AI Leadership Q&A wth Noelle

Welcome. Happy Friday. Welcome to another wonderful day at the AI Leadership Institute. It is our community call. We do this every Friday, 4 o'clock, unless I'm traveling, which, let's face it, does happen on occasion. But this is a great chance for us to get together, and if you are here live, you have a chance to ask these questions live in the Zoom, uh, call, and you can use the Q&A section that is here. Um, oftentimes I try to answer those questions right kind of in the flight of having our conversation together today, but if there is a question, um, and I'm not able to answer it, the best place to put is in that Q&A box so I can see it as we go. So welcome, nice to have you here, thank you for being here. Happy Friday. Um, it is such an exciting day. Uh, this is the last Friday before next week which is our biggest week of the year, our I Love AI week. It is a challenge. So the bulk of what I'm gonna share with you today is what to prepare for next week. And I do actually, I'm gonna try and find the link. But I will be sending out a post to everyone in both communities, the I Love AI community and the AI LI community, for you to join a brand new initiative. And it is all about Well, I have a mission. Whenever I do, you know, as I've been getting more and more excited about the work I get to do in the area of building safe and responsible systems to help companies scale, I get to do all sorts of stuff. I go to new, new countries. I get to go to new cities. I often get to teach some of the workshops that we do in our AI Leadership Institute. And so I wanted to give you all an opportunity to become part of that. Really exciting cycle of activity that we'll be doing around the book launch. So for those of you who don't know, my book, Scaling Responsible AI, it is short final for launch. I am expecting my books any day now. You definitely all will be part of my unboxing. I hope the box comes on a Friday because I don't know if it comes on, like, you know, a Tuesday or a Saturday. Am I gonna be able to wait till the whole next week to shoot— do that open boxing with you? I don't know. We'll see what happens. I'll do my best, that's for sure. But what we— so this book is coming out. The day of launch is March 25th. As it stands right now, it is still preorder, so that date could change. But right now we are pretty confident in March 25th. And the reason that I love that date is because it allows the book to land inside of International Women's Month, which is March every single year. And so all of March I'll be doing activities to help promote the book launch. As a result, I created a new team. This is a anyone-can-play— people in our paid community as well as people in our free community. Any one of you can be part of this. You'll see in your inbox a note that comes from school. It's a post that I'll be posting shortly, and it's all about how you can join what I'm currently calling the Responsible AI, or RAI, Insiders Club or Insiders Team. And the benefit of this team is that we are going to be doing so many events. Over the next 6 to 8 weeks, focusing on the contents of the book. So if you decide to join my Insiders program, again, it doesn't matter. You do not— it's, you know, I, of course, I hope AI Leadership Institute members will join, but it is for everyone. And the idea is that you're gonna get access to content that I use to write the book, my research information. I will give you some behind-the-scenes information about how I came up with some of the ideas that I write in the book. So you're gonna get kind of super secret private access. We're also going to be doing a book launch party. It will be remote because people are all over the world, but we will be doing it live. It'll be on Zoom. You'll get a chance to be part of it. You'll also get a very cool opportunity to be entered in to win swag bags that include not just my I Love AI t-shirt, but our AI Leadership Institute swag, our notebooks, our baby tigers. So you'll get a chance to be part of all of that. But most importantly, when I come to a town and you happen to be there, we can work together to host a workshop right inside your city. And so you can be my local AI Leadership Institute kind of member and, and advisor and leader. And if you choose to, not only can we use this as a chance to do something great for the local community and educate them on scaling responsible AI, but also possibly turn it into something that you can own longer term. You can build a meetup, the meetup can meet every month, and you can actually be part of that. So this is our RAI Insiders Group. Like I said, you'll be getting an email right after this call from school announcing the program. There will be a form you have to fill out, and when you get added to that form, you'll see a new classroom added into the AI Leadership Institute membership site as well as inside the I Love AI membership site. And that will be only for people who raised their hand and said they want want to be a part of this. And there are some rules. So if you wanna be part of that, you have either already purchased, pre-purchased the book, or you will commit to purchasing the book. And then you'll also commit to providing a positive review when it comes to the day of launch. Once you finally get the book and you read it, you will take the time. I know you're gonna love it, but once you love it, you're gonna commit to actually saying you love it out loud on a review page. So that is part of it. You're also going to, like I said, there's a couple other things. You'll help me, um, not only get the word out, but also potentially partner with me and bring an event to your city as I do my whirlwind book tour. Uh, and that tour has already started. I was recently, just last week, I was in, um, Arizona, then San Francisco, then I can't remember where was I? Oh, Vegas. Um, so a lot of things are happening. Uh, February is a little bit quieter from now until the end of the month. I have a couple of events at the end of February, but then come March, I have like 3 to 6 cities a week. So stay tuned. It will be in the calendar as well. If you're not part of the I Love AI community, now is your chance. You can just go to any browser and type in Noelle, Noelle.ai, and it will take you right to our community. You have 2 options. You can join our free community, which this call is a part of, and just kind of be part of an energetic group learning AI and growing with AI together. Or if you are interested in getting certified or being able to take some of our more sophisticated learning by doing courses, that is in our paid community. And you get access to our brand new AI Leadership Lab, which contains 15— I think we're more than that now— 15+ AI models that you can use to help grow small businesses in your area. So if that's something that you're interested in or something that you're doing yourself because you are a small business, It is a chance to get access to some pretty incredible models for a very reasonable price, for just the price of membership. So not only do you get all the classes, you get all the certifications, but you also get this platform. So it's pretty cool. So I do want to spend a little bit of time talking about the topics, the certifications that are going to be available this year. And everyone, you know, all of you can take them. It just depends on how you want to take them. But the first certification is one that I actually have rebranded, and many people have done this. And I thought hard and long about it. I rebranded it already twice. So this used to originally be called Building Applied AI at Scale. And so I would teach people how to use applied AI models, which means models available via API. So I would teach them, you know, in the course of a week, I'd go through 12 to 15 models and teach them how to use those models as an engineer or as a software developer or as a no-code developer to leverage existing AI models like OCR for character recognition or document intelligence or media indexing, like all of these existing models that already exist in the world and just how do you use them. That was like 2016 to 2019. Then people started getting excited about chatbots, and so I reframed it to be Okay, rather than just focusing on kind of how does a developer integrate AI into what they're doing, how does anyone integrate AI into what they're doing? So we created a chatbot builder course that allowed you to build a conversational app. At the time, I was using Alexa Blueprints, so you could go out and build in my class a blueprint. Using a blueprint, you could build an Alexa skill. No code required. No code, no limits. That's That's my jam. So that was pretty exciting. And then of course, we rebranded it again after ChatGPT launched. It's the same stuff. That's what I think is beautiful about learning this skill is that it is consistently a skill people need. We, we, we rename it to accommodate the new terms we're using in the industry, but it's the same skill set, taking technology in the form of AI models and APIs and leveraging them to deliver business value. That is the key. It is not no longer good enough to just learn how to use AI. Anyone can now use AI. It's can you use AI to deliver a very specific business value? Can you generate more revenue? Can you reduce operational costs? Can you increase the effectiveness and efficiency of the people working in the company? Those things tie directly to value in the company, and that's where you wanna focus in 2025. So today it is now called the, uh, Agentic AI Accelerator. And it allows you to build your very first agent in the course. Concept is the same. We build an agent, but here's how it's a little bit different. I am a huge believer in entrepreneurship. I'm a huge believer in building your own company and generating your own income and controlling that view of the world. Not all of you are with me. I get that some of you prefer the security and safety of a W-2, and I totally am with you on that. But for those of us who've decided, or at least wanna tinker in the idea of building their own business, the lifeblood of any business, no matter how big or how little it is, is customers, is the ability to attract customers to pay you for what you do. So what I did last year is I launched a program called the AI Consultancy, um, project. And that AI Consultancy project allowed us to go in and basically learn the 5 Ps of consultancy success. And it's everything from how do you prepare your business to you know, get people to contact you and take their money. To how do you build the product? How do you pitch it? How do you produce the right results for your customers? Like all of the things that you need to build a successful company. The blood of that, the lifeblood of companies is their customers. And so what we've changed in our new Agentek Accelerator class is that we've now added the ability for you to not just build a customer-facing conversational application, just like I've been doing for over a decade. Not just that, but now add to it the ability for it to also grab contact information, become a lead generation tool for your business, and even it, depending on what you're selling, even have that AI system close the sale for you. And this, again, whether you're doing it for yourself or you become someone who does it for others, it's a great cool school, uh, skill set. And so it's something that I thought was a really important shift for that. Now, in addition to that, we also— so that is kind of the builder class, uh, that has evolved. In April, we will be launching what's called Responsible Prompt Engineering, and this is all around the safety systems that are required to, to build, uh, and understand systems that have leveraged prompts to attain their levels of security, privacy, and transparency. So this is really big, I think, because As we move away from system prompts, right? Some of our, our newer systems are promptless, meaning that they've been trained so well that they no longer need to be guardrailed at the prompt system prompt level, that that is already part of the training. Years ago, I used to do this by what a term we called fine-tuning. And if you are not familiar with it, definitely, you know, look it up and I'll share with you a little bit about what it means. But when you're dealing with like large systems, certainly LLMs and even SLMs, these large and small language models, they still have often billions of parameters, right? When we talk about large language versus small language, we're talking about trillions of parameters versus billions of parameters. It's still a lot of stuff. And so one thing that I would often do is I would examine the business problem I was trying to solve, and then I would select a model that could most closely align to the results that I wanted to generate. And then based on the data I could access, I would fine-tune that model, which means I'd actually train first. I'd shrink the number of parameters, so I'd go from a large language model to a small language model. And then once I'm in the small language model, I would fine-tune it. In other words, train the model on the data that's necessary. Now, I got lots of conflicting advice about this over the years. So many. Oh, awesome. Great to see you, Sherry. I'm gonna be in DC next week. So hopefully we can see each other. And you have a pass. So we're gonna do that thing on Tuesday or Wednesday night. But as you know, I'm— I got so many conflicting pieces of information on whether or not to fine-tune. In the beginning, I went and took some— you know, I, I had had an opportunity to sit with some of the researchers at Stanford that were working on the foundation models concept. They ended up, you know, part of HAI or human-centered AI. So I was, I felt like it was good intel. And the researchers were saying that the research showed that it would be a good idea to fine-tune these models in order to protect the accuracy of the model. That you would basically be telling it exactly, like training or updating its training data to have the ability to get the right answers. But then fast forward, this was like, I don't know, beginning of 2023. Fast forward to the end of 2023 and people are like, ah, no, no, no, no, we're gonna use RAG, right? Retrieval Augmented Generation. And now that means that we don't have to fine-tune the model. It's actually not super effective because of the number of parameters that it's using. Because again, we were, we were mixing terms between small models and big models. So they were basically saying for all models, don't fine-tune it. Instead, just use retrieval augmented generation and tell it what dataset to use. What I uncovered was that in practice, in production, it's really great to start, like, when you're in the playground mode or baby tiger mode, as I like to call it, when you're just getting started and you're proving out a concept, you're building your minimum remarkable product. When you do that part, it's great to use a very large model with lots of trillions of parameters and data points that it's monitoring. It gives you a much better way to kind of figure out if the problem you're trying to solve can be solvable in that way. But immediately following, hey, this works, hypothesis tested, it is successful, I then go look to find a smaller model that I can transfer learning, right? Transfer the learning from the big model to the small model. And now I have the ability, rather than dealing with those trillion data points, I now have a much smaller set. And that size is completely dependent on me. And now I do have a choice of whether or not, because I'm not in this big massive model anymore, I'm in a small purpose-built model. It might make sense for me to actually train that specifically. In other words, use that dataset to train the model. And this is what I ended up doing, and I ended up being able to achieve some pretty incredible results, specifically in the area of document information, like information generation, whether it was FDA regulatory documentation generation or law contract analysis. Generation. So these organizations that do a lot in paper, a lot in PDFs, again, from life sciences to legal to even education and academia, these types of things ended up lending themselves very well to this concept of fine-tuning a small model. And that was before we had the term SML or, or SLM, small language models. But I, yeah, so I've seen the, the writings on the wall. And so I, wanted to make sure that people, you know, started thinking about these models as very finely tuned models that did a certain task. Fast forward to today, we have a new name, called agentic AI, right? Building agents. And again, the definition of an agent, which I maybe borrow or take from a couple of different sources. One, certainly the, the work that I follow very closely at Stanford, but also the work at Microsoft and Microsoft's Responsible AI team, but when— and, and the agents and assistance team, like a lot of these problems that are now getting named have been— we've been working on this a really long time, years, uh, decade maybe. So in that sense, what I always like to think about is, okay, now that I've got, right, this small model, an agent, just to define it for those of you still trying to figure out why we're calling it that, Right? It allows, right, agency. If you think about that term, right, the ability to do a thing, it allows us to create agency in a small unit of work or a model. So a model that I build to be my customer service representative now can be siloed into just doing that work and sourcing it, the data it needs to just do that task. And it can be fine-tuned to do that task very well. That's a pretty simple one. But let's say now that customer service agent needs to order on behalf of the customer. Now we have a couple options, right? We can have that single agent or that single model go and actually do an API call, which some call actions or, you know, an invocation. I can call out to that model. And the challenge with that is that the inputs and outputs are now kind of strict, right? When you call out using an API, you now wait for the API to return it back, and then that agent has to do its job. What we're seeing now is that when you build a single model and you build it in the style of agentic AI, you're now going to create it in a way that allows the model to be autonomously invoked, right? So in Microsoft, there's two different concepts, agents and assistants. And the difference between them is really this autonomous behavior, the ability to detect information about your surroundings as a machine and take action. Similar to autonomous vehicles, right? There's a reason there's like a bunch of cameras, right? Like those cameras are constantly assessing information and it's making autonomous decisions based on the inputs that are being provided to that model. Similarly though, this can be true in retail, in like cross-selling, upselling. These types of activities can actually be automated through agents that say, oh, it looks like you're in this demographic, and last week you bought this thing. So next week you might need this thing. So I might as well offer it to you now. So let me do that. And no human intervention is required. So I love this because the first step now in getting involved in AI, it will be around learning this very basic skill of building a unit, building an agent or a bot that does a very specific thing. And as soon as you have a few of those things, and their job, right? They are different parts of an existing workflow. You can now orchestrate those things together. Enter another term that's not— it's relatively new, but it will become much more important in the year— in the years to come called MLOps. And I'm actually gonna be on a podcast in a few, uh, maybe in a couple weeks, uh, about this exact concept. But scaling responsible AI is all about like pulling all these purpose-built models together and orchestrating them. For some of you who are technical or, come from kind of the cloud side, just think about Kubernetes, but for AI. And this is a product that I had a chance to work on in kind of its infancy at Red Hat, which I absolutely love, that allows people to use a Kubernetes-type approach to managing models. And the benefit to that is that these models, just like a Kubernetes cluster, could be anywhere. I might have a model coming to me from Anthropic direct. It might be an Anthropic model surfaced through SageMaker or Bedrock, or maybe it's over in, you know, Azure AI Service or one of Azure's other AI models. It might be coming to me from Google. Like, the benefit of this kind of integration platform, this operational view, is that I can now orchestrate them together. But not only that, one of the benefits of an operation platform, some of you might know, like OpenShift, same idea. Right? But thinking about it from a machine learning perspective is now I can set up responsible AI practices, groundedness detection, prompt shielding. Like, I can do a lot of this and do it once and have it apply to all of the models in the purview of that operational interface. And that— and I know some of you have already hurt your brains a little bit, but this is the future. And it's super exciting because it's an evolution. None of this is So many people always tell me, oh my gosh, there's so much to learn. New things come up all the time. I'm like, new-ish. They might have a new name, but the tech that it's doing is likely less new. And so understanding these core kind of capabilities of what does it take to build a model and scale it to production to generate a business outcome that is valuable, generating more customers, closing more deals, reducing operational overhead, growing the business without growing the staff. These are like critical talking points for most of the executives that I'm, I'm working with. So pretty, pretty cool. Oh, thanks, Bryce. Yes, for those of you who don't know what Kubernetes is, that was really meant for those who did know. But in case you don't, those in the Zoom call got the benefit. We have a lot of great enterprise architects in the house that have been working with this technology, and we're kind of delighted to see it take this turn, right, is that we're moving into operationalizing AI. And what that really means is operationalizing these models, and those models at the end of the day turn into services. So it's service-oriented architecture all over again. So all those cool things that we learned about SOA and registries and all of that, it's coming for us again. So hope you enjoyed that journey, and we're back. We're back. So it's cool because when we were doing service-oriented architecture in like the early 2000s. That eventually turned into containers, right? So what were services in SOA in 2001 became containers and Docker in 2000 and whatever, 12, 15. And now we're seeing that same evolution, same concept, same knowledge, same matrix of skills, but now applied to these services which are actually AI models. Now, I of course make it sound prob— will definitely make it sound a little simpler than it is because now when you have models communicating, there's way more complexity in the activities that you're asking it to do. But it's good to know if you have any of these previous skills, like that ramp may be even shorter for you. Yeah, ESPs all over again. Ah, don't say it. Say it ain't so. What? I loved RabbitMQ. I used to have my— one of my kids, I can't remember which one, maybe my second. No, my, my third. I can't remember. I can't. It's terrible. One of these kiddos. Um, I ended up getting a rabbit MQ onesie for them and, uh, it was really cute cuz a little rabbit. Anyway, so big fan, big fan. But yes. Yeah, it's just, I think it's really comforting, at least for me, to know that these skills that you've acquired over the last 20 years, especially in technology, but certainly even in business, they are going to be amplified in this new world. They are not gonna be reduced. They're not gonna be like, You do not become irrelevant in this new world. You become irreplaceable in this new world because AI doesn't know what to do without our guidance. And people who understand business problems are gonna be the ones that can tell this AI what to do with the most confident levels of success. And I've watched it. I mean, that's the benefit is I'm not speaking out of like inspirational, you know, foo-foo mojo. Like, this is demonstrated value that I've seen. Which brings me to one of my favorite things that happened this week. I think I posted it in the one of our communities. But Microsoft has updated my favorite blog post ever. Some of you have seen it. I will see if I can grab the link and put it in the chat. But they have taken— it started off as 100 use cases of successful implementations of generative AI across enterprises, 19 or so industries, 20, maybe 20 industries represented in that data. But then they came back to it. So normally you publish that and you're like, we're awesome. You know, we're good. And then that's it. Like, you just keep using that old document as, like, proof that you know how to do a thing. What they decided to do is make this a living document. So Satya Nadella on LinkedIn just reshared it earlier, and they've added 50 new use cases, but in the middle they added 50 more. So now there's 300 total success cases. And here's why I suggest you take a look. And if anyone has the link, I'll go and look for it. Anyone has a link, you can throw it into— oh yeah, in the chat. But I'm gonna see if I can find it. I actually have saved it on my screen. So apologies while I not look at you for a moment and grab this URL. But I've saved this, this article. It's bookmarked because it is a huge wealth of information. Here we go. I'm gonna go ahead and share my screen so you can see it. It is a beautiful, beautiful piece of work. So here you can see updated 50 new stories on Success. Here's why I love it. We now have artificial intelligence. We have generative AI. We have the ability to look at this information and pull from the information that's being shared. We can actually do some analysis of what's happening. One thing that I did is I took this document and threw it into a generative model that I own, and I asked it to categorize it by industry. I then asked it to categorize it by size. I then asked it to also categorize it by toolset. Now, you can't see all of this. It does take a little bit of work. I mean, oh my goodness, let's do some work. But I do wanna share with you what I— how I did it so you could do the same thing. This is like, Good education from data is my favorite. I would prefer that all day long. So this first link that they give you is the IDC report based on a lot of this data. And that IDC report, IDC report is extremely compelling. I use it in every keynote. Like, we now know very specifically that certain use cases in generative AI generate 3x return, and even more specialized use cases generate a 10x return. So what I always say to the people that I talk to, to my clients, is why would we not do anything other than what's on that list? Like, let's just do that. And that list is not available in the IDC report specifically, but we have a blog post that tells us the 300 use cases that they're talking about. So that correlation is available to us. So what I always like to do, you'll see here, It's pulling in the data is represented in the IDC report, which is linked here. I'm just going to— maybe I'll click on that really quick so you can see it. And I love this report because it codifies, it memorializes all of the things that most of us that are in the consulting world or in the business service world have been saying for a long time, but maybe didn't have large-scale data to prove it. Like, it validates all the things I've been saying for, I don't know, a decade. So it's Very cool. This, this first came out in 2023 and was recently updated at the end of 2024. And they'll update it again likely at the end of the year because it's just so compelling. But this, I mean, just tell that to somebody, right? Like for every dollar you put in, you get 3.5 back. Let's just do that. Let's just 3.5 back every invested dollar for every person that decides to tap into AI. That sounds amazing. The next one, usage dropped from 50% to 75%, or, or I'm sorry, jumped from, I think that's what I said, from 50% to 75%. And the year before that it was 23%, right? So it's making these leapfrogs. What that means is that no longer are you the pioneer. You are now dangerously close to being a laggard in this space. 75% of your competitors are leveraging this technology, and if you're not, you will get left behind. So it validates that FOMO is real, right? That you will, you are on the path to being behind. You know, 2 years ago you were ahead and it was kind of, you know, bleeding edge. They say that for a reason. It was a bit bloody. But last year we saw a lot of hardening of these models moving into production, operationalizing of some of these use cases, and that data is now hopefully going to help guide. But this just, you know, knowing this, um, is really, really critical. And then probably, let's see, um, yeah, this is probably, I don't know, it's my, it's not my favorite, uh, point. I probably like the first one better. Like for every dollar you put in, you get 3.5 back. And that's conservatively, cuz we've now identified use cases that do 10x returns. Um, but 24 months is about as long as it takes. However, we can deliver value, meaning I can change a metric in a company in as little as 3 months, and that is different than any other software I've been ever a part of. So, so pretty cool. I'm going to hopefully figure out how to escape out of this. Oh dear. Okay, let me— oh, let me go back. All right. So this is the article, but let me— these are all— oh, I went to— that's the IDC article. So please read that. Oh, you know what? I'll do you a favor and throw it into the chat for those of you who are joining me. Oh cool. Thanks, Sherlyn. And then this is the This is the one specific to the IDC report. I love it. Now, I got very curious in the industry and was like, hmm, do other cloud platforms have a similar article? Not exactly, but they do have similar data, which means it's harder to get, but super valuable. So Amazon has a repository of use cases that they've collected that successfully implemented generative AI on the AWS platform. Gold. Google also has this list of use cases of customers and their questions they asked before they started and the result that they had when they finished. Gold. Like, I'm just, you know, when people oftentimes will ask me, how do I get my CFO to do this? How do I get my executive team to agree? First thing you do is you build something useful, and then the second thing you do is back it up with a bunch of data that says you are late if you're not doing this right now. And that has been very successful for me. Um, but let me go ahead and, uh, revert back to that 300 use cases. So it tees it up with, hey, there's data to support why this is successful. And then like galore, lots and lots and lots of examples. So this is why you'll see they're kind of all over the place, uh, from a, a industry perspective. So that's why I, I leveraged it. But I'm just gonna go and take a look at a couple of these so you can see what it looks like. So I'm gonna click on one. This one is Credit Club. And what I like about these use cases, number one is the data, of course, that they show. But you'll also see as I scroll down the benefits that they saw, the metrics that they reached, and somewhere in here, typically— ah, yes— the products they used. So this is very useful to us as architects, as leaders, to know What kind of results correlate to what products so that I invest in the right stuff? How many of you are like, ah, I get more products every single day with the combination of my Google list, my Amazon list, and my Microsoft list. I now know what tools drive the biggest ROI. So I'm definitely gonna do that, and it's data-driven. So do that. Now, does that mean you don't wanna take your, you know, brothers, sisters, cousins, the AI GPT wrapper that they built and use that. Sure, please do. I hope it makes them lots of money. But realize the proof in the pudding for enterprise success, it is here and the data is available. Does it take a little while to, you know, click on all these and like look at it? I mean, there's some massive value here. The organizational size, right? Where they are located, the, uh, industry that they're in. It's so cool. So you'll see a lot of the things that I do based on the projects that I've worked on. I've started to create content on social media videos that walk through these case studies. Not all of these because I didn't work on all these, but the ones that I had access to. And so you'll start to see some of these names as well as some of the projects. But you could do the same thing, like look at these case studies and mirror that for the companies that you're working with for your own company. If you want to elevate the success of you as an individual in the company, like document your progress in building systems like this and then send it to your CEO. It works, I can tell you from experience. All right, I had a long day of talking, so I apologize for having to grab a drink every now and then. Yeah, it's really cool. Yeah, it just came out, so you would have probably seen a little bit of buzz around this, this article. But the key life lesson number 478 The key is to double-click. It is not enough to read. I mean, if you look at it, read the very short synopsis. So 96% per month. If I go into a meeting with executives and I just say, hey, this company saved, you know, 96% per month in auditing expenses and analyzing for 100 meetings in an hour using this tool, the very next question they're gonna be is, how did that work? Tell me more about that. So do yourself a favor if you do decide to leverage some of these, and I encourage you to do so as explanation points to people you're trying to get to realize the value of moving in the direction of AI at this time. Like, be familiar, pick a couple of these and know them very, very well. You'd also be surprised. Don't do this for like, for this one specifically, but like there are 300 cases. Like find a couple of them, reach out to the CEO, reach out to the CIO, the CAIO if they happen to have one. And be like, hey, I just saw your thing. I would love to get on a call with you. Would you be interested in being interviewed for my podcast? These are things I do all the time, and you'd be surprised how many people respond. Um, again, we don't have a gigantic number of people. Like, if this call had 500 people, I wouldn't say that because I wouldn't want 500 people to go and do the thing I'm doing that's serving me very well. But I want you to, I definitely want you to. Um, so really cool. Uh, So you can see some of the partnerships. There is not only client success, but also partnership set success, consulting success. So there's lots of really cool stories to take a look at. So this is my gift to you. If you haven't seen it already, it's worthwhile not just to look at it and go, oh wow, that's cool. 300. That's neat. Click on one, maybe two, like spend, I don't know, carve out an hour and a half, 90 minutes, and go through a significant number of these. Click all the way through to the case study. And if you don't have time to read the case study, Copy and paste it into NotebookLM, right? It's already public. So copy and paste into NotebookLM, have it read to you in a cute podcast style, but absorb the information. This is like super valuable and it's— it has not been a level of transparency that I've seen in my 28-year career. So I really think it's something special and something that will help you become a better leader by seeing how other leaders have done this. So, so I hope you will enjoy it. Awesome. Oh yeah, for sure. So Bryce was mentioning that there is another tool called the Scenario Tool, and I actually can give you some history on how this was created. But let me see if I can copy and put it in a screen I'm already sharing. Hopefully, yes. So everyone in the Zoom can see that link. You're welcome to take a look at it. There's lots of really cool tools that are hidden, but this is called the Copilot Scenario Library. I'd say about 60% of the case studies in the 300 list are Copilot case studies. And it's— the reason why is because it's a very low-hanging fruit, easy-to-adopt use case. However, most— many companies last year tried without any help at all to buy Copilot licenses and implement them in their company. And as a result failed miserably, and as a result have made the assessment that Copilot is not good, when the reality is talking to machines is a learned behavior. You have to learn how to do it. And certainly talking to a machine that's fine-tuned or custom trained on your own SharePoint directory and your data itself might be conflicting. Like, as many of you know, I have a, a philosophy of IA before AI, most people adopted Copilot and did no IA at all, or information architecture. Like, you do have to architect your information so a model can accurately present information back to you. It is— data is key. Someone asked me this yesterday. I was in a session with a client, and they were like, I mean, what about data? And I'm like, what about data? It is still— just because we are able to leverage data that was pre-trained for us in an LLM, it is still the lifeblood of a good outcome, of a good result, is how well is your data architected to be surfaced to a model to generate a business outcome that's positive. Yeah, exactly. I mean, most organizations want to use AI. They go and blindly go, Copilot, let's go. And then the data that Copilot's trying to look through, metadata is bad. The files are not named in a good way, it's organized in a very bad way. It's very difficult for even the best LLM to interpret bad data. And so there is no way around that. Your data still has to get good. Here's the good news though, is that organizations that are new and have not put all their data ducks in a row, you can instead— what, what the way that you get around having all your data ducks in a row is starting that process of doing your information architecture, your data cleaning and all that. But then identifying, usually there's some shiny area of, of your business that has good data. Someone has, or, or maybe better data than everyone else. So start there. Build an AI system in that team. And I won't, you know, I won't lead the witness and tell you which teams those are, but there are certain teams that are better at data than others. Awesome. And that, I think, is the future, right, Joe? I mean, I think you're gonna be able to, in the pioneering that you're doing in literally reinventing what a business looks like with AI from the ground up. I think organizations, more and more organizations, certainly startups that are scaling, are going to require that same skill set, and they will be very interested in that from you. So really cool. Yeah, exactly. If your data is garbage— thank you, Joe. Yeah, if your data is garbage, Copilot can't make it any better. You're absolutely right. Like, and it's funny because I think they thought it was an easy button. Like, oh my gosh, great. Everything's a mess. Let's add Copilot, press a button, and it'll be better. And the reality was it wasn't, but they, you know, sometimes naively blamed Copilot for that. So don't forget, there is a Q&A section. I am trying to monitor the chat, but the chat goes by really fast. So if you have a question or a topic you want to talk about, please don't hesitate to put it in Q&A so that we can, can discuss it. But these are some really great resources, and it does allow you to see some of the tools now. Since we're in here and I'm showing my screen, I will share with you how I've delivered extremely valuable Copilot solutions over the last year, year to 18 months. And that is by recognizing that companies are going to build a, you know, they're going to want to adopt Copilot. But Copilot by itself is relatively limited unless you've got everything locked down and dialed in. But if you don't, how do you make that experience better? So enter— I might even have it. Oh, I don't have it on this screen. I'm going to just go to— so enter Copilot Studio. So that is an extension of the link that Bryce was nice enough to provide to all of you and give you access to. But Copilot Studio is your ability to bridge the gap for yourself and for your customers. And the reason I like it is because it provides you the ability to leverage some of these scenarios in the form of prebuilt agents. So I, of course, I've built a lot of them. So there they are. But I'm going to go to the home so you can see what it looks like. Here you'll see Explore Agents. This is in Copilot Studio. Let me see if I can grab that link and throw it in the chat. It's got a bunch of other stuff in it. I want to send you to my environment, but that should do it. And yes, just to address some of the comments, sometimes the data process in the business. Yeah, exactly. Many times the data and the process of doing something and, and really even how they solve the problem is all in their head. Tribal knowledge, we tend to call it. So the question is, how do you leverage that information to be part of the preparation process? And this is great question. Because this is something many companies are struggling with. There's a term right now, and I don't, I don't own it, and I don't even really like it too much, but it's a helpful term to understand what companies are dealing with, and it's called the silver tsunami. And it's also, I heard an older gentleman talking about his experience when he was younger, coming up of age and coming into executive rooms, and his boss calling it Um, what did he call it? A cotton ball room or something like that. And all of these represent kind of the what happens to people's hair as they age. And the silver tsunami represents the amount of people, aging adults, that are going to age out of working years in the next 10 years. We've been talking about this forever, right? Boomers getting older. Now even Gen Xers are— no, I won't hurt none of y'all's feelings, but But yeah, like, there is an entire generation of people that are hitting 65, 75, 85, leaving the workforce and taking their— exactly what you're asking— that beautiful tribal knowledge with them. So there are companies that have decided to focus on this. And, and it's an adoption of a machine learning, or at least data science concept known as expert systems, right? How do we build an expert system off of these people? And so there is the ability to create a large language model that extracts this information, not painfully, from the people that are going to be leaving the workforce and preserves it in a way so that the model can be helpful to younger or earlier generation people, to help them answer the same question. So that extraction process is happening in some companies, but most companies are really quite clueless as to how to solve for that, right? They're watching people leave and realizing here's the catch-22, right? People are leaving the industry in droves, leaving these roles, and the younger people are getting, are adopting artificial intelligence systems that cut the corners of all the things that would teach them the things that they need to be that person 20 years from now. So there's, it's kind of like a, a very interesting dilemma that's being presented to organizations where the people that did all, like they basically grew up in a company So they learned all the different parts of the organization in order to lead it. Now we're gonna use AI not to replace those functions, but certainly to augment them and take away all the hard parts, right? Because who wants to do the hard part, right? We're gonna have it do our cold calling for us because who likes cold calling? We're gonna have it do our lead generation and lead, um, nurturing for us because nobody really likes that job. But the value you get when you do that work of cold calling and lead nurturing is you learn how to extract what a customer needs so you can serve them. And if you don't know how to do that and you rely on an AI system, there's a gap there that you're going to suffer from. And so in that, in that state, we, we, every company, certainly older industries, meaning industries that have been around for over 100 years, finance, healthcare, government, they're struggling with this. How do we maintain that knowledge? And I do think that there's some in the world that think, Oh, it's fine. A 70% solution is fine from an AI model. But not being able to do something that the biggest companies in the world do, which is have humans that know how to actually do the task, monitor and measure the success of the machines that are doing it. I think that gap is gonna grow bigger over time. Yeah. What? COBOL. You're speaking my language. You're 100% right. Yeah. They're running COBOL, Fortran. Absolutely. In their applications. As a matter of fact, if you all want to become a COBOL programmer, you'll get paid better than a data scientist. Little known fact. But yeah, it is— COBOL is not taught anymore. There are books available. I actually think I have my COBOL book in my bookshelf. I think it's that bookshelf, but I keep it around to remind me of exactly what happens. But yeah, it's 100%. Oh my gosh, I can talk to you about this. This is a great— I mean, it's a bit of a segue, fans. And family, but I'm gonna tell you anyway. Imagine if you will, and I may or may not have worked on a project just like this, but imagine if you will, I want to use AI to basically replace COBOL. Does anyone know why that won't work? Like off the top, asking a machine to go in and create a system that would work today from a COBOL system. Anyone? I know it is surprising that it's still a thing, but it's big. It's like ES— it's like Enterprise Service Buses. They are also still a thing. They should be dead, but they're not. Or not. Yes, exactly. Thank you very much. Someone in the house knows. So when you use, like, COBOL is a procedural thing. We are in a world of object-oriented, unstructured data. It is literally the opposite. If you go and ask an AI system to create a Java application from a COBOL system, it will fail. Ask me how I know. I've tried. It does not work. So you actually need someone who knows how to build software to ask these questions of the AI system rather than starting with the end, which is the procedural functional code. You have to start at the beginning. What was the problem that that COBOL program was originally designed to solve? And you're going to reinvent the solution. Now today we can reinvent a problem that COBOL solved by just completely rewriting the thing that was hard before with generate code generation and a good software engineer, you can actually— it's a, it's attainable now. As a matter of fact, there's a consultancy that's focused 100% on doing it. It's— I don't know if it's fun, but here's how I was able to crack the code on old legacy code, COBOL, to object-oriented code. Like, how do you bridge that gap? They're, they're intentionally opposites. The logical framework of thinking how to design these systems are completely different. So you can't just ask a model to do it because it'll try to replicate the COBOL structural thinking in an object-oriented manner. And I don't know if some of you have seen code like this or at least been in a system that uses code like this. It's basically object-oriented code that operates as a functional program, and it's gross and awful, and we don't wanna do that. So how do you get out of that? Well, what we end up doing is going back to that original COBOL program, and we build or use, find or recreate the requirements document. For that COBOL. So we reverse engineer, and we can use AI to do this. We ask it, here's the functional, like, here is the code, reverse engineer from this code the requirements document that was used to build this. And then now that we have a new requirements document, we can build a new functional spec, and that functional spec is based on a different paradigm of thinking. So it's really, really cool. Oh my gosh. Yeah. Oh my gosh. It's fun. We should do it. We should totally. We can at least simulate it, which is how I started. I did, All of this testing that I did, it was a side project. Don't judge me. But I did it for fun because I was— because I come from 4chan and COBOL. Like, that was my first love, my first language. And so I was like, oh my gosh, what if I could go back? And this is another silver tsunami problem, is all these people who know how to write COBOL are no longer doing that work. And so, as many of you know, if you're in the enterprise architecture world, there is systems that are going to sit there that you can never touch because no one knows what happens when they go down. No one knows what happens when they break. No one knows what happens if you change it. So you, you just don't change it. It gets crystallized in the architecture and now you build everything around it. It's a disaster. Um, so a huge, huge, huge opportunity. Um, oh, Pascal, I love that language too. Amazing. If I was good at college, I might have one day done that, but I didn't. So, um, really, really cool. Yes, exactly, exactly. This, Shari, you're exactly right. Like, this is It is the foundational, like, challenge that humanity has struggled with, um, for all time. Uh, and, and it's rearing its head. Here's the difference though, is I do think that in our lifetime, we are seeing this transition, like, happen right before our eyes. Where before, by the time my, you know, let's say my dad who did this work, by the time he leaves the industry, I'm not even, like, doing that work. Like, there's no correlation with what he did and what I'm doing. Doing. Now there's a direct correlation. I am impacted by those decisions in the job I do right now, and that means we actually have a chance to solve for it. It is wild. I know, right? It's so cool. It's so cool though. So I, I got a little distracted. That was a little side quest on COBOL. I can't help myself. But I did want to share with you here, and don't forget, if you have any questions in the Q&A, again, I've been answering them along the way, but if there's a question, don't forget to drop it in the Q&A in case I miss it along the way. I am trying to keep up, but just in case. But now, the gap that I often solve for, because I will go to companies, they'll be like, we tried Copilot, it doesn't work. And I'm like, tell me more. Literally, the magical superpower of any consultant is just to be willing to listen to people complain. So I go in and I love it because I sincerely empathize with the nonsense people have to deal with, with the technology. I helped build for them over the last 20 years. If I had what the tech we have today, I would have built it better, but I didn't. So it's like duct taped together and they are suffering. These users are suffering. So I go in and I say, oh, you didn't, you know, you hated Copilot. Tell me more about this. What was missing? Why didn't it work? And as they describe the gaps in the implementation, and in their ability to use this tool, I realize, oh, Well, Copilot doesn't actually do that. So usually that's where I land, right? I start with, oh, you don't like it? Tell me more. What, what was wrong with it? Then they tell me use cases they were trying to use it for. They got frustrated, and I'm like, oh, well, that, that it doesn't do that. Not yet anyway. And so out from out of that came a project that I got to be a part of in the earliest days during COVID and then following COVID as part of my MVP status with Microsoft. I got access to the Copilot Studio in like private, whatever they call it, private review or private— what is it called? I can't remember the name. Anyway, it's basically when they send an email out, tell the MVPs and say, hey, who wants to test this? And I, I try to always say yes. So now the screen that you're seeing here shows you— and let me just go to see more so you can see the entire catalog. These are all of the different systems, the different pieces of glue that were created to make the Copilot experience better. But they stumbled on the idea of like, oh wait, not only is it like this, do y'all remember, some of you remember business process execution language and like orchestration and process orchestration. So similarly, I now have a system that does some stuff and I want it to do this other dream outcome. So I'm gonna build a, like a little piece of code, a snippet if you will, that's going to help me go from A to B. But it's different now. So now I can actually build an AI system, an agent that can serve as the house for that little snippet, but then it can do so much more. So you'll notice these are all different, like, buckets of work. If you're in an enterprise, things that you might need, an IT help desk. What I like about this is if you go to the IT help desk, the thing it's gonna ask you for is to connect to whatever your help desk is, but it's gonna provide you an LLM connector, problem to solution. Right now, people are using Copilot to fix stuff. Copilot doesn't know what to do. Enter Studio. I build a dedicated agent that integrates to the solution that actually solves the problem. Now, it does require some skill. You do have to develop, you know, you have to build it. You have to know how to build an agent. But I teach that, so come and take the class. You do have to know how to build an agent, But it creates huge autonomy in an organization to build solutions they actually need and use. So whenever somebody says off the cuff, like, oh yeah, I tried Copilot, it doesn't work. I'm like, well, did you build a— did you build any agents? Like, did you solve for that? Like, that's a problem we can solve. And when they realize it, which is our methodology here at the institute, right? We believe in showing, not telling. We believe in doing more than we talk. And so when I'm in those conversations, I of course already have a bot that I can demonstrate to them. So here's what I do is I go into this. Oh, I probably shouldn't do that. Is I go in and I will build a demo of all of these. So if you're looking for ways to get experience building, this is exactly what I did at Amazon Alexa is I just started building little agents. In Alexa, they were called skills. In Bixby or on Samsung, I think they were also called skills. But anyway, I started just building these small, compartmentalized, componentized services, which were actually AI models. And now that's exactly what we're doing here. So build all of these, set 'em all up, connect 'em all together. There's no harm. It's only in your own environment. Will it cost you a little bit? Yes. So build it in a resource group that you can play around for a week and then delete it all. But this is like cool stuff. Case management, inclusivity for, enterprise resource group, or like, employee research group, resource groups in companies, right? Like Latinos of Accenture or, you know, women of UKG, like these groups, it's nice to have tools that help them foster community. And so you can build tools that they can use in Teams. So when you deploy this, the, these agents, they can be deployed as agents in Teams. They can be deployed on a website. They can be deployed— well, it's API, so you can deploy it wherever you want. So some of the things that I thought was interesting, like HR benefits. If you go to benefits, it's gonna ask you like, how are you going to establish the knowledge base of this benefit system? Right? So you're gonna add the knowledge base. Maybe it's a SharePoint directory, maybe it's a set of PDFs, maybe it's the employee handbook. Done. Or in this case, it's benefits, so it won't be the— but you know what I mean, right? We all get these documents when we start a job, right? Or we, you know, even like just imagine not just benefits, but the idea to create a copilot that solves a problem that's documented somewhere. Just surfacing that documentation in a conversational application is hugely valuable. Even we call it like repository, repository intelligence or conversational APIs. Being able to provide an LLM that sits in front of your GitHub repos and just allows people to ask what it does and why it exists and how it's used and the data that it uses and what are the risks like it's hugely valuable to stakeholders that don't have to actually go into your GitHub and look at your issues and follow your threads. Like, you can just provide this interface and the— and of course, the knowledge base for that would be your documentation, which we all write lots of in depth for everyone of the projects we built. Maybe not, but we should. We should. Okay. So just a huge amount of agents available. I encourage you as a, you know, project to yourself, If you wanna get good at this, build custom Studio agents to like, you don't have to build new ones. You can if you want to, but just build the ones that are already there. See how they work. Get very comfortable with this concept of these small purpose-built technologies, these purpose-built models. And then the beauty will come in its orchestration. We're not there yet because most companies are still struggling to build the pieces. But once we get to commoditize the pieces, all these little agents, the very next thing, it's the same thing that hap— I mean, I've seen this now, I'm on my third wave, right? So I saw it happen in service-oriented architecture. The first thing we had to do was break up all of our code into little module pieces and then turn 'em into services. And then we had to orchestrate those services, like we had to connect them all together. Then we were doing pretty good. Then come Docker. Now not just services, but we're gonna put a bunch of services inside a container and those become services. So now we're containerizing, right, a bunch of functionality. So entire web apps become containers. So that's the next level of abstraction. Now we've abstracted even further to building these agents, and these agents will be the same. They'll be like a service that you're going to need to orchestrate. So it's a similar skill set if you have it. It's gonna be really exciting times whether you know it or, you know, know that skill set or not. It's a great time to acquire it and become part of the solution. Okay, so we got a couple questions. I just wanted to ask them. Yes. So we are at 5 o'clock. So I'm just gonna take a couple minutes to talk about next week. Hopefully you enjoyed the session today and some of these new and cool AI things. Thanks for everyone for providing some of the links that we used today. So thank you, everybody. And of course, I was wondering, Verole, if that was you. No wonder I kind of felt it was you. But yeah, awesome. Good to, good to see you all. I'm trying to make sure I don't miss any questions. If I didn't answer it because it was in the chat, please throw it in the Q&A. You have a few minutes to do that. But here's some exciting stuff that's happening in the AI Leadership Institute and in the I Love AI community. So today it's Friday. On Monday, we are launching and starting the I Love AI week. It is my favorite week. All content will be recorded. You do not have to show up from 12 to 1 every single day. I always add a buffer, so I think on, on the schedule it says to 1:30, but it's an hour. 5 hours is a lot for me to ask of you. I realize that, but I will make it as valuable as possible. And what that means is that we will be learning a very tangible task in each one of these days. One is going to be personal. So how do you use AI in your personal life? I'm actually going to show you how I'm going to use my own personal medical records in order to get advice on, and you could do the same thing. You'll— and some of you have— oh, he's not here right now, but one person has also done this where you import all your medical records into a tool. And again, some of you might immediately be like, I'm not giving a model my medical records. It's up to you. I personally am fine with it. I'm not using it, but, and I think my data has been stolen from that hospital anyway, so it's fine. But I'm going to set up a model that's then going to give me actually reasonable information based on that data. So this, I don't know, I think I have it here so I can show you what it looks like. Yes. This is not the whole thing. This is just a piece of it. But I got a bunch of test results from this organization, and it has, like, blood tests. I mean, look at all that stuff. Like, what? I don't even know. There's words on here I do not understand. I know hemoglobin. That's cool. Most of this is green. Some of it's red. Should I be worried about that? What does that even mean? So I'm going to feed this into an AI system and show you how you can do that. The same for any type of dataset, but more on the personal side. Like, how do we use AI to actually get better? How do we become healthier? How do we become more mindful? How do we become better leaders? AI can definitely become your partner in that. We're also going to be covering, and so that is day 1 is personal. Day 2 is how do you do it professionally? How do you, without breaking any rules or violating any constraints inside of your own organization, how do you use AI to do a better job? At the end of all of these will be a list of tools that I personally use to do whatever I do. So it's the tools I use on my personal side, the tools I use in my professional life. And then the third day is all about how to start a business in AI. I do see the world shifting to gig economy. I've never thought ever that a big gigantic consulting company could hire me as an individual. Otherwise, I would have just done that. I didn't know that. So they can hire me as an individual, pay me very well, and I'm not in the nonsense minutia of the company. Like, I'm able to actually just give them the valuable thing that they want from me. And it's beautiful. But I didn't know that that was a thing. So I wanna teach you all how to actually build a consultancy, an AI consultancy, and how to get that off the ground. It is the first part of a much longer conversation, but it'll give you everything you need to get started if you're a do-it-yourself kind of person. On Thursday, I will be talking about how to get a role in a company. Both these things I've done very well. I've reached, I've reached a 7-figure salary at a company. I've also reached a 7-figure business as an entrepreneur. So I wanna share my little hacks for both of you. However, that's gonna look, uh, on Thursday, and I encourage you, let me go see if I can grab the URL for it. Um, there is, it's on, uh, I just reposted it on LinkedIn. But maven.com, unlock your career growth, maybe that. Oh, yeah, this one. So this class is launching. I'm the first one. It's on Thursday. So right in line with our topics on Thursday, I will be talking about how to get a job in AI or how to get a role in AI based on what I've done my entire career. So Wednesday's how do I— how did I grow my business? And I'll give you basically step-by-step guide to doing the same thing. And then And in this, it's not gonna be in our normal environment. When you click on the calendar in the community, the calendar is gonna take you to Maven. So you do have to buy this. Sorry, it's not mine. I'm a guest. But know that I will also be providing that information in our community, like a recorded version of my specific session. But if you wanna be in the Zoom and you wanna be part of fun, and more importantly, learn from more, more people than just me, like, these are rock stars. Like Pascal and John and Brian and Cass. Like, they have combined millions of followers on LinkedIn, hundreds of millions of dollars of impact, billions of dollars in impact. So really cool stuff. If you all want to be a part of it, let me send you the link. Except I actually— let me see what price it says enrolled. I can't see it. Hold on. There we go. $200. So it's a $200 class, but I can give you a 50% off coupon.. So let me do that really quick. Bitly. But this is our Thursday event. So I'm gonna go into Bitly, and I think there it is. Copy. So I'll throw this into the chat for those of you who wanna join me. It'll be really fun. So that— there it is right there. And you can also— yeah, you can ping me and DM me if you want, and I'll send you the link as well. But that'll give you 50% off instead of having to pay full price because you're my friends. OK, awesome. And I also was like, how do I make this more accessible to everyone? So hopefully you'll find that an easy, an easy yes to learn how to really optimize your career path for artificial intelligence. And my, my session at least might surprise you. And then Friday, it's Fun Friday. So Fun Friday, we're going to actually take some of the biggest learning we did in the week. We're gonna recap them, but you'll also get a chance to win a certification package. In 2025, we are gonna be launching a new certification every single month. The first one is our new Agentic AI certification. That is basically Bot Builder plus lead gen, which we didn't do before. If you took Bot Builder or in the AI Leadership Institute, you get access to all these certifications as part of your membership. Membership. So just know that. But if you're not in the community, you can take these classes one by one. We're offering them through a platform, so you can just join. And then, but that's the first one. We also have a responsible AI, responsible prompt engineer. So this is more about AI safety systems and prompting. We have responsible AI leader, which is more about leading teams, hiring. How do you make sure you set yourself up for success in the creation? AI Solutions Architect, and we have AI Startup Accelerator, which is how to actually start a business. And you do get a certification which will help you if you wanna mentor startups to qualify your expertise. But it's also like your, your certification will be getting clients and that you'll get money and they will give you positive reviews and that will become your proof. But those are the, those are the new programs that we'll be having that we'll be launching every single month. Month in 2025. It'll be a very exciting time to do that. I kind of took away from my own whatever I was doing, so let me just hit back. Oh, there it was. Update. Yeah, so there's 6 days left to enroll. So you do have to enroll by the time it starts in order to be part of it. The cohorts are very serious, and then they might do another cohort later with the recordings, but you do have to sign up to play. So hopefully you'll do that. I really would love to see you there. Oh, is it at night next week? Oh my gosh, good. So yes, check the calendar. Thanks, Roelle. Hold on, let me go to the school calendar and tell you. Oh, it says, it says 12 on this. Yeah, so it doesn't look like it's at night. Yeah, and I'm pretty sure this has been locked in for a while now, but let me just check the other community. Let's hope, for goodness sakes. Oh, good, Marion. Good to have you. I'm gonna just check the AI Leadership Institute calendar and make sure— no, it's at 12. Okay. Phew. So 12 o'clock, 12 to 1 Eastern Standard Time. All of them will be recorded. Oh, you know what, Verle? Actually, there's gonna be an automatic replay at 8 o'clock, but it looks like I'm live, but I'm not. So don't, you know, don't like comment in the chat like, oh my gosh, great to see you, because I'm not there. So the 8:00 PM one is a replay. It does say replay, but some people think I'm actually live because it looks live. So yeah, so there'll be an automatic replay at 8:00 PM to accommodate our Australian peeps. So thank you for saying that. I was like, why would that be? But that's why. So it's automatic. But if you want to be with me live, those are the dates that I always put in the community. Otherwise, you can just sign up on LinkedIn and you'll get access to both replays. So awesome. Great to see you all there. It'll be really fun. We're going to be doing lots of giveaways. You'll have a chance to join the Insiders program. You'll get a chance to earn a free t-shirt. You get a chance to buy a t-shirt. There'll be lots of fun that we do. Um, but most importantly, if you think what we do here is fun and interesting, join the community. If you're already in the community, invite someone you know to join us here. It's free. We do lots of cool stuff. There's a bunch of smart people here that provide really great advice and guidance. Um, it's just, it's one of my favorite things to do every Friday is hang out with you all. So thank you for being here. Thank you so much for spending a little bit of extra time with me today. And if you have any questions, I'm just going to check. Oh, there's another question in the Q&A before we close out. I just want to make sure I address it before we close. Oh, awesome. Yeah, yeah. So the question was how important it is to really learn the concept of agents talking to other agents. And that's how— that's why I had to— I had to really shift away from, just doing bots and just building people, teaching people how to build one single agent. I was like, oh no, no, no. We need to build an agent that talks to another agent. So now I've evolved that. It's called the AI Agent Accelerator, and it teaches you how to actually build systems that talk to each other. And you don't need to learn it from me. You can go learn it online. There's, it is, there is a wealth of people talking about it. I will maybe say that they wouldn't be as fun as me. It won't be as enjoyable, but it is definitely doable. So if you wanna learn right now before we launch that course and that certification, do know that you can search for Agentek workshops. I'd even look for what's called CBB Nation or Chatbot Builders. They're, they're my favorite people talking. They're actually pretty entertaining as well. So you can take some of those courses also. So thank you so much for being here. Thanks for spending your Friday afternoon with me. We'll be meeting again, but next week we meet every single day, and every day I will be offering something new and cool, and hopefully you will appreciate it. But I'm also open to suggestions. So please, if you see something or want us to talk about something, I'm all ears and I try to pivot as fast as I can to accommodate the feedback that I get. I don't sometimes can't pivot immediately, but I do take it all in and I do try to, I listen and I'm coachable. So if there's anything you want to see or want us to focus on, don't hesitate to reach out to me. You can do that on school, on LinkedIn, or just noelle@noelle.ai. So thanks for being here and of always don't forget if you would like link. If you haven't left us a testimonial yet, it's the way that we talk to companies and other individuals and encourage them to join us as part of this community, just to let them know that they're not alone in this journey of learning technology and being overwhelmed. And so your testimonials help us reach more people. And if you're willing to do that, we really appreciate it. So there you'll see the link in the chat. Thank you, Sherilyn. So if you are up for that, it— I'm not, you know, it's not a requirement. It. It's okay if you don't want to, and you're welcome to provide any kind of feedback. I prefer the 5-star kind, but if you have any kind, I am open to it and willing to hear it. And like I said, you all make this place worth hanging out in every week and, you know, delivering and developing content for every week. So I'm really grateful for all of you. Thanks for being here, and I'll see you next week. Bye, everybody.