I LOVE AI With Noelle

That's that.

But I always say I'm a Koye, as my daughter used to have. She had a birthday party where she was Shuri. And so of course, I couldn't be Prince T'Challa. So I was like her, her, her, all her little princess things. So I was like, I'll just be the defender of the princess realm.

So I've been a Koye since then. Good.

I love both. I love both.

I love that, though. If you're going to take on a persona, that's not too terrible. That's what I am. I believe that. What's up everybody? Welcome in, welcome in. We got some fun things to do today. Um, I have a bunch of presents that I've been waiting to open and so wanted y'all to be part of it. Um, some good fun exciting things. So this week, the last couple weeks, I have been hanging out with my dear friends at, uh, Johnson Johnson. So that's been really fun, um, finishing up that contract. And then next week I will be in DC. Sherry, who never comes to my my other call. But yeah, I'll be in DC. So we're gonna do a DC AI leadership meetup, even if it's just me and you, Sherry. Hopefully we can work it out. I am there speaking at a— oh my goodness, what is it called? Diversity and Flexibility Conference. And it's all about celebrating companies that celebrate the diversity and provide flexibility to their workers in the age of technology that we're in. So really, really cool. Uh, so I'm excited to do that. I'll be, uh, flying in Tuesday night and then Wednesday I have that event and then Thursday I kind of have a free day. I was, I was gonna, you know, I'll know me, I was gonna fly home and spend like 18 hours and then fly back to DC, but I'm back in DC Friday for another event, uh, for a mastermind that I'll be speaking at to, uh, small to basically business owners. I say small, but some of them are making a million a year. Some are making $20 million a year. For some of us, we're like, hey, that's, that's not too bad. Um, but you know, it's, uh, founders, entrepreneurs type, uh, event. And then on Friday night, I fly to Portugal to, uh, do— I get to Portugal on Saturday morning, and at noon I will be kicking off a vibathon, my new favorite thing to do. Um, so I'll be doing a vibathon with the attendees at ICCA, I-C-C-A, which is an event tech, uh, association. So I'll be working with a bunch of organizations that, um, basically deploy some of the largest in-person events in the world, and I'll be helping them learn how to implement AI, become doers, not talkers, just like us. So I'm excited to— so you'll see, as I always do when I get these little spurts of traveling and meeting people, you'll see we'll have a bunch of new members join, uh, the I Love AI Institute. So be nice to them. What's up, Steampunk? Is that it? Is that— or is it— I think it is a little bit, a little bit, right?

All seeing eye hat. Thank you for wearing it. I could see it.

It's awesome. Were you all talking about before?

Yeah, I was talking to— I'm like, we love Halloween, so we needed to make sure costumes are in.

That's right. Absolutely. I should have said something. I showed up at the earlier call and I was like, don't tell anybody. Uh, so anyway, I fixed it though. These are wolf ears in case anyone's wondering. Um, I was, uh, at Great Wolf Lodge this last weekend and it was howl-oween. Hey, it was awesome. We had fun. We had fun. Yes.

Uh, all for that.

So, um, yeah, so I'll be in Portugal. I'll do a Vibathon, and then my keynote is on Sunday. I have a day to rest on Monday, and then I fly back to, uh, the US on Tuesday. And then on Wednesday— what? I love corgi! Wait, no, is it a mouse? Is it corgi? What is it?

Well, I—

it looks like a corgi because I have corgis, but it's a fox. Oh, it's a fox! I like it. No, it's a fox. I think it's a fox. Wolf. Arctic wolf. Arctic foxes are kind of like wolves anyway, but that's awesome. I love it, I love it. Um, and so yeah, so when I get back, uh, I will get back on the 11th, I think— no, on the 12th. And then on the 13th, we kick off for the AI Leadership Institute members. We will be kicking off our, uh, a Responsible AI Leader certification program, which is 5 days, a crash course where we will be walking through this handy-dandy notebook. So we'll be going through our, the, um, Lead AI Framework, LEAD, and the Secure AI Framework, both of which have been deployed, uh, in companies all across the world now, not just the US, but, uh, in about 17 other countries. So it's pretty exciting, exciting time. So we've never done a, um, a course on the book in our community, so we're doing one. And then in December, December 6th, we are kicking off the last and final AI Solutions Architect course, which is available on Maven if you just want to come and take the course. But it is part— every member of the AI Leadership Institute gets access to that. So that is super exciting. But I got lots of presents. So I didn't know— people are so nice. I always love organizations that send me speaker gifts, like stuff. I'll tell you, the light that you see shining on my face right now is actually a speaker gift I got. Best speaker gift ever, lighting and a microphone. This one right here, little rainbow light. Anyway, it's very fancy. Um, from what I understand, it's not super fancy, but it feels fancy, and that's all that matters to me. Um, so I'm gonna start first with, uh, a box I just started to open. I was like, wait, I'm gonna open in front of you. So this, and this is more for you all as you begin to build your own brand. Here are things you can do for other people. So A couple of years ago, I got on stage and someone gave me just a cute little gift bag. And inside of it, it had a book, which was their book, and it had a little handwritten thank you card. And I was like, this is so nice. And I kept that. I still have that book to this day. And so from that, I got an idea. I was like, every time I'm on a panel, I'm going to bring my book in a cute little gift bag and I'm going to stick a little baby tiger in it. It's not as big as this one, but it's kind of like this one. Actually, I might have one around here.

Oh yeah, I do.

You all know this one, a little, little tiny little baby tiger. So I stick that in there, um, in a handwritten card on stationery. Do y'all do that anymore?

Is it a long—

yes, right? So I have stationery, um, I just did this, so I have some fresh handy made available. I can't remember where I got these printed, but, uh, oh yeah, you know what, I'll show you. That's Noelle Elisa Russell, um, That is my middle name. Anyway, so these are my little cards, and I write on them, and, and I love it because the reaction I get is always just genuine, like, you thought of me? And I'm like, yes, I thought of you. So this is someone else thinking of me. So who knows what's in here? I didn't actually open it yet, but anytime I see tissue paper, I'm like, ooh, it's a present. So that's also a signal. Oh my gosh, I love it already. But first, There's a cute little certificate. This is a good idea, right? Write your thank you note on a gold certificate. That's a good idea. I'm gonna take that. That's nice. Thank you for all you do to keep the ethical AI, um, keep ethical AI at the forefront. From Cindy and the podcast team. Amazing. So this is the Data Chief podcast. If you didn't hear me on there, um, we have just created a bunch of little shorts for YouTube, Instagram, and LinkedIn. So you'll be seeing me on that podcast, but if you wanna hear the whole thing, She would love to have you. And I must tell you, I make sure my husband isn't around because I'm not allowed to have any more of these. I love it. It's so good because you know what I'm going to do? See, it says the Data Chief, like swag, yo. I'm all about the swag.

That's nice.

And then all I have to do is add a little sticker.

You see?

Anyway, it's awesome. OK, wait, I think there was something else in there. I see I lost all track. I'm like— My new favorite cup. Data Chief. That's cool. I need to do something like that. AI Chief or something. But this is how I get all my ideas. Oh my gosh, another thing I don't need another one of, but I'm very grateful for it. A new notebook. Let's go. Anyway, so that's pretty cool. Very— thank you. Thank you so much. And of course, how many of y'all know I'm an engineer at heart, so I get paid in a couple of ways. Apparently Stanleys and Yetis now is one of my currencies. Hoodies always. This is a hoodie. I actually just got this hoodie last week. It's, it's nice because look, I could wear it with anything because there's nothing on the front, but on the back it says Great Wolf Lodge. So, uh, my husband just rolls his eyes. He's like, okay, if you need another hoodie. I don't need another hoodie, but I like them. So I got two stickers. Yay! Um, and I love stickers, so I'll add that to my thing. All right, what else do I got going on here? Oh yeah, hold up. Um, this one This one I got, and I waited to open it with you all because— do y'all know Mary Rodriguez? Mary Rodriguez, if you don't know her, go find her on LinkedIn. She— I'm afraid to set this down. Um, so this is from Mary and Stacy. They just launched a brand new company called Impressa AI. Um, and it's a pretty cool idea. And so I share this with you because it is another example, like, they just did it. And with AI, you can just do stuff, right? Just do it. Just go out, massive action, figure it out, find out if people like it. And so what she did, um, is, uh, she worked at Microsoft. She was one of their master storytellers for years, maybe even over a decade, maybe two, I don't know, long time. Um, and she just left Microsoft to go to Impressa full-time. Their mission, which I'm aligned with, I'm one of their early council members, which is basically their founding members, organization. They came to me, uh, I don't know, 6 months ago, maybe a year ago, and they were like, we want to do this, we're thinking about this, would you support us? And of course, y'all know me, I'm like, yeah, I'm in, tell me what I need to do. And so, uh, I met with them and got very excited about their mission, which basically is to empower women specifically in AI, turn them in— there's a thing they say, they've got some messaging, I'm not, you know, I don't know all that stuff, but it's something about turning them from, I don't know, to empresses in AI. But basically a million people or a billion people, I can't remember, but a lot. So I'm a big fan of focusing on a specific demographic, specifically women, and empowering them with artificial intelligence. So let's see what they got me. Again, tissue paper, you know it's gonna be good. Okay. Oh, alright.

Let's see here.

See, like branding, man. I don't do that. I have stickers, that's about it. But this seems very fancy. There's like layers of tissue paper. Oh my gosh, like for stickers. What? Uh, okay, so there's some more stickers. They know this is like my love language. Um, but this is pretty cool. It's a little Empress keychain, gold. Of course I'm feeling fancy already. And more stickers. And then this thing, uh-oh, build your digital legacy. That's pretty fancy. All right, all right. Thank you for joining us on this journey. We are building a powerful community of women uplifting each other to achieve success and pioneer change. Our mission is to create a legacy of innovation that will empower future generations of women. Thanks to you, we are one step closer. Oh, don't make me cry. Oh my God, I love it. Okay, cool. I love Mary. Let's see what else we got here. Oh my gosh, this is fancy! Anyone know what this is? It's like a symbol.

Yes, for the evil— against the evil eye.

Yes, exactly. Um, so it's gorgeous. It's like protection. What a cool thing to give someone. Uh, so I actually have been given these gifts. Oh, sorry. Um, so this, a friend of mine, Dalia, who I'm going to be teaching the workshop with in Portugal, she gave me this. Which is so fascinating. Totally unrelated, but both of them gave me this. You know, I always look at these universal things, like when, you know, like, what, what's happening? So anyway, so now I have a— the watching eye and a matching bracelet. That's pretty cool. So, all right, awesome. So when you see me with this, you'll know where I got it. Thank you, Mary. You're awesome, Mary Stacy. I'm super excited about what you all are doing. If you— you can be— they have what's called Empresses and Empresenors, which is the like entrepreneur side, but they have— I don't know how to say it, but it's like Empress Allies. So for those of you who are not an Empress or don't identify as an Empress, you can join anyway as an ally. And here's the cool thing that she's doing with the help of artificial intelligence. She is allowing people, anyone who has knowledge to share, and her actual belief is that everyone has knowledge to share. And that in this economy that she's built, every time you consume something of someone's, you get paid a fraction of that time that was— whatever the monetary value is of that person's time. So it's basically what LinkedIn does. I don't know if you know how this LinkedIn algorithm works. But when you're in LinkedIn Learning, I'm one of their author— one of the people that is done a bunch of training. I've trained like 3.4 million people on LinkedIn. And so for every, like, every moment someone spends on content I created, I get a percentage of that. But it's— if they're LinkedIn, uh, Learning customers, they paid like $72 for the year. So take $72 and divide it up across all the minutes, right? So we're talking fractions of pennies here, but still they add up. Um, but fractions of pennies. So that's what she's saying is like, hey, everyone deserves deserves. If you have something to share, you deserve to, one, have a platform in which to share it, and two, if other people watch it and are— get value out of it, that you deserve to get paid for that. And so she's creating this like really interesting, relatively new dynamic ecosystem to support, um, paying creators, paying, uh, women to do this work, you know, to share what they know and learn. So I'm testing it out right now. I'm not You know, I'm not there to get paid, but hey, you know, let's do that if we can. But I'm mostly there to test out their tech and give them advice. I'm part of their board, so I'm really excited for what they're trying to do and really inspire all of us, right? Like anyone who is here, if you have something to say, something to teach, something that you want to help women with in order to compete in the age of AI, now there's a platform you can go to that will actually compensate you for that effort and That makes a big deal, I think, for some people. Uh, I don't know, for everybody? I don't know, I don't want to speak for everybody, but it makes a big deal to me.

So I'm so cool.

I love your new Yeti. I'm so jealous. So I don't like stickers either. We are cut from the same cloth. Yes, exactly. I mean, I just got this new one. It's Caitlin Clark, and it's very— I'm a Caitlin Clark fan because she, you know, I'm sad she was out. If you don't know, she's a basketball Uh, she's now in the WNBA, but she was, uh, in college basketball, and she was— she's basically like the LeBron James of women's basketball. Um, it's a little debatable, but you know, so is LeBron. So, uh, anyway, uh, but yes, I have yet to put a sticker on it because I'm like, do I like it enough to put a sticker on it? That's my sign. If it has a sticker, it means I like it a lot. Um, so very cool. All right, so that was fun. Now the last thing, because I, you know, it's consider this, um, opening presents day. I got a new Alexa, Alexa Plus. Who has— who's gotten the new Alexa Plus? Any of you? Any of you guys? So I got a new one. First, let me tell you— oh, I don't think I even have it with me. I got a new, um, I got an advertisement that says buy a new device, get Alexa Plus. And I was like, surely I'm just gonna go and get Alexa Plus because I'm Noelle. So I just asked Alexa, um, what's— hold on, how can I help you?

That's okay. Uh-huh.

Yeah, oh yeah, for sure. But make sure you buy, don't rent, okay? Because y'all know they're gonna want to watch it tomorrow and I'll be paying $30 for a rental. Um, all right, so I bought a Kindle, the new Kindle 11 Max. One, it's our screen's already cracked. Welcome to my life. Um, two, it's bent already. I took it across the country. Where did I go? Sydney, Australia. Anyway, busted it. No problem. But it didn't qualify me for Alexa Plus, and I was like, it's the only reason I bought this thing. And so I was a little frustrated, so I had to go buy another one. This one worked. So as soon as I bought it, all of my devices in my house upgraded. Here's the challenge with that. I was talking to the AI Leadership Institute family earlier today. Um, as soon as that happened, if you have not done this yet, it automatically, without your control, changes the voice and model that Alexa uses to the degree that my son who has Down syndrome, right? He uses Alexa every single day, all day long. Like it is his buddy. And he was like, Alexa's broken. And I was like, surely it's not broken. He's like, it is not. It is not good. It is broken. And I went and talked to him. I was like, oh, that's not a good voice. Now here's the cool thing about what they did. Think through this a little bit. Not enough to not have it happen, but they thought through enough because I said, hey Alexa, I'm not really feeling your, your new voice. Can we go back to the old voice? And she was like, and instantly the old voice showed up in my sound. I was like, okay. But still now had been like threatened by it, right? Like, was like, ah, I'm not— like, he now was like a little standoffish where he had been using it every single day. Pause for one second. How can I help you? Sorry, what is your account PIN? Oh goodness, um, hold on, I don't even know really. Um, I'm gonna write it on the back of this. You need to bring this back to me. Try that. I don't know, good luck.

I hope it works.

Oh, uh, Let me see, I'm gonna give you another one. This is Live Action Family Business 3, um, I don't remember. Okay, it could be, it depends on how many numbers it's asking you for. Okay, thank you. You're welcome. I— you— yep, thank you, bye. All right, family first. Um, anyway, as I was saying, uh, my important thing, so it really messed him up. Now my dad lives on the other side of this wall. My dad also, I'm hearing him and I don't know what's happened because I don't realize all of the devices have gotten updated. I hear him screaming at Alexa and he's like, no, no, no. And he's like, what did he say? He said, I am not Noelle Russell Russell, I am Irving Robinson Russell. And I was like, uh-oh. So I'm thinking to myself, so what happened? Like, how did Alexa get broken? Anyway, uh, it all turned out like in this like transition, it's not— they don't prepare you for that. They don't prepare you for like complete upheaval. So anyway, so I bought this, uh, because I wanted the new Alexa Plus selfishly. I definitely did. My son's back to normal. My dad is also back to normal, but not without a little friction. And I feel like— so I don't know if y'all just saw, but this is, uh, Amazon Defined Created. You might have seen this in my other unboxing., a process called frustration-free packaging. And they literally brought people from the outside world and brought them into Amazon and said, what would be the least frustration to open a package? How many of you struggle with the packages when you get them? Like they're taped shut with tape that does not come off, like, or the glue gets on your hands. It's ridiculous. So this, I just tore this tiny little thing. This is what it's about, people. Happiness. I find joy in all the things. But now what? It's open. Like, I don't have to fight with it. So we're gonna see what happens. This actually requires me to have, um, a plug, but this is the new— thank you— this is the new, um, I think it's a Spot. Yeah, Echo Spot. I didn't get a Pop. Pops are the one that have color. I have lots of Pops around the house. They're all purple because I like purple. Um, oh, and I got a new shirt. Did you notice my t-shirt? It's actually not new. I found it in my closet and it's from like 2018. We created this shirt. I don't know if you recognize the symbol, but it stands for Azure AI Machine Learning. That's the symbol, or at least it used to be. And it's all of the technology in Azure. And one of my favorites, you can't really see it 'cause it's super small, but it's this word right here, O-N-N-X, ONNX. So I'm gonna teach you something as I plug in Alexa. ONNX is one of my favorite things that happened in machine learning about 7 years ago, maybe 8 years ago. Microsoft, Amazon, Facebook, Meta. It was, I think it was before they became Meta. Alright, let's see what we can do here. Alright, I'm gonna take this, put this in. Alright. So they got together and they said, what is the biggest challenge in building machine learning models and sharing them across platforms? Right? Multi-cloud AI. Now, you— it sounds probably like, of course we do multi-cloud, but it was not always this way. When you built a model on Microsoft, you couldn't move it, so you were stuck on that platform, and you, you had to do inferencing and training all on the same platform. So ONNX, O-N-N-X, for those of you who want to go research it and look it up a little bit, when you build or customize an existing model using ONNX, it separates training from inference. So now you can train your model on Microsoft and then deploy it and inference it, meaning use it on any other platform, including hosting it on your own machine.

Wow. Okay. Okay.

Oh, scan to open. This is new. Did you guys ever see that? I never had to scan before. Scan the QR code to open the Alexa app. That's the old Alexa. Escanea el código QR para abrir la app de Alexa. Okay, I'm getting confused. Hold on, too many languages. That was for you, Sandrine. Did you hear? Did you hear any of that? That was a little like, um, a little French for you. Okay, I know it's a scanner for ouvrir l'application Alexa.

Okay, to open the application.

Okay, I did pretty good there, right? Yes. Alexa is online. Just a moment. All right, cool.

Let's see.

Ah, it is nice because it's the old voice. I really like it. This is definitely prettier than all of my other devices because my devices are old. I have the very first Alexa device that was launched, the first batch of 100. I have one of those. All right, it's connecting. I don't know what it's connecting to.

So Noelle, I have a—

I want to make sure I understood what you said. Um, So like, I have 3 Alexas. So if I got the new one, then all of them would have the new software? Yep. If you get one of the new Alexa devices that has— But if I buy a—

that's what I'm saying.

If I buy a brand new one, then my Dot and my old Alexas will all automatically update? Whether you like it or not. Yeah, that's kind of the rub. So if I'd known that, I just would have warned my children.

Yeah.

It also doesn't give you like real-time traffic anymore on the new one, and a couple of other things it won't give you. It's very odd. What— it's very odd. It's, it's like two steps forward to go two steps— whatever that saying is. Um, I guess it's one step forward to go two steps back. Uh, but yeah, so which device would you like to set up? This better not be like— I think I'm set up. I should not have to do this, right, Alexa?

Are you good?

I mean, if you say so. See her voice change? I think I'm at least decent at what I do, helping folks like you navigate their day without too many digital hiccups. I feel a digital hiccup coming on. She sounds like a young adult. Yeah, ageism, Sherry. I said ageism. Like, okay, uh, she stopped listening to me because I was talking to her, but Um, but anyway, so now one of the features of the new Alexa Plus is— Javier, I know. Um, one of the features of the new Alexa Plus is that when you start talking to it, it keeps the context window open. Where before, because of the way the model was working, and I don't know how many of you know this, so I get to share like my long knowledge of how Alexa works in my little tiny window of, uh, chatting with you about the new Alexa. Um, but it would, as soon as you said the wake word, right? In this, I'll say computer, whatever the wake word is. When it wakes up, it listens for 6 seconds, and as long as you keep talking, that window will stay open. But as soon as you stop talking, it stops, it counts 6 seconds, and then it shuts down. And what would happen, especially with people who had either cognitive delays or speech delays or just were processing like slower thinkers, which is not a disadvantage, it's just the way we are. If it takes you a second to think, And specifically 6 seconds to think it would just shut down the conversation. And if you remember, it would have you start over like it had no short-term memory, no long-term memory, no con— cognitive memory. Um, and so with the new model, that is no longer the case. It'll actually stay open longer expecting you to have a conversation with it. But even if it shuts down, it is storing cognitive memory on device so that you can ask it a question and it will have the context of what you just asked it. No, I did not ask it any questions, but it's kind of neat. I don't know what to do with this screen. Uh, oh, okay, there's like a home. Oh no, that's not right. I don't know, this is new for me. Um, brightness. Let's see, settings. I don't know. Oh, so one of the things I did on TikTok, which I— now that LinkedIn is video friendly, I'm gonna do them on— I'm gonna do them again on LinkedIn. But I used to do Alexa Easter eggs. And are you familiar with Easter eggs? Have you heard of this term before? Oh my gosh, I get to change some of y'all's life. Um, so there are about, I don't know, 10,000 of these Easter eggs. They're things you can say to Alexa that are not like the data scientists put in there, like we as humans put in there for you to find, just like an Easter egg in a game, right? Where if you've ever watched Ready Player One, one of my favorite movies, um, and books, uh, where there's Easter eggs, you find them, and then it gives you the gift of being like, oh my gosh, you know this. So there are— you can Google it— there are lots, thousands of them. Um, and so I would just do them on social. So again, when you're thinking about like, what content should I create, this is one of them. Uh, find a device and just do that. Now another thing— well, let me finish this thought. But anyway, so it seems to be working, but did you notice like the voice is already different and the setup voice was not that voice? So it seems confusing to me. Alexa, how many voices do you have? I've got 8 voices to choose from, 4 feminine and 4 masculine. Want to try one out? Is there a gender-neutral voice I could use? Beep beep. Nope. So I love that. I love that some things still stay the same. When you say something to Alexa that no one's ever said before or they did not plan for, that's what happens. It just goes, did you say something? Alexa, is there a gender-neutral voice to choose from? I don't currently have a gender-neutral voice option available. The voices I have are categorized as either feminine or masculine. Would you like to try one of those? Uh, no, I'm good, thank you. You're welcome, Noelle. All right, I like it, I like it. So what you can notice is that that is 100% an LLM talking to me. Um, and I don't know if you know, but the LLMs that Microsoft— or that, um, Amazon was working on to build into Alexa Plus for the last 3, 4 years imploded, did not deliver, could not reduce latency in enough time for a real-time speaker experience. So Amazon specifically, I guess Andy Jassy, decided to go in a different direction and adopted Anthropic. Did you know that? Anthropic.

What?

I mean, good for Anthropic. I'm happy for them. But I will tell you, there's a lot of data scientists that are like, my baby, it's my baby. Like, they worked hard. We gave away moments of our child— you know, of our kids' childhood. We sacrificed to deploy for to now 17 language models on Alexa. Another fun fact as we just wrap up this portion with Alexa is that there was a decision that I often find organizations making right now when it comes to AI. Do you build one big model to rule them all that does all the things, or do you build a bunch of small models? And on Alexa, this was specifically for languages. Do you build one huge model that knows all the languages, which is what we have now, by the way, ChatGPT, right? Knows all the languages. LLMs. Know all the languages. Do you do that or do you build one model that you can then carefully guard and, you know, basically audit and ensure that that single model for that single language is as good as it can be? So that was a decision, a leadership decision that they had to make at Amazon around, do we build one language model per language or do we build a huge language model? We did not have the benefit of transformers. Right? The transformer architecture didn't get released as a paper until 2017. Alexa launched in 2014, which meant it was in development 10 years before that. So we didn't have transformers, we didn't have short-term memory, long-term memory, we didn't have any of the beautiful things we have now. So as a result, they made the command decision to actually build one language at a time. So English, French, Spanish. I'll just tell you, when I deploy my Alexa skills, I deploy them in 5 languages. All of them are English. So it's like Indian English, Irish English, American English, Canadian English, and English English, like London English. Crazy. That's 5 just for English. And a different model for each one. So when I was, I was just recently, you know, in at Johnson Johnson, similar questions, right? Like, do we build one big model that handles all of these things, or one little model, or a bunch of little models? I actually think it's a great question to take into our Q&A little, uh, breakouts today. I'm not gonna give you— I'll give you my opinion after, but I think it'd be a great conversation to have is like, what do you think? Like, what, what would you do, A, if you could— we could do it all over? Um, maybe it's easy to say now, but do you think you should build monolithic agents? That you, you know, like these are two different architectures. There's no wrong answer, right? One architecture is you build an agent and inside that agent it might call out to a bunch of tools, but it does it all in a single agent. Or do you build a planner architecture, right? Where you have an agent that doesn't integrate directly to tools, it calls out to a planner and the planner then dynamically in real time figures out who to talk to. Like they're all different and You know, different benefits for each one. So I'd be curious what you'd go with, one big model or a bunch of little models. And if you think there's a difference, why there, there are some, definitely some interesting differences. Some of us will know more about that than others. So I'd love for you all to, to share that with each other. But, but yeah, so I shared with you all of my toys. Happy, like, it's like Christmas here for me. Um, I did get one thing I could not open in front of you. Which is like a sign. I feel like when I was, um, earlier in my career, I would watch people on YouTube and they would show presents that they got in their mailbox from like fans, and I was like, that must be so cool. So I got a— a fan sent me a gigantic box, two of them actually. Um, one person sent me a gigantic box, it was like 4 feet tall, and I'm like, what is this? And inside of it, it actually is something you install over, it's AI-enabled, sensor-driven. It's deployed over, you know where your stovetop is? It's deployed right on top of the stovetop and it provides a sensor-driven fire suppressant. So if there is a fire that triggers a sensor in above the stove, it will immediately zoom. I don't know how many of you have ever experienced like an oil fire in your kitchen. It's a bit terrifying. I guess maybe some of us in a certain culture, we maybe, uh, cook with more oil than other people, but, um, but, uh, what I, whatever you cook with, uh, but if you ever have a fire on a stovetop, like, and open flames, right? Especially if it's your kids who are cooking, um, or your kids are around, like, it's, it's very, it triggers panic. Like you're kind of like, oh my gosh, we're gonna die. It's kind of the way I think about it. And so when this person heard me speak, and the real reason that they sent it to me was because they were like, you have a son with Down syndrome, if you really want to allow him the freedom to like cook, then this is a great thing to have because if anything goes wrong, you know that he would be safe. Um, it is like a fire retardant kind of thing, right? Or, uh, you know, like, um, the stuff that's inside a fire extinguisher, right? So it does make a mess and all that, but It solves the problem. Uh, and, uh, so I got that. I didn't install it yet, but that was really cool. When I opened it, I was like, what? Because we talked about it, and then he's like, oh, here you go, I sent you one. I'm like, oh my gosh, this is amazing. Another thing he had was super cute little discs. I didn't get one of those. I think he— they might be in the box. Um, but there were little discs that did, uh, moisture detection. And what I liked about them is that they would— you set these discs Okay, I live in Florida, so a few y'all know what this is like, but sometimes in Florida your air conditioning will break, and when it breaks or, you know, it gets clogged or whatever, as soon as that happens, the condensation that can't leave starts pooling up in this plate, like this basin. Some of y'all know this, um, and when that basin fills, it basically will— like, it just— it's supposed to stop because there's a little plunger thing that's supposed to go up. It never goes up in my house ever. We always— my son who lives, who is downstairs will come up and be like, so my bathroom is leaking again. And that's because he's right above the air conditioner. So if it overflows, it's right into his bathroom and it leaks through his, his vent system. And so this is amazing because now I can stick that in there and I have an app and I have notifications turned on. I don't turn them on for everybody, but I turn them on for this thing., and it now alerts me and says you have moisture as soon as the first drop. That means I get to know as soon as it happens that the air conditioner has stopped working. Um, sometimes it is hours before I know, and you know, water damage isn't cheap. So yeah, what was that? The new 340GS military-grade— ah, One Hour Storm here, and today we're reviewing the Amazon— oh no, okay, hold on, hold on, hold on, hold on. What am I— what's, what's happening right now?

I'm like, wait, What?

Um, okay, I'm gonna mute y'all, but unmute yourselves. But I feel like somebody's accidentally muted. Um, so cool. So anyway, so that's exciting. That thing, I haven't seen it. If I actually do decide to— well, when I install it, I will do a quick video so you all can see what I have, um, what I've installed there. But those— I think this is an interesting time because guess what? Those little things were 3D printed. On demand. That's, that's the world we live in right now is that I have a 3D printer right here. I don't know if I could 3D print a sensor that detects water, but I probably could. Right? So there's so little distance between, man, I wish we could have X and then actually being the person that provides that X to the world. And Nitin is this guy, right? Like Nitin is building stuff with 3D printers all day long. Um, so I mean, every, every week we're like, yes, what's up, good to see you. Every week we're like, what, what cool things have you built today? Um, but it is cool, right? Because you have an engineering mind, right? You are an industrial engineering mind. Like, you, you now have the ability to build things without needing a $100,000 or $100 million infrastructure to support the development of that. I think that is That's one of the most inspiring things. I am not like a physical builder, like a— what do they call it— like a, um, civil engineer. I'm not that kind of an engineer, but I'm very tempted, you know, now, because I'm like, wait, I could build something. Like, I've many times— what was it the other day? It was, uh, the Millennium Falcon. Y'all know, right? You can actually 3D print all the pieces to the Millennium Falcon and like build your own. And I was like, might want to do that. Um, I don't need to do it, but I might want to. Uh, so anyway, so I think there's— and of course I'm not obviously wearing Invisalign right now, but I probably will be on the retainer system when I get my braces off. And Invisalign is another great example of AI plus 3D printing meets an industry that was ripe for, you know, innovation. And so all of us, I think, can look around and find industries that just need some help And I don't think we're far off from just taking an idea and turning it into something that, that actually could scale. So wonderful. Nice to have you. I hope you guys enjoyed my little, my little, uh, collection of cool new things that I got. I, I certainly did. Um, any questions before we break out into our breakout rooms? We got about 20 minutes left. I like to take a few questions if there are comments or updates, wins. If anyone has a win to report, you can do that. Otherwise, I'll be telling you my wins, but you know, that gets old after a while. Noelle Russell, okay, I've already gotten my Amazon description updated now. So there's a lot of security. Oh my gosh. Well, this group might not know. So over the weekend, I picked up Renee's new book called The AI Imperative. It's on Kindle, and it was excellent. So it talks about security in a way that just, you know, it's the hard stuff, right? Like, it's the stuff that nobody really wants to talk about. I think, Talisha, you're the kind of person that asks these kind of questions. Uh, when, you know, it's the person who at the beginning is like, so, um, have we, have we thought about how this could go wrong or what could happen or what could break? And, uh, and so it was, it was so good though. Like every, every chapter I was like, oh my gosh, yes, yes, some more. And it takes a little— I always tell people, you get a, you get one chapter for free. Like I will read, I'll buy your book no matter what because I support you, but I will read the first chapter. And if you got me then I'll keep reading. And I kept— I finished the whole thing in the weekend, so it was really, really good. So I wanted to encourage y'all, if you're into kind of security and red teaming, I think it fills in a lot of those gaps. It is literally security for AI leaders. I loved it, it was so good. Um, and there's already a part 2 coming, so I'm super excited for that. Uh, but yeah, Joe, do you have a comment, question? What do you want? Yes, come on in. Oh yeah, I— so I have a win of the week.

Oh, what?

Let's go. So I can't announce too much yet, but with my product development, I have now trademarks going in 5 different regions: the US, the UK, Canada, Japan, and China. Wow, that's amazing.

Yes.

Congratulations. What's up? That's amazing. Wow. Cool. Okay, well, we can't wait to hear more about it as soon as it's like unleashable. We would definitely want to be the first. Consider us your like OGs, founding members. Like, come bring it. Let us know. What? That was cool. Somebody just did some exploding clapping. I don't know how that happened. Me? That was you. Oh, of course. Talisha's always teaching us stuff. How's that? How'd you do that? I want to do that. That was pretty awesome. But yeah, really cool. Really cool. All right, I'm trying it. I'm trying it. I do have a question. Oh, yes, please. Come on in.

So I am on an AI project in my company, and we are looking for agentic AI benchmarks. And I'm looking for that information. So here is the perfect place to ask that question. So any resources would be greatly appreciated.

So benchmarks for like how models are performing today, or—

Yeah. For anything that has just benchmarking qualifications. So how well they're performing, token utilization, anything of that nature so that I can get— I'm trying to benchmark a product against standardization, which again, obviously changes every 5 seconds.

I'mma let my people. See if, if anyone has an idea before I tell everybody what I always tell everybody.

Okay.

Um, anyone got, got some ideas on benchmarks and like who does that?

I'll tell you the reason why we're doing this. This is really just more of a starting out again. I'm in Civic Tech at this moment, the company that I work for. So obviously when you're building something, the very first question is like, well, what's the standard? And we're like, there is no standard. But you can't say that, right? You can't tell people there's no standard. And I'm from Pittsburgh, Pennsylvania, and my coach always says the standard is a standard. So that's my answer to the question.

Oh, Nanda, I like that. That's a good answer.

Um, but this is, you know, that doesn't work either because some people aren't Steeler fans. So I was looking for, you know, something more, uh, you know, cognitive, something more, you know, for, for those— my, my audience. You always have to address the audience's question. So that's what I—

absolutely. I love it. Um, and I'm so glad because it causes us to, uh, rethink and restate some things. And who knows, we maybe never covered it here. I never know what we've covered because we've been doing this a long time. So, um, but yeah, so I love that. Um, what? Really cool. Yes, exactly. So these leaderboards, before we talk about Joe's reference that he just threw in, these leaderboards were founded by Stanford. You're familiar with HELM, right? The Holistic Evaluation of Language Models. That's the benchmark of benchmarks. The benefit of that though is that you can go get HELM Lite and download it onto your computer. It is a GitHub project, so there's a little technology involved, but those leaderboards have now shown up across the board in public cloud provisioned environments as well as, uh, the artificial— what is it— analysis.ai. So they all have kind of grounded in— and now this goes to something why I said this earlier in a different call, but one of the challenges with some of these benchmarks, which is like I think what you're saying, Talisha, like there is no standard in the sense that it's very subjective what these numbers even mean specifically around tokenization, because the art of tokenization is completely subjective. What a token is, is different on every cloud platform. I don't know if y'all knew that. So the best way to test this out is to just take a prompt, same exact prompt, go into a platform like Poe.com or, oh, what was I using today? Fall.ai, F-A-L-A-I, I guess.

Fal—

how do you say that?

Fal—

I don't know. There's some phonetic thing, but, um, fal.ai. These are model aggregators. You take the same prompt and put it in, and you'll find out the number of tokens that you use is different even though it's the exact same prompt. And you're like, well, how's that possible? And that's because tokenization itself is completely subjective, right? Some, or, and, and there's, uh, one of the things I loved about my educational process of learning about tokenization at Stanford was that we went through this activity of actually building our own tokenizer. And, and, and anyone could do this. It's a class, it's a free class on online. You can go to YouTube. But when you build a tokenizer, you realize like it's your job to decide how you're gonna break up a word. And this is all comes down to cost. So for those of you not familiar with tokens and what tokens do and what they mean, like tokens is the currency, which is why Talisha, you're probably so curious about it, like, what is this gonna cost us and how do we estimate what that cost looks like? And the unfortunate thing is we're in such infancy in enterprise deployments that every company has their own tokenizer. There is no standard tokenization. It's like cloud computing in the olden days when every cloud provider priced everything differently. Now that's not the case. Things have been commoditized, but in the early days they were not commoditized. You could have like a $300,000, uh, cloud bill on Microsoft and it would cost you like $45K on AWS and you'd be like, well, it's the same. I'm doing the same thing. I don't understand. But they're all, they're all up on the, they, they all got smart on us there and now they charge us the same higher number for everything. Same thing will happen probably. But when you build your own tokenizer, there are some trade-offs that's happening, right? So think about, and, and this is just a quick quick crash course in tokens. If any of you are, and if you know more or want to contribute, please feel free. But when you build, you know, we call them token strings, right? Or streams, right? These collections of tokens basically are the words, right? I think if you go to any class, it'll tell you like, the big brown bear eats cheese. That's probably brand new. I think I just created that one, but they all have a sentence that they use and then Tokenizing is how do you turn words into numbers so an algorithm can interpret it? And that's chunking it up. Now there are a bunch of ways to do that. You can actually do it by word. The challenge with by word is that the chunks are big and so it takes a long time to process latency, right? The, so now we're like, okay, well why don't we break up words? Well, you can break up words, but if you break them up in the wrong way, you could lose the context, which means again, you're going to get a faster processing, but longer time to actually figure out what was actually meant. One of the values of tokenization is that you want to break it up in a way that allows you to also maintain an understanding of what's being said and high levels of accurate predictions. One of the tokenizers that OpenAI developed, which they use to this day, actually accommodates misspellings and miscasings. But in order to do that, that means when they tokenize a word, they have to identify whether or not it's spelled correctly or incorrectly, what that means is that it's gonna take longer. It's gonna create longer token strings. So it's more expensive actually to use that, but it's better because when we misspell something, it doesn't fail. And so we don't complain cuz it's not failing and they get a higher, they, they have more tokens to, to process. But I think it's interesting because it is a bit of a, I think like a little gray cloud. Like I don't even know, Maybe it's not gray, but it's definitely opaque, this tokenization process. And most people don't truly understand it or the trade-offs that are made. And the reason for that is that you don't ever do it, right? You're gonna hire a company like Amazon, Google, or Microsoft, NVIDIA, and they're gonna tokenize for you. But it is important as a leader to know that that tokenization is different on every platform and is subjected to however that company has decided to do it. And so that's a question we get to ask. How do you tokenize? What is the philosophy of your tokenizer? Because they should be able to tell you, oh, we did it this way in order to achieve these gains and as a result took these trade-offs. And you should know that. So that's, I think, gold from that perspective of your question. Most leaders do not have any idea, like they're just flying blind. But Helm also turned into leaderboards at Microsoft, NVIDIA, and Google— or I'm sorry, Microsoft, NVIDIA, and AWS. Google is different. We'll just leave it at that. But, but if you go now into Azure AI Foundry, right, you will see now when you go to pick a model, a leaderboard pops up. Now there are like 70 leaderboards all on different metrics. So I always tell people the first thing you're going to do when you start an AI project Right? The first thing you do is figure out how you're gonna measure success. And so those would be the metrics I look at. And it might be accuracy, it might be— I mean, if you're building a system that is supporting, like me, I have, uh, JusticeGPT. It's an LLM that I'm building for nonprofits. Well, we built it already, but we're constantly working on for nonprofits to ensure justice, right? To ensure that when they use it, that they are basically getting supported. Let's say it's a person in prison and they're trying to figure out what are their rights. It's a model that would give them their rights and help them understand them. It also would be given to their families, right, for their families to understand. Some of these models are predisposed to not provide information to certain demographics, to certain people in certain situations, right? Like, every model is trained. I don't know if y'all know that. I hope you do., but every model's trained. I, I just did this. Where was I? Oh, yesterday. Um, I was on a call with Ensono. Uh, I was with their ERG group for women in engineering, and I was talking to them and I shared with them something that I hope you— I want to ensure all of us are on the same page about, which is one, when you use a model for free, you aren't using it for free, right? Everyone know that the currency upon which we transact, if we're not paying money, Everyone know what that is? What is that? Mention us.

We are the—

yeah, yeah, your soul. Exactly. You're paying with your soul. Um, but your data, right? You're paying with your data. You're paying with information. You're paying with, you know, your questions, your queries, your answers. All of that is the currency upon which you're— so it might make more sense for you to pay $20 and own that and keep it. It makes more sense to me. I mean, it makes a lot more— I'd I pay a lot more than that to keep my data private. I set up a Microsoft tenant. Do I need to do that for Noelle's little tiny company? I don't, but it allows me to guardrail my environment and keep my data close. So that being said, that, that portion, uh, I always like to flip the script, like, hey, make sure if you are using a model, don't use a free one. Now I know, Talisha, I'm preaching to the choir. I know. But the second part is, is for, I think, this this benchmark perspective is that when you use a model, the instructions you give it are critical, right? So going in and custom training that model on the metric you're gonna use and how you're gonna use it. So those benchmarks are based on open models, so they're not even super relevant, which is why there actually shouldn't be a benchmark for what you're doing cuz you're custom building that model for your solution. Mm-hmm. So you can use those big benchmarks to make sure you're at, you're not using a model that's sexist, racist, hurts people on purpose, is inaccurate, um, promotes self-harm, right? Like you can at least pick one that's high in ranking on those type of characteristics. But when it comes to accuracy, like maybe you wanna know how accurate it is with no training, but I've never deployed a model we didn't train. So what I told these, uh, women yesterday is I said, if you are not the one, and every model has this, you have the ability, we have the ability to go into ChatGPT and we can write, We go into custom instructions right in the settings and we tell it here the 3 Ps of prompt engineering. It's the shortened version for those of you who know my 5 Ps. But every model, I should be telling it what its purpose is, right? The metric I'm gonna use to measure its success, who it's for, who it's not for, and the portfolio of data retrieval, augmented generation. That's the baseline. If you're using a model and you didn't do that, it's important to know someone else did that work. Someone else has told it who it's for, and it probably isn't you. Someone else has told it what it's supposed to do, what its purpose is, what metric it's driving for success. Those are all black box to you if you don't actually overwrite that training, right? So I think, you know, yes, benchmarks are good and you can at least avoid big rock problems of like buying into a sexist model that'll hurt your business. But at the end of the day, I think it's more important to build a, build a model, customize it with these instructions, and then create your own baseline, right? Snap the chalk.

Well, that's what it's for. It's for, I always say too, you gotta show, then you gotta tell, and then you gotta snap again. So I'm gonna show you where it's going to be. I'm gonna show you how mine is a differentiator, and then I'm gonna show you how my customization worked for you.

Yeah, yeah, exactly. I love it. I mean, that's like solution architecture. Like 100%. So yes. So yeah, thank you for asking that question. I mean, I over-answered it, but I found an opportunity to share some other stuff.

Always.

Always. Alli, there you go.

I threw it up there for you.

Perfect. Perfect. Awesome. Javier, come on in. Do you have more advice or questions or comment?

Actually, on the same topic about evaluating the model, I was doing actually some LLM testing last week with my counterpart in Chile. And we also were testing it not only from the part of the evaluation of the standard, but also the response time and the performance of the response time. So it's like, okay, you run a test of the API and it's like, okay, it gives you an answer. Excellent. So you have— there are certain levels of performance that's supposed to— they're supposed to actually go by, you know, based on the API architecture. And this is highly technical, but we figure, okay, you get a response, but was it the right response? Were there issues with the response? And in the analysis, you could see, like, because LLMs are probabilistic in the responses, yes, you get a response, but the response might be incomplete, maybe actually words missing on the response. So that means also that the performance of the LLM to the API, especially when it's on a high load, then it's not, it's not going to be totally accurate. There's going to be some level of performance. So that also falls into like, okay, if it's not effective in this response, that means there's less tokens that is actually consumed. Therefore, you can see the value of, you know, how important it is to test the LLM and whether you should consider that in your evaluation of the LLM you're using for whatever purpose or business case you have.

Absolutely.

I love that. Thank you. And it also reminds me from AI evals. So is it in this community? I hope it is. I think I posted an AI evals YouTube class that recently was released. It was very good. As a matter of fact, they created a Maven class. It's like $1,500 or no, $3,800 for a Maven class. I was like, I need to raise my prices. But it was on, on just evaluations. So, but the YouTube, like, podcast style version was also very good. So that's in the, in the community. I'll see if I can find the link and pop it in here. I'm looking through the chat to make sure I didn't miss anything. Oh, here we go. Oh yeah, good. Yes, me too. I also Kindle Unlimited. That's how I got it. Yeah, it was awesome. I'm such a big fan. And then Deb, yeah, I see what you meant, what you put in the chat. Now I will tell you And I'm just going to say this, I'm going to actually tell you a story about Alexa, and then you can like attribute it to however you want to this, this perspective on whether or not this option is, is true. So on Alexa, there is, uh, this little button on the top, and it's a mute button. It allows you to basically put it in do not disturb. However, you can also— now when you do that, you actually physically press like you can feel, uh, some kind of switch turn on and off, which is hardware inside of this. The reason that that is there, because in the beginning, if some of y'all remember, we could just tell Alexa— oops, I should probably not talk about it, um, but we could just say, Alexa, mute, right? Or Alexa, do not disturb, and it would turn off. But it was a software turnoff, which meant that, uh, I mean, if subpoenaed, We might have information that we weren't supposed to be listening to, but we were listening. We just weren't actively listening. Like there's a little bit of gray area when it's a software decision deciding whether or not it's training data, right? And what we end up finding, ask Snowflake, that later somewhere down the line someone's like, oh, an engineer inside OpenAI is like, oh, I thought I did that. I didn't do that. I'm working 18 hours a day, you know, like I thought I closed those, you know, blinds. I thought I shut down access. Like if it's possible, meaning if it's a slider in your environment, data leakage is 100% possible. I'm not saying, I'm not saying it's happening, but if it's a slider, it's literally like me being like, Alexa, mute. And it's like, sure, I'm muted. But how do I unmute it?

If you're— that's good to know. Thank you, Noelle. But let me ask you then, so now Alexa has a hard button that you can trust?

Yeah. Oh yeah, good question. So, um, this button was created because Amazon wanted to sell a bunch of these to the military. And if Jason was on here, he would agree. Um, and they required a hardware schematic that showed that when they pressed mute, it was muted. Like it was a hardware mute, not a software mute. And so they changed, they re-architected the entire thing to support that contract.

That changes everything because we don't use any of those devices because we don't trust that it'll shut off. But now you're saying we can rethink that.

Yep.

Absolutely.

Thank you.

I'm going to go for the—

Go get it out of the closet. All of you get them out of the closet. Oh, we have trick-or-treaters. Hold on, uh, oh, hold on, I'm gonna mute myself and scream at my kids. I'm gonna go get these kids candy. Hold on one second, talk amongst yourselves, I'll be right back. There's candy above the fridge. To ask everybody, did you guys all watch the Stanford videos that, um, Noelle shared?

Mm-hmm. I thought it marked them.

Yes, I actually, um, the day that she shared it, I watched one of the first one, and that same day I had class at my, uh, natural language processing class, which the professor went over it, and Stanford was a hundred times better than my professor explaining it. So I thought it was really interesting. Oh, Noelle Russell Russell, what candy bars are you handing out? Little ones or big ones? Oh no, we do, um, I mean, I always give candy that I like to eat. And so it's like Willy Wonka stuff, like Gobstoppers and, uh, all like no chocolate. Although, I mean, I don't mind chocolate, but I really— yeah, so I will show you. This is like a big tell. So this is my candy lantern from Great Wolf Lodge. Luckily, I mean, I came back like 5 days ago and it's still pretty full. I mean, I might have refilled it, but it's fine. Um, anyway, but yeah, so it has, uh, Runts. Do you know the Runts, like candy with all the little— anyway, some of y'all know. But yeah, so we give away that kind of candy. Yum. Um, and so these kids came. It's like 4 o'clock, I guess it's 5. So our community, um, anyway, it's very interesting. They like block off the gate with like these metal gates and they're like, only people from inside our community. And, uh, I was like, I didn't know what they meant. And then I saw the kids, like huge gaggles, like 50 kids going around from community to community. It's nuts. I don't, I don't even know. I'm turning into one of those people, get off my lawn. Um, but we decorate it. It's beautiful. Uh, I will have to show you all. Um, awesome. Okay, well, let's get to know each other a little bit. So this is our chance. Um, what was the question I posed earlier today? Oh yeah. Big or little. If you don't have a topic you wanna talk about, this is your chance to just like, think about like if you had to build a model, um, and let's say it was solving a big problem. Would you build one model that has a bunch of tools? Like, this is a design pattern for agents, tool use design pattern, agentic RAG. Like, would you build it all in one, or would you build a bunch of autonomous little pieces and assemble them, uh, through an orchestration? How would you do that? Happy Halloween. Um, okay. And so I'm gonna throw y'all into, uh, breakout rooms, get to know each other, say hi, what you do, uh, who you serve. How you serve them. Usually it's something like, I'm Noelle Russell and I, you know, help people who are in transition in their careers learn a little bit about AI and have fun doing it. And I do that through the AI Leadership Institute. See, 30 seconds, that's it. Just so people know how to lean in, how to contact you, what to contact you for. If you have something you need help with, also feel free to just be like, I don't need you to know about me, I just need help. And we're here to help you. All right, here we go.

Here we go.

Hold on, I'm assigning automatically. I created one room for all of you. Okay, here we go. If you see a bot in your room, I will do my best to eject that bot. Oh my gosh. They almost all went to the right room. That was pretty, pretty impressive. All right. All right. Cool. We're doing pretty good with our bots these days. So I'm gonna put you over here. Um, and you over here and all right, here we go. So I'm gonna give you 10 minutes. We'll come back and say goodbye and then go trick-or-treating, or you can live through my trick-or-treating experience that I'm about to go on. All right, here we go. I'll see you in a few minutes.

Bye!

Click the blue button or whatever color it is. Say hello. And I'll move people around if they don't join. Okay, we got good, good group there. Good group there. Okay, yeah, got it. Well, there's a lot of people in that room. Oh, room 4. I'm gonna move Tara, I'm signing out. I did the wrong thing and got 3 vaccinations this morning, so I'm kind of feeling it right now. Yes, well, I'm glad you came. Thank you very much.

See you on the next call.

Absolutely. And send me any— if you have questions, just let me know. I'm happy to help.

Thanks.

Bye-bye.

All right, I'm actually going to talk to my LinkedIn family. If y'all are here, thank you for being here. It's so nice to see you all. Thank you for, for, uh, following us and hanging out with us. This is our weekly call where we get to just talk about all things artificial intelligence. We have lots of people of all different capabilities and varieties, um, This is a community, it's called I Love AI. You can join it. It's very easy to do. Just head on over to noellemyname.ai, N-O-E-L-L-E.ai. If you wanna be on the Zoom and be able to break out into breakout rooms, connect with other people live, get to know them, have them know you. You never know. We've had hiring managers meet with people. We've also had, um, entrepreneurs meet other entrepreneurs, business partnerships. Like, it's amazing. It's all about who you know, and that is the gift I get to give here every week. Every Friday we meet here. So thank you all for being here. Um, I am going to— hold on, let's see. Um, okay, there we go. Um, so thanks for, for being here, and, uh, I hope you are enjoying. I mean, I didn't talk a lot about the latest news. Some of you come just for that. It's been an hour, so you probably have already left if you were interested in that information. Um, but I got a lot of surprises this week, so I wanted to share them. Uh, but some really cool news is around an article you'll see in the AI Leadership Journal. If you have not subscribed, that is available on LinkedIn. We have over 15,000 subscribers. Uh, and this AI Leadership Journal, this week's edition is all about what we need to think about as leaders. Like, what do we need to think about if we have companies, right, attaining trillion dollar growth, right? We have a $5 trillion market cap company, right? NVIDIA has done some incredible things. Uh, but what do we need to be thinking about in a world that is achieving this growth this fast? What are the questions we should be asking? What are the things we should be doing? And so I wrote an article, uh, check it out. Uh, you'll also see it show up in your feed, hopefully. Um, and you can always find me on Instagram, noelle.ai, TikTok, Though I don't dance on, uh, also noelle.ai. And of course, just visit our website, join our communities at noelle.ai. So really cool stuff. Hope you're having fun. Uh, go out there, be brave, build something awesome and tell us about it. We would love to support you. So thanks for being here. Thanks for showing up.