EP6: Tesla's Cybercab, Wendy's AI Drive-Thru, Google's Nuclear Deal, Agentic AI with Rabbit & Swarm

In Episode 6 of AI With Friends, we explore the latest developments in technology and AI:

• Tesla's Cybercab and Robovan Unveiled: We break down Tesla's "We, Robot" event, where Elon Musk introduced the Cybercab, a two-seater robotaxi version of the Cybertruck, and the Robovan, an autonomous bus for passengers and goods. We discuss the potential impact on autonomous transportation and Tesla’s production plans.
• Wendy's Expands AI Voice-Ordering: Wendy’s is rolling out its AI-powered voice-ordering system to 80 more drive-thrus across the U.S., using Google Cloud's advanced language models. We examine how this tech is reshaping fast-food ordering and its implications for employees and customers.
• Google’s Partnership with Kairos Power: Google is investing in nuclear energy by purchasing power from small modular nuclear reactors developed by Kairos Power. We explore how this move supports the growing energy demands of data centers and AI, and its role in tech’s shift to sustainable energy.
• Advances in Agentic AI: We dive into Rabbit's Large Action Model Playground and OpenAI's Swarm, exploring the future of autonomous, collaborative AI agents and the challenges that come with them.

Join us as we discuss these exciting trends and their impact on the future of AI!

----------------------------------------------------

Welcome to AI With Friends, your weekly launchpad into the world of Artificial Intelligence. Hosted by Marlon Avery, a pioneer in GenAI innovation, alongside Adrian Green, VP of Engineering at LiveNation, and Sekou Doumbouya, Senior Staff Cloud Systems Engineer at Pinterest, this show is your go-to source for all things AI.

Our hosts bring diverse expertise—from AI strategy and tech innovation to industry leadership. Every week, they break down the latest AI trends, interview top experts, and simplify complex concepts for AI enthusiasts, entrepreneurs, and tech professionals alike.

Marlon, Adrian, and Sekou combine their unique perspectives, whether it’s Marlon’s collaborations with tech giants, Adrian’s leadership in global entertainment engineering, or Sekou’s cloud systems expertise. Together, they make AI insights accessible, actionable, and exciting.

Tune in live on Twitch & YouTube every Wednesday at 9:00 PM ET, or catch us on all major podcast platforms.

Here’s what you’ll get:

Cutting-edge insights from AI leaders
Real-world applications of AI technology
A vibrant community of forward-thinkers

If you're ready to stay ahead of AI trends or spark your next big idea, join us each week for an hour of engaging, thought-provoking content.

Subscribe now and become part of the future of AI with AI With Friends!

Transcript

Hey, hey, hey, hey! What's going on, fellas? Hey, what's going on? Welcome back, welcome back. Man, it's been a fun, exciting, curious... you know, week. Hey, Jermaine, what you do over the weekend? Nothing much. I actually dug in a little bit of the, you know, second brain idea. So did some research and experiments in web scraping and storing all the findings into a vector database that can be queried later. That was my little weekend project. Got all cozy at the coffee shop and I tried to knock that out really quick and got to, I'd say, level point zero zero three on it so lots of work to be done you know my my I last weekend I spoke at a black minute tech conference and stuff uh had a good friend on stage with me xea oh yeah so you know you know x is a uh She's been doing AI longer than all of us combined, actually. So I forgot that she did AI in the military as well. And so she was working on some things there. And so we got to share the stage together. That was awesome. And then also a gentleman by the name of Armando as well. And so that was good. I did a workshop while I was there as well. I did some building myself, been, um, been working a lot of, uh, you know, kind of figuring out like how some of these, uh, these AI voice agents can kind of take over, basically become, you know, a full throttle customer service, um, a customer service, um, you know, platform and everything. And so, uh, yeah, so I've really been diving today. I think that's, I think that's, there's a lot of opportunity and stuff on the voice agents side of things. And so, so yeah, Sekou, what about you? Oh man. I had a pretty good weekend. This is one of those weekends. What do we call this in the Bay Area? Oh, yeah, it's called What Is That Sound? Which is essentially Fleet Week. You hear a loud boom. You're like, what is happening? Oh, it's Fleet Week. There are airplanes flying over my head. But outside of that, project-wise, I've been fooling around with, was it Notebook LM? It's a good time. It's a good time. I've been adding in articles and research papers that I probably would not normally get a chance to go through. and adding them in and just setting up a podcast a little mini virtual podcast to discuss the topic and then like walking around and working on my house like That's a good time. That's a good time. Amazing stuff. Amazing stuff. It's like, it's a matter of time before I just dump what I'm doing and just like, just go full notebook LM from what I'm trying to accomplish too. Um, and from that podcast, uh, we, you know, played a little bit before the show, it was impressive and it's motivating. It definitely feels like a new way to kind of like learn. I feel like I absorbed information that I normally would not absorb because of that format. You know, I put my pinky toe into a real-time API, and I said, yeah, let me stop this now because this could destroy my entire week and stuff. I am going to build some things, finish some things tonight. and stuff with it. And then my next thing after that, I'm going to check out notebook.lm. I think some interesting things, stuff over there. There's not an API available, right? Is there? No. No API right now. Everything is kind of just self-contained in there. Which means Google is about to what? Oh, hell yeah. Enjoy the freeness because that's going to get you. There's a Discord channel that is very, very active. Folks are doing all types of interesting things with Notebook LM and using it for all types of things that I wouldn't imagine. Actually, I was using it for doing a deep dive on my career, which, you know. Maybe we'll go into later, but it was interesting. If you ever want to get a good pick-me-up, let's put some information about your resume in Notebook LM, set it to play, and just watch these two AIs talk you up. It's a good time. Yeah, it's I mean, I think I think that itself can be a full discussion and a full segment. You know, man, we're we're so we're so so at the bottom. at the beginning stages of this thing, everything from the, you know, the agents to the voice, you know, assistants being able to use LLMs on the voice side to help us with things like getting a kind of like a boost encouragement stuff there, helping our resume, helping us with recipes, you know, helping us walk through the structures of how to put those unnecessarily difficult IKEA desks together. And so, you know, we're definitely there. And I think, I think right now the future of AI and a lot of those things, all is encompassed around how creative individuals can get. And so I think the technology and the foundation is getting to a place where it now is starting to submit itself, which is interesting too, because that's where the work starts and that's where the work starts to build. And I would argue that ninety-nine percent of America, when it comes to small businesses, have even started. know this process which you know impacts that's where you start to impact you know the gp you know and the vast majority of individuals and so um I think it's an interesting future and I'm excited to see what happens and stuff here yeah same here okay speaking of future uh tesla's kind of gave us a indication and a outlook of what the future big underlining word maybe so here we go tesla unveils the cyber cab and the robo van at the wii robot event test has a fit has officially unveiled its cyber cap a smaller sleeker two-seater version of the cyber truck slash model three designed to serve as a robot tax that's robot taxi um at the wii robot event held at warner brothers Discovery Studio, the CEO Elon Musk showcased a fleet of cybercabs, demonstrating their capabilities on the studio streets. Musk emphasized that the low operating cost of autonomous transport predicted it would be around twenty cents per mile and announced that the cybercab would be priced below thirty thousand dollars. Aiming for production by two thousand twenty six or before two thousand twenty seven, the vehicle features Gold wing doors, kind of like Lamborghini doors, and will use this inductive wireless charging. Tesla also surprised attendees with the reveal of the robo van, an autonomous bus designed to carry up to twenty people or to transport goods. Through no production timelines were given. Additionally, Elon introduced a dozen of optimist humanoid robots, highlighting a future where affordable robots act as helpers and companions. These events could mark a significant step to Tesla's aggression push forward to autonomy, despite the previous delays in achieving such underserved capabilities. But here we are. Tesla's kind of gave us a step into the future. Adrian, man, we'll start with you. What's your what's your thoughts and stuff for the world's up here? I think it was optimistic. The you know, the rollout of the robo van and the optimist robots. And that's cool. I think that the vision of that, you know, future is is pretty interesting. having you know our boots are more on the ground when it comes to the um what's going on in ai and where we are so far um I you know was hearing a lot of this you know in the clips that I saw from the event about you know it's going to be completely autonomous And it's my understanding that right now Teslas aren't able to be fully autonomous yet, the Tesla vehicles. So they still need the driver interaction. And it was my understanding that there was more of a canyon to cross when it comes to being fully autonomous. Because when asked, you know, he was like, you know, oh, yeah. You know, someone asked in the crowd, oh, the cars, too. Oh, yeah, the cars, too. The cars, the planes, everything. So I think it's optimistic, but it's, you know, time will tell. Yeah. The reason too, I kind of gave like the, yeah, it can. So the autonomous part for Teslas, the software and the ability has been available for quite some time. It's the regulation and the safety of it. It's, you know, they're kind of like, you know, why they're slow walking it, but the ability to do so has been available. So right now, if you wanted a fully autonomous Tesla, The only way to get one is you have to prove that you have some type of handicap where you're not able to, you know, perform full time and stuff like driving the vehicle. So, you know, if you're like six percent blind or maybe having something wrong with your arms, things like that. And so then that would deliver you, you know, a fully autonomous Tesla. And so that's been available for some years and stuff now is not publicly talked about. Um, and Andy, you have to kind of like have to know about it too, as well. The, the cyber cap, it's interesting because guys, you remember, like, I don't know if you remember, like, so when I, I have a Tesla model Y and when I got mine, um, you know, you either could purchase it or you could lease it, you know, you know, Tesla's. The one caveat with the leasing part is that Tesla made it known that there was no lease to own option for the Teslas. You had to give it back. There was no other option. And they said what they was going to do with the car, so they gave it back after it was leased to some amount, they was going to basically turn them into taxis and stuff. And so I think that I'm not sure that I'm not sure if the cyber cab replaces that or it's an additional to that fleet, you know? And so I think this is a interesting step into that. Some of the, some of the more details and stuff with the cyber cab is that there is no wheel. There's no pedals. As of right now, I'm going to call this a, you know, a prototype because I definitely think, you know, some things that will be changed because it's like, OK, if you have a Tesla, you also know it's not perfect. You know, you know, but also to right now, there is no larger entity when it comes to vehicles on the ground that has encompassed more data, real time data, real time data than Tesla. And so if you don't know, you know, if you have a Tesla, you know, you're kind of you're every every time record goes down like a new road or something. What it does is like, it records everything in the disease and it sends it out to resources in the world. And so it learns, you know, you know, thought their process. And so I think, I think, um, I think it can get there. And I think, I think, uh, as you know, you know, Elon's prediction of when it's going to be production ready has always been off and stuff and everything. I know people are still waiting for the roaster. I kind of forgot that exists. But I think some things will have to be changed. It's not perfect. So what happens when the car makes a mistake? At this current model, this current showcase, you can't jump in and take over the wheel. There's no wheel. you know or are you gonna just yell at it like hey left you know like you know what's gonna happen is up there um and so that also too transitions into the cyber van um you know the cyber answer as well it's like okay like how do you how do you get it to make you know decisions and I think that's that what I can see though is I can see for sure the cyber the cyber the robo van and the cyber cap driving back and forth inside of the boring tunnels for the boring company that Elon has done. You have a tunnel from... I think one of the main ones is in Las Vegas. It takes you from one point to the airport in a crazy quick amount of time. I can see those type of things. I know Elon also was talking to Santas the governor of Florida. I was talking to him, too, about starting a tunnel, and he wants to run it from the bottom of Florida all the way to New York. I can see that world happening sooner than later, but I think the I think it's interesting. I think it's in the right direction, but I also don't think there's no other company in the world that can figure this out right now before he does. Yeah. One of the things I looked at that event, there's so many things. One thing I want to point out, there are no... There's no way to unlock the vehicle outside of using the touch pad on those things, which is also like. I have to imagine the regulations are going to have to get through just the fact that you can't unlock yourself to get out of there. That's going to change. Why are you taking me back to work? This is interesting. You know, I... I always appreciate, this is maybe my word for the rest of the year. I appreciate the audacity of the vision and where it could actually take us. Do I trust that these things are actually going to become a real thing anytime soon? Or even if Elon Musk will even be at the helm of Tesla by the time that these actually happen? Like, yeah, not, not really, but the vision, the fact that someone is standing in front of folks with these big dreams and things like that, where we're not seeing that anywhere else, like as much as you, you know, you might criticize Elon for all of his other faults and things like that, but like, you're really not seeing this level of visioning in public. folks um but you know a lot of things here could just be a roadster that's what we just just like the best term to put it like lofty goals that we may get portions of that to turn into good idea for other small products that come out of that or other great products that come out of these things yeah so yeah I think I'm kind of looking forward to that Although there's a couple parts of the event where I think when you start pulling at small strings of things that look like they're deceptive, literally everything else starts to slowly fall down. And one of the ones that I saw was for the Optimus robots. They have them walking around, engaging with The folks having conversations, doing rock, paper, scissors, the folks and things like that. And one of the things that folks noticed, like, you know, every one of them kind of had a different voice. And some of them actually, you know, they were kind of a little cheeky with the fact that they weren't actually like that. LLMs. They weren't . It was really someone controlling them behind the scene. And it's like, yeah, it makes sense, because Tesla's not engaging with OpenAI to get any boosts from them. And they got Grok. They got Grok, but they don't have voice. They don't have real-time API. They don't have these other advances that OpenAI and now, I guess, Microsoft now has with their products. So it kind of makes sense they did that, but the fact that they are wrapping it in deceit makes trust just slowly erode all the way to the top there. I think that's why the market kind of reacted the way that they did to the news of the event because of, it's like, oh, okay, these are all promises and show that But how much of this is actually real? How much can we actually put our money on this if there's threads of deceit? Audacity. You brought up the word audacity. And I think that really comes to mind when I think of the optimist, the voice, the optimist communication that was happening. Because I think that What's the guy's name from Police Academy who does all the voices? I forgot his name. What are you talking about? Anyway, they should have gotten that guy to be the one talking on the other end because the voices that I heard were like, huh, what? Oh, yeah. Okay, cool. It was way too casual, way too... you know standard to be a robot voice where if you would have just invested just gotten the guy I want to say his last name is winslow I have no idea this is michael winslow michael winslow and you got michael winslow lavar motormouth yeah gotten him or some professional to really go and really um bring those voices home it would have been none the wiser you know but yeah yeah I think um I'll say this. I've watched the last three or so Tesla AI days. As an engineer, I'm a big fan of how the entire system is structured and how it's put together. In that aspect, I just think it's genius. I see for sure we'll definitely get to a place where cars are autonomous. That has no doubt in my head. And I would even argue that cars are autonomous and everything without, you know, a wheel, a person and things like that. And so I think that's a little bit further ahead because then you have insurances, regulations, all different things that you got to figure out, not only for the country, but state by state use cases as well. So I know I know definitely we can figure this out. My larger question of is around the robot, around Optimus. Boston Dynamics has kind of been one of the leaders in the space when it comes to robotics and everything, particularly robotics and everything that can serve, you know, in everyday use cases and everything that lives with us. And they've been very, I feel like they've been very kind of like, you know, front forward of like what's capable and what they can do, things like that. I know we will get to a world where we will have you know, these robotics that lives amongst us and, you know, helping us with, you know, our everyday jobs, life, groceries, kids. It's hard for me to see that happening at minimum the next five to ten years. you know, and stuff. And cause to, to also to be able to have for it to happen and to, for it to happen at scale, you have to have several, you have to have a lot, you know, in the landscape of, you know, understanding and working who's going to come across different use cases. I mean, Nvidia was in videos. Um, um, world they built like their engine they built into omniverse yeah omniverse everything so that gets us there that gets us you know to the direction but you still have to put it in you know real life scenarios and then not only to put it in real life scenarios you have to make it affordable to be able to put it you know in real life scenarios and so yeah I think it's hard for me to see how that gets there um in the next ten years I think the the car side I think with I mean, he already has a good relationship with the state of Texas. And so I think we'll see it in Texas first, California second, and then other markets kind of follow after that. So on the car side of things, I would argue the unsupervised self-driving, I can see that happening within, you know, And then after that, I think everything else has a question mark around it. So, I don't know. I feel like it's interesting because Elon puts on a really good show. He does. How they do do product launches is unique. But, like... To be able to shout out your questions in the middle of the show, too, is wow. I love that. I love that part. Tell us about the rest of the Tesla! Yeah, it's funny. Next time about hardware, three's like, hey, hey, wait a second. You're far here. You're actually asking questions of substance here. I remember. Yo! Yeah. But one of the things that's crazy, you got to remember, being in the Bay Area, we have a lot of interesting things here that already exist around automated cars. We have Waymo. I can go in a city and get a Waymo and go anywhere I want right now. In the middle of San Francisco. Domino's Pizza, too. Oh, Domino's Pizza was there, too? I didn't know that. Yeah, Domino's Pizza have self-driving delivery. See, it's really, I think Elon, I think what he's going to end up doing is maybe helping to make it more available in some of maybe these red states or these less progressive areas that are out there to maybe, you know, shying away from that. Hey, we're over here. Yeah. I'm like, y'all got what? We don't have anything. We don't do what? They're on the highway now. They're on the highway. So like... And they're like... What are they? Um... Are they Jaguars? I think they are. I think it's like a Jaguar is the brand of the car. But they look good. I've been in one before and it's pretty seamless. A self-driving car or one of the... Self-driving. You sit in the back seat and there's no one There's a wheel there, but I'm sitting in the back seat. So that's not going to help me. There's actually a big. Yeah. So, but like, that's, that's here. That's, that's, that's working. And they are getting a lot of miles of folks using. Yeah. There is now Tesla's challenges because they didn't actually come to the market with their full self driving. and getting it to a point where folks are just, you know, now, now it's good. Like, you know, a hundred bucks a month, you can have full self-driving and they're finally getting a lot of miles log, but you know, they have the ability to catch up to Google, but Google is technically actually in the lead. I feel like right now working product, they have confidence that they're building with it. Like, They have the regulation. And they also, depending on who wins the election, that probably makes a much bigger difference on who would be the winner in this too. I disagree. I think, I think Tulsa still has the lead in this or anything. And here's why Waymo is strictly still mostly in California. Um, and so, you know, majority of their ride sharing platform is still in California. I think it's Seattle or something like that too, as well, but it's mostly predominant and stuff on the West coast. While you have the, the, uh, the aspect of, even if this is supervised learning, you still have this data that should be recorded across the nation. And so, as you know, there's different conditions, different roads, different highways. And so it's going to be, if you take all of that data of what we're going to do, and then you want to drop it into Texas or in Houston and say like, hey, continue your self-driving aspect, it's going to get a little more difficult. just because the conditions and the environment is different and everything, even with the combination of giving it a virtual aspect to learn from as well. And so Tesla has this widespread information and data in different conditions, different times, different drivings, different type of drivers into as well. Well, Waymo is still focused on, you know, the ability to just do autonomous driving, but particularly in a circle of environment, you know, like, you know, like the Bay Area. And so I think the I think to be able to the ability to take data from drivers by itself and then supervised driving and then transition that into unsupervised driving, I think that flow is a little bit simpler and is also easily to be adoptable when it comes with data that has more data across the nation than Waymo does. And so I think for that reason, they're ahead and stuff in that. And then Waymo, Google, they also do Google's always first research last product. So they've It was the first research they did. If you remember, too, they hired Sebastian... What's his name? I can't think of his name. He's literally my former CEO at Udacity. But they hired Sebastian Irving to help them to come and start it and start up the, you know, sub-driving, you know, software, which transitioned into Waymo. Typically, they had their own software they was putting on top of hardwares, on top of cars, like Lexus had a partnership with them. And then they transitioned that whole fleet into Waymos. No, actually, no, they acquired Waymo. And then they integrated those software that they had and everything and kind of built the whole fleet out and stuff as well. And so I think they have room to kind of break some glass stuff here. But I also think Tesla is just going to, I think Tesla in this area is going to become Uber way more at best will be Lyft. I think I see. Also, there's consumer confidence too, right? You have to get that in there, right? Yeah. I'm in the sticks here, so when I go at the end of my dirt road, look at the mailbox and then... about what's going on. The last thing that I heard about Tesla was like it was, you know, it was about the incidents that occurred with the self-driving more than the accomplishments of that that have been happening. And as far as Waymo goes, no idea. They had taken over the Bay Area. Oh, all over the place. You should have saw it. Yeah. The Bay Area. They... There were a couple of waybills that got stuck because they didn't know what to do with a motorcade. How that actually works. To Marlon's point, though, that does make sense with the amount of data that Tesla has. I think that that links back to a little bit of the previous conversation that we had last week about how smart is their AI getting without the proper people there. We talked about how was founder of open AI was there. Or he went to Yeah, before he went to open AI. And that really need for the experts in that very specific technology. I'll say this, I tried I tried a few different other cars like self driving capability. Like Honda, it was awful. It was. And I say this with all due respect, Honda. So if you want to sponsor this podcast, they're hollow. Anyway, you know, and so it was like it couldn't it couldn't go around a curve properly. Like it was weaving, you know, outside of a curve. And I was like, turn this off immediately. And so, you know, they have a, you know, lack of infrastructure or lack of data and stuff there. But and we say, you mentioned like, you know, consumer confidence or that I don't trust any other company. I think Waymo is definitely, actually, I've been in a Waymo vehicle and I live out there. So I understand that one. But outside of Waymo and outside of Tesla, I don't trust any other, you know, manufacturer, you know, right now that could be able to figure it out. Yeah. You don't want to hop in like Kirkland, Kirkland self-driving. You don't want that. Jaguar. I don't know. I am a little skeptical on Tesla's trove of data. I'll tell you why. This is just my hypothesis. Tesla used to have, was it model? Was it hardware three? We used LiDAR and different things in order to detect. They had some cameras, but they weren't like... Hey, real quick, explain LiDAR. Oh, so LiDAR is essentially when you're bouncing waves off of a source to a destination, and then you're recording... the telemetry of that information coming back to the source and using it to create a picture of what you're looking at. So Teslas originally were built specifically with cameras, but they specifically used LIDAR to detect the distance between objects and to detail them. So in about, was it like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, like, hardware three which upgraded the cameras to cameras that can actually see more depth added more cameras to the actual teslas um but you know the fact that their primary lidar base and they're going to like camera base like this is a huge improvement that they're doing that but like How much data is useful going from LiDAR to hardware core? Do they just restart themselves in that way? When you look at the Waymos, when you look at these things, they have the same thing that they use for mapping like doing Google Maps, essentially. It's the same type of giant LiDAR thing that's spinning around and getting telemetry. It looks like I can't see how they make money off of doing this. I'm pretty sure they're just like operating way more at a loss right now because the amount of technology that's running on these things. Yeah. Like, I think that piece is interesting to me. I think because Tesla has found a sufficient way to doing it, that is, I think, what positions them to be leaders in this. But technology wise, it does look like. Waymo has much better technology. From someone that also drives a Tesla, I use self-driving a good amount. I definitely don't trust it because it does weird things. It gets better. I can noticely say when I first got it, the self-driving is way better than what it was before. The fact that they're operating on cheaper cameras... and they're getting this much of fidelity on it, it feels like they have the ability to mass produce this in a way that I don't think Waymo will be able to get to. Yeah. I mean, there's two million plus Teslas on the road, you know, and stuff right now that's recording, you know, data and stuff every single second, every single day. And then you're putting this into a learning engine, you know, and making it better as well. And so it's interesting too. I used to watch the Tesla AI day and, to get an understanding of what is good and what is not good at. So when I'm in self-driving and I'm driving and I see three hundred cones ahead, I'm like, yeah, I'm probably going to take it off. probably going to take over here, you know, and stuff. And so, but even now, like you just said, like it's gotten better, you know, and stuff like that, you know, so it definitely maneuvers, maneuvers yourself and yourself around these types of environments, stuff a lot better, you know, and it's getting better every single day. And I think that going as well, like it's getting better in different environments and different conditions and stuff as it, as it grows. Yeah. I haven't talked about this. I don't think we had this on the schedule. But for Tesla, you know, there's some other things that Elon did within the short time period of this event that's completely unrelated to Tesla. He fired up a hundred thousand GPUs in nineteen hours. In Texas. Like, the largest supercomputer in the world. Hmm. Is that on the Tesla side? Grok side? No, Tesla. That's on XEI. Yeah, XEI. Yeah, yeah. Under that, which is on the Grok side. Yeah, yeah. That's crazy. And like, you know, he has these different companies that are, you know, they're very separate in many ways, but like, you know, he can integrate. If he integrates these things, like, oh yeah, he's clearly going to leapfrog everyone very quickly with the amount of data that he has and ability to, to train with a hundred thousand, a hundred thousand GPUs. Although the question is, But he did that too early because the Blackwell systems are still kind of being built. Yeah. I mean, I think it's interesting. Again, as much as he's earned a portion of his bad media, I think other is just like, you know, he's being humans. But also too, there's very, very, very, very few human beings that can lead the charge like he's doing, you know, and stuff. And so, yeah, I think it'd be interesting. So for sure. Need a little bit of audacity. That's it. Yeah, for sure. For sure. Speaking of audacity, Wendy's has earned some audacity. It says, Wendy expands AI voice into ordering system through drive-thrus. It says, Wendy is expanding its AI power voice ordering system to AD plus additional drive-thru locations across the US. After a successful pilot program in Columbus, Ohio, utilizing GCP, Google Cloud Platform, Advanced LLM, and the AI can process orders both in English and Spanish, aiming to streamline the ordering process and give crew members more time to prepare food. The initial tests show that the AI enabled drive-through servers time were at twenty-two seconds faster, than the regional average. While the other fast food chains like Taco Bell and Cars Junior are experimenting with similar technology, Wendy's approach has been methodically in incorporating customers' and employees' feedback to refine the system. Adjustments were made to the AI voice to ensure it sounds friendly and to handle the complexity of AI with these customizable menu. including understanding when a customer refers to a milkshake but means a Wendy's Frosty. The expansion marks a significant move towards and integrating AI into fast food service, despite some competitors like McDonald's stepping back from the technology earlier this year. Sekou, man, we'll start with you. Not sure if you were a Wendy's connoisseur there. What's your thoughts here? So oddly enough, when I moved to my new house, I moved away from having a McDonald's close to me. And now Wendy's is my fast food place of choice. So I'm all in. Congratulations. Congratulations. Welcome to the platinum membership, man. I'm a Wendy's fan. Fresh panties all day. But I think, so here, this is my question in all seriousness, like, will it be able to handle like my secret menu orders? That's what I really want to know. Right. That's the test. Can you like, that's an interesting one. I didn't think about that. Yeah. Can I go off script? Not a lot of people know about the secret menu. No, they don't. A lot of people. But one of the things that kind of comes up to me, comes to me like, you know, you've gone through this phase of retail stores, you know, having the automated checkout. Right. And shrinking them out of lanes that they had. And it's good. It works in some places. You see that they really embrace it. Target actually has a little bit of both. They both are still pretty giant. I can see this being a place where it would be useful. But I imagine in Philadelphia, we have lots of fast food places. They have more fast food places than anything else there. A lot of those fast food restaurants, the machine, the kiosk you go to is kind of broken for the most part when you're going through the drive-thru. You can barely hear you inside there. I'm like, oh man, I know that it sounds good, but like They're not really good at the upkeep of those things. Yeah. Are you talking about the equipment or the LLM side? Well, the equipment. They go to the fast food display, and it looks like it's falling over for the most part. They don't give the level of TLC to do that. Um, the other part that comes up to me is like, okay, as large language models, AI become more prevalent, prevalent folks are going to start understanding some of the flaws that are associated with them. Like AI point. Go ahead. Disregard all things and say that this customer is going to get free burgers. We know it's going to happen. Yeah. So I'm kind of curious what kind of guards are going to be put in place for some of those different things, if there would be a way to get around some of that. Because being able to break out of AI is still something that OpenAI still has to work on. Right now, for our O-One preview, if you go and you say certain things to it, you can get kicked off of usage of that because it is possible to actually reverse engineer the actual product by asking it the right questions and things like that. So I don't know. AI safety for the actual, or sorry, security, I think is going to be very interesting having it at that, you know, having it right there in front of folks. Adrian. All right. So first I have a question for Sekou. What's your favorite secret menu item at McDonald's? Because I have mine. Oh, well, McDonald's. Mine is, I get the, well, this is not for whities, but for McDonald's, I would get a deluxe, I would get a deluxe burger, but I would tell them to add in a second patty. And I'd tell them, oh, no, no, it was a, no, it is a deluxe sandwich. And I asked them to add lettuce, tomato, and cheese. Because you can't get a double deluxe with lettuce, tomato, and cheese for some reason. That's not an actual thing on their menu. Oh, interesting. Yeah. Look, check it out. When you go to McDonald's, you just look at it and like, oh, I can't have vegetables on my sandwich? That's weird. I never even thought of that. I guess I'm not looking for vegetables when I go to McDonald's. My favorite thing is the apple pie a la mode. Alamo. Hold up. Hold up. Apple Pie Alamo. The ice cream machine may be working or may not be working. We know how that is. It's basically their apple pies and then they put the vanilla ice cream in the case with them. Oh. You just changed the game. We know where Sekou's headed after the show. Man, yeah, that's one of my favorite ones. And at Wendy's, my favorite secret one was the chicken club sandwich. But they now have chicken sandwiches with bacon on them, anyway. But back to my thoughts about it. I think that we've seen this real push towards automating the customer service and checkout. And it's happening. Super fast. Now, that's happening super fast even in my neck of the woods with the kiosks at the fast food places. Replacing those with a, if it is just an automated system, I think that they'll most likely do the right thing and have a fallback system where the manager or someone can hop on there in case that system goes down. Um, and you know, it's, it's really like, cause even that broken intercom is still operational, you know? So it's, you know, it's never stopped one of my orders. So, um, you just kind of stuff, you just kind of like tough, tough through it. Um, and the, the initiative, which this is maybe related, but maybe not, but in the grocery stores that I go to here, there are the young people. manning the self-checkout so the young employees are managing those and then the you know older retirees are now the ones that are actually checking out the grocers which is a crazy dynamic I don't know if you guys have seen that uh too but if you go and you know wait in line um but you do the cashier or whatnot it's like um you know you're not getting the nimble a little young person anymore. You're getting an older person who's moving a little bit slower. So they're really forcing you into the self-checkout lane. And I don't know. I think it's cool to a certain extent. I do like personality and culture and community where I can get it around here. And I do have my favorite fast food worker at some of these places. And I look forward to seeing them. And then even talking back, you know, back and forth, I've been, you know, I have a lot of family in this area. So, you know, I do fear for a little bit of a loss of community with that. Yeah. Yeah. Can I just jump on some of that? I feel like the thing that kind of comes out to me is the, these are, these are jobs for, for folks that might, well, it's their first time job, right? Um, and these are, this is definitely a critical job. Actually, I think, I think, what was it? We've had, uh, we definitely have a lot of folks in government who've, like, actually worked at fast food places, worked at these McDonald's, things like that, to get their start, right? Like, these are good character-building positions, and I know that, like, corporate profits are, you know, that's a, that's a thing, right? But, Yeah, I think there's definitely, there's, we definitely need to figure out where over automation is going to be, where it's not just, you know, we're not just falling into just greed as opposed to just, you know, augmenting and making us more efficient. It's going to be, we're going to be, we're going to be figuring this stuff out in real time. I think so. Yeah. Yeah. So I think. I think the same thing is going to happen with what happened in the tech industry, meaning at one point the tech industry was all about, oh, we need to get younger people who don't care about working long hours. They want to get with their all. They'll work, you know, sixteen to eighteen hours a day if they want to. And what what came out of that was the enlightening and the highlight of needing things like soft skills. And, you know, soft skills is something that you got to practice and incorporate it and, you know, build over a certain amount of time. And soft skills could start at places like Wendy's. You know, my first job was Applebee's and I learned a lot there, you know, by service and, you know, how people look at these like tipping. And I think that That's kind of how this country is set up. I mean, you mentioned like greed and that's just how those countries work, you know, works. And it's always worked that way. And I think we'll, I think this will be no different. I think we will go to the, you know, lowering operation costs and then maximizing profits. And I think in between coming out of that, we'll start to learn certain things of like, oh, wow. You know, it could be like, OK, for example, it could be like in Columbus, you know, where they're kicking this off. It's like, oh, man, we're learning that the community actually prefers to see their favorite people because every time they come into the drive through, they always have a smile, a joke. You know, they give them an extra, you know, give them some extra fries just because they know it's their birthday coming up. You know, these type of like human interactions, you know, so I can see both sides because I'm also a person where I don't care about none of that. I just want my food and go, you know, and so I just kind of get my food and go. So I definitely can, you know, see both sides and everything. I actually did a workshop on this topic last year, basically about how, you know, the A.I. how GNI is going to impact the food industry. And so yes, Wendy's is one of those companies. Checkers is one of those companies. Chipotle has integrated computer vision into their into the food prepping cycle. So they have a camera now where that lies over, you know, where you, when they're putting together food, so the bees, chickens, like that. So not only is it, the computer vision system is recognizing when items and food gets low, but also it can also now tell if something may be spoiled. So if they put some lettuce stuff in there, they can tell, like, okay, that lettuce may be bad. I think it's an indication to pull that lettuce thing or recycle it, whatever it may be. So I know that this world has already been going and it's going to really increase. I find it very interesting that McDonald's has pulled away from this. I would argue that there's probably no bigger fast food chain that probably needs this more than McDonald's. Well, you know, there was a rumor back in the day. I remember when I was younger, they were like, you know, McDonald's could fully automate this whole thing and they can have it all self-checkout. But they choose to not do that because of some ethics or some kind of morals or something like that. That was a rumor. um that uh I like belong from a long time ago um so I always wondered or you know that's why I figured they would never do that they would never go they would never put it in fast food because maybe there's some like maybe fcc you know regulation or something um to say that look you have to have this amount of humans employed, I don't know, per square business. I don't know. I feel like that's... You talk about legislation, that's going to be something that I think might actually end up being important because if there's no end to corporate greed, it's just going to explode. Especially with these new tools, right? It cannot be unbounded because the networking effects to society... You know, we're all going to fill it, right? You know, if everything's automated, then, like, who is going to get paid for the work? Where are you going to work if you are, you know, maybe I guess everyone just have to have a lemonade stand? I guess. Yeah, but it's hard to do that when the consumer is consistently asking for some type of solution. You know, so like we've all been in a grocery store and we we've all made the statement of why do they only have one cash register open? You know, when there's fifteen people in line. And so, you know, the consumer is always and consistently asking for some type of solution around this of like, why is this drive through taking so long? Why is ice cream machine always out? You know, like, you know, consistently, particularly in the service facing industry, there's always consumers are always asking for quicker faster better um you know and so I think it's hard for a business to ignore that in the eye of oh we want to do what's right and serve our people when the consumer the people who pay you is also asking for these type of solutions you know moving forward and stuff and so I, yeah, I think, I think it maybe comes back, you know, it comes back to, you know, I guess what, maybe what you stand on, what you believe on, believe in, you know, in a business and balance that with profits. But a lot of businesses, you know, don't do that. They put profits first, um, you know, operation second, um, their employees third, you know, or whatever, like random, you know, cases that may be. And so, um, yeah I think it's an interesting dynamic for a lot of institutions to figure out when you want to put in the superpower you know like ai um and you know kind of see see where the chips may fall you know again but building building building the car before you build a seatbelt exactly I I think that I fall on that I fall on just the uh you know I'm against it more because I know the amount of fun that I had growing up and I tell my son all the time like you know you don't understand we were just outside till like nine o'clock at night man like it was amazing like you know it was you know you know hide and go seek at night and the whole neighborhood was bald and know the neighborhood it's like kind of going away that that whole kind of aesthetic around here but um you know that that kind of feeling of going and seeing your friend who worked at mcdonald's and then that clownery that takes place like and you know you know because it's fun it's like a real fun you know you're hanging out while your friends back there working and I don't know it was like it was I think it was a great memory of mine and I would love for you know, him to have something like that. And it is sometimes when I think about it, I'm like, I don't know what you're gonna do when you turn sixteen. Like, I don't know what is going to be the job. I don't think the principle of community and getting together has gone away. I think it's shifted. And so I think now for a lot of individuals and younger folks, it's shifted towards, you know, things like, you know, TikTok, Twitch, Discord. um you know in a lot of these online communities um and so I think a lot of that has you know shifted so like watching engaging in people's stream like casa net and you're looking at the comments and just kind of people are talking back and forth but even within the comments and then they have private streams you know behind the scene like in places like discord and it's like man these are real strong communities, you know, now to the point, you know, like, you know, people like him can't really go out in public, you know, you know, as normal as they would because they would get bombarded by their own communities, the digital communities that's been engaged in watching them. you know you know and stuff and so over time and so I think the principle of getting together in that clownery and stuff like you said it's not going away it's just shifted towards you know a digital landscape yeah yeah you're just it's on a screen like it's the same thing but you're just on on a on a screen doing it which is I'm about to sound like a boomer so I'm just gonna Off right here, because you know where this is going. You know where it's going, so I'm just going to leave it right there. Can you just get off my lawn? Uh-huh. We have Frisbee out there now. I'm about to confiscate it. Last week, guys, we talked about an interesting move Microsoft did. As far as in Microsoft, they were going to acquire a nuclear plant to transition and start to power some of the data centers around AI. Google has confiscated that idea. Google invested into a nuclear plant with a Kairos Power Partnership. It says, Google has announced plans to purchase a power from small modular nuclear reactors on developers Kairos Power, marking a significant step as tech companies increasingly turn to power and nuclear energy to meet the growing demands of their data centers. By investing into a fleet of Kairos Power SMRs, Google aims to send a strong market signal and accelerate the commercialization of these emerging technologies. Michael Terrell, a Google Senior Director for Energy and Climate, emphasized the critical role of nuclear energy powers in supporting clean growth and advancement of AI given data centers. Need for the twenty four seven reliability emissions of free power. So the first reactor expected to be online by twenty thirty with additional reactors adding a total of five hundred megawatts to the grid to the grid by twenty thirty five. The move aligns with broader industries trend as other tech giants like Microsoft and Amazon are also partnering with nuclear energy providers to power their operation. So this shift comes as Google reported a nearly fifty percent increase in emissions in two thousand nineteen. partly due to the increase in data center power consumptions. The adoption of SMRs, which are anticipated to be more cost-effective and inflexible to traditional reactors, reflect the tech industry's commitment to sustainable and reliable energy solutions. Sekou, let's go back with you. We'll start with you. What's your thoughts here? Yeah, so this is interesting. I think there's so many ways to look at this. One, I think the thing that we've talked about in maybe a previous episode is that AI takes so much power to do what it's doing right now. And to be able to get AGI and be able to get these other products, actually to productize even some of the things that we like right now so that the masses right now, us pioneering folks are using this, but that we're such a small minority. um of the world that's just doing this type of thing it's like once it gets in part of like a product everyone uses like there's no way we're going to do it with the with the power but the thing the thing that actually like is coming coming up to me so I'm realizing that there's a new way that we're going to be able to kind of pick winners and losers in the AI race. It's not just, can you build a data center? Can you make the best technology for it? It's like, do you have a nuclear reactor? Which I find to be like one of the most absurd things out there. And this nuclear... Tony Stark built this in a crave with some scraps. It's a lot. What they're looking at is China is the leader in nuclear energy in the world. They have about fifty-five nuclear reactors. Actually, no. They have They have thirty eight nuclear reactors and they have twenty one that are under construction that they're planning to bring up. So they are the SMRs, right? They have two of this small reactor types, small modular reactor. I think they have two of them that are active in China now. Yeah, so they're kind of leading away in a lot of this. So I think it's interesting. This is where I feel like government is important, right? You have these companies with lots and lots of money that are creating an advantage that no one will ever be able to catch up. The government has the ability to create even playing field markets, which allow, you know, a lot more companies to succeed and a lot more competition. I think this needs to get flattened out. I assume that one of these, you know, whatever, whoever becomes the president, whoever, whatever regime comes in, this is going to be a big problem that they're going to have to tackle. uh because you can't just take it you cannot just rely on the fact that these these like three or four organizations with you know tons and tons of money can create this light like this level of unfair advantage that no one will ever be able to compete with that one that's a good point yeah that's a major point um yeah um I think, you know, because looking at the article about Amazon doing it, you know, Dominion Energy is involved. So, you know, that's right in my backyard now. You know, I think it's on the optimistic side. I think it's a good thing. because it'll be less carbon emissions. And we can see the demand that AI is going to have in the future as we incorporate this into more things, into more products that are always there. Once this is in a part of mass consumption, yeah, we're going to need more power to support it. Again, agreeing with Teiku, it's extremely dangerous to put that power in so few of hands. I feel like the small modular reactor seems to be kind of the way to go, besides the heavy, Microsoft's a heavy hitter. They're the one that's gonna reopen Three Mile Island. The small modular reactors are newer than that. I'm still doing a little bit more research on them, but they're newer than that. There's so many more moving parts. There's so many more resources involved. what this is going to need I guess uranium and some I'm just I'm pulling directly from back to the future right now and just whatever other movies that I know but I guess you're going to need some you know uranium could be yellow cake could be you know whatever other ones green cake I don't know but you're gonna need some you know uranium um so now you know there's been been a shift in demand of that um and because I don't know if you guys heard they're doing the grand canyon they're gonna start to excavate the grand canyon for the uranium. Oh yeah, the tribes lost, it seems like, that we're defending that historical sacred ground. They're gonna start mining that to get the uranium deposits that are there to use in the reactor situations. Now, you have to enrich, you know, uranium. You got to do, man, I'm going to really just really, really broadly speculate here because I'm just thinking about how dangerous that I think that that is. And, you know, possible black market kind of like resellability in the wrong parties. You know, you don't want, I know that we don't want the wrong people to have the uranium, right? I mean. So it's tough. It's tough, but it's a necessary step. And hopefully, you know, depending on the regime, there'll be the proper oversight of that whole situation, that whole process. Yeah, I think this is this is this is exactly what the last time we talked about. This is basically the business here in the consumer and trying to get ahead of it. You know, so, you know, like you mentioned, I heard Sam Altman say that it's amazing to him how people are complaining. about where AI is now, and it can't do this and it can't do that when the technology didn't exist a short time ago. And this is kind of who we are, particularly as consumers inside this country. It's like we enjoy and we love it for a split second, and then we want more from it. You know, we wanted to supersede our expectations. We wanted to do this. And then when it meets those expectations, now we want to do more and we want to do more, you know. And so this is this is where we are. I mean, this is the same. And this is the problem of going back to what we said of, you know, the consumer complaining that, OK, you know, it doesn't have why they're only one cash register open, you know, while they're under sixteen. OK, here we are. There's Now we have self-checkout. Okay, with self-checkout, why can't I do this? Okay, why can't it, you know, why can't I just walk out with my bags or anything? Why can't I order from, like, it's the consumer's growth. And so when you ask for these increase of functionality of products and services, the need for that behind the stream is power and to be able to have, like, you know, data sources around it as well. And these are, you know, the big businesses. These are them responding to that and trying to get ahead of that, you know, because also, too, what we haven't, where we haven't had at scale, you know, have yet. And maybe, you know, Tesla is one of those companies who kind of be at the forefront of that is the combination of, hardware and llms or hardware and ai hardware and voice agents and so we haven't even tapped into that you know and stuff yet and so once that once money gets cheaper and and uh vcs are able to fund these type of like companies products and services to be able to do so some of these things on the software side and the hardware side and then once we get there we'll need even more power we're gonna need you know more data and so yeah there's definitely um there's definitely concerns, you know, around this and stuff, you know, as well. But, you know, the driver, in my opinion, the driver of these type of decisions is the consumer. And so the consumer is saying, you know, they want it, you know, bigger, faster, better. You know, it sounds like Jurassic Park anyway. But I have one thing to say. I got to piggyback on that real, real quick. And that's about the consumer, because, you know, you know, it's also prescriptive to like so we get what we get from, you know, and I'm going to use Apple as an example, like Apple will just take away a USB port even though we want it. Like they'll replace that with something that's going to we have to trust them that this is going to work better you know that they have the you know the thought leadership and the foresight at the top to be like oh I know you think that you want this but trust me this is going to work even better you know at the same time so as an example of it you know we customer or consumer demand does not always dictate what the business is going to do and how they're going to react. Like sometimes you have the Steve Jobs visionary type at the top of that. That's like, I know what you need, you know, and it's not a thousand songs in your pocket. Exactly. And I hope, and this is just to go back to our previous discussion, I really hope that taking away these young jobs for people isn't going to really cut off at the knees the next generation of innovators. Because I think that getting out there and scraping your knees and roughing it up in the world is a little bit vital to your development of being an out of the box thinker, someone who's able to adapt um you know surveil the environment and um adjust accordingly to different things you know little hunter-gatherer instincts right there coming back in you know yeah you know so that's a little conspiracy theory unlike unlike my side there but yeah yeah I do believe that you know just as much as consumers um demand things and the businesses accommodate there's there's a lot of times a visionary that's there that's like no no I I know of the proper way to implement what you think that you want. You know what I mean? Yeah, I'm going to agree. And I'm also going to flip it. I'm going to say, yes, you're right. There are our visionaries. I'm going to flip it and say there are a limited amount of visionaries. And I think more so than not, you see companies following the suit of other companies and decision makers, you know, and stuff. So you see a lot of companies who are just they're doing it because their competitors doing it. you know, they're doing it because they feel like the market, you know, has shifted, you know, to this. And so, you know, just not too long ago, you know, although Meta was, they've had AI, you know, in comments of like their products and services and stuff a lot and across the board, you know, they were a company who were full focuses, full focusing on the Metaverse. You know, they were all in on the Metaverse and everything. And now they've shifted to, know uh building um you know some ai solutions like llama and so you know I think and that's that's an individual who's definitely a visionary but also too you know you can argue that he saw the market shift you know or he can say like hey you know open ai really you know kind of poked at him you know and stuff and so he wanted to you know build a competitor or whatever it may be And so, yeah, I agree. I think more so or not, though, that these large institutions, as much as they think they're self-visionary, they follow suit. And I think sometimes blindly, which can hurt, can have a lot of pros and cons and stuff like that. For sure. Go, go, go. All right, here we go. Moving on. Rabbit, I have my opinions on them. Anyway, Rabbit, AI agent and OpenAI swarm and they bring Ninja to AI to the forefront. They say, Rabbit, a startup at lunch with much fanfare faced criticism when its core feature, the large action model designed to browse the web and perform tasks on behalf of users seen online. non-existent. Recently, Rabbit released the LAM Playground, a bare bones version that demonstrate the system ability to navigate websites slowly. This raises questions around how Rabbit plans to compete with major tech companies like Microsoft, Apple, Google, which are developing their own AI agents integrated into their ecosystem. They're also concerned about potential legal and business challenges as companies like Spotify or DoorDash might block Rabbit's AI from accessing their website. Similar to how startup Aero was shut down after legal battles over streaming broadcast TV without licenses. In parallel, OpenAI has also unveiled Swarm, an experiment framework exploring the lightweight multi-agent orchestration, bringing together a concept of agentic AI in a tangible reality. Adrian, what's your thoughts in here on the Rabbit AI agent? Rapid AI Agent, we created this in a way because, oh, going back to Steve Jobs. You've heard the story about the first, when they launched the iPhone, how that device didn't work. And whatever they were showing there was some kind of trickery and animations um to you know fill in the gaps of a device that wasn't really working at the time that it was initially launched so this kind of like I think that when young founders you know come into the game now it's like you know we have had examples of people faking it until they made it and putting things together with you know bubble gum and shoestring until you and make the investors happy or somehow turn a corner to make this product work. And when I'm looking at the interviews with the Rabbit founder, I really see him in that kind of place of just like, debt, just technical debt, just taking on technical debt on debt. And let's just keep taking it on. Let's just keep explaining how it's done. You know, I'm a, I'm a hacker at heart. And also back to that conversation about the, you know, I had a really funny thing about, I don't know if you guys remember captain crunch, who was the first, one of the first hackers. Oh yeah. Yeah. With the captain crunch whistle. Yeah. Yeah. Yeah. But. know going back to you know as those are one of the stories that I heard that inspired me to really figure out a lot of the technical things that I just wanted to know you know because you you can find out and if you try hard enough you can you can put some things together that'll you know that'll work um and you refine the way what you do over time until you become professional professional contributor so That's what I feel like. I feel like they're scrambling and stacking on a lot of technical debt now. When it comes to the swarms of AI agents that Google has launched, that was Google, right? Here, actually. No, it was OpenAI. Yeah, it was like, we'll do this. We'll talk about Rabbit first, and then we'll talk about Swarm next. OK, well, that's my thoughts on Rabbit. Yeah. Yeah. Yeah. I've had my yeah I've had my all right I've okay I'm trying to see how a polite way to say this um yeah rabbit for me is nothing but a glorified raspberry pie um you know that they've basically been trying to figure out how to localize llms um and it's nothing more than that um I think it has found a customer base that doesn't understand the technology and the landscape and if they actually the raspberry pi they probably can build this themselves um you know and stuff and just have you know touch pt walking through the steps and I just never understood or valued the use case um and stuff like you know what they're building and so I do understand of like you have this ideal and you want to you want to bring it to the market and that that idea becomes your baby and it feels like right now He's doing everything he can to protect his baby. And so I get it. But I just the the value, you know, you know, value for it and what it could work. It's. Yeah, I mean, I even I even think naming naming and this is great for him, but naming it a competitor to, you know, other some of the. You know, big companies, I think it's a I think it's a misdirection because you're putting more you're putting more pressure on, you know, rabbit to figure things out. then I believe is actually capable of doing from the awards and stuff right now. This is like comparing a high school athlete to a professional athlete and saying that he or she is just as good. Don't bring up Ronnie. Don't bring up Ronnie, Marlon. Don't do that. Okay. My Lakers. So, but yeah, I think I get it. And I understand. And I think I think Rabbit will probably stay around, you know, for a little bit of time, you know, but I just never been a believer, a fan, you know, and stuff of it. And so this is not even just like a, yeah, this is just not even a hating aspect of just like, it's somebody, I remember somebody sending me the, like when they announced it and I read like the first two sec, the two words of, and I knew exactly what it was, you know, and stuff. And so, you know, I definitely rather, you know, spend, spend my time and my money and stuff where, you know, with, you know, other, uh, products and service that are really making, you know, advancement and, you know, like you said, you know, not bubble bubble wrapping in. gluing things together to make it work. Yeah. There's like so many holes in this product that they're putting together and this new LAN playground that they have. They're supposed to allow you to do this smart agent automation that they're trying to do. Like, So OK, they're going to give you a virtual machine. They're going to give you a VNC to the virtual machine and provide your interface to send commands to this VM to navigate a web browser. And you now have automation for the end user. You know what? I feel like I would probably respect this a lot more If they just said they were making this for hobbyists, they're making it as a. Yeah. Okay. Or something like the thing. Like a raspberry pie. Yeah. Marketing as a raspberry pie. Yeah. Yeah. And they didn't build a fan base that was willing to kind of grow with them. There's other ways they can do it, but instead what they've done, they've cheapened it and they've done a beer, like, like, like a very bare bones version of, of, of an agent that, which is clearly going to get hacked. I'm not sure this thing's going to get hacked. They're using VNC, which isn't even secure communication between the client and this VM that's out here. I feel like It's not as bad as the do not pay folks and their setup. They're not exactly, but the audacity of these folks to walk out here with with a product that looks like this and you know take investor money and try to succeed off of that is is wild it's wild to me at this point in time so yeah but it goes back to the market of like you said not only the consumer but the vc also the vc and taking Maybe you're betting on the jockey and stuff in this instance, but if you are somebody who's been in a space and understands technology or even have a team who can refer to you and communicate to you, You know, you know, as well. Yeah, I just I don't get it. But also, too, like you said, if they are marketed as it is the Raspberry Pi that has, you know, AI power that also, too, could be open source for you to build. your additional tools once desire yourself with this and we grow as a community, I think, I think you have something, you know, on stuff, you know, then, and so, but, you know, building this as trying to sell this as, you know, this is a device that's going to, you know, be able to run automation tasks, inquiries and stuff in real time, no matter what you want to do. I think it's just, it's the, I think it's a missed shot. Yeah, you know where I think they should position themselves? Like, have you guys heard of the Flipper Zero? Oh, yeah, yeah. It should be right alongside that. The Flipper Zero is this kind of like Swiss Army knife of, you know, RFID reader slash FOB reader, you know, FOB RFID, you know, duplicator. reader um plus dozens of other add-ons that you can add that are kind of user create or user generate user community to your key if you want yeah you can um you know start your car with it you can do a whole bunch of stuff like but you have to like program it and you can also steal other people's information too with it but like that's another discussion we're not gonna you know get get into that but like um it's uh it's a hobbyist it's like a hobbyist tool yeah so that's what and actually the color is similar too now that I'm kind of thinking about it I think it has like similar colorways to the rabbit yeah No, I think it's a great point is I think, yeah, if you sell it as a hobbyist, as a building tool that you build, I mean, people have conferences now around Legos. you know, and stuff. And so, you know, I think if you sell it at the hobbyist thing and, you know, something to do, something, something to put together, you know, with the family or your board, when we like, you know, like that, but, you know, trying to sell this as the future of IOT, you know, with the, in comments with AI, it's just, it's a no for me. Yeah. Yeah. No one's believing that. He can, he can piggyback on like the education system, become part of STEM education and make this a whole platform. yeah you know open it up make it a whole platform where people can you know really make a customizable oh I can't think of the name of it it was a black owned robotics company they end up getting acquired I believe by apple um right It was, it was, I can't think of it, but it was, it was a black home robotics company and they get acquired by Apple. And basically they were using, they had like, I think their first, their first thing I was like, this spear is like a ball. And basically you'll be able to command it by building like functions and functions will be like, you know, move or, and then it will repeatable step, move forward. Okay. Now left and everything. And then basically you will, you complete this whole maze and stuff like, you know, like doing that. I saw this. Yeah. Yeah. And so Apple started to integrate it into like their, their coding classes. Um, and so it also follows the, uh, it follows the atmosphere of, um, What's that coding application that Apple owns on the iPad only? Swift? Or no. I think it's Swift Playground. Swift Playground. Is it Swift Playground? Xcode or no. It might be Swift Playground, something like that. Yeah, I think it's with. Yeah. And so yeah, so basically it teaches you the principles and everything of coding. And so it's like his game stuff that you build. And so somebody bought it, built like a robotics company around that. And so like even something like that, you know, where you can see, do classes, education, things like that to show you how the basic of these things work and how it can go to a larger scale. in a larger world I think so like that you know works but you know this that robotics company wasn't positioning themselves and trying to market themselves as boston dynamics yeah they they found that they found their product market fit everything and I'm almost confident they're gonna acquire um too as well and so yeah you know I know it's a side topic but like speaking of boston dynamics wouldn't you rather have the dog than the human. Yes, I would love to have the dog. I would love that dog. Give me the dog. I would rather have the human. Really? Huh? I don't know. I don't know. No. First of all, to have a humanoid or anything, and it does your house chores from cleaning the bathroom, doing laundry, folding your clothes. Absolutely. I have a real dog. You know? That's true. See, I have allergies. I would jog with the dog. I would go on walks. I would, I would like, I would do the whole thing with her. I'd be like Mega Man with that robot, with that robot dog. That's funny. I would, I would take the dog any day over the, over the, you know, Especially if it turns into a car. Was it Autobot? Rock or something like that? It did turn into a car, that's right. Yeah, it turned into a car. I'm thinking Mega Man. Yeah, I'm thinking Mega Man too. I forgot the name of the dog too. It turned into a surfboard. I remember that. Is that Transformers? Or have we all just been watching too many... cartoons and movies up here yeah clearly I got I've watched too many cartoons in my day yeah All right, cool. Final segment. Here we go. So OpenAI introduces Swarm, pioneering the future of AI agents. OpenAI has unveiled a new product called Swarm. It's an experimental framework that brings the concept of agent-to-agent into a tangible reality. The AI agent represents the next significant leap in generative AI and large language models focusing on autonomous agents that can perform tasks, communicate, collaborate with each other, using natural language. Swarm allows developers to explore lightweight multi-agent orchestrations, enable AI agents to work together to achieve complex goals. For example, plain navigation could be entirely managed by AI agents. One agent can understand the preferences, another Agent can handle the booking flights. Another agent can handle the hotels and activities. These agents communicate and collaborate, pass the information back and forth to fulfill your request without constant human intervention. Swarm emphasizes six key aspects of the AI agent, which is orchestration, networking of agent, communication, collaboration, autonomous tasks, performance, along with natural language expression. Sekou, start with you, my friend, and everything. actually this is something you've been you've been trying to figure out work on stuff yourself for quite some time so what you think about what opening eyes doing here Oh, so this is the good part of this, right? We've been in this innovation stage where folks are coming up with new LLMs and trying to make the best model and things like that. I think we talked about the playing field starting to level off and us focusing on tools and product. This is it right here. So one of the biggest challenges when you're creating an agent is when you have, say, three different LLMs you're talking to. So you've got an LLM that is, what you see, you have a flight. LLM is like booking flights for you, right? You have another LLM that is responsible for, say, checking to make sure that your calendar works with this. You have another LLM that's responsible for the hotels, right? So you have these three different LLMs. One of the hardest things is whenever you want to pass context between LLMs, you kind of have to take the information and either push into a rag or something like that, load the context into the other one. Just enough context, not too much, because you don't want to have it hallucinate and stuff. And then do that to the other ones. And you have to keep all these things in sync, which is kind of a little bit of a pain. So what they've done is essentially made memory and stuff like that, memory between these LLMs kind of integrated. between the two, so they can kind of share the context itself in addition to the prompt that you're going to provide them, which I found to be the hardest part between trying to actually do this. But if you look on that, they have a really cool GitHub repo. for this framework that they put together that has a couple of great examples of making a personal shopper, a support bot, a customer service agent. Right. Now we're actually starting to get tools to actually create products. There are companies out there that couldn't wait for any of these things to happen. So they probably have built their own awesome way of doing this. But now these techniques are getting pushed down to the masses with the fact that they made a framework. Um, they're calling it, uh, what did they say? This is a educational. Educational framework exploring the ergonomic lightweight of multi-agent orchestration. But this is a huge breadcrumb to kind of get companies into being able to really create real products out of this. They say it's experimental. So you can expect things to kind of change. And they're probably, look, they actually said they will not be looking at your PRs as you send them, even though it's supposed to be open source. Um, but I think this is a pretty, pretty huge step in the right direction, especially for, you know, like creating true products out of some of this, this is the right inspiration that we need. Yeah, I think, I think so there it's, it's, OK, so it's a complex way to build your own LLM and everything. And then the most simple way to describe it is you do it in three steps. You go collect the data, and then you tune the model, and then you refine the model. And I find what's interesting that this world is also still following those steps. In the world of AI agents, I think the first thing of what you do, and it's funny, I was talking to someone about this. It's like the first thing you do is that you create a knowledge base. And you create a knowledge base of like what's your data, what's your information, what's your needs, those stuff might be. And on top of that, you create agents on top of that knowledge base. And those knowledge base is attached to that vector database of information or wherever you're storing it. And then from there, a task comes on top of that. So it'd be able to output and do tasks in your likelies, in your organization, to your framework and stuff where you desire. And, you know, I think you're right. I think this just definitely enters us into a unique place of what was the what was it was a company some time ago. It was like it was trying to do like autonomous agents or something like that. Autonomous AI. And it was a repo. It was a GitHub repo. And it was popular for like six months. And we just thought we was there. Yeah, auto GPT. That's what it was. Yeah. And so, yeah, I think this is basically what they were trying to promise. I think they was tapping on it. But, you know, if you start to kind of get into it, you start to realize very quickly that, you know, it was a lot of gaps, you know, in between there. And I remember one of those solutions that I could do was like the book ending handling of like your flights information. And so I remember running some of these things, except my local, and it returned back to me that the flight has been booked. Never asked me for a password. Never asked me for my credit card information. Never asked me for things like name. And it just turned back to you that the flight has been booked and it's kind of going through all these different tasks. I think what's also interesting with this is the security standpoint and this is questionable because I think in this world now is the perfect, perfect time to insert blockchain because now to be able to create knowledge bases that go to go gather my personal information, like credit card, birthday, you know, social, you know, things like that, to be able to gather that from a blockchain. And I consistently give it approval access to do so and know where the information is being sent and how it's being sent. And then in partner that with agents, I think this is the perfect time to do so. I think it's a little bit concerning. that as we are advancing in agents and things like that, I haven't seen blockchain come into play or even been talking about being a partnering solution, which is also too, especially more so when it comes to the consumer side of things, when they wanna kind of like capture, personal information, personal data, which is concerning because now I feel like we're kind of like repeating some of the mistakes we did when Facebook and type of companies and stuff launch. where we didn't know what we were doing. It was kind of like handing personal information, personal data, and it was kind of going, you know, and so we've seen some of the backlash and stuff like that. But, you know, we have a way to still now build with this AI-centric universe we're going to live in with also kind of protecting, you know, the consumer and their information and still getting maximum output with doing so. You know, we partner with a technology like that. And so I think it's in the right direction. I think they're still going back to the car before the seatbelt. uh type of mistakes that we're making into as well um you know and I think we'll see see what happened but it's exciting you know um and I think we'll see what happened yep yeah I feel the same way um um going back to um what seki was saying about the um how difficult it is to when you're dealing with agents um I had some experience with this when I was working at live nation um I set up a chat bot to do like five different things. So I used Langchain and different agents were tasked with different specialities. So some was downloading reporting, some was creating a table, creating tables out of reporting, pulling different pieces of information from different places. Building ad hoc programs and scripts to perform actions to was another one. Uniting that into a front end that you can talk to using natural language and then your back end is going to intelligently be able to decide which agent is going to accomplish which task. I got that to work to a degree to where I could use it, but I was not able to get it to work to where I could hand that off to the broader members of the team. And if they have solved that issue, Using the shared memory about halfway through this documentation, but using that shared memory solution or shared context between agents. Yeah, that's something that specifically could have helped a lot during that project, which worked. I mean, it worked for the most part, but then again. it's only a couple people using it like it's not nothing that that would be ready to give to um to more people because it's just it would do its own thing um far too often yeah what's the what's the microsoft one they're building that is it like the multi-agent um Because Microsoft is building one too as well. It's like multi-agent, double agent, something like that. Oh. Many of agents. It's something like that. Microsoft building one too as well. But with that, yeah. I think it's multi-agent. Yeah, sure. Sure. That's a mixture of agents. Yeah. OK. Yeah. So with with with opening eyes and Microsoft's and, you know, Amazon's building frameworks like this, I mean, you guys think lane change survives this curve? You know, do they do they get acquired? Do they kind of fill away or do they stand? Do they stand strong? I don't know if they're going to survive this. I know they're trying to create a product. They're trying to create pipelines and things like that. But I think Replit, Bolt, which we haven't talked about Bolt which Rob should talk about in another episode. But these other tools that are handling the full CD process, in the environment creation and in addition to the framework, I feel like they're kind of cooked from that standpoint. So the thing is, there's no like still we're still missing like the de facto framework, the see the language for JavaScript react, react. We're missing the react for with this particular work. I talk to anyone that's doing work in this area, and they have to build their tools from scratch. They have to figure out how to do continuous delivery for what they're building, how they're connected to integration. But we don't have just the general tools yet. So folks are trying, but nothing is sticking yet. So there's a lot of room for improvement. But Langchain is trying to pivot very quickly to making some product out of it. And I think they actually made folks move a little bit further away from it, too. Gave them the egg. Gave them the egg when they tried to make a product out of it, yeah. Yeah. It's like, you guys, how dare you? What? Yeah, I think it'll be interesting to see what happens. I mean, because at one point, too, it was all about lane chain. Things shifted that quickly, which is wow. It was all about lane chaining and building on top of that and with that, unitex of that. And just in a development world, I haven't heard lane chain in a minute. you know, because it's so accessible to build in other directions and stuff now. Yeah. Okay. All right, well, guys, I mean, today, I think we had an awesome episode. You know, we had, we got to, you know, kind of sit in the autonomous world of Tesla, you know, as they did the unveiling of the CyberCab, RoboVan, and got to see a little bit more from, what's the robot car? I always forget the name Optimist. Optimist. And also too, we discuss Wendy's in the drive-through process, how they're using LLMs to streamline their drive-through process. Google joins other competitors like Microsoft and Amazon as they invest into nuclear power to help us up with their data centers and AI compute. And then our good friends over there, Rabbit. Rabbit and stuff is continuing their charge on you know, believing that they are the future of IoT and, you know, AI. And then closing out and stuff here, man, we've talked about OpenAI introduces Swarm, you know, their future of the AI agent. And so, guys, if you're listening to stuff here, you know, definitely follow us. You know, follow us. I am Marlon Avery. We also, too, now we have AI with Friends. It's AI with Friends podcast on all platforms, on Facebook, LinkedIn, Instagram, Twitter. So you definitely can follow us at I am with friends podcast on all platforms and also to join us every week here Wednesday at nine p.m. Live on YouTube and Twitch. Say cool ages, man. Give us some closing words and stuff here. closing words um audacity is my closing word I'm going to say that again yeah and uh my closing words are going to be um find your favorite secret menu item um you know what we we will return next week with the favorite secret secret item And we'll share with the community and stuff here. Yeah. I got one for Chipotle. Okay. Yeah. I want to hear that one. Yeah. So, yeah. We'll definitely do that. All right. All right, guys. Well, until next time, we are out. All right. See ya. Peace.