AI Leadership Q&A wth Noelle

Hi everyone. Good to see you all. Sure. How's it going? Happy Friday. Oh, wait, Monday. Monday feels like Friday. There we go. There we go. There we go. What? Hello, good to see you all. Welcome. I'm actually coming to you live from— I'll go this way— from San Francisco. I am here for a couple of events. One, I'm gonna be talking to some video professionals in Silicon Valley later, early tomorrow morning, and then I will be headed over to AppsFlyer's MAMA conference where I will get a chance to talk about mobile and marketing. So a bunch of people are building apps in the mobile space. It's been growing for, I don't know, almost a decade. And so I will have an opportunity to share with them how AI is likely going to implement, impact the mobile marketplace, but also talk to them about really what a great and unique opportunity that they have. So hey everybody, Welcome. Good to see you all. I always like to make sure that I can see you and you can see me. What's up, Sherry? Amazing. Jessica. Oh, alright. So I got— I was just talking to someone earlier. I don't know how many of you went through this, but about maybe 2 years ago when people first started using like Firefly and Otter to join phone calls, it was kind of weird, right? People were like, ugh, like What is this bot doing? And now, like, bots join. I have just as many bots on my calls as I do humans. But I always say, like, if you're— if you're— if your bot shows up and you don't, then I have to kick your bot out. So no offense, bots, but I will be highlighting and prioritizing my human interactions, human-to-human interactions. But yeah, so super excited to be here in San Francisco. We actually, my husband and I this morning went to a very cute breakfast place called Sweet Maple. I don't know if you all know it. I'm staying in Japantown. So very, very cool little hotel, boutique hotel. So very excited about tomorrow. I will be on stage almost all day from 8:00 in the morning until about 6:00 at night. But, but I'm excited for it. Should be a lot of fun. Welcome to the AI Leadership an I Love AI community call. This is meant to answer questions that you might have about AI, responsible AI, about things that you're hearing. I am doing a lot of— having a lot of conversations across different industries. And so my hope is that with this call, I will share with you some of the things that's working for myself, how I'm staying up to date and separating the signal from the noise. Also though, how I'm helping organizations implement this technology at scale, uh, in companies across different industries from supply chain to retail to, uh, finance to healthcare. I have bridged across, uh, regulated industries, non-regulated industries, even the federal government. So anything that I have a chance to experiment with, work on, productize, I wanna make sure that you all have some insight into what's happening. There's been some really exciting things that have gone on or that have been happening. Hello. Yes, no problem. Thanks, Sherry, for joining. And, and your note-taker is taking good notes, hopefully. So let me know what they look like. I don't know if maybe some action items will come out of it. But great. Nice to have you here as well. So good. So we normally have this call on Friday, but today it's on a Monday because on Friday I was headed to Scottsdale. In Scottsdale, Arizona. And on Saturday, I had a chance to speak with MHI, which is the Material Handling Industries Executive Board. And it was really exciting. I got a chance to, to not only, something I shared with our AI leadership members earlier today was that when I talked to that room, you know, a few years ago when I would come into an organization and talk to them about AI, I'd ask them how, you know, raise your hand if you're using AI right now. And a few years ago, very few people in audience would say that they use it and that they know it. But on Saturday, I'm like, how many of you have used a generative AI tool like ChatGPT? And like 98% of the hands went up. And I feel like the 2% are people who probably would never raise their hand anyway. So I feel like it was a very, very high percentage. And that's something I wanna share with you today is that using AI or, or getting your fingers on ChatGPT isn't the same as really understanding how it can help you. And what we're starting to see, or at least what I'm starting to see in the conversations I'm having, is this shift from going from enthusiasm to focusing on execution, from playground to production. Like, how do we take the excitement that we've all seen grow over the last 2 years and start to really see how we can use that to generate business value. And that is exactly what I focus on and what I've been seeing in the communities that I've had a chance to work with. And so myself, of course, many of you know, but in case you don't, Noelle Russell and I lead a company called the AI Leadership Institute. And over the weekend, I had an opportunity to speak to this, you know, materials handling or executives. And one of the questions that kept coming up was around value? Like, how do we— and some of these questions were very simple, but it encouraged me to think a little bit more broadly about the application of artificial intelligence in these very specific industries like materials handling. Or, I was recently at the Adhesive and Sealant Convention doing a session for the members of that association talking about how AI can be used And right now, all— almost every industry has AI working in it and probably has been for quite some time. The difference is that now we're seeing AI become accessible and available to everyone else in the company, not just the technology teams, not just the research teams, but actually become a tool in the toolbox of, you know, actuaries in insurance or front desk personnel at a medical practice. Or, um, I recently just signed a contract to do a, a workshop with a law firm. So helping lawyers now not re— not require a data science team or an IT team to come in and help them leverage AI, but really teach them how to use AI that's already at their fingertips. A big session we had success with last year was called the Microsoft Copilot Power User Training that we did, and many of the members of our AI Leadership Institute took it. But it was how do you go from just like casually using ChatGPT to make your email sound more professional or help you design, you know, some ideas around it. So maybe an idea that you have for a work project to actually partnering with it every day on all the tasks that you want to delegate to it. And now as Microsoft Copilot has evolved, there's now what's called Copilot Studio. So in our AI Leadership Institute, we created a live learning lab, and we had about 25 people go through this live learning lab learning how to use Microsoft Copilot, but not as an end user. How do I start creating agents that will facilitate end users to use the right model at the right time for the right thing? Now we have an opportunity. That's why I do like the world that we're moving into. Some of you probably have already started to see this term. The new hype term for 2025 is agentic AI. AI and building an agentic system. And to be, well, candid with you, it's not a new topic, right? I've been talking about building AI systems for a long time. It almost always starts with a single AI model doing a single thing. I'll, give you a playback from like way back in the day, you know, 10 years ago when I was working on Amazon Alexa. I remember being really frustrated that I would go into an organization and tell them, hey, We've built a user interface, Alexa, that you could use to talk to anything, do anything with your voice. You can ask it to talk to SAP. You can ask it to talk to Salesforce. You can curate, uh, revenue and insights. You can identify, um, for example, you could survey and then analyze results from your employee survey data, your pulse data. Like, there was so much that you could do with Alexa. And then with a voice interface similar to what we have now with ChatGPT. And the struggle that most organizations had was they couldn't see past that very simple use case. Like, they were narrowly focused on, I wanna build a knowledge agent, right? I wanna build a single bot that can help me find an answer to a certain type of question. While that was helpful, it was a very, very simple tip of the iceberg type of application for Alexa. And most companies never really got beyond that. And the burden was not really on the tech. The technology is available now for you to create these incredibly integrated reasoning AI systems. And you need to know very little, if any, code to build them. What you do, however, need to have is clarity of thought. And so today, I want to share with you a little bit about what I learned from talking to the executives of the material handling organization, but as well as the team that I'll be talking to tomorrow around AI and marketing in the world of mobile app development is that it's less about the tech. Now there's hundreds of apps that can facilitate your use of AI to accomplish a task. But which task is it? What's the most important thing that you can use this AI system to do for you? So what I, I've come to, as a matter of fact, big news. We are just a few weeks away from my launch of the book that I wrote a couple years ago actually. But it's finally coming to market, which I always, you know, find it very cool the way the world works. But my book is coming to market, I think, of course, at the perfect time, though I could never have planned it this way. But it's called Scaling Responsible AI. It's available on Amazon. So you can go just type in Scaling Responsible AI, and it'll pop right up. I also have a website where you can see more about the book. But most importantly, you can find a lot of the frameworks, including the one I'm talking about right now. You can find a lot of the frameworks on our— on the website as well. And it's responsibleaibook.com. Very easy to remember. Responsibleaibook.com. And so you can go there. You can find a link to purchase the book, preorder the book. If you are a member of our AI Leadership Institute, which right now, actually, I think on Monday, Oh, it is Monday. So like at the end of today, we're gonna be increasing the price because we were doing a January, kind of kickoff 2025 special to attract new members to the organization. But now we're, we're, that discount I think ends today. So if you want to jump in, please feel free to do so. But everyone in the AI Leadership Institute will get an autographed signed version from me. And this is gonna be an incredible year. Here. But in this book, right, it's all about enthusiasm to execution. As a matter of fact, I characterize it in a couple of ways, right? One is this concept of day one AI. Now I came from Amazon and some of you are Amazonian people and know how Amazon, how its business and leadership principles are organized around this concept of it's always day one. It's something that when I was going through and meeting with the leadership and when I was in sales kickoffs or in employee all-hands where Jeff Bezos would speak to the members of the company, he would say things like, it's always day one. And what that meant was that you never kind of become complacent in the work that you do. And in AI, this is extremely critical because it, it, it goes, it's in alignment with the way that I build software., right, is that you don't ever just build it and launch it. You fire and forget is kind of how we've learned to characterize it in like app dev, right? That you build the app and you, you know, push it out of the nest. But you don't really need to monitor it or look at it. You just wait for, you know, you collect a bunch of bugs and then you fix them and then you release V2 and V3 and V100, until you rebrand. And so, this is one of the most interesting and exciting parts of the world of AI is the ability for us to actually do that, to, to go from day 1 to— I have actually called it day 2, that there is a day 2, that yes, you don't wanna be complacent, but also on day 2, you're asking different questions, right? Day 1 of AI, the hype cycle last year, everyone was like, what is this? I'd say the top 3 questions executives were asking me was number 1, how do I do this safely so I don't get fired or end up in the news? Number 2, how do I pick the right use case? Like, how do I know what to start with? That was a huge question. And number 3, what about my workforce? Do I train them? Do I retrain them? How do I engage them? Do I lay them off? And regardless of my advice, companies made some bad decisions last year, which I imagine this year they will grow to regret. But Some companies though are realizing, and now more importantly, there is IDC research that supports the number one bottleneck to massive growth and results with artificial intelligence implemented and in production is not the technology stack. It is actually the humans that will adopt that technology to grow the business. And that I think is a huge opportunity for everyone, not only on the call, but who might happen upon this. So great to see you all. Nice to have you here. Hi, Tara. Hi, Sherilyn. Nice to see you as well. And Diane, I haven't seen you in a long time. So glad you're here. And of course, Sherry in your car, so please be safe. Nice to see you all. Okay, awesome. So I'll keep the chat up. If there's a question, let me know. I'm happy to answer the questions as we go through. And Sherilyn, for those of you who are on the call, when you join the I Love AI community, you will have an opportunity to get access to the Zoom so you can ask live questions. You also have an opportunity opportunity to get the links that Sherilyn is putting in the chat for you all. So that is helpful as well. If you are not in the community yet and you're listening or watching this, you can always join it very simply by going to my first name, Noelle, Noelle.ai. And that's an easy way to see all of the community happenings that we have also. So great to see you both. Cheers. So fun. Okay, cool. So That being said, right, the, the day 1 to day 2, I mean, I'm a huge believer in day 1 and kind of harnessing that energy and excitement. You know me, I'm very excited, very zealous. I love the hype. However, hype is only as good as the implementation that follows. And so what we're now seeing is organizations that spent last year answering these questions. What is it? How do I get started? What's the best use case? How do I start seeing ROI? Are now pivoting to, okay, we've done a bunch of POCs, what I call minimum remarkable products. Not everyone built minimum remarkable products. But that's what I, like, day 1 is about building small remarkable use cases, things that people are gonna be like, well, that's pretty cool. I like what you did there. Encouraging you to now go and do more, maybe more complex or harder things. But once you get past day 1 of, like, building what I call baby tiger models, right? Like, excited, cute models that do a specific task. In order for you to move to the next thing, it's really important. The 3 things now that I'm getting asked when someone goes, alright, I already built a POC. Now we're trying to put it into production. Then 3 top most important things that people run into is number 1, security. Understanding how to secure a model once it goes to production or before it goes to production. This requires hardening. It requires understanding of infrastructure. Again, when most people build a model, they build it with their words. They usually will train a model by just using it, using chain-of-thought prompting and responsible prompt engineering. That is great. However, when you actually go to acquire and, and get users to adopt this model in the business or your customers to interact with the model, you're going to need to make sure that that model doesn't have access to everything in your company, doesn't share information it's not allowed to share, doesn't even have access to information it shouldn't have access to. And so that security paradigm is not a lens when you're in baby tiger mode. You don't think about security. You just think, oh my gosh, it's so cute. You don't think, oh my gosh, it's gonna kill me or hurt my customers or, you know, violate terms of my partnership agreements. Like, you don't think about any of that. And all of that are— those are viable concerns that usually show up once you get out of that excitement phase and you start really thinking about how to use this technology. So the first thing is security. The second thing ends up being around latency. Right? So when you start to scale an AI model, most organizations start with a model, like, a generative AI model. Let's call it Azure OpenAI Service. And they want— they then create a prototype, a minimum remarkable product. It's pretty remarkable at answering employee benefits questions. This is awesome. Let's put it into the world. Now you don't wanna be like some of those organizations. I'll just give you something to Google later. Like, an airline recently put out a conversational app for their customers, and that conversational app ended up generating refund policies on the fly that violated some of the profitability policies of the company. So we don't wanna do that. But instead, we want— again, on day 2, you now need to think about how do we onboard users and control how well that model performs against those tasks. So what this often will translate to— many of you know the term RAG, but for those of you who don't know it or are new to it, it stands for Retrieval Augmented Generation. And it does a couple things. It's an architecture. It's not software. It's not a thing. It's a way of thinking. It's an architecture, a design pattern that you can use with the large language model to reduce the dependency that model has on the data it was trained on, which is the internet's data, which is contradictory and often causes hallucinations. So we wanna reduce the dependency of the model to find answers in its training data and increase the dependency of that model to find answers in data sources that you provided as the developer or as the builder. Right? So RAG gives you an architecture that says, do not use your training data to answer this question. Use this data source instead. Retrieval augmented generation. We're going to augment the training data you have with a new data set that you're going to use to answer those questions. This allows you, if you are very careful, specific, and persistent, allows you to create models that don't lie. You can finely tune a model. You can shrink the number of parameters that model is trained on to actually create a model that does not hallucinate. It is possible. However, most people don't even know that it's a question they should ask until it starts hallucinating like that model did when it was hallucinating refund policies to make— to satisfy customers. So that the first one, security, nobody really thinks about that on day 1 or on day 2, right, when we're starting to move to production. The next one's latency. Like, nobody really thinks about how the model interacts with the user. So to talk more specifically about the latency question is to think about, imagine, right, your, your end users are trying to request anything. Let's say they're asking for employee benefit information. Now chances are, this is something I learned actually during my time at IBM. I became an expert in a product known as IBM WebSphere. And when I say an expert, it means I knew every stack trace that that application server could throw. Every single one. If I saw it, it was like The Matrix. I could instantly be like, I know exactly where that is. I can go— I mean, it's what makes a good consultant. Right? It's not what, you know, it's, it's that I can go into a problem and it's not that I know to use a hammer. It's that I know exactly where the hammer needs to hit in order to get it working again. And so similarly, though, there is a— there's a finite list of stack traces that that application server could throw. There were thousands of them, but it's a list. Like, there is no more or less than, like, there is a list of the number of problems you could have with that product. And so once you know that, I, it actually became easy. I didn't have to go research anything or test anything. Like, once I knew all 1,000, and it was probably more, but once I know 1,000 of these problems, it's fine. Same thing happens with your employees, your customers. Eventually, you will understand through customer intelligence, which is one of the biggest use cases to get ROI in, in, in artificial intelligence, Once you get those, once you understand, like, what your customers are asking, you actually don't have to hit an AI model directly. You can reduce, and that's why I call latency, right? You can create a middleware. There's actually a company that coined a term called cognitive caching, and I love it. It's not actually a product, though they have a product that does it. But for those of us that have a history in software engineering and, and product or products and APIs like Hibernate and these, like, middleware products, these pieces of software that sit— would sit between software, perform caching capabilities, you know, second-level caching. We have the same ability to do this between language models. And again, this doesn't become interesting or even worth asking about until you've passed that kind of exciting, enthusiastic, zealous hype cycle phase, and you get into, like, how does this actually work? And how do I do it at scale? At scale, you're not gonna wanna hit the model for every single request. You're gonna wanna go, oh, this is one of 1,000 problems I already know the answer to. And then you're only hitting the model and getting that reasoning for the more sophisticated, complex questions that are being asked. Now in contact center solutions, which is one of my areas of expertise, it's where I've deployed probably the most AI models, is in contact center or customer service or customer success. Use cases. In those use cases, it's extremely valuable to have a model because many times I'm using a model or I'm using an architecture that has an LLM in it to respond back to a real-time interpretation of a call that's happening. Right? It's not post-call analytics where I'm looking at transcripts of calls that have already happened. I'm listening in real time to conversations that are happening on the phone. And when I do that and I listen to those calls, I want an AI model to give me real-time feedback. Oh, the client just, or the customer who you were talking to just mentioned this drug. So it goes and searches that drug and provides me a very, very concise explanation that triggers my understanding of that so that I can elevate that conversation with the customer or the patient in this case. Right? So it gives you an opportunity to, now when I know that customers— so here's another fun model that I built, as well, I built many versions of this, but it's called global trend watching, right? Where you actually build a model and its job is to just to detect patterns in real-time communication. So you're looking for— you could do it like on— yeah, I totally agree with myself. You could look at like Twitter as an example or X. You could look at Instagram. Like when you start to see things trending, It's really important that you can identify that trend and either hop on it or comment on it or contribute to it, right? Well, in contact center, oftentimes when calls start coming in, especially let's say in airlines, when airline tra— like calls start coming into the contact center, it's usually based on some kind of trend, whether it's a, a regular operations trend that's happening, whether it's, you know, maybe there's weather in a certain part of the country. And so this volume starts coming in and people are asking the same question. Without a model to actually listen to those calls and go, hey, everyone's talking about weather in Detroit. Like, without that, you have to just— like, every agent is, is treating that conversation as if it's the first time they've heard it. When we're talking about latency, these are some of the things you can do to say, oh, we're, we're seeing this huge trend, global trend analyzing, right? Huge trend in Detroit weather. I now use that AI model to identify what the trend is. To write a synopsis. And then I can surface that to everyone in the call center. 75,000 agents now all get to know before they even hop on a call with someone that this is something that's trending. And we get to say something super intuitive and very familiar with the user who's calling and say, hey, I know we're having, you know, some issues in Detroit with weather. Is that why you're calling? And 90% of the time, right, or 80, you'll know that percentage based on the analytics. That's the problem. But it allows people to feel like, you know, before they have to explain it to you. And that context is exactly what AI models like LLMs are really great at. So that gives us, so exciting. It, it's such an exciting time because that transition, right, from, oh my gosh, this is a cool model that does a cool thing to how do I actually use this model to generate business value? It requires us to think differently. And it, it also bubbles up new questions. So security is one. Latency or increasing the efficiency of model invocations, that's another. And then the third one that often companies run into is cost, is tokenization. And so I always like to refer to my friends at Stanford. They have a really fun project. It's actually on GitHub for those of you who are GitHub people called FrugalGPT. And it's just an example, again, of an architecture similar to RAG that you could implement where an LLM is educated on a bunch of different models that might solve a problem. And it allows you to investigate an incoming request and analyze it against what the cheapest— frugal— most frugal, effective, and accurate model would be available to solve that problem. And when you have this layer of middleware kind of in the middle, right, this LLM in the middle, it allows you to make sure that you're leveraging the right model for the right task and not hard coding those together. So that kind of enters us into the state of the world we're in now with agentic AI, right? It's the new buzz term of 2025. So with agentic AI, it gives us a chance to talk about, like, how are these systems becoming agents? And what that means is, are they autonomous? That's usually the distinction between building an AI bot, right? A chatbot. Many companies have chatbots on their website that are not autonomous and not agents. Right? They are scripted. They are trained to answer very specific questions. As a matter of fact, we have an autonomous or a, a bot builder workshop that is available that people can take as a step 1. First step, learn to build a single system. Single AI system. Second step, learn to choreograph those systems together. Agentic AI is all about the choreography. How do we create agency in single AI systems to connect with other AI systems on the fly? And you'll see lots and lots of companies coming up with and delivering agentic systems, one of which of course is Microsoft. They led the pack last year with their launch of the Assistance API, which was an agentic AI platform. And you can still go look at it, Assistance API. There are lots of great tutorials that are available there that you can look at and research. It is more for the coder. So it's an API. So that would require you to actually use it. Sorry, there's a tiny lag. So if it lags, I apologize, but I'm in San Francisco, so we'll take it. So those, those allow us to actually see the different labs and, and tutorials that are available. And I like the Assistance API because, again, it'll teach you the difference between chatbot that just answers questions and autonomous agents that actually do the work. So that, I think, is a really exciting opportunity. Now let's talk a little bit about some of the news. And oh, I see a question. So the, the best thing, I'm so I'm glad you're asking. So someone's asking about our AI Leadership Institute live learning labs. So in the AI Leadership Institute, we do live learning labs. We do different ones every month. In December, we actually did 3 because I was super excited about some of the tech that was coming out. But we do at least once a month. We do these. We also offer an AI Leadership Lab platform, which includes 15 right now. I create new models all the time, but 15 models to take you through if you are a small business owner, an entrepreneur, or just someone who's looking to build, like, get experience, fingers on keyboard, building with these systems, I created a 6 Ps of successful small business, small business operations. And so it takes you from, like, building a website to creating email copy, like, some of the most common use cases. And if you're not familiar, the IDC actually released a report at the end of 2024 They also did one at the end of 2023. So it's cool to look at both of them. But the most recent one identified a couple things that I think is super important for you all to, to know. And you could just type in IDC. I think Microsoft was part of that study. So maybe IDC Microsoft or IDC Business Opportunity of AI. That's, I think, the name of the report. It's also available inside our community. I posted the links in the I Love AI community as well. But when you look at the business opportunity of AI based on this IDC research that was done, what they uncovered was that there are certain use cases that generate a 3x return on your AI investment consistently, 2 years running. But more interesting than that, we are actually seeing a 10x, 10.7x return on even more specific use cases. And those use cases tend to be in the area of, well, 4 different areas. One is customer intelligence. We talked about that already. Right? How do you use what you know about customers to, to actually accelerate their ability to do a thing, whether it's buy from you, get help, get support. The second one is internal process bots. So this is— someone had mentioned to me, is this RPA on steroids? Kind of. And I, I like the, the, the analogy actually because steroids could kill you. And, so can RPA with AI if you're not very careful. Right? And I don't mean you personally, but it can definitely hurt your business, hurt your customers. So you have to be very careful with it. But it is an accelerated form of robotic process automation, specifically leveraging AI. Now the dream of RPA was that you would be able to automate these processes and create autonomous reasoning between processes. That was very hard to achieve 5, 10 years ago. So we built the, the bones of that. I remember becoming an expert in a language called BPAL, or Business Process Orchestration Language. And we were writing the code that would tie these systems together. And while it was very successful time in my career building these integrations, it was also a lot of plumbing that I was writing and I was hard coding into these architectures that now I don't have to do. So even even though the activity or maybe the result is the same, the way we orchestrate systems to talk to each other, it's very, very different. It's one of my favorite things that came out of the LLM kind of excitement is that we now have the ability to never have to really write data transformations ever again. Like, I never have to write that code. I never have to, like, concatenate in logic information from two columns into one. Like, I never have do any of that because a large language model can do a— I mean, I've asked a large language model to do extremely complex million, uh, record deep transformations, uh, and it's been able to do it successfully. So obviously, it's not an overnight success. It's something you have to work on. But I do like that it's not nearly as brittle as writing hard-coded logic into an application that says, this is how this system is gonna talk to another system. LLMs are creating a really great opportunity for us to let the language model do that disambiguation and mapping. And, I met— some of you know schema mapping is, well, it's not the most fun work to do in the world. So if I, as an architect, can leverage an LLM to do that mapping so I can focus on higher value tasks like strategy and implementation, that is certainly the way that I'll go. But yeah. So we do in the AI Leadership Institute every single month, do different types of execution labs. So thank you for asking. That's amazing. But yeah, we do all sorts of things. So this month, we actually did our— for January, our live learning lab was all on building a digital twin of yourself. And we used Prafina, P-R-A-F-I-N-I, Prafina as one of those tools. And I also suggested people go and try out the digital avatar, which is actually a visual, representation in HeyGen, another platform that I use all the time. So we spent 2 hours together learning how to create a responsible, scalable digital representation of ourselves to help scale our brand. And that was pretty fun. Next month, which is a great segue. Thank you for that question. Next month we actually have a really fun, exciting time, which is the week of Valentine's Day, which Valentine's Day is on a Friday. Yay, date night. Valentine's Day is on a Friday. But what we have is the opportunity for us to scale what I have done in these little live learnings every single month. We're gonna actually do an I Love AI challenge. And during this week, you're going to get a chance to actually take over 5 days our Responsible AI Leadership certification. So by the end of the week, if you go to all 5 sessions and do your homework in all 5 sessions, you will have an opportunity to obtain from the AI Leadership Institute your Responsible AI Leadership certification. And it'll include, of course, my book, my new book that's coming out. So you'll get a copy of my book as part of that. It'll be really, really fun. HeyGen is actually spelled H-E-Y-G-E-N, like generative AI. So hey, GENitive, H-E-N or H-E-Y-G-E-N. Awesome. Okay. Okay, awesome. So hopefully that's helpful. So yeah, we're gonna have a lot of different things happening. I also, just to announce, I'm going to be delivering a class as part of the I Love AI week. And I'm also in this class is on Maven. It's actually— we discounted it intentionally for anyone who wants to get a role in AI, in the world of AI. 5 of us have come together, all of us getting into AI from different avenues with different backgrounds. And it's literally called How to Get into AI on Maven. But if you use my QR code, I can give you a 50% discount on the course. And so I'll throw that into the chat. Hopefully, I can find it. I don't know if I gave it to Sherilyn. And I wonder, yeah, I'll have to, I'll provide that as a link also in the community so you all can get access to it. It's not my course. I'm just one of the featured speakers, but I will be a featured speaker on Thursday, which is the 13th, as part of our I Love AI Challenge. And I'll be talking about my journey into tech, how I broke in and what made what I did something that you could also do to get to gain some of the same results. So hopefully that would be, oh yeah, thank you. Yeah, there's a— yeah, there is also a link on my Bitly page, Sherilyn, if you want to share the Noelle AI Bitly link. Yeah, let's see here. So another question came in about Maven. Do you think— yeah, so you had asked me this question separately. I think I saw in the community as well. So TopMate and Maven are very different platforms. So TopMate is meant for you to publish yourself as an expert that someone could connect with individually. They do host the ability for you to, like, run webinars and sell digital products. But their focus is on someone searching for, I need some help with starting a business, and they would find you as an individual and schedule time with you. So one-on-ones. Texting, that type of thing. Maven is just a course platform for creating courses. So if you want to deliver a course, then yes, I think Maven's great. They, you know, obviously will take a percentage of your— of the revenue. But it's— they're very different platforms. So Maven, the nice thing about Maven and what I've told others as well is I really like Maven because they are training the world of engineers people on LinkedIn that you have to pay for this education. Like, so many times, you know, we— I certainly, when I was earlier in my career, I kind of looked for ways to learn everything for free and didn't realize that there was value in actually learning from people who had done it. Not just people who talked about it, but people who had real experience. And so I do find that Maven is a great platform for finding people like myself that actually have real experience that they're willing to share with you. Relatively reasonable price. So I think that is a great opportunity. Their audience is much more of the professional audience as opposed to, you know, like Udacity or Coursera, which is more a much broader audience. But yeah, I really like what they're doing at Maven. So definitely take a look at it. Let's see. What is— what link did you just share, Sherilyn? I don't know what that is. Anyway, so yes, and, and I also decided to use a platform called Intro, and I'll just share this with you so you can see kind of how I've evolved my own use of these platforms. But one of them is— I'll just throw it in really quick— intro.co/noellerussell. I think that's it. I hope I type that in correctly. There we go. There you go. So if you click on that, you will be able to see my intro profile. And again, these are different audiences. So one of the things I, I talked to our AI Leadership Institute community about is the importance of really understanding your audience and knowing who you're talking to and making sure that you're creating relevant content for that audience. As opposed to trying to be everything to everyone, which is kind of what we saw this last year. Like everyone who was doing anything slapped the tag of AI on it and said, called themselves an AI expert. What we're seeing now in 2025 is a real disambiguation between kind of a separation between doers and talkers, and rightfully so. Like it, it depends on what your needs are, right? Some people only wanna kind of hear like anecdotally or they wanna hear kind of metaphorically about artificial intelligence. But there are some people who actually wanna hear from people who have deployed it in production. One of my friends, Susan Slye, she's one of the only executives I know that has actually deployed computer vision in production. I mean, there's a handful of them in the world that do what she does. And so she's also one of those people. She started a company recently. But you can go and take a look at her and her profile and she's on LinkedIn. To see kind of what she has done as well. But doers are— we're coming into kind of a new era, this day 2 concept of like, how do we actually use this technology successfully? And unfortunately, people who just talk about it and have never built it are gonna struggle in that transition if they don't actually start building with what they're talking about. And so no one platform will get give you business opportunity, right? It's what you do with it. I always say, like, tech is just a tool. AI is just a tool. Models are just a tool. Like, it requires a strategy. But here's the good news. Like, leveraging AI, you have an opportunity, and I will share with you a couple tips that I use for myself, like, to make sure that I'm constantly focused on the most relevant use of my time to grow my organization. The first thing I do is I make sure, as I just mentioned, that I know my audience, right? That I know who I'm talking to. So figure out, pick a lane is what I often say. I say it on stage, like, the world is not looking for people who understand AI. They're looking for people who have leveraged AI to do something very specific, to, to, you know, perform actuary tasks, to manage an emergency room, to run agile projects. Like, what is your thing and how can you use AI to do that better, faster, stronger? That is the direction that people are going to be moving in in 2025, and that's why we're calling it agentic AI. Your, your The reason why we have, you know, our community is all about AI leadership and responsible AI leadership is that over the next 3 years, you're going to start managing. And some of you may not ever, like, have only ever been an independent contributor. But even as an IC, an independent contributor in a company, you are now gonna be leveraging machines to do some of your work. So being a leader, having clarity of thought, being very clear about the deliverables, the metrics for success, like, all these things that leaders need to do. Everyone now is a leader in the world of AI because you're gonna need that same level of clarity that you would use with a human who works for you. You're gonna actually need even more clarity for a machine that's doing that work for you. And more oversight, more coaching, more advice, advisory services, right? These machines are going to require additional levels of leadership skills. So if you thought you could get away with, like, kinda not really focusing on the leadership part of your career. I think we're actually seeing, you know, those days kind of come to an end from that perspective. All right. I know. That's why I got it. Yeah, you're right. I should put I heart AI or like AI coach right on the top. As a matter of fact, I was just at a store and I bought cute little AI embroidery letters. So yeah, you're right. You're 100% right. That's why I bought this shirt because I'm like, I'm a coach. So I appreciate that, Sherry. You're 100% right. I thought the same thing. So Diane asked a question about companies. Do, do you see companies standing up a few simple deliverable tasks to be a viable business? It depends on the task, right? Like if the task is generating leads and closing sales, 100%. I think there is a whole market of companies. There's Millions of businesses in the $1 million to $100 million range, millions of companies in this gap, all of them. Well, I shouldn't say that. There is a very, like, 86% of them have less than 12 employees, and 50% of them have less than 3 employees, which means everyone's working really hard to generate the results that they're getting. And people is their bottleneck. Hiring the right people, getting people to stay, getting people to work for the amount of money that they're willing and able to pay. Like, there's a huge challenge there. So helping these organizations maximize the humans they have in order to grow their, their business with sales, marketing. I mean, those are, those are two. So yeah, it, you absolutely can grow a business with just a few very specific tasks. But what tasks are there? As a matter of fact, you can build a multi-six-figure business on one task alone if it's the right task for the right customer, right? Like, you need to get into the mind of the customer and find out what's the biggest thing they're worried about, what's the biggest thing they struggle with. And if you can build with AI as your copilot, if you can build a solution to that problem, you will not go hungry, right? The challenge most of us have, have is that it's very hard. I mean, I struggle with it too, picking that audience and only going after that audience consistently, like forever, like that, that becomes your market. And that's why I think it's really good to have that clarity of, of thought around what you do well, what is your superpower? Or a, you know, pick a superpower and let that be kind of your guide of how, how do I use AI to complement this task? Now let's say you don't care, like business investors like myself. I buy certain businesses and I don't actually care what they do. Dry cleaner, Taco Bell, KFC, like I don't really care what they do. Apartments. I will invest my money as long as I know that I'm gonna get a return on that investment. I don't really care that much. So if you are able to identify an audience with a problem that you know they have, and a product or service that solves for that problem, your ROI will be like— your ROI is evident. You're gonna get a return on that investment. Then you can hire people or complement your team with machines to actually deliver on the promises that you're making in those, those sales conversations. Yeah, 100%. Thank you, Sherry. I'm like, have we talked? Yeah. So Sherry said, am I getting hired to create plans for companies that are— yeah, my getting hired to create plans for organizations or what I like to call strategies, right? Building AI strategies to implement right now to save money, or I actually see them asking me to come in and do what we're calling operational efficiency, right? Especially in some of these older industries, like, I don't wanna say older, but maybe more, less internet-based. We'll call it that way. Right? So distribution, manufacturing, not like retail, retail banking, customer service, like, these types of use cases are much more easy to identify use cases for and build for. But these other industries, like supply chain, distribution, Automotive. Right? And I'm— right now, most of the events I'm going to are talking to these kind of more traditional businesses that have been around for over 100 years that have process after process that is not digitized. They are looking for ways to, number one, use AI immediately to see results. But they're also very concerned about reducing costs and creating operational efficiency. And a company that has 700,000 employees, How do we do more with those employees? I don't— they don't necessarily want to lay off those employees, right? They don't want to create efficiencies to lay off staff. They want to create efficiencies to grow more revenue. Thank goodness. Like, that is the direction I think every company should be going in. Oh yeah. So speaking to the materials handling organization, they also were very operational efficiency focused. But the number one use case from like multiple people pulled me aside and said, so, so wait, can I build a system? You're gonna love this, Sherry. Can I build a system on my website that would interact with people who come to my website and get their leads so my sales team can follow up with them? Like that, that was like, that would be amazing if we could do that. And I'll enter Bot Builder. I'm like, well, yeah, we can do that right now, actually. It's a good, it's a good entry point, what we call in kind of the sales world, tip of the spear, right? It's a tip of the spear that says, LaserPoint can solve this problem for you. I can generate more leads, which ultimately generates more revenue. But that's not where the big opportunity is in the company, right? The big opportunity is doing what I call an AI impact assessment, right? Once I get, I give them this, like, candy, Here's your leads. Likely, they don't have a good sales strategy following up with those leads. So AI can help train their sales team to be more effective with those leads once they have them. But more importantly, I can say, let's just look at how you do business. Let's look at how you— once a customer— once you get a customer, how do you onboard them? Right? You can use generative models to create entire customer onboarding sequences, which most companies never have time to build, right? Like, they literally— you buy the product and that's it. You're kind of on your own to kind of figure out how to use it, right? Training, documentation, like all of these things that have been just kind of like the bane of existence to most companies in this $1 million to $100 million range, they don't got time for that. They're trying to close business and deliver product, taking care of customers, retaining customers, improving customer service, increasing customer success. That has unfortunately become deprioritized in their effort to grow more, get more new customers. So they're losing customers as fast as they're gaining them. And it's literally a rat race, like a hamster wheel, because customers aren't staying, so they have to go get new ones. And so AI allows them to actually, like, someone like me or you can go in, build them a strategy for implementing AI technologies that'll fill out the gaps in their organization already. That's why we call it an impact assessment. We're not gonna come in and tell you what model you need. We're gonna look at how you do your business, and then we're going to say, oh, well, right here's a big hole, and over here's a big hole, and over here's a big hole. So let's use AI to fill those holes. We're not impacting anything. Keep doing what you're doing. We're just gonna add AI as an augmentation of your already working, you know, business process or workflow. And when you augment it now, the same thing they were doing, that baseline of productivity, automatically when you add these little gap fillers with AI models, their productivity shoots up. And that's enough to now get you to start talking about much more real, complex, elaborate use cases with them. So I think that is a, a really big part. Bot Builder is different from RAG because RAG is an architecture, Bot Builder is a workshop. And so So we talk a little bit about RAG in, in the Bot Builder workshop, but that is in the, the AI Leadership Institute. You were asking about these execution workshops. That's one of them. We actually will be deploying it on demand so people can just get it on demand and get Bot Builder certified. But it is stage 1. Like, the first thing before you build an agent, you need to understand how to build a safe, responsible AI system, just a single one. So we build one system. And then we tee you up for connecting that system to other agents. And that's when you can take, like, the AI agent workshop, which we just started in January, and we've seen a huge amount of interest in because Agentic AI is, you know, is the talk of the town at Davos, talk of the town at CES. But I always say, like, it's very exciting, but it's, again, it's another baby tiger. Like, people are super excited about it. But if you don't know what you're doing, if you haven't built a safe, responsible system for one application, you do not want to blow that up and connect 50 of those things together. Right? You need to really understand what does it mean to build a human AI experience, to build responsible— another certification we have coming out this quarter is the Responsible AI Prompt Engineer. Right? How do you prompt responsibly? There is guidance for that. There are principles behind it. You don't just, Not everyone can talk to AI and get the same result or even get a positive result. So there is a method to the madness, and becoming a responsible prompt engineer will allow you to create and harden models for production. And then, of course, once you've got that done, now what about model selection, right? How do you know you're picking the right model for the right task? Do you even know that there's a difference? How are you evaluating the success of those models? And then finally, Infrastructure, right? Of course, I would be, you know, remiss if I did not mention one of the biggest news stories of the week, which was the announcement of Stargate, right? So OpenAI has partnered with a bunch of other companies, Microsoft, NVIDIA, to, you know, grow. Sam Altman mentioned this, like, I don't know, 18 months ago that he was gonna build a half of a trillion dollar infrastructure, and everyone kinda laughed. I was like, you're crazy. Like, oh no. I don't know if that's gonna happen. Turns out it's happening. And now they're going to build these massive centers. And it's gonna require all sorts of things from architects to engineers to— that have nothing to do, you know, AI infrastructure is less about— it's nothing really about LLMs. It's all about cloud architecture. My, my stomping grounds, right, where I grew up. So building cloud infrastructure, building well-architected infrastructures like elastic scaling, you know, water-based cooling, like all of these things that are required. So I do, I do agree to some, some level that it will produce a lot of new jobs, this, you know, half a trillion-dollar investment in AI infrastructure to support the reasoning engines that will be coming to market. However, I also recognize the massive digital divide that's being created in these roles. These roles are not junior roles, right? These are AI engineering, applied engineering, principal architects, infrastructure architects, network architects. Like, these are not like— you're not gonna get out of high school or college and go get a role like this unless you went to school for mechanical engineering or maybe even, you know, systems architecture or systems engineering. But either way, those are, you know, not everybody does that. So I am very, very focused and interested in growing the base of understanding and education in high schools in underrepresented areas, right? Like, not just schools, not just cities that are feeder schools to Ivy League colleges or even universities. Like, what about everyone else? Everyone should understand how to use this technology to make whatever they do more effective. It is truer now than ever that you will not, you know, be replaced by AI, but you will be replaced by someone who uses AI. And now it's a bit unfair if you go to school and they've never even taught you AI by the time you leave. Like, that's not okay. So as a result, I ended up creating a nonprofit called the AI Leadership Council. And the members of this council sponsor the development of these programs for K through 12 or schools as well as community colleges all across the United States. And I'm very passionate about it because, like I said, if, if we don't teach them, no one will. If we don't go in there and provide a course, a program for them to educate 4th graders, 8th graders, 12th graders on how to use AI just conversationally in their daily life so that that daily life usage grows into how do they use it every day on the job, they're gonna be at a disadvantage when they start interviewing for roles and as they leave the education institutions that they're a part of. Yeah, so another question came in around leveraging AI to do surveys. So one thing that I like to do, and this is similar to— it's an agent. It can be considered agentic because there's multiple models involved. But it is in its simplest form. You can use Claude, for example, to create an artifact for you. And that artifact could become a widget that you can host on your website. It can collect data and store data in a Google spreadsheet. And so that integration will happen and you will— don't have to write any code at all. You do have to copy and paste the code into your website. It doesn't do that for you. You do have to test it. You do have to make sure it works and that it's collecting data correctly. But the infrastructure is gonna be be completely built for you in a tool like Claude. And so Claude, you could just ask it like, hey, I'd like to do a survey. You give it the questions. You can even give it the audience and the problem you're trying to solve and ask it to generate questions for you. But you can create that survey. I've done this for companies who wanted to create an employee pulse survey. And so we used, you know, knowledge from previous pulse surveys to as, as a, like, as data, grounding data to train a model on how to build custom questions for specific groups within the company. But most importantly, aggregate the result set into like a Google Sheet, but then use that Google Sheet as grounding data to a model that could actually analyze that data and provide feedback. So your ability to go end to end from ideation to execution, right? Like, you can now do that in the span of a day, really. And then move into the rest of your time, the next 2 weeks into fine-tuning and training and advising that model on how to do a better job getting what you want. But even so, you're now with a fully functioning online survey that you didn't write any code for, you know, in less than a week. But yeah, that's the product I would recommend. But you can also do— there's another tool called Gamma, G-A-M-M-A. You're of course welcome to use any, any of these tools. There's lots of them available. But these, these are artifact development tools that actually build things for you. Let's see. Oh, yeah. Okay. I think I did it. I answered that question. Let's see. Oh, let me go into our Q&A section. Yes. So the— one of the questions was the I Love AI Challenge, which is the week of the 10th through the 14th. Is that open to everyone? One. It is open to everyone. There is a charge to be part of the certification. So there's a fee associated with it, but you get a free book out of it. You get a t-shirt that says, I love AI. So it's, it's worth it. You'll get massive, massive value, thousands of dollars of value for less than $100. So hopefully you won't find that to be inhibiting. But we also have, like I said, if you can't do that, there's lots of other opportunities to take little pieces and you'll be able to watch the videos or attend the webinars for free. But if you want the certification, if you want to be able to actually, you know, get something out of that week of spending this time learning other than just watching and being part of the calls, you will have an opportunity to do that as well. But they are all— yeah, our leadership members, because they're already members in the Leadership Institute. Today's the last day that it's at $47 a month. So if you're interested, that would be great. Like, you should do it now because it's about to double in price. And then it'll double, it'll go up again because we are launching an AI model platform. So it has 15 AI models inside this platform. It's called AI Leadership Labs. There's 15 models from building all the assets to create a company to doing marketing, sales, content creation, social media content, social media management. So all of these different models. About 15 of them. And you get that for free as part of being a member. So it's a great time to join. But as a result, that by itself is about $500 a month, but I'm giving it to free to everyone who joins the AI Leadership Institute from now on. And so that is a huge value. So we're gonna increase the price of the Leadership Institute to at least be comparable because you're gonna get a, a whole collection of private models that I built for you. But yeah, so that's available to you. Feel free to check it out. You can go to the AI Leadership— actually, if you're in the I Love AI community, you can just click on the little links on the, on the right-hand side of your screen in the community. You'll see a link that takes you right to AI LI, and you can see the video that I did around that topic. Oh yeah, just earlier I did a video. I think I pinned it in the community about that as well. Let's see. We all sorts of things. So another question about what can you do with these widgets? If you look at— did I give you all my Bitly? I don't know if I did. So let me— I'll just type it in again. If I did, I'll just repeat myself. But it's basically bit.ly/m/noelleai. And so if you go to this website that you'll see in the chat, you'll see all of my lead magnets. And when you go in, if it says calculator or blueprint or risk assessment, any of those things are all artifacts I built. Granted, the model didn't build it by itself. Right? I gave it, just like I mentioned earlier, I gave it my audience. I gave it the problem I was trying to solve. I gave it my frameworks. I have 2 frameworks I built that are in my new book. One of them is called LEAD AI. And, yeah, LEAD AI is awesome. LEAD AI, and what, which stands for Leadership Ethical Foundations, AI Governance and Policy. And the D stands for design with inclusivity. So that's kind of the framework that I use to build responsible AI at scale. And then there's also another framework that I have called Secure AI, which is all about securing your infrastructure. Now, that being said, these two frameworks, you know, I created the framework, the ideas are mine, but I wanted to create really cool ways of demonstrating that and giving it as a tangible piece, asset that someone could use, whether it's a checklist or blueprint or a quiz. So every— if you go to my website or you use, you know, if you play around with these lead magnets, you'll see all of them take you to a widget that I built using Claude and that I embedded on my website. I try to, in practice, use the things that I'm talking to you about very successfully, mind you, but use the things I'm talking about and see that it's an example. You know, I hope to share it as an example that you can use to grow your own business or help you get some of your ideas out into the world. Alright, awesome. Cool. So, oh my gosh, it's already done. Time. What? It's— oh my goodness. So let me see. Okay, I think we answered all the questions. Thank you all for being here. Any final questions? So just so you know, I am today in San Francisco talking to the MAMA conference, which is AppsFlyer's mobile marketing conference. Super excited to do that. Next week I'll be in Vegas. So if you're in Vegas, come say hi. And then the week after that, I will be talking to a legal community about using AI in legal— in the legal office. And so I will be productizing a platform for that. And we're also going to be productizing a platform for event planners. So just think about, like, whatever your specialty is. So I'm partnering with these organizations that want to help all legal offices. Right? It's not enough for me to go and say, I can teach you about responsible AI. They really want me to say I can teach them responsible AI through the lens of a legal practitioner, through the lens of a medical practice, through the lens of a chiropractor. And so the more specific you can be, the easier it is for you to go and help those organizations with their specific problem. And now you can build models that know those problems. You can ask for questions. You can do research and study what kind of problems do chiropractic offices have that AI can solve, and then make a decision on whether or not you wanna be the one to help them solve it. So super exciting. Thank you all for being here. Thanks for all the amazing question. It goes to Firefox. Because you have Firefox. It actually is just a URL, so you can open it up in Safari, Chrome, or Arc, which is the one that I use. So, so I will keep an eye in the community. I'll be sending out a note announcing the funnel, the, the webpage you can go to sign up to be part of the challenge. Super excited. The $47 a month is actually for the AI Leadership Institute. So you can go, I'll maybe Sherilyn can add that link into the chat for you all. The AI Leadership Institute is actually our other school. So we have school, the free school community, which is where you all are now. And then we have a paid school community. And again, the re— you don't just— it's not just to join so you can pay for another subscription. The idea is that if you wanna put your fingers on a keyboard and actually build stuff, if you wanna get access to an AI platform, if you wanna get access to building chatbots for customers. You get access to this software through that subscription. So it costs me to deliver it. So I put a very nominal price on being able to, kind of cut the cost of ownership for me to extend it to you all. But most importantly, you get me. So if you start using these tools and you get stuck or you don't have help, I, I have a whole separate set of calls that we do with our Leadership Institute members to make sure that you're using this technology safely, responsibly, and most importantly, profitably. So it's been my pleasure to be with you today. I hope you enjoyed the session. We'll be right back here on Friday. So just 4 and a half short days away. We'll be back here on Friday, to talk more about the very interesting events that are happening in AI and give you a sense of some of the things that are working and not working. And of course, give you an update on the events that I had this week. Thanks for being here. Totally. Yes. Very, very helpful. Yeah, you could see. Oh, yeah, of course. If you have any feedback and want to share a quick testimonial, Sherilyn just put that in the chat and you can just go to noelle.ai. That's the website. And that'll give you access to both the free community and the paid community so you can take a look at what's offered. But thanks for being here. It was fun. Nice talking with you. And your questions were awesome. So I can't wait to see you at the end of this week for another Q&A session. Talk to you soon. Bye, everybody.