I LOVE AI With Noelle

We'll see, we'll see, we'll see. My door's open. Hey everyone, it is that time, Friday. If you guys want to head out and shut the door, that'd be awesome. Um, I'll be done in a little bit. Uh, so nice to see you all. Thank you for being here. Welcome, happy Friday. It's, you know, that time, that time of the week. We're gonna do some fun. I have a few note-takers in the house, so thank you for being here. AI Systems, nice to have you. So this is our call every week, 4 o'clock. Maybe we'll change it, but for right now, it's every week at 4 o'clock, and we chat about the latest and greatest that's going on in the world of AI today. We've got some interesting— I don't know that we're gonna talk about it in our breakout sessions today, but we do have some interesting news that has come up. Recently, some really cool stuff is happening, of course, around the world and in AI. So we'll be talking about that, and then we'll be breaking, as we always do at the end of our calls, or at least as I try to do at the end of every call, we spend just a few minutes getting to know each other. We break out into small groups, and we have a chance to just understand what each other is working on, and most importantly, share either what we're learning or learn from those who are building. And that is kind of my favorite part of the whole session. So I am going to first go over here. So good to see you all this week. Um, I've had a lot of fun over the last couple weeks. I just got back from Sydney, Australia, which I talked to y'all on Monday, I think. Oh no, I don't think I did. Um, so I haven't talked to y'all since I was in Sydney, uh, but I took a ton of videos, uh, so you'll be seeing my Sydney escapades, uh, in videos on Instagram, LinkedIn, and YouTube over the next weeks because I took so many, so many videos. I think to the point where my husband was like, um, I feel like we're good, I feel like we've captured the moment. Like, no, there's always more. And, uh, it's funny about, uh, building AI systems. Like, once you start, it's very— when you start building like AI content, everything you look at, it's like called the reticular activation system. Everything you look at, you're like, I could talk about AI about that. So, um, like today I was on my way to pick up my, uh, my child from school. And I pick them up and they walk out and we walk home, and they did not come out when they were supposed to. And then everyone came out, and then I'm there by myself. And I'm like, so I asked the, the teacher— or, you know, wasn't the teacher, it was whoever was monitoring the, the release of children— and I'm like, so, um, yeah, I expected my child to be here. And they're like, Huh, are you— this is what they— are you sure they came to school today? And I was like, yeah, they're in third grade, so yeah, I'm pretty sure I dropped them off myself. And, uh, you know, it was funny, like instantly put the burden on me, like, are you sure you even brought them? Like, yeah, I brought my kid to school. Um, anyway, like time goes on, I'm kind of freaking out a little bit. I have some issues, triggers. One of them is my child not being able to be found. Um, Anyway, I finally, 15 minutes later— so there was a whole bunch of nonsense, right, where they go and they call the school and the school's like, yeah, we found her, we're sending her back. Turns out they didn't actually find her, they were just saying that they found her. Anyway, it was like craziness. So I'm, you know, losing my mind a little bit. I get, you know, I— she comes out, I hug her like I've lost her for ages. Thank God. I might have cried. Um, And I mean, it's ridiculous in hindsight, but anyway, this is me. But I get going and I start walking back to, you know, on our way home. And my son, 'cause of course I'm like, did anyone else pick up this child? Is like, is there any explanation for this? And my son is like, you could always check her Apple tag that you made us put on everyone's backpack. And I was like, oh, I can. So AI in action. I could have like zero geotagged my child But here's the thing, I didn't do that. I freaked out instead. So that's, I guess that's like a classic lesson in AI adoption, right? Like you can implement the tech, but if you don't use it, it doesn't solve your problem when you need it most. So that was my moment. But I did feel a huge sense of relief that now I know that. And I'm hoping, which is also true about adoption, the first time maybe it's not a go-to, like you directly use the thing. But as you realize, right, as you make it a habit, Um, like, I anticipate if this happened again, I'd be like, oh my gosh, I have a tag now. Uh, where I implemented that tag at the beginning of the year, and I trained my children that if they ever were to leave their backpack, that they were to take the tag off and put it in their pocket. Like, just don't, don't judge me. This is like, I'm not a helicopter mom, but I'm definitely like a drone mom, maybe, where I like to follow everything that they do. Um, so anyway, uh, but yeah, so I was like, oh my gosh, that would have been awesome if I could just have thought to open it up in the app and be like, she's actually in the cafeteria, if you could just go, you know, like that would've been, that'd been a bit of a flex on my part. But that's not how it went down. But yeah, a lesson in AI, in AI adoption. So that's why, like, I see AI adoption everywhere in all of the experiences that I have. So this, at, in Sydney, I got a chance to do a couple really cool things, things that of course inspire inspired me to do things on my part. While I was in Sydney, Sydney, we launched our brand new program, uh, for 2026, which was hard. It was a lot of work we did— you did— to put it together for you. Um, but we did put together a program for 2026 to help basically shortcut my 10 years into a single year. And honestly, I will tell people, you could do it in 12 months, 12 weeks, 12 days. It's going to kind of be up to you, your consumption of this. But my favorite part about the program is that we are going to be moving into helping you not only get certified for my company, which they're reputable industry-known certifications, like AI Leadership Institute is a known entity, but I think it's even more exciting because that's cool. You should do that. But it's even more exciting when we bundle in the ability to get certified by Amazon, Microsoft, and Google. Like, that's, in my opinion, the ultimate flex when you walk in and you're like, yeah, I have all three. Very few people have all 3. So that's kind of what we're doing in our new program is like, become an AI Solutions Architect, become an Agentic AI Practitioner, become a Responsible AI Leader. Of course, you should do that anyway. But then also become a GenAI Leader on Google, right? An AWS AI Practitioner and a Microsoft, uh, AI Fundamentals, like, certificate holder, like, all of that. And then my goal is, of course, at the end of all of this, when you achieve these things, I submit you, nominate you as a Microsoft MVP, and that is like icing on the cake. You get a little sign, like, and then we have a party and I celebrate you with a big trophy or something at some coordinated event in the future. But it starts first though with us doing the needful things and getting this education. So I'm really excited about it. We actually have a— I'm— I just created a really cool video that explains all the things, so stay tuned for that. By the time next week comes around, you'll be able to see what that looks like. Oh, I should probably keep an eye on my comments and chats. So that being said, I want to— I do want to spend some time going over some of the latest and greatest news of the day. I do want to tell you a tool that I use all the time to give me this news. And I mentioned a couple of weeks ago, maybe just to my leadership people, maybe to you as well. I think it was quickly I did it last time we talked, but it is called Hux. H-U-X-E. It is built by the creators of NotebookLM, and it's basically a custom NotebookLM built for you. So it— and Skye actually started using it and was like, hey, it's pretty cool. So it's been validated a little bit. But this tool, it's available on Android or on iPhone, and it's just great. It allows you to custom train this model to the news that you want to hear. And then when you do decide to hear it, Like, I always, when I go to get my kids, I just put on my headset and I open up that channel and just let it tell me what it thinks I need to know contextually based on my email and my calendar. And it has served me very well. So I, I just love it. I think it's really cool, really cool tool. I mean, it's in its infancy, so I anticipate that they're gonna make it agentic and that that single feature of No, I mean, it's already a bit agentic, right, where it's using my calendar and my email. But I anticipate that it's also going to allow me to do other things, maybe connect it to Slack or whatever, and allow me to take the things I'm listening to and learning and actually action them, which would be an amazing, amazing thing. I'm very much looking forward to it. I'm trying to think of other things that came into my purview. Oh, of course. Microsoft has an event coming up in Seattle, November, the week of November 18th, and I'm currently considering speaking there. We'll see how it goes. Fingers crossed, maybe. I don't know. It's, it's a lot. I've been traveling a lot, y'all. So we'll see how it goes if I decide to go there for that. But I do love this event. It's called Ignite. It's meant to— it's scheduled at this time for a reason. It's meant to kind of ignite a fire under you around all the new capabilities capabilities that are available as part of what was pre-launched at Build. So lots and lots of AI capabilities. And this is a good segue. So that Ignite event is really useful. If you can get there, you should. AfroTech is next week. So AfroTech is one of my favorite events. If I can, I try to hire people who are at this event because they represent, one, an underrepresented group, but two, those that are like pushing the edges in technology. And those two things just make really cool people. So I have hired at this event. I've identified people to hire many times. I won't be there next week because for goodness sakes, I need to stay home for a little bit, but I will be. But that will be a really cool event if you have an opportunity. I don't know where it is, but it's Afro Tech, Afro, so go check it out. Events that are coming up for me, so where I will be in the world next week, I'm actually virtual, so I will be speaking virtually to a an event, a client, and then, which I love virtual events because I'm excited. It's like this, like I'm excited about them. I feel like we're talking to each other. So it's super cool. The following week I will be headed to DC for an event. And then I— so DC people, here I am. I'll be coming. I think it'd be great if on the 6th or maybe the 7th we do an AI leadership meetup. But I need someone in DC to partner with me to do that. So let's go. That'd be fun. And it doesn't have to be complicated. Like, when I was in Seattle, we just met at a restaurant. So let's do that. Plus, I get to eat. So that would be awesome. I don't know how many of you know this about me, but I don't typically eat on days that I talk. Not the whole day, but usually it might depend on when I speak. But I don't eat until after I speak for obvious reasons. You might think, right? Like, nobody really wants to see me with stuff in my teeth or digesting stuff on stage. Nobody needs to see that. So I usually just don't eat. I'll drink my food. I'll have a protein shake and some Active Greens, um, AG1, but, uh, but I don't usually like eat physical food. So after the talk, when my adrenaline starts like declining after being on stage, I'm like, where's Taco Bell? You know, like I usually eat horrible food and watch horrible TV, like, you know, a romantic comedy from the '80s or something. But anyway, so that being said, we should do that in DC. We should go have something to eat, break some bread together. So for those of you in DC, that's coming. So that's next week. And then I'm headed to Portugal, to Porto, to speak at an event conference called ICCA, ICCA. So I'm super excited about that. So if you happen to be in Portugal, let's go. I'll meet you there. We could do our first International AI Leadership Institute meetup if you're in the area. But I don't know anyone in Portugal, but maybe, maybe one of you are from there. So let me know. But that is all upcoming for Where in the World is Noelle Santiago. I will be in those locations. That just takes us to mid-November, and then all heck breaks loose again, and I start traveling. But it'll be fun. One thing that we are doing in the AI Leadership Institute, but as well as on maven.com, you can go and check out one of our most successful classes is the AI Solution Architect course, we have opened up a new cohort. It's called the Rise Up Epsilon Cohort. This is the 5th time we're running this class. So the Epsilon Cohort is launching. We— the class will run December, I think it was 4th or 5th. Anyway, it's the Thursday of the week of December 4th, and it runs for 5 days. It's a bit of a crash course, but we found that it's working for people. So it allows you to just No, you know, do the work. It takes you about 2 weeks to do the capstone, but it's good to just do this, like, accel— I call them accelerators for a reason. You're gonna accelerate through this content, like, absorb it as much as you can. It's— that's why we do it over a weekend. So you have Thursday, Friday, weekend to digest, take a breath, build something awesome, and then Monday, Tuesday, Wednesday, we round out your knowledge, and you then have 2 weeks to do your capstone project. Which is super fun. So that is one thing. On the other side of that, we also, or I should say, on the earlier side of that, we are also launching for the third time. I've taught this 3 times before, so I guess it's a fourth time, but never inside our school community for our school people. So this is called our Responsible AI Leader certification. So that's gonna be running the middle of November. So right when I get back from Porto, it's like the 13th of November. It's a Thursday. We'll be launching the Responsible AI Leader program. And again, it's 5 days, but you might— we do Thursday, Friday weekend to digest, think, ideate, rest, and then we start up again on Monday, Tuesday, Wednesday. And we've just found that this model really works for acquiring knowledge quickly, but also giving your brain a little bit of break time in the middle to, to rest and absorb and think about what we're, we're learning. So, so that's coming up. And so if you, and those are right now. So if you are interested and you're not part of our AI Leadership Institute, you can go over to AI Ally on Skool, or you could just Google AI Leadership Institute or noelle.ai. All of them will take you, and you'll see you have to apply to become part of it. But if you join the 2026 co— like, program, the AI Leadership Essentials program, if you join, 2026, you get these 2 certifications as part of that. So that's kind of exciting. So you'll get like 8 certifications for the price of 5 or whatever it is. So that is what's coming up. I'm excited to do it. It'll be basically every time I teach these classes. That's why it's kind of nice. Every time I teach them, they're different because the world's changed, right? Everything's different. Everything's different. Technology changes. And so I always teach it live, but sometimes, right now, for example, Agentik AI Practitioner, we did over the summer. We're, we're, we've already prepared to do it again. We'll do it in the beginning of next year in March, I think. But like right now, you'll be like, but I don't want to wait till March. So you can go back and listen to it. But as you know, it's not nearly as fun as if we do it live. So I want to take a look at some of the news that's happening around the world of AI. Let's just, you know, ask what's happening. I always, like I said, I use Hux, and it is the lens upon which I see the world is through kind of what's happening in the world of responsible AI and how, you know, one of the things that I ask all the time, um, like these systems, right, is like, where is— I ask for AI gone wrong stories. Don't— hopefully that doesn't say anything about me. I'm an optimistic person, but I want to know, like, I want to learn through the mistakes of others so that I don't make those mistakes or people connected with me don't make those mistakes, right? I always say Success leaves clues, but so does failure, and both are valuable, and both are extremely, you know, if you use that information you learn, both of them lead to doing better work on the other side of that. And so that is what I think, you know, one of the ways that I learn is through the practical applications of others. Whether for good or for bad. Now, this week I got a chance to work with a— I'll call them a pharma company. But basically, most pharma and life sciences organizations that build, like, drugs or things like that, like, they're mainly a distribution company and ultimately a supply chain and manufacturing company because they're manufacturing, right, the drugs that they sell or the items that they sell. So I was in that type of a company this week. And one of the things I thought was really cool Hold on. One of the things I thought was really cool was it— they— the reason they asked me to come work with them was that they were really looking for— and I think this is true for everyone I see, I get asked this all the time— they, they had already heard and acquired their own knowledge on like what is AI and everyone, as most of us have, right? Like whether you wanted to or not, it's kind of like by osmosis, the world kind of knows what AI is. But what they didn't know was how do you translate what you know about AI into an actual thing that delivers a— some value. I know I mentioned this before, uh, in some of our previous calls, but 90— you know, uh, Forbes released an article. I don't know if we have that available, but Forbes released an article that said 95% of projects were failing to see any kind of production value when they got deployed, if they got deployed. That's a lot, like 95%. And so what I always think about is, well, why? What did the 5% do differently? And luckily, I know because I've been on the 5%. Like, 5% include Copilot, regardless of what you might think. Claude, like these big AI companies, right? They are doing production AI, and they are doing it profitably. So that puts them in the 5%, and I've had a chance to work on these teams. So I know at least a little bit about what that looks like and the decisions that we had to make. But at the same time, I don't know. I feel like when I went to this organization, I feel like everyone feels like they, you know, when I ask them, so do you know AI? I feel like their instant answer is, yes, I know AI. Even though they don't know how to translate what they know into anything valuable for their business or their career. Or their household. And so that's what I think is the differentiator of many of the people like in the AI Leadership Institute, right? We have lots of consultants and advisors, and that's the kind of the demystification that we're focused on is how do we demystify the hype of AI and turn it into something tangible that can be measured by business outcomes. And so I don't like to be 100% outcome focused because then we lose sight of kind of the holistic cost of AI from a human perspective. But I do think you have to speak in the language of people that are gonna use this technology, and it isn't data scientists, right? Data scientists are building it, but the business is using it. And that's what I ended up going in and doing. And so in 2 days, we ended up doing a bit of a vibathon, if you will, right? We put our heads together. We identified— some of you know my pattern. I do this whiteboard. Where we go from what's the first step in an AI project to get it going on the right path. And that first step is problem solving, right? Like, identify the problem you're gonna solve, then figure out the data you need to solve it, then identify a model that you can use to train on that data to get going. And then the hard part starts, right? Like, that's the easy part. You can do that in 2 days. So that's what I do is I teach them the easy part and the hard part. But what gets them stuck in, like, or not stuck, but, like, attached to, like, wow, this is working and this is cool and let's keep going is that we talk about the entire holistic architecture, AI solution architecture of what we're trying to do with them. But we do it in a way that at the end of the first day, they've built something that kind of almost works, right? Isn't that the nature of ChatGPT, right? When you use a large-scale language model like an LLM, like ChatGPT or Claude, like its job is to kind of be okay, like be pretty good at whatever you ask it to do. And in the same sense, that's exactly what happened for us. When we built it, we built an AI system that would validate a certain business process. And we trained it. I taught them the 5 Ps of prompt engineering, as I tend to do. And they built it. And it did do it, kind of. So we gave it the process. Its job was to validate it. It gave us a validation. But, and this is at the end of day 1, right? So at the end of day 1, we've got a, a model. It's working. They're all super excited. We're taking selfies in front of the, the, like, agent created sign. Like, we're super excited. But the answer it gave us back was totally wrong. It looked good though. How many of you have had that experience, right, with AI where you, you give it a question, it returns you an answer, and at first glance, you're like, that's not terrible. I agree with that. Oh my gosh. But it was funny because As they started reading into the details, they were like, wait, where did it get that information? How did it know that? What about this? And so now we enter into where all of these projects, the 95%, is it starts broken and then you fix it. So now you've got to fix this program and it allows you to kind of turn your eye towards what I call SAT, right? Security, accuracy, and trust framework. Like, you have to— those are the 3 things that will burn you when you take a model that you're excited— a baby tiger— and you try to deploy it. It won't be secure from multiple lenses. It will not be accurate. So you're going to need to figure out how to train it to make good decisions. And how do you build it in a way that you garner trust in the very people that are going to use it? And the best way to do that is to involve those people early. So that was a little crash course. But the cool thing is, is that we did that in 2 days. And at the end of the 2 days, the people were like, what just happened to my brain? Like, oh my gosh, I'm a new person. And I love that. I love that sensation when I can watch them go from I'm not sure about this stuff to, oh my gosh, I'm the builder of this. I'm the business owner. I'm the one who owns this business process. But I know how to tell this model what good looks like. And they are now responsible to do that, right? As opposed to the way we used to build software where we'd lob, you know, technology would build something, we'd lob it over to the business, the business would look at it and maybe they used it, maybe they didn't, but they definitely weren't involved in building it. Now those lines are gone. Like now it's they, your business and your implementation team need to be hand in hand. So that was really fun. That was kind of where I spent my work week and now I'm documenting all of that work. For that team so that they understand kind of the decisioning process to do the SAT moving on, right? Like, I love the SAT. The reason I like this acronym is because SAT, that it's a test you take, or at least you used to take. I don't know, some of you with— I don't have kids that have gone through this process yet, but it used to be a test you take to like determine your college readiness. And I use SAT, same idea, but for AI readiness to go to production, right? What your security profile, your accuracy profile, and your trust profile. And all of those are the— if you look at the top reasons AI projects fail, those are them. Different lenses, but pretty much them. Okay. So let me see what I wanna talk to you all about today to see if there's any new research. I like the legislation side of this, of this, of AI, because I think we, when we get into our little echo chambers, we don't really understand what's happening on the Hill, right? On Capitol Hill, in the Senate, in Congress. We don't, we're not in those rooms, not many of us. Some of us are though. We do have people in our community that are in DC, having our part of some of these conversations. But not only at the federal level, but I think it's also really interesting at the state level. So this is what, like, my little Hux platform educates me on because I asked it to, is like, are there bills that are coming up for a vote? Are there— is there legislation that's currently being talked about? And so there was a recent bill that was passed in California. I'll just, you know, California, I love California. They're good people. So they passed a bill recently. It was called, I wrote it down, 243. So it was signed earlier this month and it, I'm just gonna read it exactly, requires that companion chatbot AI systems clearly identify themselves as AI so that users aren't misled into thinking they're human. Have we not been talking about this for like a long time? Before, it was a choice. If you were a good person, if you are a good builder, if you are a responsible leader, you will disclose AI when you're using it. What that means is that real estate agents in California can no longer use a system. I won't call out the brand in case they want to pay me as an influencer one day, but you can't use a system that simulates human outbound calls. You can't do that anymore unless that outbound caller says, I'm an AI assistant. I'm here to help you do blah, blah, blah. Here's the danger in this for businesses that have adopted using AI for outbound calling or using AI for conversational apps, and they've tried to make it look human, is that when you do that and someone thinks they're talking to a human and then they uncover that they're no longer— like, that they've been duped the whole time, right, that they've been lied to, your distrust goes to a point of really unrecoverability. You can't recover from that, right? Like, you can't recover from, you know, basically lying to somebody. I mean, maybe with some people, but that is what this, uh, codifies, right? The legislation is putting in place— we know that this is going to hurt you. That's what legislation does. When we know that this is going to hurt you more than help you, we're going to put in place a law that will protect you from hurting yourself. And so that is what it— what this, this law ended up providing. So take a look at that. I always like to look at these early laws that are passed because they do end up sometimes showing precedent that will be established and leveraged by federal action. So I— that's why I like to hear about it. I like to know. Now, the 2025 AI Index Report, which some of you know was released by Stanford, Inside of this report, it shows— I always tell you, I mentioned multiple times that there's about 60 laws that are currently on the floor being evaluated for whether or not they get voted on. And so if you're interested, I think there's 59 now because one's gone. But anyway, if you're interested in what they are, you can go to the Stanford AI Index report and it lists them out. So that's how I got— that's how I got into this in the first place, is that I was at Microsoft Responsible AI in New York and the head of policy there. I think her name is Natasha. Crikey, I can't remember last name. But anyway, you can look her up on LinkedIn. But she had cited a couple of these laws as her— as she was talking, and I was like, where did you get these? I want to know what laws are available. And so now Stanford puts them together and stores them in their report on a yearly basis. So pretty cool. So interesting. Interesting developments overall. So one of the— I always push myself to do things that are just crazy. So one of the things I did recently was launch an initiative to build an e-commerce store, and that's only because I don't have exposure to this. I don't have people who do this, and the premise is that I can use AI to do this. So I'm launching an e-commerce store now. As I launch that e-commerce store, I now get involved. And now, granted, I'm not selling, you know, cell phone cases and pencil cases or whatever. Like, I'm thinking about— I'm trying to set it up from a perspective of if an e-commerce client came to me, I want to have thought through all the things. And so while I'm going through this research of setting up this e-commerce store, one of the things I immediately run into is communication with suppliers and supplier trust. And so one of the things, that Alibaba, one of the big distribution channels for all the nonsense that we buy— these little tigers, I'm almost positive, come from Alibaba. Um, but, uh, these, like, these little tigers, uh, you know, anyway, so Alibaba distributes a bunch of stuff. Um, and while doing that distribution, right, they become a supplier. So if I wanted to become the Baby Tiger store which I do— I'm not, but, but that sounds cool. Uh, I would need to contact a supplier through Alibaba, and they built an AI system to support— it's a chatbot, basically a conversational agent, but it supports this very painful process of supplier identification. And now I understand that painful process because I just researched it in order for me to do it for my own company. Like, This is, it's, you know, it's a numbers game. I have to call like 200 suppliers, and then once people start saying yes, I have to negotiate terms with them. It becomes the hill upon which my dreams of starting an e-commerce store die because I'm like, I don't know that, I don't have time for that. But now Alibaba, Shopify, they're now implementing these AI systems to help you in this journey, which goes to my, the reason I care about this is that it changes accessibility, right? So now Shopify, Alibaba, Amazon will create tools that you can just ask these questions. 2 years ago, maybe 5 years ago, I couldn't ask a question. I would send an email, which I just did. Some of y'all know my Lucky Lip Gloss story. So I just sent a question to one of my suppliers and they didn't answer me. And it's a pretty easy question. I think they know the answer. But there's some person sitting on the other end of that inbox and they haven't gotten to my question yet. And so I'm like, let me help you help me. I'm going to build you an agent that can answer these questions because I know it's documented somewhere. I just don't know where to look. So let me build you an agent that uses cognitive search to semantically index all of your information. So at 2 in the morning your time, 8 AM my time, I can find the answers that I need. So that is, I think, some cool stuff. So these are— so Alibaba and Shopify both launched announcements. If you're in the e-commerce space or have e-commerce clients, it's worthwhile taking a look at what that looks like. Alibaba also announced Quark AI glasses for presale. You know, I'm gonna get some. So what that means is, you know, you can imagine getting glasses that— I mean, it's Alibaba, so it's a shopping thing. So imagine getting glasses, and as you see things, I will correlate this to another thing you should be aware of because it's cool, but you might not be because it's not very popular, popular. But this concept of X-Ray, right, on Amazon, like on the Fire TV. So I have done this multiple times where I'm just gonna combine all the tech that I use, but like X-Ray is the capability that allows you when you're on a specific scene you can tell who's in that scene. It uses computer vision to annotate, right? I don't know if you've ever seen a Fire TV, but when you go on Fire TV and you play it, at the bottom, when you pause it, it tells you who's in the scene. It shows their picture. And then when you click on them, you can find out who they are. Well, imagine, if you will, not only do you see actors, but we also can use computer vision to identify what people are wearing. And so I— if it's a dress or they're you know, on the table, product placement, there's perfume there. The API exists that if I chose to do this, meaning the tech is already built to do this, that I could identify and tag something in content, and then I could give you as the consumer the ability to say, hey, buy that. Now, this has been a dream, I think, for a very long time of, you know, the Fire TV world and the X-Ray developers that we'd one day actually use that. Quark glasses are an expression of that, right? That we're now going to have glasses on. It's going to use computer vision to identify things. I will take a picture just like I do with my Ray-Ban Story— what is it called? Storytel— no, Ray-Ban Wayfarers. Um, take a picture, but now it's analyzing it through the lens of I want that, and it finds where I can buy it, and I literally can just say buy that. Now Alibaba is similar to Amazon in the sense that they have billions of people's one-click purchasing power, right? Billions of people have given Alibaba and Amazon our credit cards so that we can just say buy that and it already has a credit card on file in which to do it. So just— I just think it's really interesting to think about consumption, but I'm going to throw a little ironic twist to this as we're talking about like Q4, the run-up to Christmas. Buy, buy, buy, buy. Money exchanging hands from consumers to creators. I like to draw your attention to something that we talked about in our AI Leadership Institute call, which was the impact of AI on the workforce and how some companies— we're seeing it already— are moving towards this direction of laying people off in light of AI systems that are removing their jobs. I don't actually think that's happening based on what I'm seeing. I think they're proactively getting people off their payroll in hopes that the projects that they are deploying actually deliver value. What I've seen though in the 4 years of Gen AI, 3.5 years of Gen AI, right? Like what I have seen is that people get laid off The company's like, yay, we got this. We can build AI systems to do their work, no problem. And then they're— as soon as they go to production, or as soon as they start trying to use that and get value from those systems, they run into SAT, right? They run into security, accuracy, and trust. And accuracy probably more so than most, where they realize, wait, the model doesn't— I thought when I pressed the button, it would do this thing, and it doesn't do that thing. Who knows how to do that thing so we can tell the model what to do? Oh, we let them go. So in a world where we are creating mass capability to, right, buy things, what happens when we decide to lay people off and they're not gonna buy as much? So Jeff Bezos has a really good perspective on this. Like, I remember when I was working there, he always would give people, hiring managers, guidance. Like, no matter what happens, if someone's not a good fit for Amazon, they're not a good fit. But when you let them go, you make sure they— when they leave, they stay a customer, right? I want to just like reiterate, like, when they leave Amazon, that they— and of course I'm one of those people, I didn't leave, like, I didn't get pushed, you know, I didn't get fired. But, um, but when I left, I loved Amazon, you know. Obviously I continue to talk about them in very positive ways even though my experience was not amazing. Um, but I got to build Alexa, so I love it. But I— whenever anyone left, our guidance was you make sure that they don't leave disgruntled, that they stay a customer, they keep their Prime account, right? Like that was his long-term vision. Like people are going to come and go, but at the end of the day, we need customers to survive. And I think there's a little shortsightedness in the industry right now around the societal impact of laying off 40% of your workforce at scale. If we all decide to do that. And our machines aren't good at doing the jobs yet. So I remember a company, they laid off 20,000 people in 2023. They ended up opening 80,000 headcount at the end of 2023 into 2024. I still don't think they've replaced— like, that they've been able to hire for that. So I don't know. I just feel like there's a really— there's a little weirdness happening in the industry. Would love to hear more from you all about your perspectives on that. I'm gonna keep the chat open so I can see what y'all are saying. But it's really, really interesting. So back to the news. Some— oh my gosh. So did y'all know that Loom got bought by Atlassian? So I'm a huge Loom user. Apparently it happened a while ago, right? Yeah, I didn't know. So I get in there and I'm— purple is my color. Some of y'all know this. I love purple. And Loom is purple. So we're friends in that way. But they got bought by Atlassian, and Atlassian is the Trello people. They also bought Trello, and they're blue. So Loom just took a big brand remake. But I always worry when these companies that were awesome and gave me so much stuff for free, when they get bought by these companies, some go-to-market strategists would be like, you gave that for free? Oh no, we're charging for that. And so it's only a matter of time between all the, all the things that I've done on Loom becomes very expensive. So I'm a little sad about that. Um, question from the— oh my gosh, I use, I use OSB. As a matter of fact, let me show you. Um, OBS, you mean, right? OBS? Yes. Yeah, yeah, yeah. Okay, I'm so glad you asked. So this is, um, I have 3 of these. This is OBS Bot. I think you can see its name. Oh, it's really small. There it is, OBS Bot. So I bought 3 of these, and this is an AI topic, so don't— we're not getting too far off track. But I bought 3 of these because they are— inside of it is a small little model that does computer vision identification, and it's trained on gestures, right? Gesture control. So when you turn it on and it faces you, you can hold your hand up and it will zoom in, and then you can go like— you could put it in an L and you can go like this to zoom in. This to like zoom out, um, this to stop, right? So there's all these like really cool gestures that you can use. But what I liked about it the most is the reason I bought 3 is that it also has an AI editor in the OBS bot, which is OBS software. The AI editor will take in all 3 feeds and dynamically— have you ever seen those cool videos where it starts here and then it goes here and then comes back? Like an AI system, while you collect those 3 feeds from each one of these, It allows you to do that. Now, I didn't— I never got that to work. So the dream that they sold me was not super achievable, but it's not to say it can't be done. I just lost patience. So I have 3 of these if anyone wants them. I'm just kidding. I have it here because I'm renewed in my interest in getting this working again. That's why this is a new camera. I don't know if you can tell the difference, but This is a new 4K camera, and so we'll do it. But yeah, I, I believe you will. And Darren, once you figure it out, like, teach a girl something. I don't got time for that. But it is all it said it would be. So I've used these cameras for one-on-one, but you can see it's a gimbal, so it kind of goes up and down and it— so if I was standing up, which is why I used it initially, many of you know my early, which I'm going back to next year, but my early classes I would teach in front of my digital whiteboard. So I would just move around and walk and into the it would just move with me. It was really, really cool. So yeah, I like it. What? I like baby tigers. So that's really cool. Oh baby, I love it. Mama and baby, that's a whole thing. I love it. That's cool. Okay, cool. So, so that was in the news. I just wanted to mention the whole Alibaba thing. Go check it out. The Atlas browser. Anyone actively using the new ChatGPT Atlas browser? So we're not here to like open baby tigers up on you. Like, you don't have to, but the Atlas browser is a new browser experience. I use the Arc browser, and it took a minute to get used to it, but it heavily controls how I work and the amount of distraction I get into when I work. So I went into the Atlas browser, I maybe lasted 25 minutes, and I'm back in Arc. But It would be cool if, because they're now in the browser business, if they went and bought Arc, that would be awesome. Sam Altman, if you were out there, that would be pretty cool. So the nice thing is, is that the Arc browser, of course, being an LLM-based browser, allows you to actually execute tasks, right? Go out and do this, grab this, put this in the document, create a presentation, generate an image. So that's kind of cool. I just, I leverage the system Arc provides me so deeply. I don't, you know, if you've never used Arc, then I think it might be an easy transition from like a Chrome or a Firefox or a Safari. Who's the other one? Edge. Microsoft Edge. It might be an easier transition, but for me it was pretty hard. So that was kind of neat. And yes, yes, yes. So those were kind of my big 3 announcements, because I think it's cool that California is now requiring you to disclose you're using AI as a law. I wonder what that means for global nationals, right? Like, that do business in California. Like, if I sell— because I don't know that my AI salesperson discloses that they're AI. So maybe now I have to— I don't even know. I have to look and find out if you are going to work, if you're going to sell something to someone in California. From a digital e-commerce perspective, does that mean that your AI system— what do you all think? Do you think— thank you for that, David. Perfect. Awesome. Likely so, right? Yeah, I think so. I don't know. It's one of those things where I'm like, don't ask, don't tell kind of thing, but I probably should. Right? Oh my gosh, it's a great idea. 100%. As a matter of fact, I think that's a whole business. So here I want to share with you. I was working with a manufacturer this week. So they were using a third-party piece of software called— sure. OPB. No. OBP. Oh my gosh. Anyone know manufacturing software? I think it's OPB. Let me go look. Let me go. I took notes. OPA, I think. Oh God, I can't remember. Anyway, it's an old system. It's not that old. I get— 15-year-old system, but it is— it's just like any software built back then, not easily customizable. And therefore, businesses over the last 15 years had to build customizations around it. Are you familiar with this concept, right? Like, SAP is a good example too. People build all these, like, hyper-personalized customizations to the way they do business. So now when you go— so what the reason I wanted to mention that— I'm gonna have to Google it so I can see it. Um, If y'all don't know, I'll ask. I think it's OMB, OMP, OMP. That's it. OMP. Thank you. All right, cool. So OMP, I don't know what it stands for, but it is like, I probably need to give it a little bit more context. That's too funny because OMB also is the AI policy, or I should say the management policy division of the federal government, OMB. And this is OMP with a P, but it's manufacturing software. But the reason I say this is because there are tons of these systems that these manufacturing systems and manufacturing medtech, health systems, legal systems. Like, every company has, like, a piece of software they bought into that they then built around. And what I— and what they're doing, a bunch of the friction that they have in their workflow is associated with that. We called it enterprise integration. I think we sold it on purpose this way, but enterprise integration. And what I think there's an opportunity to do now is to go to those big— those companies, the OMPs of the world, and say, hey, either can I help you, or become someone who can write new LLM or MCP-driven integrations for these. I mean, this is like, you solve one problem in that space, that's a billion-dollar problem. Like, it's not a billion-dollar company. But it's at least a billion dollar pro— that's enough. You can, you can eat pretty well solving that kind of a problem. But that's what it made me think about that because when they were showing me the software, I was like, why don't we just fix it? And they're like, oh, it's too hard, too expensive. Nobody knows the code, right? That's— and it's a very common theme in enterprises is that they've bought into a software. The people that did that work are long gone. Like, they don't work there anymore. They were likely laid off. No, I'm just kidding. But they don't work there anymore. So there's a bit of like Black Bocknet boxiness to this solution. But I think that is an opportunity to build LLM solutions. So I wanna give you an example of one of my favorite, startup ideas that I was a judge for. And they're doing pretty well. I mean, they're making like millions, which is pretty well, not hundreds of millions, not even tens of millions, I don't think. But the premise was something that I worked on actually at a consulting firm. I went in and I was like, wouldn't it be cool? How many of you are familiar with a long ago known technology called COBOL? So COBOL is the language, right, that runs mainframes. These are very— they're functional programs. That's probably the best way I can describe it. You can look it up if you don't know what it is, but it's old code, old, old, old code. And there aren't very many COBOL programmers. The ones that do still do that work, I think, are paid pretty well because that's what happens over time. But there was a need, right? Now the world has moved to client-server and web applications and mobile, end-tier architectures, right? Separating the client from the server. COBOL, the client and the server are all the same thing. Like, you work for— you work on the mainframe with this code. And so I had this vision of like, oh my gosh, do you know how many companies run on COBOL on OS/390 or on AS/400, like these old mainframe systems? Systems, and they're locked into that. Like, it's locked. They don't have enough staff to understand the code. They don't know how to rebuild the code. They don't have enough time to rebuild it. So in the early days of GPT-4, or actually, I guess, 3.5, I was like, oh my gosh, we could use an LLM to build new systems from old code. Doesn't that sound like a good idea? Like, I could just go into an, a COBOL program and ask it to build me new software. I immediately ran into problems with this philosophy because the mindset of the developer that built COBOL is very different than the mindset and the, even the objectives of a developer building an n-tiered service-oriented architected application. And so what I had to teach the model in order to do this successfully was actually not to take COBOL and turn it into a Java application. But instead, I had to take COBOL and have it define the requirements of that COBOL program, which is how we got it ultimately to work. So we reverse engineered, we basically said, look at this COBOL program and define the requirements of that program. So basically, you're building a requirements doc. And then from that requirements doc, generate a functional spec. And then the entire time I am validating that against my unders— which I do know COBOL. It's one of my first languages along with Fortran. But like that, that was the process. But what changed is between the business requirements and the functional spec, the functional spec is when we actually start saying how we wanna translate it. That translation was much more accurate because we did— it's almost like the thinking steps, right? Reasoning steps, multi-turn interactions with an LLM. End up generating better overall outcomes. It takes longer, but the outcome is better because you've given it a chance to think, right? You've given it an opportunity to kind of reason its way to the right outcome, as opposed to going, hey, insert COBOL program, output Java program. You all can probably imagine what an LLM did in this case. Like, it created a Java program that looked like COBOL, which doesn't solve the problem. So So we ended up doing that. A startup ended up taking this premise and building an LLM, a sequence of LLMs that would take COBOL. And the benefit was is they built different LLMs for each stage of the process so that the LLMs themselves wouldn't be like leading them, leading the witness, right, leading their own work. So they had one that did business requirements. And just like in a real company, you'd hand business requirements to a technology team to do the functional specification. And then you hand that to an implementation team, an engineer to build it. And so they built LLMs to represent each part of that process to get an MVP of a COBOL program that actually was built with more recent technology. And it worked, which is, I think, the most surprising thing. So it worked. So they have a bunch of clients that they're migrating applications for. And now their struggle is not, can I do it, but how do I scale, you know, doing it? I think they have 4 developers and that's it. So they haven't been able to really scale that. But interesting, right? Pretty cool, pretty cool stuff. All right, so yeah, exactly, more verbose job output. Exactly, exactly. Let me see if there's anything. Now, I, I wasn't gonna say this because I'm on, like, you know, we're on LinkedIn, YouTube, but there was a really big announcement from OpenAI on allowing a new type of content. Are you, are you Are you picking up what I'm putting down? Um, do you know this story about the content that they are allowing? So Sam Altman made a deal, um, to allow for erotica to show up. Spicy content. I should have used that word. Dang it. Yes, thank you. I'm like, I had to say the word, now my algorithm's all jagged up. No, uh, but yeah, a little more spicy content, uh, which would make sense There's money there from a capitalist. I mean, that's his job. He's serving people who want to make money with this company. But it's a little worrisome. Obviously, we're seeing some societal— negative societal impact before we've even allowed this to happen. And I just worry about addictive behaviors that we are well known about in humanity, amplifying the ability for that to become accessible to humans. Anyway, the only restriction is that you're an adult. So I don't know, that's something like— again, I always just— I— it's a Harvard Business School case study in the making, right? Like, and what I mean by that is that when in Harvard, when they analyze a case study, they think about who each person in the decision-making process is, and you sit in their seat and think What questions are you asking, and what strategies are you using, and what sacrifices or trade-offs are you making in order to get to the outcome that you got to? So it's kind of like reverse engineering. You know the outcome. We don't with this yet. But we know— I have a feeling, right? We know the outcome. How do you— how would you as a leader do something different to generate a different outcome? Or could you? Like, what forces are at play? And so I am curious. I think you're welcome to— we're gonna take the last 10 minutes of our call today to just get to know each other. I always encourage you to just hop into a breakout room. I use a statement called the I help statement. So just tell people like what you do, right? Like you help people do something, or you might need help, right? Just let people know who you are, you know, kind of what you do, and how, you know, you know, how you can be of service or how you can be helpful. I always like to lead with how I can help people before I lead with, can you help me? But these are trying times. If you need help with something that someone in this room can offer, like, you should ask for it. I encourage you to do that. But what I thought would be interesting is to stay away from the spicy content and focus more on, you know, if you just answering the question, if I was, you know, like, if I was the head of a very large AI company, what kind of questions would I need to be asking myself as these oppor— you know, you get basically, here's a bunch of money. Under what lens do we evaluate that decision? I have my own codex for evaluating decisions like this. I, of course, as a mom, would not— I would not have done this. But at the same time, It's a multibillion-dollar decision. So I'd be curious if any of you are, are, you know, actively contemplating, like, how do you talk to leaders? How do you— how do our leaders talking about this? Like, just share your experience. But I think also just empathetically, what would you do, right? How would— how would you handle— and it's the same with Shopify, right? Like, Shopify is now going to, you know, help people who don't know anything about building an e-commerce store to use AI to build an e-commerce store. That's not cheap. It's highly risky. What is the responsibility of Shopify in protecting accessibility? I'll end with this. Like, the Microsoft process of AI deployment I think is really interesting. They didn't say AI for everyone. Maybe they started that way, but they very quickly went, not AI for everyone, AI for approved use cases. So if you remember in the early days of OpenAI service, not everyone got a chance to use it. You had to apply. And that still happens today to these services. You apply to get access. And Satya's driving force is, I want you to explain to me how you're going to use this. So as a leader, he chose not to say, AI for everyone. If you've got a dollar, you can spend it with me, which is kind of like Amazon and Google's approach. Instead, he's like, hey, I'll take your dollar, but I want to see what you're doing with it first. And that's a decision that they made. So what kind of leader are you? Are you, you know, and it doesn't say anything either way, but we love your take on that. Like, how do you, how do you handle these types of conversations when it comes down to making decisions that might have a really interesting societal impact? It's interesting, actually. So it depends. It's kind of like You know the story of Jeff Bezos. So David makes an interesting point. Are they going to raise their pricing given their profit deficits? I think it's an interesting play because their profit deficits are not concerning at all based on their outlook. See you later, Keyshawn. Thank you. Happy Friday. Um, I'm actually more interested in like their long-term strategy because Jeff Bezos was a big believer in decades of lost profits before making money. And I think Sam Altman is taking a page of that book of like, we're not here to make money, we're here to make impact, right? Like create relationships, establish— like AWS, the reason AWS is still the number one cloud provider in the world is because they got a 7-year head start on certain adoption patterns. And that's it. It's only because people started on AWS that they stayed on AWS. And I still think that that's true with, with ChatGPT. So whether they raise prices or just create more opportunity for revenue. So I know you, some of you may have heard about the Shopify agreement, right? So he's, he's creating these, which is Jeff Bezos's style too. Like, ChatGPT is not the moneymaker. That might always be in the red. But this relationship with Shopify where they're going to get a cut of every sale of every product, like that's going to turn them black pretty quickly. And then Alibaba is another, right? Like all of these different ecosystem plays that I think are happening right now, I think we don't really understand that the financial impact that that will have. So no, I don't know that he'll raise prices on the individual. I think he will get his money through these ecosystems. Oh my gosh, 100%, me too, David. Um, yes. Yes, exactly. Yeah. And, um, I mean, and that blew up, right? Like, and it hurt a lot of people and people lost like their livelihoods. I 100%— I don't see anything different, honestly. There's no like level of, we know how this went and we don't want to hurt anyone this time. It's literally like, we moved fast and broke things back then, and that's even before technology was the driver. Like, I'm really worried about all of that. Um, yeah, I have lots of shares in AI, but I do buy— I shouldn't say that— I buy options. So if you're gonna do it, options are great because I'm just buying the option and selling the option. And so I, yeah, I guess make money on people who want to play in this game. Um, it's not really selling shovels. I don't know how you'd characterize my options trading, but I like, I like to play in the variability of the AI industry. Um, and I enjoy it, and I do well, uh, because I, I just know the trajectory of these businesses so well. Like, I could see it coming a mile away. Like, I know where you're— I know what's gonna happen to you. Um, and so I, I, yeah, I do a lot of options trading based on that. But yeah, holding shares in AI is super— I mean, the good news is, is that at the end of the day, long-term strategy, you're going to do better than a CD, right? Like, from that perspective. So it's not terrible. Um, all right, options at least, like, it's worth it. Look into options because you could like trade on margin. Anyways, I don't mean like— never mind, I'm not giving any financial advice. I'm not a financial advisor. I have no— like, I know nothing about these things, but it's an interesting world because of the perceived volatility. Um, anyway, yeah, yeah, exactly, that's it. And then actually what you should do is just build a GPT on this topic and then sell it to people that do what I do, you know, sell it to people like I would pay $49 a month for something like that. I drop $49 a month on all sorts of nonsense. So, um, all right, so if you're up for it and you want to meet each other, uh, I'm actually just gonna split us up into 2 groups. I'm gonna hop in one room because I never get to do that, and we're a nice small group today. Um, sometimes we have 40 people and it's much harder. Uh, so I'm gonna break us up, and if you want to chat, let's do it. Um, so really quickly, do, do, do— if you don't, don't join. Like, if you're not gonna, don't like join and just turn your mic off and camera off. It's awkward because we're like, are you there? Are you not there? So it's really nice if, if you just have fun with us. So there is a couple of note takers. One of them is not here anymore, so I'm going to kick you out. Move to— oh, I don't know. Okay, so here we go. I think we're all set. There's 2 note takers in the room. I will see you in just a few seconds. So yeah, let's just meet each other, have a quick conversation on, you know, what leadership is facing right now, and I will be back in about 10 minutes if I don't lose track of time. Okay, here we go. How can I— I can't remove these note takers. I don't know why. I guess I needed an extra session, an extra room. Move. Okay, that's Afwa Fireflies. I don't know how to move you. Oh, it's not joined. Okay, cool. Gabriella, one, two. Okay, go. I'm coming. Join.