have it or you might have to build it custom, but Make was built before GenAI. And so therefore it has a lot of more traditional connectors, uh, that we see and use. And I, again, don't make your decision. This is what most people do. They go to the website and they're like, well, this looks good. Right? I want to give you these li— the questions we just went through, give you that, you print it out or use your iPad and you just go through and go, okay, does it do this? Does it do this? It's a little bit of work, but it's worth it to pick a tool that you don't regret within 3 months or 3 weeks of building it. Um, Stack AI, here's another one. So Stack AI was, uh, provided to me because I do happen to know, uh, the people who built this. It is very similar concept of node by node, but here is what I like is that N8N and Make, they're not great UI tools. Like they don't create a beautiful user interface for engaging with those processes. This takes into consideration, do y'all remember on Monday we talked about human and AI experience design? Stack does that. It thinks about the whole stack from the time a user engages to the, the experience they see when they engage with it, the data, the access, and the return of a response, all of that. And hey, you get a— oh, I'm like, is that free? I would do that. I always mystery shop these, these organizations. I do like to look and see who their customers are. Gives you a good sense of the industries that they support. I don't know if you can tell, um, there is, uh, quite a bit of financial here. So this system is highly supportive of regulatory and compliance-driven workflows, which is pretty important. Thank you, Sherilyn. Um, yeah, exactly. I would too, Donica. Um, that's where I started, uh, and, and of course I went to Lovable too. We mentioned yesterday Cursor. It's not my favorite, but that's just me. So if you like— if you're an engineer, you will likely find this better. You will likely find more control, higher levels of flexibility, right? Those are the trade-offs that you're making. But it's a little harder for you to use if you're not an engineer, if you are not— if you don't understand kind of the core, you know, underpinnings of building engineering work. It's good. I like it. Um, you could see, and again, this is why I look at the teams that are using it. These are technology teams. Datadog, NVIDIA, Figma, Ramp, Adobe. These are all technology companies, not companies that use technology. They build technology. So kind of use, these are all things that I look at before I use a tool to be like, okay, who are you serving now? And who, if, is who you're serving someone like me? Because if not, I'm gonna end up being a one-off for them and they won't take care of me. So I want— these are the lenses that I look at it. Okay, last two, last two. This is of course lovable and, uh, uh, uh, yeah, yeah, exactly. Yes, I agree, Javier. Okay, so this is lovable. I'll go back to my dashboard. And Lovable, in my opinion, is a great first. Actually, let me go. I think I put a slide on this. Um, do, do, do, do. Yeah, this is the ecosystem we're going through. So you'll get, you'll see this in the deck. These are all the tools. They're not hyperlinked, but that's a great idea, Noelle Russell Russell. So I will hyperlink them. Um, but the thing that I wanted to, uh, make sure that you were doing as you're going through this is actually testing out these tools. So Lovable, I like because I can go into ChatGPT, which is kind of what I did for VIP. I went into ChatGPT and I said, hey, I already have this PDF that I was gonna give them that I've used before. I usually use it on site with clients. Here's a PDF. Can you turn this into a tool? And that's all it took. Can you turn this into a tool? And so here, I think I called it the Strategy Hub. So I'm like, can you turn my PDF into a tool? And now you could see it here. Right? So it has different sections. Here's the sliders, right? I mean, it's V1, so we'll see what happens, but like, I'd encourage you to do the same thing. I'm gonna— in, in for everyone here now, you're gonna get the questions and you're going to get the tools. And so just take those questions, drop 'em into Lovable. You get 5 credits for free. As a matter of fact, if you use my link, um, you get 10 credits for free. So like here, let me share you, copy the link., and put it in there. So if you use it, do, do, do, do. If you use that link, you'll get 10 credits. So that means you got 15 credits. That's enough to do some damage. But don't prompt in Lovable. Prompt in another cheaper tool, ChatGPT, Claude, Gemini, and ask it to build you a Lovable prompt that you can use to create what you want. Um, awesome. Okay. Uh, and then finally, my favorite, which is what we're gonna build tomorrow, I'll give you the link to this as well, which is doo doo doo doo doo. This is Marbleism, and most of you are going to find, if you haven't used it before, that this satisfies a lot of what we're doing. Oh, thank you, Sherilyn. This satisfies a lot of what we're doing. So let me click on it here. This is the demo that I showed you earlier of the data that— let me go in here— of the data that I I mean, I'm up to 323 hours cuz I use this every, all the time, every day. Um, but things that I like, for example, uh, is earlier today. Yeah, here we go. Is this today? I think it is. Good morning. Oh, that's the 12th. Let me go down. Let me go down to today's. I think this is today's, right? Is it? Oh, is that today? I don't know if that's today. Is today the— yeah, today's the 18th. Oh my gosh. I think I told someone it was the 19th today. Um, but here's, you know, it gives me my schedule. I configured this. It tells me things that I need to be aware of. It gives me my daily snapshot. It tells me, hey, would you like to— I, I sent it a message being like, hey, I need to book a call. Notice my request. Can you book a call with first name? It knows because it has context to my email who that is, where they're from, the call, right? Um, so these, I want you by the end of tomorrow to have a similar Set of tools that allow you to get immediate support based on the words and definition we use. However, this is just so that by Friday you're like, well, that was pretty cool and we built 6 agents and that's awesome. You have the opportunity to use any one of these tools in their free tier to build out what you've, what you've developed. Right? Me too, Danica. Me too. I love it. So this is, I just did this yesterday. But I'll just give you a sense. These are the posts that I've created. I'm going to make it a little bit bigger so you can see. I'll go back to last week, right? I was showing you, but oh yeah, I was going to do something fun today. So right now we're at Wednesday. All of these posts went out at 9. I'm going to go ahead and grab this one, conduct, and I'm going to drop it over here and you'll see it's scheduling the post now. Now, granted, I went in and I reviewed it before I got on the call with you, right? I— that content is based on my training that we did earlier this week. Right. So thank you for that. Thank you for that list. And now it's going to schedule it. And so hopefully, unless I had too many hashtags, which does happen because I'm a hashtag person, but then it'll be scheduled and you'll see this post on LinkedIn. Their pricing model is ridiculous. So remember— oh yeah, don't forget, if you sign up for this, use #TEAM70. As your promo code, um, because that'll give you 70% off your first month. It's like $248, $248 for the year, for the year. Um, and 68 anyway. So this is like, I, my entire business model is based on this platform. Now, granted, I have a very, I did the things that we talked about. I investigated them. I met with their CEO. I looked at their leadership team. I, uh, figured out what, what, uh, LLMs they're using and where they're hosted. I also asked questions about responsibility. So I did a lot of lot of work. Um, right. Uh, oh, that's okay. Oh, well, yeah. Sorry, Kim. Um, yeah, I don't know what I could do. Um, but, but at the same time, Maurice, like, it's good that I've added them, but at the same time, like, feel free to do the same thing. What I think is nice is that I can at least let you know they were receptive to my inquisition. And that I think is half the battle. Are you using a tool? Like, that's why we, we don't think about these as tools. They're operating systems. Like when I invest in Microsoft, like that's not a short-term, I might use it today. Like I, it, it seeps through everything I do in my company. So I wanna make sure that I'm much more particular, right, about who I welcome into my technology family. And right now at least, I've seen that all the tools that I've shared with you today give you the opportunity to do that, but do your own homework. So after today's call, we're gonna end right now. I'm a little, I'm late again, dang it. Um, but after this we're gonna, uh, jump into VIP. So all of you with, uh, Q&A, please jump over to VIP with your Q&A. Um, yeah, uh, that was Stephanie, that was for Marbleism. So Marbleism is, uh— oh, good, good, yes, good. Well, oh yes, nope, I hope so. I hope Team 70— and you don't even have to use my link for Team 70, like, just use it. If you don't want to give me credit, that's fine. I like the CEO to know that I'm like help, you know, I'm sharing my— and again, the only reason I'm doing this is because I am finding success. So it is working for me. Uh, and my— the next thing, what I'll show to you on Friday is the governance, which is how do I know that these 600 drafts were good? How do I know how many of them— how much editing I needed to do? So there are ways to measure your success, and that's what we'll do on Friday. Um, But what I'll do, so I need an hour with my VIPs, and then I'll be posting the replay, the full tool list into the community, and the list of questions to ask. You don't have to ask them to a human. You can ask them to the chatbot on that app, or you can just look through the documentation for those responses. And yes, some people— oh, I don't know. I don't know if, Sarah, that's the question you were asking, but yeah, that's the AI Leadership School. School. But if you want to, this, I think you're asking about the I Love AI School, which is where all of this is stored. So that'll give you access to all the replays, all the content. And then we have a call. So AI leaders, don't forget, we're meeting tonight, 7 PM, if you're in the AI Leadership Institute. If you are in I Love AI, we do calls every Friday. So that means this Friday, we have 3 calls together if you can take it. Right? We'll do our calls, for the challenge, uh, and then we'll also have an I Love AI call and it's a free-for-all. You'll get to break out into meeting rooms and meet each other and find out what people are working on and demo your own agents should you choose to accept the challenge. So that's a little sneak peek to Friday. Uh, if you have an agent, William just posted his. If you're up for demoing it live, we would love for, uh, for you to do that. Anyone else who builds an agent that is working and you're up for demoing like I'm doing here, We would love to have you. So we're gonna do a little demo showcase on Friday. All you have to do is let me know you're up for it. So thanks everybody. By the way, don't forget you can go up into the chat. There's 3 little dots in the chat. If you've broken up the chat, there's 3 little dots. When you click on those dots, it says save chat. Feel free to do that. You'll see all the links. What I usually do is I save the chat, throw it into ChatGPT or your favorite LLM and say, pull the links out into a list. Not because you have to, because we're gonna give you the list, but because it's a good use of your AI skills. Like, just partner with AI to help you get immediate access to what you need. And then tomorrow and Friday, please don't let me forget. It's too late now because we gotta jump. But I do wanna start with everyone, you know, throwing in their LinkedIn and us having a chance to actually connect with each other outside of this challenge. So hopefully one of you will remind me to do that tomorrow, or just throw your link in because that'll also remind me. But right now we're going to jump, go take care of our VIPs. Thank you all for joining. Hope you enjoyed today's session. Again, it's a little bit of work, but it's so worth it because you'll end up picking a tool that will serve you much, much better than just picking the cool tool that's trendy today or the tool that's the cheapest or most expensive. These— this rubric will serve you well. So have fun tomorrow. We're literally— bring your laptops, don't join from your phone. We are gonna build these 6 agents out. So I'm gonna encourage you to go and you can either watch me do it. I'm gonna build a whole new company and I'm gonna build it from scratch leveraging the work I did on Monday and Tuesday. So it's gonna be super fun. By the end of tomorrow, we'll have 6 agents working and, uh, hopefully you'll be really excited about that. So, all right, have a great day everybody. We'll see you in just a few minutes. VIPs, head on over to the VIP link. I'll see you there in just a second. Bye everybody. Oh, Gabriella, I'll see you on the next call. And you remember, you're first. Hopefully you got that question. You— I saw your question. Hopefully I answered it. Um, all right, thank you everybody. Talk to you soon.