DAY 4: I ❤️ AI Challenge: Build Your First AI Employee

All right, so let's get this party started. Before I forget, it is one of the most important things. It is completely optional. However, I will tell you, this world is very small. People who have the skill set that we are learning today, but, and some of you, this is the beginning of your journey. Some of you, um, it is, you know, the middle of your journey, but in either case, this world is really small. Uh, so for example, I right now have a hiring manager who's reached out to me and said, do you know anyone? And specifically asked if I knew anyone. Oh my gosh, Maurice, this might be you. But if I knew anyone in Canada, so any of my Canadian friends, it's a very specific role. Not that you want to go and join, you know, the workforce potentially, but it is pretty cool and these are cool people. So I'm going to, I'm going to introduce you. Or I'm going to at least let you— this is how I roll, right? I want you to see them first. You, you, you check them out. But, but I get these kind of requests all the time. And as you all will, as you're exposed to this space, you will become leaders in organizations or running your own company. You're going to need help. So the best you could do, best thing is to build your own network. This is a huge part of my success. I, of course, I use LinkedIn to grow my community, my network, my relationships. Uh, and I did not know this, but there is a limit to how many connections you can have on LinkedIn. No limit to followers, but like, I appreciate followers. It's awesome. I like people who just follow what I do and eventually one day they're like, I've been following you for a long time, but I better would more appreciate being actually connected with someone who I can email and say happy birthday or congratulations. Uh, anyway, so I hit that limit. It's 30,000. I encourage you all to hit this limit because What's better than having 30,000 people, uh, that you could talk to? But it also, it kind of like, I don't know, it kind of made me really appreciative of the people that I had connect to up to that point. Uh, so I'm gonna encourage you to set a goal for yourself to just increase the amount of people you're talking to on LinkedIn. Create opportunities for them to connect with you. One of the easiest things to do is to have an experience like this and then set up a LinkedIn Live and just invite all your friends, all the people you're connected to. I'm not sure why, you know, you would connect to people if not to, like, invite them into new experiences you're having, new ideas you're learning. I do it all the time. Granted, it's not for the faint of heart because I just go live and I might not— it might not be my best hair day, but I don't, I don't mind because I'm so excited about the technology. I'm so excited about the opportunity, and more importantly, I'll share with you a story. Some of you know this, but like in my early days of Alexa, I desperately wanted people like me on the team. There weren't very many people like a female, Latinas, moms, caregivers. There were no one who had my lived experience that I could be like, oh my gosh, wouldn't this be cool? Most of the time when I'd say, wouldn't this be cool? My response from my team would be like, eh, I don't know, that doesn't seem very good, but you go, you do you. And so I actively hunted like people. I'd be like, oh my gosh, you'd be awesome for Alexa. Come join me. And this is before it took off. But my, like me joining Alexa and then of course me getting ambitious and building a bunch of stuff for Alexa changed the trajectory of like my whole life. Like just doing it at the time that it was offered to me, which is what we're doing here. You are here at this time right now. The whole world is gonna start shifting to away from what cool tools can I use to what outcomes can I build? And that's literally the definition of employee, contractor, AI, agentic teammate. Like, they're— we're looking for, can it do a thing for me? And I'll just tell you, like, right now is like a moment where a lot of companies are gonna start shifting to, can people help me build these workers? Not just, do you know a cool tool? Can you teach me AI 101? They're looking for help. Most of them are actually like, you know what, on second thought, I don't really need to know the details. I just need it. To work for me. And so some of those people that you'll partner with might be right here in this community. So I want to encourage you to connect. I'm going to give you my secret, like, way that I do this is that every platform, every Zoom call I join, I will throw my LinkedIn into the chat, encouraging others to do the same. And then I don't get super distracted because we're going to do a lot of hands-on stuff today. And so I'm going to hope you can stay focused, but As people throw their LinkedIn, and you could start doing it now, just give me an example, but you just click on every link and it'll open up in a tab and you just go through it later. Like, just open up every single one and just commit to yourself that you're gonna look at the people that have shared this experience with you. Somebody— some are here for the first time. People are continuing to join, which I think is exciting and good. But I also realize, like, if you join on Friday, like, you're gonna be like, what? I, I, I missed something. But the good news is all these are recorded. And based on the success of the challenge, we'll do it again, of course. So you'll have a chance to repeat some of the things too. So thank you all for doing that. So just click on them. But here's— of course, y'all remember from yesterday, you can at the end of this call when everyone's thrown in their LinkedIn and you don't have to play, but if you do, you could just do save chat and then a little flex. You could take the chat that has all these LinkedIn. You can actually go into ChatGPT and ask the new operator capability to go connect with people on your behalf, which is kind of fun. Or use your agents that we're gonna build today to take that information. You don't have to, you can do it the old-fashioned way. Personally, I like to actually know the people I'm connected to. So whether I've met you in person or I've met you virtually, like we've had a moment, like it's not like I'm randomly connecting with people. So if you haven't had a chance to actively connect with someone in this room, like, that's your chance. Like, open up their LinkedIn, see what they've done. You'd be surprised. You'd be like, oh my gosh, I went to that school. Oh my gosh, I like dogs. Oh my gosh, I'm running a half marathon. Um, so who was I just talking to? Oh, a good friend of mine from, uh, she lives in, um, Melbourne, Australia, and she's doing this crazy cycling ride. And I had no idea she was into that, but I'm like, oh my gosh, I like doing crazy things too. So I felt more kindred knowing those personal details about someone. Um, and then of course my first question was like, where's your page? How can I give, you know, money to your effort? Uh, you'd be surprised how many people wanna help you. Uh, but most of us just don't have the courage to even say hi, let alone ask for help. So, and she didn't ask me either, and I was like, send me your sponsorship form. Let me help you raise money for your cause. Um, so it was really, really cool. So I wanna, I of course extend that to all of you. I am here for you. If you are doing a cool thing, I wanna be part of it. If I can amplify it to my audience, I will. I have about a million followers across all the things. So like TikTok, I have about, I guess I'm now like 4 million views of my videos. So people, I don't necessarily count if you follow, like I said, following does not mean that much. I'm like, do you engage with me? Do you send me messages? Do you buy things that I sell? Do you ask me to help you sell things you're selling? That is how I evaluate the engagement of my audience. And I encourage you to do the same thing though. Followers is cool. We call it a vanity metric. It makes you feel good, but people can buy followers. Sadly, I don't, but you can. Um, and so if you wanna feel good about yourself and have 20,000 followers, it's like $300. Like you could just go buy it, but it's way better when you look at your followers and you look at your connections and you're like, oh, I know them and oh, I help them and oh, I, you know, that person, a lot of my, uh, people in LinkedIn, I've helped their kids. Like, they're like, I have a kid who's in college, can you talk to them? And I'm like, yes. You'd be surprised how many people have told me, um, when I connect with them, they're like, I'm shocked at how responsive you are or that you respond at all. Um, but even more so when they ask me for help, if I, if I can in any way, I will. And so I've talked to a lot of people's kids. I've talked to, you know, some people are nice and they go in and they buy my time, but that's not, that's not the metric I use to measure my success, right? I measure it through impact, through, you know, like I want somebody to be like, oh my gosh, you helped my kid forever. I will be your friend. Uh, so I think there's a huge opportunity for us to do that. But I will tell you, the only way I have time to do that is by implementing technology that frees me up. From the nonsense of my day-to-day, and that's what we're going to focus on today. I will tell you, I do have a bias, a little bias today. Of course I'm going to show you marbleism because I love it, and I'm going to share with you what I love. Um, and it's easy and it's cheap, and it's— I mean, I wish I could do it for free. Is it free? No, I don't think it's free. Anyway, it's super cheap though. Um, so hopefully you'll have a chance to grab it. Don't forget, um, I'll throw in the link, but it's got Team 70. So if you didn't grab an account yesterday, some of you already did it, Team 70 takes whatever the cheap thing is and gives you 70% off. So do that. Like, it's only for the first month, but you only need it for, you know, 16 hours. So do that. Yeah, $259 a year. Uh, I don't know. I— oh, well, if you use my link, you get 10% off, so it becomes like $249 or something like that, or $239 anyway. Um, and then for the first month, $13. There we go. What? Like That's a coffee and a half. Maybe it's a coffee for me and a coffee for you. So I would encourage you to do it just 'cause it'll be fun and it will give you, in my opinion, one of the most important things that we can do is see end-to-end results. Like not just learn a tool that does one thing, but like how do you build something that's really valuable to a company? What I wanna do first, so I'm gonna show you that, but then I'm gonna show you Microsoft. I'm a Microsoft MVP. Again, I am biased. That's why you take these challenges. Like I take, challenges all the time from different people because I actively know that they are extremely biased in what they're telling me because they found success in this way that they're doing something. But I, I want that. I want to know what— how did you do it? How did you— you know, one of my favorite books is that— what's— I always forget his name, which is terrible. I want to give him props. Dan Martell, is it? Anyway, long time ago, 10 years ago, he runs not challenges, but he runs these workshops. I take them over and over and over again. One, because I like the people in his audience that he attracts, but two, I still haven't figured it all out. So I'm still, still waiting for the golden age of me, you know, not being overwhelmed on a Monday with all my inbox, uh, disasters. It's getting better, but, but I always look for ways to get better and, and optimize. So hopefully in either Marbleism or in Copilot Studio, I will give you an opportunity. Now, the reason I like to revisit Copilot, we talked a little bit about it yesterday, but the reason I wanna revisit it is because So many businesses are running on Microsoft and Copilot as of March, like days away, are going to, with their license, depending on their license, they're gonna get Copilot for free and they won't know what to do with it. So you might even have a Microsoft license and you're gonna be like, I get Copilot for free. This is amazing. Um, so it's kinda like yesterday with Lovable. People heard about Lovable and they already had preconceived notions because of what they heard. 6 months ago or a year ago, they're like, oh, Lovable can't do this, Lovable's not secure, like all this, you know, stuff that came up. And that was true, but now it's not true. They create— of course they heard on Reddit people were complaining, so they built features to make it better. I can't keep up to date on all the things, but I'm a Microsoft MVP, 5 times actually. So this is the cute award that I get. I always like to like— this is my one moment to brag a tiny bit about myself. Like every year you have to apply. And you have to ask them like, am I worthy? Um, so every year I get, and the first year I got it was during COVID which I felt, I don't know, I felt like, um, I felt like it was a really interesting time to get an award from a technology company because it's not like they could see me like this. This award is based on how much are you learning about what Microsoft does and then sharing it with people. And so during COVID it was a bit of a weird time where sharing it with people wasn't as easy as it was before COVID where I could just, go to a meetup and see 300 people and tell them what I know, or I could do a lightning talk at an event. So that's when I really like laser-focused on, I mean, some of you can go back and see my TikToks from, from COVID They were cool. They were good. I didn't dance or sing, so don't worry. But they were like, I did this whole series on how to think like a programmer, which interestingly enough is extremely valuable as we move into AI because you thinking like a programmer and actually writing code are two completely different things. But Thinking like a programmer changes the way you build agents. But I also wanted to demystify it for young women, right? Like, I wanted women, and I will tell you, I'm very popular with 7th graders for some reason. So I have a lot of 7th grade followers, about 2,000 of them, which I love. I don't know why. They're now, I guess, older, but I have a big demographic of, like, middle schoolers. Um, and, but I like it. Like, it, it, it, it's very interesting when a woman meets me at a conference and goes, oh my gosh, my daughter loves you. And I'm like, Oh, that's nice. Is she here? She's like, no, she's in school. And I'm like, yeah. But I used the platforms I had available to share what I know. And so as a result, Microsoft said, we loved what you're doing. We love what you're, what you're about. But the benefit of that is I got to use technology early. And so I've watched the evolution through practical experience of Copilot and Like some of you might know, Copilot has gotten a pretty bad rap, deservedly so. It's been a rough ride, right? Like they promised the world and didn't— it didn't really do a good job in its first couple of years. Um, the worst thing you could do is give someone an agent in a product and the agent doesn't even have the ability to help you in the product you're in, right? Like you're in Outlook and you're like, Copilot, help me with this email. And it's like, what email? And you're like, huh, what are we doing? So that was Copilot's like initial, a little bit of a disaster. Also, they didn't teach anyone how to use it. So I'm here to kind of revisit it to give you all a sense of like, it's pretty cool. But most importantly, 9 million businesses are running on, are leveraging this technology. And they, it's an opportunity for you to help if you are a consultant, if you are a business owner. If you're going to do anything related to AI, it's an easy way to create value for them to allow you to open the door to creating more value. Cool. Oh, awesome. Is this it? I think so. Yes, but did you use Team 70? Oh, oh my gosh, I've been sending messages to waiting room participants, which nobody is in there. My bad. Anyway, so Team 70 is the promo code. So make sure you use that. That should work for you. That's the link. Thank you, Maurice. And then I'll put a reply, Team 70. I did not get this when I joined. I just paid whatever it was. I had no discount codes. I got no promo codes. So when I found this code and I had the call with the CEO, I'm like, how do I make this free? And he's like, well, I can't make it free, but I can make it really inexpensive. $13. That's cool. That's cool. I like— I don't— I'm not hurt by that. All right, so let's get this party started. We are all here. We're all ready. Today is the build day. Now, it's not just what I'm hoping you will see across both Marbleism and Copilot. You will see a pattern, which is the work that we did on Monday defining the profile of the agent we want to use. Remember, the goal that we started out with was if I was doing my work and all of a sudden I needed to leave. Family emergency, or I'm going on maternity leave, or I just want a vacation, what is something in your daily life that would cause you stress while you were gone? And typically people land in the area of like email, calendar, um, you know, keeping up to date on social, like keeping present in social media. Like a lot of business owners especially have this stress of like, oh my gosh, my inbox will be a disaster. I remember I used to take 2 weeks off when I worked in corporate America. I would take 2 weeks off and then 3 or 4 days before my vacation was gonna end, I would be like, like this anxiety would build knowing that I'd have to recover from my vacation by like digging out of the hole that I created for myself when I got back. And so that's why, um, I thought these use cases that we're going after will work. But the pattern is exactly the same no matter what it is, which means you define the purpose of the system, the, the problem you wanna solve. You figure out what success looks like and you define it. You hard code what that success is into a document, whether that's on Word, whether that's— I mean, sometimes I do, I use GitHub, right? So I'll use GitHub to document my agent profiles. I didn't call them agent profiles back then. I called them personas because we didn't have the word agent. But I built personas and then inject, right? You all remember, some of y'all remember. Um, but I injected those personas into my LLM. So So depend— it's kind of like some of you might do this today where you got trained to say, you are a marketer and your job is to build good marketing content and be highly responsive. And but you would do that in vending machine mode, right? You would, you wouldn't train the model. You would just say in a single prompt, you are doing this. Here is your job. Build it for me. And then you'd hit enter and expect it to give you magic. And so what we're doing is an evolved version of that. But the process is the same whether you do it. I'll show you a couple of examples, but whether you're doing that in a platform like Marbleism or a Microsoft Copilot, or I think on Tuesday I showed you Meta AI, right? Meta AI Studio custom GPTs. They're all exactly the same. They ask for the same thing. What's the purpose of your bot? What are the instructions you wanna give it? That's what we did on day 1 and day 2. And then on day— end of day 2, we also talked about what is the data we need, right? That's a critical part of the process that equates to the knowledge base. So I'm gonna, like I said, today's all about walking you through building everything, but also giving you a couple demos of different platforms just so you can see the skill set you developed over the last 3 days is transferable to all of these different platforms. Now I say transferable, but my hope is that you will actually just go try it out. Take what we've built, the documentation that we've created about our agent, and just start sticking it into a Meta AI, uh, agent. Stick it into a custom GPT. See what you can get done, uh, with different tools that I'll offer you. But we're gonna focus— I say that and I'm like already thinking about a third one that I wanna show you, but we'll see. We'll see what happens. All right, let's get this party started. So, um, okay, cool. Uh, don't forget, Uh, let's see, I tried to connect. Oh yeah, good. Okay. Oh, let's connect. Okay, good. So don't forget to, uh, by the end I'll try and remind you again, but don't forget those three dots, save the chat, grab everyone. You know, like, we're family. You don't know it yet, but we're family. Some of us, we met a few months ago, and now I'd be like, I don't know what my life would be like without y'all. So you, you just, you, you're in a new world. I would welcome every friendship, relationship, even just business connection, right? Because like I said, you never know. I just got a request from a hiring manager. I'm like, I know people. And I told them it was super funny because I was like, I know people, but not just people I know. Like I said to her, I was like, I know your people. And because her people are my people. So it's a really cool company. Anyway, I can't wait to share with you. So if you're in Canada, let me know. But let's go ahead and get started. So I'm going to share my screen. Hey, Noelle, I'll jump in real fast. Please do. Is that Tara? Hi. Yeah. Hi. The, um, well, somebody's LinkedIn, um, connection that was in there, and I don't know which one it was because I had to restart my computer suddenly and you lose the previous chat, but it put a virus on my computer immediately. Oh no. So before anybody else just clicks on the links on the things, read them and make sure they're valid or cut and paste them into the URL thing. Yes, thank you. I just want to give you a heads up. Absolutely. I'm scanning through them right now, but if I see it, I will delete it. Um, or someone could put like big stop sign, but that would be terrible. Uh, yeah, I don't see anything here yet, but good to know. Good to know. Um, but yes, thank you for that. Thank you for that. Let's see here. Uh, uh, uh, yay. I would love to stay connected too. There was something else in here I just saw that I thought was really cool, but now it's gone. So, oh well, I'll have to come back to it. I'll see it in the chat later when I'm going through for my notes to you. I'm just going to scan through really quick. Oh well, it's nice to meet you all though. So connected, just know that I can't accept any new connections, so I will be following you. I'm a follower of you. Um, so, uh, start posting good content so I have something to follow. Uh, but I'm here for it, here for it. So thank you for that. Okay, let's go now. Let's see. Oh yeah, I'm just gonna jump right in. So this is, uh, I'm actually, I just uploaded a bunch of media, so we'll come back to that later. But I'm first going to walk you through kind of the, the intention, right? And the problem we're solving. So this is, I, I want you to kind of picture Noelle, right? And my ideation around who I can serve in the marketplace. Um, I'm gonna put this in slideshow mode, also share these slides with you. Uh, oh goodness. It did. Can you see that? All right, it took over my whole screen. I'm gonna try and make— I don't know. Okay, hopefully you can see it. Um, but this is like the pain point I was trying to solve, and this was me, right? Like, I'm very busy. I do a lot of speaking. I travel a lot. So getting consistency in managing my inbox, getting on social, writing content— I have a newsletter with 17,000 people on it, and I barely— before January 1st, reached out to them. I mean, the only time I reached out to them, which is not good, is during Christmas. I would do 12 Days of Christmas, so you wouldn't hear from me all year, and then you would get a post every single day for 12 days. Now, some of you would be like, and some people did, about 100 people on my list were like, okay, delete, right? Even though it was good stuff and I wrote it, right? Like I, I was pretty proud of it. I gave a gift every day, but they were like, I don't wanna hear from you every day. So I, I lost about 100, but I gained like 700 people. So it was worth the 100 and those people aren't my people anyway. But it was pretty interesting, um, because I looked at it, I'm like, nobody wants to have like, you know, start and stops. They did all, everyone just wants consistency. They wanna know that you're trustworthy, that you show up, that you stay invol— you know, stay engaged. And so that consistency was a big challenge of mine because I would travel and I have lots of business, 3 businesses that I'm running. And so it was a challenge. But I also, one of the challenges I had was just understanding, not just can I filter and categorize my email content, but like, would I know what's important? So many times I would not see, and to my dismay, I would not see when a request comes in for me to speak at an event, a good one, like an event where I'm like, oh my gosh, I wanna do that one. I would not see it until much later. And then I would go and I would find it and I'd be like, is it too, too late? And they'd be like, oh, we just booked the person. I'm like, and I don't even know if they did. I would say that to me if, if I can be like, yeah, we're good. Like you waited 3 weeks. Like we're, we, we figured it out. Um, so I understood, but it was so painful because the opportunities that I'm looking for were in my inbox and I didn't have the bandwidth to filter through everything. And so I hired, you know, I have a, an executive assist, like a human. Um, but that human, I was like, do I really want them sifting through my emails or do I want them helping me grow my business? Like, so I ended up repurposing my human to do other things. Um, and so this just consistently nagged me. And so that's what I put myself in. And this is what I encourage for some of you who are kind of looking for how do you serve an audience? Like I just put myself in my own position and I was like, How do I fix my own problem? So I fixed my own problem and then I was like, I should show other people how to fix this problem. And that's how we got to what we're doing today. But the, but I consider it what I probably learned in sales at some point. This is like a tip of the spear opportunity. It's really cool. Uh, who was talking like during VIP or maybe it was during our leadership call last night. Art was saying that she today, I think it was today or maybe I think it was today, um, is delivering a talk and they asked her to deliver a talk on agents, like building 3 agents in an hour. This is a great talk, right? Even if you never actually use it, though I do, um, it's a great like way to show someone what's possible and get them excited about what's possible. Now I am, you know, I've been doing, working with this long enough that I now have clients that are like, okay, I want that. And just last night I was on a call. Not only did they say they wanted it, they were like, I've already pre-sold it to like 4 of my friends. And I'm like, let's go, because these are the problems, right? So when I'm thinking about offers that I'm making to the world, and so for those of you who are building a business, I kind of look for signals of problems that people have that I personally have had and then have fixed. And then I just teach and show and even maybe deliver as a service the solution to that problem. And the, the best part about that, of course, is that I enjoy it every minute because I know what it's like to be them and I know what it's like to be me after having fixed it. So I feel very confident in my ability to kind of carry them through that painful process of overwhelm into a world where things are done for them. So even though I'm demonstrating a very specific product, I'm gonna go through the features that I implemented first, and then we're gonna build it together. Hopefully it won't take too long. Um, but I think it's really important to narrowly focus on that persona, that problem that you wanna solve for. And that's how I built— I always like to give a little transparency to why I even built this challenge in the first place. And it's to solve everyone who is me 6 months ago. This is for them, right? How do I bring them forward? Remember I talked about that a couple days ago, like how do I bring people Who, I mean, I know your problem, I was there. How do I bring them forward to now in the most inspiring and enthusiastic way possible? All right, so the first one, um, well, first thing, I always kind of look at how people are solving this problem today. Um, so I'll go and I'll research my ideal customer profile and find out what their— so again, if you're gonna become an agentic agency where you're building agents for other people, or you're going to build agents for yourself, you should understand the scope, right? You should know how are you solving this today? What is it costing you today? So that's what I did. I went in and I searched for all of those contracts that I had missed, that I did not answer as a result. And there were a lot, a few, a lot more than a few, less than a lot, an embarrassingly large number that I just never got back to because I was Busy. Now I got back to the people that needed me to help them with their kids, but the contracts, not so much. Um, so that was something that really focused me was like, how do I not lose $220,000 in a year because of my oversight? Like, how do I not let that ever happen again? And how do I help other people with it? So that is why I think there's this huge, uh, directional shift towards agentic workflows, right? Building systems that do stuff, not just can I go in and ask a question, which is Again, we're in the 1% of the 1%. Those of us who understand that AI is capable of doing so much. Most companies, 90% of companies are not here. And I know this cuz I've talked to them. I've talked to trucking companies, I've talked to real estate brokers. I talked to, you know, I've been at the National Association of Broadcasters. I've spoken to the, uh, International Builders Show contractors. I went to the lawn, uh, what was it called? The Landscapers Association. None of these people have any idea how to get started. They think it's really complicated 'cause we in AI tend to do that, right? Like they, they distance themselves from being able to actually use this. So that is why I think people are enamored with this idea of, hey, could I just buy an employee from you? Could you configure it and I just get it? And that's why the best way to sell AI, uh, agents is to actually build them and use them yourself. Because, and I just, uh, for the purposes of today's session, I just added a video. I recorded it. You'll see today. Um, but I wanted you to see an example of me on the fly building a video that just demonstrates what I'm about to walk you through. But that's what will encourage someone to try it because they're like, oh, it's real. Oh, you're showing me what you did. Oh, I see myself in that problem you're solving. But AI, I say AI employee, but really it means agentic systems that do a very specific task on your behalf consistently, safely, securely, and most importantly, accurately. Uh, and this is something that people are hungry for that not a lot of solutions offer these days. So we're gonna build Eva first. Um, and I'll show you, I'm gonna onboard an entirely new workspace. If you have that open, let's do it together. It's going to connect to your Gmail, which I'm gonna connect it to mine. I have lots of accounts. So I'm just going to pick one. It's going to manage your inbox. It's going to create filters that you design. It's going to take notes during your meetings if you let it. Some of you know I am not a big fan of having agents sit in my meetings and watch me and watch my client. So instead, uh, I prefer to use, uh, ambient listening tools like Plaud, um, P-L-A-U-D, uh, but you're welcome to. It's a little checkbox, and then Eva will show up in your meetings and actually take notes for you similar to any other note-taking process. I don't use that service, but you definitely could schedule meetings, organize meetings, reschedule meetings, um, avoid double, uh, booking meetings, conflict management. I mean, oh, amazing. And most importantly, drafting email responses. All of this, it might take a day, uh, for it to get to the point where it's jet, but that's, that's why we're doing this today. So you, by tomorrow you should see some, some increase in activity, in Ava. This saves about 2 hours a day, maybe more for me, cuz I so heavily rely on it for day-to-day operations, which is what I'm gonna share with you later. Now, Sunny, I talked to you mostly about this this week cuz it, it's so cool. Um, I generate all my content here. What I have noticed though is that the content that I have handcrafted, I've provided it my branding, my voice. Here's some of the coolest things that I do a lot of video. Like this, I can take one of these recordings, day 1 of our challenge, feed it into Sunny and say, hey Sunny, take this YouTube video and create social content from it. And it can create me 10 to 50 pieces of social media that represent this content. Like, can you see? You do it once and you build a system that allows you to like amplify. Oh my God, so cool. So we're gonna do that. I'll show you what I'm doing, of course, but I'm gonna show you what I did to train the model. Penny is SEO optimized. I actually like to say AEO optimized because SEO is still working. Google's still a thing. But what— I don't know if you know this, but like ChatGPT, Perplexity, like these models are scraping content from search engines in order to generate their real-time training data. So if you do things that are SEO optimized, they also help you become AEO optimized or AI engine optimized. And so AI engines mean like LLMs. Where are element LLMs getting their content? Well, they're getting it from the internet. So the better you can do— so sometimes people say SEO is dead. I would beg to differ, that these models are training on SEO content, on where you level up, right? They're training on Wikipedia. So your ability to get recognized and ranked, uh, is still a valuable part of that scenario. And most importantly, for small businesses, it allows you just to be seen when people ask for you. Now, Stan, we're gonna do really quickly, 700 million leads across about 15 different databases. You can ask questions of Stan about how it gets its leads. That's what I did. I just was like inquisitive. Um, but it's gonna write emails, create sequences. It only does one, but to our benefit, I, I, like I said, I think I said this yesterday, or I definitely said it in my video this morning. You wanna walk before you run, dare I say crawl before you walk, um, when building automated sequences, meaning you're building an AI agent that's gonna operate without you. So start small so you can monitor how it's doing that job before you put on the gas. Um, and that is exactly the process I use. So I'm gonna walk you through that too. Now, Rachel, the receptionist, we talked about this yesterday. Kim Gaskell, who you'll see in here, she, she was like, wait, you get a number? This process can be difficult and has a bunch of like technical aspects to it that have been, have created a lot of friction for us entrepreneurs and business owners. So the fact that you within seconds are going to get an assigned number to your business that you can immediately train with your website data. I'm gonna hope some of y'all have a website, or you can use LinkedIn, but you have a source that you can immediately say, hey, when someone calls, first of all, turn every call into a lead. Second of all, go use my data from my website in case they have any questions immediately, like day one. This is super powerful, especially when you're talking to a small business that hasn't figured out how— one of the, um, one of the principles of our AI Business Accelerator is you need to— the first thing we do Of course, I called it the 6 Ps. You'll see a pattern in how I name stuff. Um, but the first P in business acceleration is preparing for people to pay you. And the first thing you have to do is create a mechanism for them to talk to you, to call you, to get their questions answered. And so this is a really interesting but very quick, uh, opportunity. By the end of today, you'll be able to call your receptionist and test it out. Um, so that's the rundown of what we're going to do. I am now— oh, did I do that? Oh no, no, no. I didn't mean to do that. So what I want to do now is just kind of show you— oh, approve. Sorry, Sandy. Show you the process. Now, of course, I've been using AI Leadership Institute for a long time. I'm going to go, go in and build a brand new workspace so you can see what this looks like. So we're going to do this together. Are you ready? I'm going to say 15 minutes, give or take, for us to build this whole stack out. Now, the only way you can do this in a way that adheres to my policies around safe use and development is if you did the work on Monday and Tuesday. So Monday and Tuesday, we wrote down our— what we wanted. We wrote down the definition of success on Tuesday, right? We got very specific about what it was going to do. Some of you built different agents, and that's fine. Hopefully you'll find this at least entertaining while we go through it. But the idea is to replicate those definitions, descriptions, guardrails Safety, uh, good practices, bad practices, all of that is going to find its place in the guardrails and guidelines within these agents. And then I'm gonna switch over. So maybe I'll try to do it in 10 minutes and then I'm gonna switch over to Copilot so you can see the ex— not the exact same process, but you can see I would even maybe take it as a challenge, right? I'm gonna give you prebuilt ones, not a lot of control. Remember the trade-offs we make, right? We're using a platform that's completely prebuilt. So your ability to have high levels of complexity or flexibility or control, you're trading that off to have a system you can have up in 10 minutes. So the alternative is to maybe go for a system that takes longer to set up and configure, but gives you more control and more, more capability, more flexibility. So that's why I thought it was good to give you kind of those two lenses of marbleism plus Copilot. But that could be, There's any number of tools like we went through yesterday that could also be supporting this, this outcome. The most important thing to do is to just start testing and building. So I'm going to go ahead and get started. It asks me what my role is. I'm going to say CEO. You can say whatever you'd like. When this— it is using this role, though, to determine its pre-training for these elements. In other words, if you're the CEO, what are the expectations of an EA? A social media manager, a legal assistant, right? So it is gonna do some pre-training guidelines based on what you write here. I wouldn't get super fancy and like create your own role name, like Chief Visionary Strategist Officer. Like, I would stick to something that would be helpful to an LLM to shorten the amount of time you have to train it. So I'm gonna say CEO 'cause that's me. Do I have a website? Indeed I do. So, um, ah, let's see, I'm gonna go to https— oopsie. Um, ai-business-school.com. Now, ai-business-school.com is a website, it has a bunch of content. I actually built it on Lovable. Um, and it's going to— now what, what the reason it's asking for this is to custom train it on your website. Again, you have unlimited, you have workspaces here. You could just put anything you want for the exercise that we're doing and then go back and create another workspace and do it for real. Spend time, right? This is just to kind of go through the motions so you can see what's possible. It's also to help you later. This is recorded for you to go back and be like, what did she do? What was that? So you can go back and watch the recordings, which we'll break up per step. All right, this is just setting up the entire business. So now it's gonna take a few seconds. It's gonna go out to that website. It's gonna fetch some content. I don't— I've never done this before for this website, so fingers crossed it actually does something useful. But it has for every other website that I pointed it to, some on Lovable, some on GoDaddy, one on Wix. It's done a pretty good job. And then, yeah, I'll just let it do that. So while it's doing this, you'll notice it's going to— here we go. Provide an intuitive description of the business or industry that you're in, which is pretty close. If it doesn't match, it's a dropdown. You can select it. It's also going to take content from your website and put it in. But this is for the business. This is your mission statement. Who do you help? How do you help them? Right? So in this case, it's an accelerator, helps business owners deploy AI workforce. Um, to handle daily operations, we offer these agents. Um, let's see. Uh, yeah, we do it in 5 days. You get 40 hours back a month at minimum. Um, and then of course, who's my ICP? So it takes a good stab based on the content in, right? It takes a good stab at who my target customers are. You can change this. I would just take it as is if you, you know, hopefully you pointed it, you could point it to your LinkedIn. Any website you choose, um, or you can take what you wrote down on Monday and plop it in. But this is not for an agent. This is like the overarching business, um, information, usually informing the agent on how to respond to questions when the agent is request— is, is asked. So now it'll ask you, right, like, so here it'll tell you all the things that I just went through, and I'll say, oh, congratulations, you get all this. Am I ready? Yay! And it'll say your subscription includes unlimited Workspaces. Kabam! I'm in. 6 agents ready and willing to do my bidding, right? But going back to how we started the week, these are employees. They are specifically trained to do a specific job. So the first thing we need to do is onboard them with the job description. Now we've already got an idea of what each one of these do, but you'll notice each one of them. Has an onboarding. Now you don't need them all if you don't want to, right? But let's start with the EA, which is my favorite. Um, we could just choose to onboard. Now there's some words here, you should read them. It's like instructions. You should read instructions. Uh, I'm ad-libbing them for you, but please feel free to create another workspace and go through this. I honestly, if you're going to do this for other people, I would even create a bunch. Um, but you'll notice here I'm onboarding my employees, so this allows me to tell it what I want it to do. Right? So draft email replies, yes. If you do not want it to do that at first, maybe you just want it to organize, you wanna see how that's going, just deselect some of these options. I'm all in. I'm like, yes, yes, yes. Organized email with labels, meaning you're gonna ask it to create labels for you. Archive unimportant emails. Notice one of the guardrails that the platform has implemented is that this agent cannot delete anything from for you. So intentionally, so it is a restriction of the platform that it can't go in and accidentally delete a really important communication that you need for any other reason. So the only thing it can do is archive. Manage your calendar. Yes, please organize all my existing emails, right? It's going to sort the most, the 50 most current to get started, but you can have it. I have 44,000 emails. Don't judge me in my inbox that may or may not have ever been read. So I really appreciate this perspective. So I'm going to say yes. Now I get to choose some of the things that are in my inbox. Now, right now it's doing the most common that people ask for. However, I know, for example, that, uh, what one of the ones that I ended up adding to it was contracts. I want to know, is there a contract in there? So I created, um, another one called contracts. Maybe I'll spell it correctly. And notice when I put in contracts, it's like, ooh, people who needed contracts also needed, right? It's a recommendation engine. So let me say enter. Um, so it's like, do you also need receipts? Follow up? Do you wanna mark things as urgent? Do you wanna track all your orders? It's up to you. You get to define what this looks like. Then I go to next and it's like, I'm almost ready, but I need an inbox. So now I already have my inbox configured, but you'd want to go and connect an account. Doesn't take long for Google. Um, let's see, do I have one I want to connect? Um, I don't know. Um, I should use a new one, so I'll do this one. This is my personal account. Oh, goodness knows it has a lot of stuff in there, right? So now it's connected to this. However, if you're not Gmail, go to Outlook, or you can use IMAP and you can connect these. Totally up to you. Um, you also will want to give it some guidance on how you want things managed. Now you can move, which is what I do. You can move things from the inbox to a filtered folder, or you can copy them and keep them in the inbox, which I don't do. Um, ultimately I decided I didn't want anything moved or copied, so I only use labels, and I'll show you that in just a second, but I only use labels in my inbox. So that way when I'm looking at my inbox, anything that's green, contract, I know to pay attention to. Everything that's blue, newsletter, I can read if I have time, which means maybe never. Right? So I'm going to go back here and choose next. Now I do have to give it authorization. This authorization will reset every 2 weeks or if you have a multi-factor authentication that, that triggers a reset on your authorizations, which sometimes it does. So you'll— it's just based on your security policies of your inbox. And if you don't know, you have some. So I'm going to authorize my inbox. It should just say, yes, I would like to do that. And then it'll show authorized. And then I'm going to authorize it to also draft email responses. And I'm going to use my AI Leadership, uh, one to show you what that looks like. But now both are authorized and now almost there. We're ready. I first got it connected to my calendar. So I'm going to go to— there's nothing on my personal calendar, so I'm going to use my existing Noelle.ai. But again, you could do the same thing, connect an account, point it to a calendar. Notice, uh, here depending on Google or Outlook, same idea, but I'm going to keep it with the one I have. All right, so now Ava is thinking. It is autonomous, completely separate, not required for you to stay in a linear fashion. So we're gonna go right on to Sunny. So that's gonna keep going. It will do its magic. It'll say, thank you. I'm all set to dive in. Let's crush it. Yes. Now I will tell you very quickly, guidelines is how you will control it. Notice there are none. My handy dandy notebook. That contains all the rules I said on Monday and Tuesday, I'm now going to go in and apply. If you chose an email management system, like, what are the decisions you wanted to make? My notes in AI Leadership, if you saw yesterday, I have lots of guidance and that guidance is organically generated every time I answer something or use a draft. So just like it happens with memory management in an LLM, it's— it can be organically generated, but you always have the opportunity to type it in. Now you get to use the benefit of traditional email management processes. You can see here, use the love, love, uh, use the label legal. Don't use a label on any email that's auto-generated. Uh, apply personal to emails from friends or family. So you can either use the ones that are pre-selected or we handcraft them ourselves. All right, so I am just going to leave it as is because I got lots of agents to set up. But you'll want to apply those instruction sets that you have. Now some of you only built one of these, but I'm going to encourage I encourage you to light them all up. What, what's the harm? Just so you can see what's capable out of the box. Then Sunny, of course, has to be onboarded. This is my social media, so the first thing it's going to ask me is where can I best learn who you are, right? So I'm going to put my Instagram, but you could put your, your LinkedIn, wherever you've got the most content that represents you. If you have no— you definitely have a LinkedIn, so I'm just going to get I just saw it. I saw y'all put it in there. So I would do that. But I have a good Instagram account. It's where I put most of— oopsie, I should spell it right though. Instagram. Oh gosh, exposition typing. All right. And then Noelle Russell is my name. Instagram. Does that look right? It does. So now it's going to go out to Instagram. I do not do this, but you can upload a logo. And when you grab that logo, so for me, I'll just show you what mine looks like, but, um, A-I-L-I, right? So I have one of these. It's cute. I like it, but now it's gonna put it on every single post and I'm like, mm, not always. So I disabled it, but at first I did it because I was like, that sounds cool. But it turned out I didn't want a little box on every single post, so I, I don't do that now. But. That's there for you. Now I say submit. Sunny is on its way looking at my existing content, which is funny 'cause it'll be using Sunny content. So now it's gonna pull together a few content ideas and get going. Same thing with the blog writer. So we, you'll see this is just onboarding, meaning initial grounding of the agent. I always encourage you not to spend a huge amount of time in these models because it's often that it's intuition about what you, do, say can be provided without you being super explicit initially. Then once it's up and running, now you can go into the guidelines and provide very specific instructions based on the work we did on Monday and Tuesday. Y'all with me? All right, running through. Okay, here we go. So I'm onboarding Penny now. This is long-form content. Notice now it's shifting to, hey, I'm gonna write some content for you. It already knows based on what I gave to Sunny and my business information, what it's gonna write. I want it as early as possible. It is identified again for my own safety, I guess. Uh, it sent the earliest time it'll send me this message is 8:00 AM. So I'm like, great, perfect. I will do that. That's all it takes. It's gonna send me ideas at 8:00 AM. Now it's gonna actually start right away with ideas for my first blog post. I'm interested to see what it has to say. All right. I'm not going to go back up because these agents are going to start sending me responses. So I'm not going to go back up until they're all onboarded now. Oh, I skipped right over Stan. Now, normally when I do this with a client, I do Stan last because this is cool and very engaging and people are like, wait, what? But for the purposes of us getting through them all, I'm just going to do it now. So this is your sales associate. Its job is to find your ideal customer profile. Which you saw identified on the first page of the on, of the onboarding, remember? Um, so I'm gonna onboard Stan and it's gonna ask me some questions. When would you like new leads coming into your inbox? Remember, uh, Bill, right? Bill created, uh, used a, uh, Google agent script, I think it was called an Apps Script, to go in and find new leads. So very, very similar activity. I want it to come into my inbox. No, earlier is better. And what kind of emails do I want? If you're B2C, maybe you want personal emails, maybe you don't care, maybe you want the best available. I'm marketing for a business opportunity, so I'm going for professional. Then same thing, which email account do I wanna use? I'm gonna use the same one connecting my inbox. What this means is that Stan is gonna send emails on your behalf to the people that it finds, right? So you wanna make sure that this is the right— like, in other words, when somebody gets an email from you, from Stan, it's not going to say Stan because Stan isn't, isn't a thing. It's an it, not a who. So it's going to say it's from you. But as I— and this is going to go out as part of an email sequence, which means it will go out autonomously. You will not review it like your executive assistant. You'll review it right now, but you won't review it before every single lead goes out. Because it's the same message. I'll show you that in just a second. So I'm gonna authorize it, of course, to get access to these systems, and then I'm gonna authorize this one. Remember, I already did the authorization, so it's just capturing that cookie, uh, unless it's in this case for a different service. So it's added an additional service allowing me to send emails, right? Because, um, my executive assistant doesn't have that allowance. This allows me to send it because it's restricted to only sending the sequence that's about to be created. Now, the second thing that Stan does is for every connected lead it identifies, it also is going to send a connection request to LinkedIn. This is where it gets sad for Noelle, because I can't connect with anyone else because I hit 30,000 leads. So that, or 30,000 connections. So that's sad. But I mean, you know, oh goodness. Yeah. Oh, let me, let me, let me take advantage of my, population, but I'm going to go ahead and sign it up anyway just so you can see what it looks like. But as soon as it gets to this part of the sequence, it breaks for me. Almost all of you will not have this problem. Oh goodness, dual factor, triple factor authentication. Let me go into LinkedIn. To do, to do. There we go. Yes, it's me, it's me, trust me. All right. So this should hopefully do its own thing and authenticate. It kind of looked like that worked. All right. Okay, so now my LinkedIn account is connected and I'm just going to keep the free account. But notice if you get a lot of great traction, sales is a numbers game. The more you send out, the better your results. I do free until I get 300 leads to return value. Like if no one is responding, do not increase this number. Change your offer, change your sequence, change the messaging. Um, so I get 300 working for me, which I expect, let's just say super conservative. I expect th— out of 300, I expect 30, right? If you wanna be super good, 3, but I expect 30 people to be like, hey, that's pretty cool, I want to know more. And if they do, then maybe I choose to increase this amount of my investment. Because if I get 30 people to respond and one of them buys my product— this is pretty intuitive setup. So she's just walking through how to set each one up once you get them. Um, so this gives us, um, yes, and it is, it is pretty intuitive. Um, okay, so submit., and then now it's gonna go and start based on my ICP, which you can define and redo. So I let it assume my ICP based on my website. Some of you were not that thoughtful maybe about your website. Did you design it for your ICP? So if that's the case, you're gonna wanna go back to that first day and rebuild the purpose statement of who is this for? Remember the audience in the 3 Ps, right? So Thank you. It's setting up the, the sequence. All right, let's go look. It's still working, so I'll come back to you. Oh, see, I can't even help myself. I'm not supposed to look yet. I'm giving it its time. Now, the legal associate, it is ready to go. All I have to do is ask it a question. So I might go in and provide a legal document that I want its perspective on just to test it out. Now, my favorite is building a, a phone number. You get this phone number for free, so you might as well do it. Onboard Rachel, your new phone-based receptionist. You do need an area code. I type in mine. So 954 is Florida. Let's see if I could do 305. 305. Um, so type in your wherever, cuz you want it to be located near you, or at least relative to your cell phone number. Um, so my cell phone number starts with 954. That's what I ended up creating. It's gonna check area code availability. Maybe you have one of those numbers, an old school area code, and you'll have to pick something adjacent. So another. Miami-based area code, or Fort Lauderdale, 786. So I could have used that as well. So it's asking me where I'm based. It then allows me to choose how it's going to answer the phone. Be thoughtful here. How do you want someone, when they call for you, to get this message? Now, I just start with what it gives me, and then I call and listen to how Rachel answers the phone. Um, and then based on how it answers the phone, I might come back here and modify this. Right? So now this is pretty cool. There's ambient sounds that Rachel can have in the background, right? Call center. That's what I pick. Coffee shop. I don't want people to think Rachel's sitting in a coffee shop answering my phone. Convention hall, summer, outdoor, cicadas, right? Like maybe you're a hiking company. That's kind of cool. Um, you, anyway, there's some fun you could have with this, or you could have no background noise at all. I picked call center. I didn't find it offensive. So, and then remember from the beginning, what is the tone? Now you might not have thought through the tone you want for your agent, so it's going to be pretty generic, but this is an opportunity for optimization and for you to guardrail how you want. Notice here, how would you like me to handle the conversation? This is the constraints and the opportunities. So we wanna tell it, here's what a good conversation looks like, here's what a bad conversation looks like. You have 3,000 characters to use, I would try to use as many as possible, not just to fluff it up, but like use as many instructions as you can fit in to get. And if you have more instructions that fit here, just take it into ChatGPT and say, great, crystallize that into 3,000 characters for me. Cool. All right, I'm going through, going through, going through, going through, scrolling down to the bottom. Okay. Oh my gosh, so many comments. How would you like to end the call? When the caller says goodbye or bye, or when the caller appears to be unsolicited or spam. And that just saves you, right? If it's like an unsolicited call or a salesperson, which people— salespeople have called my sales, you know, my AI agent. If you detect that, you want the call to end so you're not being billed or paying for that. Now we're on a free account, but still. Uh, okay, cool. I'm gonna scroll all the way down. There we go. Oh, good, good, good, good. Okay, so I'm gonna say— oh, and you can also add additional conditions. I'm trying to think, did I— I had an additional condition I added, um, but I can't remember what it is. So anyway, uh, I have one, um, Noelle Russell. One of the— if it's like when you get a real spam call, at the beginning there's It's white space. So I put in mine, if it— there's no sound, just hang up. Yeah, perfect. If there's no sign, I would give it, you know, quantify it, right? If there's no sound in the first 5 seconds, hang up the phone. But that's an excellent one. I love that. Absolutely. And so now you're gonna ask if it needs to transfer the call. This is the magic that Kim and I were talking about yesterday, that this is kind of hard to configure. It takes time. Some people think it's, you know, it could be we were using Twilio for those of us who are Twilio people back in the day. Like, this used to be complicated. The fact that I can now, um, transfer the call and summarize in private, meaning it's gonna transfer the call, Rachel is gonna summarize it to me before it activates the caller. Like, that's kind of cool. Um, and so I'm gonna put in my actual phone number. And it's fine, don't worry. I have that, uh, Apple thing where if you call me and I don't know you— yeah, so I'm not worried about anyone calling me. Some people immediately, like, you put your cell phone on a video. I'm like, it's okay, Apple saved me now. Um, so that's okay. I know, right, Dantasha? It's your instant instinct. I love that though, thank you, because you care about me. Uh, but yeah, so don't worry about it. Um, and, uh, but I can also send it to more than one person., right? So this is in the case that I have a call center, right? So, or I have 3 sellers that I wanna forward to. So really, really cool. You can also not summarize it and transfer it immediately, right? Christina, I try to be. Now this one I don't do because my calendar is still under duress. But why wouldn't you do this, right? Give your receptionist the ability to say, Hey, just wanna, you know, if you wanna talk to Noelle, especially if you are in lead gen mode and you wanna talk to everyone and try and get them to become a customer, you can have your assistant book calls for you. If nothing else, turn it on so you can just test it and see how it works. Um, because again, these are processes. If you think about N8N, Make.com, like these pro— these, these were all nodes you would tie together and each node required separate configuration. So as I started using this, all I'm thinking in my mind is how much time I spent configuring this for other people. Like, I was just like, it's magic. So that's why I'm so excited about it. Okay, so what number should we expect testing calls? That's me again. So that way when it— when you call it, right, it's able to instantly identify it as a test and again, not calculated against you. So you get 15 calls per month for free. Again, if it's useful, I just use conversions. If I can get, if 15 calls come in and I convert one of them to a product and it pays $29 a month or more, and for those of you who don't know, build a product worth $29 that you can sell. Um, but that way, and then it goes up to 300, 1,000 and higher. Um, you can always reach out custom, which most of you, when you sign up, you're gonna get a custom seller who's like, what can we do for you? How can we help you? Uh, but I always start with free, get it to validate its value before I end up paying for anything. Cool. Oh my gosh. Love it, Renee. Love it. So Renee went back to the legal assistant to be like, hey, help me build an NDA. That's— I think I did NDA and, um, mutual NDAs, and then I also had it review a contract for me and I was like pretty impressed with its outcome. All right, so the executive assistant has come back. It's all set. Um, social media manager. So right now it's like, all right, I pulled together some options for your first post. Uh, and these are, and I would say, yeah, I like this. Okay. Now stop buying AI tools, hire AI employees. That's right in alignment. And then I can ask it now, it'll say quick question. So I nail the angle. Which platform would you like to go after first? So it's not just gonna build generic content. If I ask it for LinkedIn, it'll be a little bit more verbose. X will be super short, right? So I can ask it this. Um, and do you want to point people to the accelerator or a specific agent? What? Multi-agent, inter-agent communication. Again, it's not that it's the end-all be-all, but it is accelerating what you would likely have to configure manually on your own. So I'm all about acceleration. Uh, and then blogs, same thing. Uh, here are for a few blog titles. Which one do you want me to start with first? I'm not even gonna read 'em. I'm just gonna say 3 so we can see what that looks like. Uh, but now we've got a sequence. So thank you, Stan, for this sequence. First email's gonna say, if you are wearing 12 hats, which many of us do, uh, we help owners build a workforce, uh, and basically it's including inbox content, SEO, sales calls in 5 days using plug and play standard operating procedures and templates. Most customers get 40 hours back per month. Which thing would you like to remove first? Now, I would edit this probably a little bit, but it's not terrible for the first one. And honestly, sometimes I just run this against leads I didn't create just to see if it works. If it works, I'll use it. Um, but usually I, I'm a copywriter at heart, so usually I'll go in and mess around with that copy myself. So what ultimately it will, uh, start searching for leads. You remember my first lead sourcing won't happen till tomorrow morning. So for those of you who've gone through this process, you'll see that. Then you can also import leads. So I have, um, I always have lead databases that I configure. I think I created one earlier for this purpose. Uh, uh, uh, this looks like a CV. Here's one, right? So these are a bunch of leads. So I can pull all these in if I want to, but I'm not going to because those aren't the leads for this business. But if I wanted to go buy a lead, or some of you know Apollo, Um, Apollo, you can go and curate your ICP audience and pull them in. You can even identify buying intent and, uh, identify those who are, have consented to have you reach out to them. Like they're already looking for you specifically in the area of AI and AI automation. So you can import those leads and have this same sequence serve them. This is not to replace a CRM or go high level, right? I have those systems in place with very advanced detailed LLM-driven customizations. This is to get you up and running and getting value to your business immediately, and more importantly, give you something you can demo to a client about the kinds of things that you can do across different platforms. So this is now running. We'll revisit it super quick tomorrow morning before we get into governance. But I do wanna share with you some of the in-house, like similar to Marblelism, you get 6 agents. If you're in Copilot, there are a set of prebuilt agents that you can use. I am currently in the process of delivering what I call the CEO Accelerator. So it's basically if you're the CEO or leader of a big organization, what's the minimum Copilot implementation you need to just be like, oh, my life is better now. Agents are helping me. And so this is meant for that purpose. Like, go in and use these agents. These are already predesigned. You can see There are existing managed agents, which means that they're like the ones we just saw. You need to provide a little bit of data, but it's already prebuilt to satisfy a specific result. You can also start with these templates. I mentioned these briefly yesterday, but website Q&A, right? A voice, voice. We just built a voice agent, right? A travel companion, financial insights. I am not using, like, I built my own financial model, the hard way. But using financial insights, especially if you have that data inside your Microsoft tenant that is available to you, all of a sudden this becomes pretty interesting, right? HR benefits, having a benefits buddy. I've built multiple of these for clients. Citizen services, local, state and local government. I have 20 events this year that are teaching how to use AI in state and local government. I do a keynote. At the end of my keynote, I'm like, if you would like, I can build you an agent in an hour. I do what I did here, right? I do a little pitch, I show them the opportunity, and then I build this right in front of them. And again, these organizations are not like deep data scientists. They haven't built their own. They're not N8N people. They're drowning. They're federal age, you know, federal or local or state employees. So they're just like, please help me. Uh, and so this becomes a really cool platform, but same idea. You go into these, different agents, and I'll just show you how similar the process is, right? So you're gonna write the purpose, you're gonna give it all the instructions we designed, right? You're gonna write suggested prompts. These are the, what do we call it? Conversation starters, right? They're the little links that show up that give them a chance to not have to think through what do I ask this thing, right? And then the knowledge base. In this case, it's demoing Redmond, which is the county, I think. No, city, uh, township in, uh, that Microsoft is, uh, based in, Microsoft headquarters. And so I can now create knowledge bases for these services. I mean, imagine if you will, like, I actually created— I think I bought the domain, which is my, my own nemesis is buying domains, but I bought the domain like Small Town AI, Small Town AI, and I was gonna go and just pitch the ability for me to build an agentic system that would help them serve their residents. And it's— I was going to use this, I don't have time for that, so I didn't do that, but you could, someone, one of you could. So thank you, Sherilyn. Uh, City of Redmond, thank you, Christy. Thank you, thank you. Um, I was like, is it a city? Uh, but yeah, City of Redmond, they should definitely have that in their URL. Uh, but huge opportunity, right? Any one of these is a single opportunity. And as I mentioned before, these agents are the tip of spear. Once you create trust building a, like an agent like this that delivers value that you've thought through the instruction set and the, uh, starter conversations and the problem you're solving. When you've provided that level, like the, the key here is this is not super valuable unless you've defined the system prompt to control its behavior. That is what people are asking for. Oh, let me give you, uh, this. Let's see here. It's actually just copilotstudio.microsoft.com. So I'll put that in there. I think you probably tried to grab it and you got all of— let's see, let's make sure that is working. Yeah. Now it might ask you to log in. I think there's a free trial. There might be a— I'm not asking you to build it. I'm not asking you to spend any more money. I'm just saying, I'm just encouraging you to realize that this is true for, Microsoft. Google has Vertex Agent Studio, so they have the same exact idea, pre-built agents for Google businesses, right? We have Salesforce that has, um, Agentforce, a bunch of pre-built agents on Salesforce that map to their marketing cloud. So all of these ecosystems sound disparate and different, but they're not. The hardest part is the part we spent the first 3 days going over, which is How do I identify the problem? How do I work through it? I mean, that is your differentiator is do you know how to narrowly focus this agent to deliver consistent, accurate, safe, secured responses? And then you can get paid if you choose to do this work. I call it AI MSP, right? Become an AI managed service provider. You can become the managed service provider that monitors these things over time. Time. You become the red teaming agent, right? That monitors for bias, inaccuracies, uh, model drift. You can get hired to do that for $1,000 a month, $2,500 a month. Like it just depends on how many agents they have. This is just the tip of the spear. One, to do it for yourself, please do that. Use it to your own benefit, but then use your experience. For those of you, again, there's a percentage of you that are not going to go into business doing this work, but this is meant to help you see either how you can use them for your own business or potentially build a business to help these agents, you know, help other companies. So it's totally up to you. Um, yeah, absolutely. So, uh, Claude as well. So Claude, um, and OpenAI have Agentic. I mean, but they don't like to make a distinction because they want— the go-to-market strategy at OpenAI is everything's an agent. Right? Like everything you build, and I'll just go and show you, let me get out of there. Show you a little bit about what that looks like. Let's do new chat. Right? So if you go in here, you have the ability, right, to, oopsie, leverage a whole collection of tools. You also have the ability to create, what we, what we used to call operator, but you can actually build autonomous activities, threads, if you will, projects that do work while you're sleeping. Open Claude, I think we talked about it yesterday. Um, let me see if I, I don't remember if I have that up anymore. Um, there it is. Right. So this is a Claude plugin for an Agentic system. This is custom built. Uh, for those of you who weren't with us yesterday, I'll just share the link with you so you can go take a look at what an example of this looks like. Here's the cool thing about this, for example, is a lot of people aren't gonna wanna build this. You get good at building. I, I don't know if this is the most, the best one, but I teach one of my most successful classes is called AI Solution Architecture. It gives you a certification as an AI solution architect. It's my special skill, selling AI, the bridge between sales and execution. Like, that's my, my sweet spot. And I know that AI solution architects need to sell. Stay on their game. Like they need to keep track of all the new things their clients are gonna ask them. Well, I just saw this thing on Instagram. Is this, is this a thing? Should we use it? So the better you can stay up to date, that's why I use, uh, a tool, um, called, which many, some of you who are in the program know, hucks.com. This is my distiller for me, right? Maybe I'll throw it in here. Um, Right? So this, every morning while I'm getting ready, throwing my hair in a bun, um, I turn on Hux and it gives me a rundown, but it's connected to my email. It's an agent that is connected to my email, connected to my calendar, and tuned to the topics in the world that I care about. Some of you are not AI people, right? Some of you might be a restaurant owner. If you're a restaurant owner, now you wanna know like, is there anything in this business, any, any new trending topics happening on TikTok or Twitter that you wanna use for yourself. Huxley's, I think it's free. Someone help me with that. I think it's free. Um, I don't remember. Yeah, thank you. It's free. It's like, I don't remember paying for it, but it— yeah, thank you for, for that. Um, yeah, and it's— some of you, some of the people that have used it would tell you, like, it changed how I spend my day. Like, I, I have an agent, of course, that gives me— oh, that's the other thing, is that in Marbleism, what I ended up doing You'll notice down here, it might have escaped you, but there's this thing called tasks. These are autonomous tasks you can give any of your agents. I have— oh, I'm not in the— I'm gonna go to the one where I actually have some so you can see what it looks like. But in my task, these are things, these assignments I've given to my system. So for example, Sunny, my social media, right? I've given it a requirement for daily engagement. So I never want a day to go by that it hasn't created daily content based on exist— like whatever's cool. I think I said trending topics and contrarian perspectives, cuz those are really highly engaging. So it's building me a list. That's why it has, that's why I have 600 drafts, cuz I haven't been able to use all the things that I'm asking it to build. My morning brief, this is what I'd encourage you to do, right, is have a morning brief. I want a morning brief, uh, of emails that need my attention. Um, my daily schedule, a snapshot, a social post. I didn't even give it this much data. I just said, I need you to help me get prepared for my day. That's what I said. And this is a version of what it built. I messed around with it a little bit. Um, but now every day it's gonna do this. Every day it's gonna send me a very— it's not as good as Hux because it doesn't read it to me. Um, but Hux is like a— and oh, just to give you context, hux.com is based on NotebookLM. It's built by the developers of NotebookLM. So it's meant to be an example, a reference architecture for what NotebookLM can be. Um, so, uh, oh yeah, sure. Uh, let's see. Very good. I will actually copy and paste these if you are interested. So that's Sunny. Let me put Sunny on. Oh gosh. Oh goodness. I'm going to throw Sunny in here. And these aren't like, again, don't use— oh, Typed in my own name. S-O-N-N-Y. Um, I would mess around with these. Like, this is the hard part. How do you craft these prompts to deliver the purpose that you want, the metric? How are you gonna measure its success? That's why one of the reasons I also am enamored with this platform is that it provides me metrics. It tells me exactly what it's doing. So many people hire resources into their company and they never put those resources on metrics. They never say, How many social requests, you know, how many social media posts did you build this week or today? How many outreaches did you do? How many phone calls did you make? How many emails did you process? How many emails did you draft? Like, those are not things that we, some of us, some of you are the, that kind of leader that's very meticulous, operations oriented. But as a solopreneur for much of my career, I have like not gotten that level of, you know, I kind of, in my mind I'm like, I know what good looks like. So, Um, awesome. Oh yay! What? Hux was like, you got the AI challenge coming up. That's awesome, I love it. Um, but yeah, so I like having a tool. So no matter what tool you use, make sure— that's why we did it on Tuesday— make sure you have a measurement stick, right? How do you know if what your agent is supposed to do is actually happening? How do you validate and show proof to either you or to your clients that this is actually working. That's why I like this, um, option. Um, yeah, great question, Angela. If you are in a regulated industry or government, I would encourage you to use a platform like Microsoft and Microsoft. So micro— or AWS or AWS. AWS has a different platform, which I didn't really talk too much about. Um, called Bedrock. Bedrock is its automated agentic, uh, they have an agentic platform as well, but Bedrock will lead you in that direction of building an AI system that is kind of low-code, no-code. But again, it's going to require you be very specific in what it does. So these are all kind of, if you've looked at the direction we went in since Monday, the goal was to identify a problem we wanted to solve, figure out how we'd measure that it had been solved correctly, and then provide guidance to the model on what good looks like, what bad looks like, and how to manage and mitigate risk. That was our goal of writing that, those prompts. So now we've built this, but none of that information is in here. So for homework, if you choose to accept it, I will encourage you to go in and just start, like, you'll notice, oh, this, I'm back in my AI leadership, so I have a lot of guidance that I've provided. But if I go back to the one that I just built, right? If I go into my assistant, Like there's very little in here, right? This content just came based on the configuration that I built. I will want to go in and be very specific about if, if you did this work on an email, like an assistant that's going to render new emails for you, I think you should be very specific. This is where you are going to mitigate risks. And when in doubt, tell it not to do something. Tell it no, do not manage this. I will manage this myself. Now I have— I start with no and then increase my yeses over time as I build trust with how the AI system is reacting to my, my guardrails. Same thing is here though, right? So what are your content themes? It's extracting this from my website initially, but I can go in and one, delete this or add to it. And again, this is work we didn't do. So this is homework. Go in and apply the restrictions, the risk mitigations, the guardrails, the guidelines that you developed over the last two days and paste them into the appropriate agent. And then we'll just take a look. Oh, that's not terrible. So it created this cute little graphic. That's not terrible. I don't hate it, but I'd probably do something that looks a little bit more like me. Go ahead, Krishna. Hello. I— there's a lot of information I'm trying to gather here. So is the fundamental difference between marbleism and lovable that Lovable builds the app and Marbleism offers these agents that I will configure? Yeah, the best way to think about it is that this is a layer of abstraction, right? So someone had to build Eva, had to create, like, that there is engineering work that went into building each one of these agents as well as the platform that it's on. This is a layer of abstraction so that you are not concerned, like, you're using a SaaS product. Right? So Lovable is you're building a SaaS product or you're building software, building an app, building a, a mobile experience. So you are not using something that someone's built for you. And that's the, the trade-off, right? Is I, I kind of think of it as like, this is, you know, done for me. Everything's kind of done for me and I just have to pop in, you know, words. Lovable is more done with you kind of experience where you're gonna tell it in words and it's gonna generate code and you're gonna test it and have a conversation to get the right outcome. And then, of course, do it yourself is you just write it the old-fashioned way in VS Code, right? So those are kind of the 3 levels of abstraction that we're working with. I wouldn't encourage you building an agent from scratch, though it's good experience, as some people on this call have gone through. It's good experience to build it from scratch. So maybe do it once, but— So if one of the 2 ideas that I have is a, say, personal assistant, chief of staff, whatever you want to call it, then I think it sounds like you use Ava and configure it. Sounds better. Yes. Than building it. And the other one that I was thinking of was, well, we're gonna go through this, uh, crazy 1-year AI literacy ride. So the thing that I want to do for myself is a plan architect. You know, I want somebody to be able to look at all of my resources, the links, the teachings that you have, the transcripts, and tell me, okay, here's what I want you to do this week. Like, do learning by thing. That's one of the ideas that I have, and I don't think I can do that on Marbleism. So that would be a lovable thing then. Yeah. Or, um, let me, uh, I posted about this in the community. Um, you also can create a project, uh, I'll just show you Gemini. This is also available in ChatGPT. I just like to well-round us. Um, but there's this concept of guided learning. So you feed it the sources and it creates it's this guided learning path. So guided learning, again, it's also available on ChatGPT. Um, oopsie, I keep clicking it before I'm ready, right? So study and learn, same idea. So you could actually create a project in, in Gemini or in ChatGPT, or maybe in both, and kind of, you know, you're, you're in the business of building, um, and use the study and learn process. And what that means is that it's never gonna give you the answer. It's gonna, guide you to an answer rather than supply you an answer. But I'd also consider, which we've talked about many times, right? NotebookLM. Because when you're in NotebookLM, you now have the ability for a project, right? You create a new notebook that has all this content. Let me go into— and once you— I'll go into Stu. Oh, let me go back to a previous one that I would have built. Let's go into— yes. Yes, perfect. I think I built this when you asked the question last time, right? So I've now got all of these resources, but I can go in and build a mind map or an overview or flashcards or a data table, or even ask questions and ask for a guided plan. So I think there's several different ways to do that. So, so I'm playing with NotebookLM, and I was wondering if, if this can do the synthesis for me and an action plan can come out of that employee. And again, just for the sake of knowing how I can build one. Because there's a lot of these things that are already done. It's a question of knowing which, which one works better. I don't want to reinvent the wheel, but I also want to know what it is that if I want to build something at work, I need to know what are the steps, the artifacts that I'm producing so I can go ahead and say, okay, there's a better way to do this. Let me go try that. Yeah, absolutely. So the, the good news is there's not just one way. But the way that I distinguish them is through how, like, the levels of abstraction. How close am I to the engineering work that's necessary to do the thing? So Lovable is gonna allow you to use words to create software, but you get the code. So you can actually go into the code and see what you've built. So if you're an engineer, I find that that's a big trust-building activity with me. But if I don't wanna get caught in the code where I'm, may spend too much time. I just want the outcome. Then I look for higher levels of abstraction. So that's when I go to maybe using, you know, a tool like Gemini or Copilot. But if I want high levels of customization, now I'm looking for what can I build that's custom to me. And that's like an example here is like I build a, a project. Now I can actually tailor it. I can tell it, like when I just asked an action plan for, uh, for NotebookLM, wherever that was, right? So it's gonna create an action plan here, but I might want to build that more as like an LLM, right? Where I, the LLM every day says, hey, welcome to day 3 or day 45, we're gonna be doing this today, right? So I want a little bit more control over how that experience is, and that would be more of a, a more technically in-depth, like a lovable type problem. And I say lovable, but please know everyone, it's, any vibe coding platform. So that could be Lovable, it could be Cursor, it could be, um, Voiceflow. I mean, there's lots and lots of them to choose from. And just to be clear, Marblism can't do that. Not yet. We only have those 6 that we have to work with. Okay. You know, for the longest time I thought it was you who created those, and I just realized that it's not you, and that's, that's what Marblism offers. Sorry. Yeah, yeah, exactly. Yes. I try to say it. Um, but they're— I guess they do a good job because they don't really say marbalism here. Um, but yeah, I try to say it every so often, but it also is one of those words that escapes people for some reason. I don't know if that was for me. Okay. Oh, thank you, thank you, Christian. Yeah, no worries, no worries. Gabriella, hi. Hi, thanks very much. You're welcome. Um, could I ask if there's like a, a way to, um, send like an idea of, uh, what, what direction, if it's going well or not? Or could I email it? Or— Absolutely. Okay. Um, absolutely. And totally. Um, I'm just— if anybody else later, I, I promise that if it, if Noelle approves. I will try to see if I could share like, uh, like a recipe, little, little part for design. Okay, because I know you want to like— I love it if this is okay. Yeah, yeah, definitely feel free. You don't need my permission. Like, as long as we're not selling each other forex trading tools, like, please share your prompts with each other. No, no, this, this is just because— no, I'm just kidding about the platform. So yeah, just— well, if it's related, like, this is what this community is all about. So please feel free to share prompts. I try to share freely Everything I have is yours, so please do the same. That's amazing of you to even be thoughtful enough to do that. So thank you. Thank you so much. Awesome. Okay, Stephanie, you're up. Hey, happy Wednesday. Okay, so in building an affiliate social machine, because I think that's what it basically is. Yep. I'm going to use product website. It isn't about me at all, any of the stuff. Right, right. So I just go to product website and that's The content. Okay. And then I'm gonna use something like Marbleism. You can, or is that con— too comp— not really necessary. Because I just wanna gener— I wanna do the research and generate the content and then post and analyze, right? Yeah. That's all I wanna do. I would use just, I would just start, I'd spend an hour testing your hypothesis in Marbleism, and if you can get it done, like it's easy, I would use it. That's how I started. I started just using social and then I was like, Wait, and then I found the phone number. Anyway, it became a slippery slope, which is a good sign, a good design in a product. But the other thing you can do, uh, because you're doing, I mean, you have access to GoHighLevel too, right? Yeah. So GoHighLevel has, you know, agents for building content as well. It's just not the thing I like about this platform is that they, they thought through the complication of building a bunch of siloed systems. That you then have to figure out how to connect and you can figure it out. That's what N8N and Make, that's the business of those platforms is to connect these systems together. But it can take a long time. Uh, you have to manage inputs and output. There's just a lot of work. So I like that I can have an idea with my executive assistant and then I can tell the assistant to tell the social media and blog poster what to do and that I can then even ask the executive assistant to update my sales messaging to the new thing that I'm talking about. So, So I like the interconnectedness of this that did not require me to draw one line from one box to another. Um, and I, I grew up in Make, like I've been building workflows for very long, over like, I don't know, maybe two decades, which is very sad to say, but a very long time. So that's why probably I was so interested in this is I was like, wait, I don't have to do that. Let me just do it. Especially if your problem is just that, if you have just one thing you're trying to do, I would try it here first. Um, and it's cheap, you know, like unless you have a cheaper way to do it. Um, but remember yet on Wednesday we went through this measurement of like, how do you pick a tool? So for me, like ease of use, uh, quickness to outcome, cost structure, team, uh, of, of like team coordination, like all of those are the reasons I pick up one platform over another. So. I also encourage you to just do a little bit of work, maybe on a platform you're thinking about, and just go through that rubric. What's important to you? Because you might, you might not choose what everyone else chooses because you have a very specific problem you're trying to solve. That's why that— and it can be, it can be a good tool even if I don't use most of it. Yeah, because as long as I could do the most simplistic thing, which is what I'm trying to accomplish, it's okay. Yeah, I mean, it's okay because if you look here If I look at this, right, I'm only using 4 of 6 agents right now, right? So, gotcha. But it's still saving me 340 hours of work, so that's great. Pretty good. Cool, thank you. You're welcome. Christine— Christiana, I almost did. Hello, beautiful. Oh my gosh, I love your enthusiasm. It's like I was having panic attacks after a while listening, going, I am such a newbie. Oh my God. I'd have to like go outside or take a nap or do something to calm down and be like, okay, just do one thing, just one thing. Yes, just one thing. And then I love what you say in the replays that like, hey, we all start in the same place and it's just like working through it. So thank you for like helping me manage my anxiety over all of this and not knowing. So two things. One, when I went through, I kind of did it before you did it, and so it didn't give me the legal Associate. Okay. Oh, maybe it did. Maybe it finally did. Okay. But, um, the other one too, Eva has a note taker built in. How do you use that? I don't use it only because I don't like, um, I don't like having like a note taker. And I say that and there's one in here because if I check that box and then I forget, I show up in a meeting and they're like, oh, I see Noelle, your Eva is here. And I'm like, it's just awkward. So I prefer not to use in, in like bots that are inside my meetings. I prefer ambient listening like that. It's happening outside. I still disclose, right? And I'll tell y'all, if you're on a meeting with me, I'm recording you. I'm just letting you know right now. Always. I always record everything. But yeah, I don't really like to have it. What are you doing? But I think it would be really cool to try it out. Awesome. Awesome. Thank you very much. Oh, is that someone else talking? Um, yes, I'm trying to find them, but I don't see them on my list, so keep going. Okay. And then, um, so going back through, so, uh, what I did with, uh, the executive assistants, I want to make sure that that one's working before it starts writing emails. I want to make sure I have all the guardrails in there first. Yep. And then go through and test it and Um, how do you go when you're, when you're first starting and you're starting to test it, is it obvious testing it or how do you go about testing it? Yeah. So immediately it's gonna start doing this. So these are all of my, like, you know, labels that I've created. And then when you go into drafts, you'll see these are all, these drafts are AI-generated drafts. So it'll immediately, if you ask it to, it's gonna immediately work on the first 50. Drafts, or if there is one, like you might— your first 50 emails might only require 2 responses, so it'll do 2. But it's going to first grab the first 50 just to give you a sense of how it's working, and then you can optimize it. Now, I got very specific, so I'd be like, in emails like this, and I would reference it. It has context to the entire inbox, but I would say in emails like this with like this person, I would name the person or the company. I would really like you to do this. So it would, through the conversation, it would optimize its own guardrails and its own guidelines based on what I'm asking it to do on a per email basis. But you can also just go into, uh, the guard— the guidelines themselves and manually write it in. Perfect. And then one last quick thing. Um, so in my email inbox, it's like, you know, a bazillion thousands and thousands of messages. I do have old people. PII information in there and I wanna like extract it out so I can delete it so it's not just like living in the cloud somewhere where somebody could hack it, you know? Yep. So how would I write a guideline for that? Yeah, I would ask it to filter out any information it, I call, I would call it sensitive information. Um, the challenge with that of course is it's probably going to grab more information rather than less, but in that scenario that would be okay. Uh, but it'll ask it to do it for. Like maybe a certain number of emails, right? So find the first 50 emails with sensitive information and categorize it this way. That way you can scope like, is it working? Are these the type of emails? And then you can confirm with the agent, yeah, you're doing a good job. Keep going. Oh, perfect. Thank you very much. Yeah, you're welcome. And I've done that for lots of different reasons. Sensitive data, but also contractual data. Financial information where I'm like, I do not want to just have that out in the open. So yeah, great, good questions. Gwen? Hi, Noelle. Don't want to appear weird, but I know you understand this. I'm hiding in the bathroom from the kids. No worries. Yeah, I do. I do. But I wanted to get a little bit more of an understanding on the— you know how it shows how many hours that you're saving? How is it determining that? Is it based on the KPIs that you set up or? Yeah, so these are, you can adjust these KPIs con— like in the guidelines I took, like these are all part of the, the platform. So it, its job is to process emails and draft email responses. And so this is, these metrics are defined by the platform itself. However, you can also, which is what I do, I also ask it in its daily result, right, to give me metrics I care about. About as long as it has the context to see that information, it can generate that for me. So this is pre-built, so you're going to get this whether you want it or not. But you can actually ask each agent to give you its own metric, which metrics you define, its own metric-driven results on a daily or weekly basis. Basis. And you would do that just right in the, like, right in the chat right here. You'd be like, I would like you every day. If you say every day, it's going to turn— or every week, it'll turn into a task that you can manage. Okay, that's great. Thank you. Yep, you're welcome. All right, Bill, hello. Hi, uh, I had a quick question about security. When I connected it to my G— I was going to connect it to my personal Gmail account, which really needs to be cleaned up. That's like 20 years old and has a lot of personal stuff. So I had that moment of hesitation, like, what data is it going to be accessing? I would imagine a customer would have the same hesitation. So I was curious, how would— how would you recommend, like, reassuring them? Yes, this is a great question. So oftentimes I spend a little bit more time. So let's say, let me go into the executive assistant and change this to connect an account. So this is the reason that it's limited to 3 types of connections. Google, when you say like, I'm gonna pick a different one. So this is a, an extremely old, um, account, right? I walk them through this process. Now I would, if you're going to offer this, I have extended the privacy policy. Like I, I'm that person who reads privacy policies and terms of service. I throw them into ChatGPT. Is there anything in here I should be worried about? And you'd be surprised what nonsense is in some of these, but I actually read them. The other is it, it is typical, um, security policies provided by Gmail, right? So you are opening up a gate to an existing platform that you don't own, nor do you secure. So it's really just giving your client a level of awareness that this is the connection they would make, like if they were using Outlook. Right, as a platform, it's the same exact connection. But if I say continue, it will actually— oh, in this case it didn't do it for me. Let me try this one. I'll see if I can get it to. So then when you've done it, because I've already done it. So when you do it for the first time, let's see if I can— I don't even know. We'll see if I can try it. Oh goodness. Let's see. All right. Almost there. But it will pop up. These are the services and their access levels for each service. So one thing I would do, if especially if you're gonna share this, this is, this is a good example, right? So it's gonna do this, it's gonna do that. So I like to walk them through and maybe taking like a snapshot, uh, of this, right? A, a screenshot so that you don't have to go through this process. Um, and I'll do that just so I can send it to you all. But this is specifically for my Microsoft, but this is the same screen they're gonna get no matter what they connect it to. But oftentimes they don't know what they're giving away when they make this type of connection. So I like to explain it to them and let them know that these permissions are revocable at any time. Um, they are also limited by the guardrails that we put in place. So there's, uh, a bit of a story arc that I share, um, before we hit accept where most of us, cuz we're, some of us are tech people, we're like, next, next, next, finish. You know, like, cuz I connect Gmail to all sorts of things, but if you are in a regulated, you know, space, Now, the other side of this in Microsoft, for example, is that you actually can control what a Microsoft tenant— so for me, anyone, Skye, Padmini, all have AI Leadership Institute emails. But when they connect, that, that email they're connecting to this system doesn't have access to everything in my environment. But that's a role-based access control policy I defined as the Microsoft tenant owner, right? So you can kind of see there's like a, It's almost like a cascading level of security responsibility that doesn't— like, marbleism is just the client. It's how are you protecting your inbox from sharing information it shouldn't, um, or giving away access to things that it shouldn't. So that's, that's how I explain it to them, is that if you're worried, why don't we take a look at how Microsoft and Gmail are set up from a security perspective so that I can help you configure that. Does that make sense? Yeah. Thank you. Yeah. Yeah, you're welcome. But having those screenshots is really helpful because they usually, they'll be like, ooh, I've seen that before. And they'll be like, I've never even read this page. So it creates a lot of trust, just giving them visibility and awareness of the permissions that they've already likely granted their systems. And then if they're worried about that now, you can also offer them the advisory services of, hey, let's take a look at the security profile of your, your existing email system. So great question, Sophie. And then I'll come back to you, Gwen, if you still have your hand up, or I don't know if maybe you just didn't put it down. But go ahead, Sophie. Thanks so much. So I have a couple of questions. First of all, is it— can it access a Mac calendar, or is it just only for Outlook? And so is it working for Mac? Yeah, I don't think it works for Mac as a calendar, but that's just a sheer market size challenge. So that's something we could add into our feature requests. Let's see if it's in there. Okay. I'm just going to look for it and see what happens. Back calendar. Calendar. Yeah, actually it's there. Yes, quite a few. Yeah. Great. Other languages. You're clicking on that every day. Okay. I saw in the already discussion that other Languages are supported. I assume also other countries are supported. Um, and then, uh, for, uh, so for the— can I change a draft email? So if there is a draft email, can I ask the agent to change it or shall I change it by myself if there is a draft email? Yeah. Once the draft email is created, it actually is stored in your Gmail. So you can ask it to update it, but I would— what I do is every day I have an hour. Set in the morning where I read. So every morning, and this is a mobile app as well, so I'll show you. So every morning I have, you know, it's— I've got this little icon right, right here. So every morning I click on the icon and it goes right to my executive assistant and it gives me my morning briefing, right? You've got 4 emails you're supposed to respond to yesterday. You better get on it. And so from here, I will look at what I need to do and it will tell me, like, I built— I created these 3 drafts. Now, granted, I wrote that guidance, right? I gave it the task of building me this, you know, awareness. If I wanted to, I could say, oh, well, that draft with, you know, so-and-so, actually I already delivered that. Um, can you update it to say, et cetera? Uh, and it would update it. But what I typically do is I go through it on my phone and then when I get to my desk, I just go into my inbox and I look at my drafts, right? Yeah. And so that, that's the, I'll go into this one. So you can kind of see what it looks like. So that was the draft that it wrote. And I could, you know, like at this point I'm like, well, it's easier for me to just edit it, but I absolutely could do it conversationally as well. Okay. Thank you. And can we ask to attach specific files or links that are in, so does it have access to our file system? So to select, select the files and links? Good question. So it doesn't have access to your file system. However, It does have the ability for you to— I just did this for our purposes. I loaded a bunch of images that were taken of me. And you could add any, like, any— this is for Penny and Sunny to leverage. So if I wanted something to be in, like, my social media or my blog post, additional, like, I usually have a call to action or a, you know, PDF. I could add that here if I wanted to change the behavior of my— oopsie. Oh my goodness. Um, of my executive assistant. I also have the ability to give it context through either my emails that have attachments, um, or I think it's down here. Yeah. So you could see like this. Oh yeah, this is the checkbox for taking notes that I don't have selected. Um, and then I also can add my signature, uh, as well. Uh, let's see, what else was I going to mention? I lost my train of thought. Oh, maybe it was in settings. Was that it? Yeah. So I think that was it. Yeah. So this is the most control that I have and the media files is the only way that I can add additional resources that my agents can use. Thank you. And I think you mentioned yesterday, but just to be sure, we can connect it to Airtable so that it can export afterwards, that we can export afterwards conversations or documents or whatever it can go there. It— yeah, if you want to, you can create— it's a webhook actually, so there is an ability to just take the info. It's kind of like I use it to send things to Slack. I use Zapier, but you could— yeah, I use Zapier to send things. A backup mechanism. Great. Thank you. Thank you so much. Yeah, you're welcome. All right, last but not least, and then VIP is like, my gosh, what did I just— anyway, it's fine. You see, this is my problem. I want to help everybody. But VIPs, if you're up for it, we're going to start it too. Olga, how are you? Hi, Noelle. I think showing how much time each agent saves is very valuable, and especially for corporate clients. And I was wondering if something like that exists in Copilot, or can we easily build it because I'm working for an AI incubator of a large company and we're building loads of agents for different teams. And what would be really valuable looking at this is to show this agent will save so many hours for your team of 300 people. And that's how much value our incubator is bringing to the organization. Yes, there's a couple of ways to solve that. One, Uh, you would have to build an agent that collects it. Like you'd have to build a system that's going to collect, you know, for example, understanding this is what I do for existing clients is that we establish a baseline of how long it takes you to do the things you do today. How many emails are you sending? How many contracts are you processing? How many invoices do you get? And then that becomes our baseline. And then we create an AI system that looks at the delta between what we were doing at the baseline and what it it's doing now, but you would have to like build it, right? So they built this like little, you know, agentic system that like just collects this information, but it's all just baseline to active, like current state, um, aggregation. So you could do that in an agent, but I'd also consider like maybe building a lovable dashboard because it's not, I, I like that it's very specific to the things I'm actually doing. So what that would mean is that your agent would have to report out somewhere to a database, to a table, to a Google Sheet, to a Excel spreadsheet. It would have to report out, I did this today, so that you could then aggregate that into a visualization. It's a really great idea actually. I think from a, you know, if I was gonna invent something like that would be a really cool thing to, to create and sell to companies. Um, but yeah, you would have to build it. There is no, it also requires kind of cross-agent, right, awareness and knowledge. And most importantly, those agents all have to report out the data you're collecting. So these agents in Marbellism were built to do that. We would have to manually have our Copilot agents do that, which I don't think is hard, but you'd have to go in and actually tell it your job is to report how many— like, let's say I build an agent for, you know, email triaging. Your job is to report how many emails you triaged and how many urgent emails you saved, right? You would have to give it those instructions to capture that data and then create a service, an agent or some kind of service that would then visualize that in some kind of pretty graph. Yeah, and then you compare how much the agents spent on it, how much time and how much humans used to spend on it. Yeah. And that's like a, it's an arbitrary mathematical equation, right? So you'll just make that decision. Just like, for example, I looked at, um, like Marbleism's numbers, right? Mm-hmm. I actually think this is probably true. I easily can spend 30 minutes creating a post, especially if I'm doing what it's doing, which is researching it and making sure it's got good copy and all of that. Mm-hmm. Um, a legal document, like building an NDA, I don't know that it takes 60 minutes, but I do know my paralegal would bill me 60 60 minutes. So, so that makes sense to me. So like understanding the reality around the economics, right, behind those metrics, I think that would be just, you'd have to decide on what that is. Yeah, at the moment we make assumptions when we're putting a business case together. Exactly, exactly. We estimate that will save this amount of time, but what you are suggesting is to actually collect real data And then, yeah, show the real outcome. I just think, and when you do it like this, right, where you can show them, here's how we're using it and here is our real data, right? It's just, it's so much more realistic than what we used to do is when we build these business case studies, like there's no fundamental truth to that, right? We're just kind of guessing. I mean, granted, it's educational guess, but it's not grounded in anything where that's why I like to build an agent that I can use ahead of time. This is what we do in AI solution architecture. How do we build something that can demonstrate that? Like, we wanna show versus tell. And these business cases typically are telling, but they don't really demonstrate a real live version of what we're talking about. And so I think now with the, the amount of time it takes, and I don't mean this, but like building a lovable that has its own internal metric system, and then you just use it for a week before you show it to your client. Is a really powerful way to demonstrate like actual value rather than just talk about it. And can you use Lovable on top of Copilot of Microsoft tools? Yeah, I mean, but if you were, I would, I would find out what you wanted to build and decide whether or not you wanted maybe to build it as a custom Copilot agent, which gives you some additional flexibility, or if you actually do need to build it as a completely separate piece of code. Oh, okay. But there are multiple levels, just like we were talking about, of abstraction, right? There's Copilot, which is the highest level of abstraction. No code, no flexibility. You get what you get. And then there's Copilot Studio with prebuilt agents, and then there's Copilot Studio with custom-built agents. And so each one gives you higher levels of control. Yeah. Yeah. Thank you. You're welcome. And so for those of you joining us this weekend, AI Leadership Institute, we are going to go through that. I'm actually going to just teach you how to build this CEO accelerator, right? How to build a set of agents inside Copilot. It's called the Copilot Power User Masterclass or something like that. So, so you'll have some fun with us for those of you joining us this weekend. It is Saturday. Yehan, Jahan, hello. Sorry. Hello. How are you? I have a very basic question. I don't know anything about the business stuff that you were talking about, all those documents that I have to create. Do you have any training inside of the school platform that I can visit and get an idea of what you are talking about? Tell me more about like the question, like what are you looking to learn? Well, I don't have any of those documents that you were mentioning, the SAS or I don't know, all the documentation pretty much. Yes. So there's a couple different things. There is a— now what is, what is it that you're— what are you, what are you doing? Are you— do you have a consultancy or like what is your focus? I want to build a website for Qigong classes that people can pay through it. Awesome. Okay, cool. So, and so you're wondering about like building the foundational knowledge that you would use to like inform the development of these things. Is that what you're thinking? Yeah, what is the information that I need to give to these agents, right? Yes. Oh, that's a good question. Mm-hmm. I think so. I feel like I do. So one, um, it's actually in the school. Let me go into school so I can show you for any of you who have not gone through this yet. Um, and I'm gonna just open this until the end of next week. So remember— oh, this isn't the right one. Sorry. Uh, uh, uh, so let's go in here. So in the classroom, I opened up the Design Your AI Solutions, which is a paid service, but I opened it up and then I was like, I'm just gonna keep it open until the end of like next Friday. So you, if you could take it, I would encourage, it's good, I like it, but you'll have it free and then I'll lock it back up. Um, but the other thing is this one, Bot Builder Certification.. So this allows you to go— it, it's the process of end to end. It's kind of nice cuz it's, um, I'll just give you a, a taste, um, of what it looks like, but it's studio recorded, right? So it's nice. People have given us rave reviews, but it allows you to see end to end what you need to do to train a model. And that is, will help you fill in the gaps between I have an idea and that you build an agent. Like how does it know? What truth is, how does it know what good looks like. It'll give you much more, um, detailed approach on how to build your knowledge base and how to educate. I actually think it's, uh, around here somewhere. Yeah, selecting— oh no, one of these is like building the, the knowledge base system. But this is, um, definitely I think what would help you because this is literally Agent Development 101 from, I, you know, 2 years ago or something like that, maybe a year ago. Uh, but it'll be, it'll help you define the knowledge base for sure. Let me go in and make that available to you really quick. But that's if you've never done that exercise, it'll be very helpful to help you figure out what that is. Edit. So there for all of y'all, enjoy while you can. So I'll, uh, same thing. I'll, uh, I'll just, when the challenge wraps up, which is of course tomorrow, I'll give you another week. Hopefully you'll have a chance. It's only 4 hours. These things are not super long. Um, the design session is like 10, 10-minute sessions, so it's like an hour and a half. Like, these are not long courses, but they're, they're meaty. Oh, you're welcome. Thank you, Lauren. Enjoy. I hope you do it and you can get certified. So if you actually build an agent and you ground it and it's responsible and you build, I'll tell you, it's hard. Uh, you have to create a video of yourself talking to a camera. I know right there I've lost half of you, but you have to create a video explaining what you built and why you built it the way you built it, and you'll see the instructions in there. But if you do that and you publish it inside our community for all to see, which will take a level of bravery, I know you all have, we will then award you with the AI Bot Builder certification so you can add it to your LinkedIn and share it with all your friends. Okay, thank you so much. You're welcome. All right, I think we've done it. I know it's terrible. I'm so sorry for going over. Sorry, VIP. Thank you for all your questions. I hope you found today helpful. The most important step you can take is going from learning to building and like idea to execution. Like just start trying these things. The cheaper environment you can try things in, the better. So you're not feeling like, you know, anxiety around investment. That's why I try to look for these platforms. You'll notice in Bot Builder, the one I just opened for you all, I use Chatbot Builder to do it because it's free to use. It's not about the tool. It's about the process, right? It's not about the specific, you know, platform like Marbalism. It's about the words you have to learn to craft in order to create a safe, secure, and responsible agent bot platform. I actually in Bot Builder have you build a customer service agent, which I think everyone should have, but it's, it's gotta be a good one, right? How many of us just are very frustrated with the, the state of customer experience? Experience today. Like, we can do a good job if we actually design it. So understanding the human AI experience design process, all of that is included. So there's a lot to doing what we're doing. But my goal was in 5 days that you'd just be able to build a bunch of agents that deliver some level of value to you. And I think we've probably already exceeded, you know, you could save 5 hours between today and tomorrow with just the agents that we've deployed. But take the exact same thing for those of you who want to. Go into Copilot, see, like, look at the default. I think I showed it to you, but like, look at these default agents that are available. I think I did it here. Let me go back. Right? So if I scroll down past all of the agents that I've ever built, right? Like, all of these are prebuilt. Just pick a couple. IT help desk. Every company needs one, right? Sustainability insights. Somebody's building a sustainability agent. Maybe this gets you a little bit of the way there. So I get it if you're not, necessarily a Copilot user, but I think the more tools you have in your toolbox, the better. And you'd be surprised how very consistent and similar they are in configuration. Once you've done the work we did on Monday and Tuesday, you can actually use that work on any of these platforms, and the configuration is relatively similar. If you get lost, if you get overwhelmed, just remember we're here for you. This entire community. We're all— I always say we're like grapes on a vine. Some of us are like ready to turn into wine, um, some of us are super green, uh, but either way we're all on the same journey. We're all here because, uh, I picked you all. You're here because I was like, you are my people, we are the same, um, you don't mind the way that I teach, uh, and hopefully you found today interesting and hands-on and got an opportunity to to see the words you chose earlier this week in action. And that was my goal. Tomorrow it gets a little serious. I'll land the plane in— right now we're, we're in baby tiger mode, experimentation mode. Tomorrow is all about how do we take the experiments we've built and actually like, what if I give this to a client? How do I ensure that I'm doing something that's going to help them and not hurt them? There are, uh, policies, standards, right? So we're gonna talk about how do you set up an organization, even your own, with an AI strategy, right? With a governance policy, an AI use policy. So many people don't have— how do you just think through that? So I'm gonna give you that so that you can do it for your own business. But more importantly, maybe you go and help other businesses do the same, cuz this is something that comes back to haunt, haunt businesses that get super excited about AI, get out way over their skis. You know, we're in Olympic time. And then they end up, you know, doing something that, that hurts their business, hurts their customers, or worse, hurts themselves. And so I don't want that. I want to make sure that we do, do right by those who are trusting us with their business and, and especially our clients. So hopefully you enjoyed today's session. Thank you all for being here. Don't forget to save the chat. I remembered. It was a long day. I will try to break up the, the video so that it will break up the Q&A from the rest. So if you don't have to look at it and go, 2 hours, I'm not watching that. Um, but hopefully you enjoyed it. Thank you so much. Um, hopefully I answered all the questions that came in. It looks like I did. VIPs, if you are up for it, I'm up for it. I'm sorry, I, I kind of did some Q&A here, but let's just go and have a conversation because I do have some tools, uh, and resources that I've used that I want to share with you that's specifically for you all. So I'm gonna hang up here. We're gonna head over to VIP. Thank you all for being here. I'll see you tomorrow. Uh, where the rubber meets the road, as they say. So, uh, go have fun, build out some agents, and let us know tomorrow how it went. Thank you everybody. Talk to you soon. Bye. I guess I could stop sharing. Bye everybody. Thank you for being here. Y'all are amazing. It's so nice to see you all. Okay, bye.