AI With Friends

EP2: OpenAI o1, Slack AI Agents, AI Automations, AI Second Brain, Intel + AWS, AI-Focused VC Fund's

• AI With Friends LLC • Season 1 • Episode 2

Share episode

Send us a text

In Episode 2 of AI with Friends, we’re diving headfirst into the hottest AI topics of the week! Join Marlon, Adrian, and Sekou as we break down OpenAI’s latest bombshell—the Omni1 and Omni1 Mini—and explore how this new model is pushing AI's reasoning to the next level. We’ll also be chatting about AI agents like Replit that are shaking up the game, transforming everything from coding to social media. Speaking of social, we’re digging into how AI is reshaping platforms like YouTube Shorts and TikTok, and the team has some big thoughts on whether AI could become your second brain. Plus, we’ll talk Intel’s new AI chip deal with AWS and the massive AI VC funds that are fueling the next wave of innovation. Tune in for insightful discussions, practical takeaways, and, of course, a few laughs! You don’t want to miss this one.
-------------------------------------------------
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 week on Wednesday at 8:30PM 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!

Support the show

Follow the Hosts:


Affiliate Links:

Yo, yo, yo, yo. Fellas, how you doing? What's up? What's up? How's it going? What up? Yo, you know, real quick, before we get started, you know what we should do? Number one, we're definitely going to add in, well, actually, if you're listening to this in a podcast, you'll definitely hear kind of like the AI intro they've created. We should let the fans, you know, decide if they like it, love it. like hate it or anything and then it was a completely ai generate the music was uh created uh with uh what was the music order order the music through was it sooner sooner no uh-uh the other one um starts with ass wow going blank Well, the voice was Eleven Labs. Yeah. And the music, the tag, the prompt and stuff was created for the Chachaputi. And then the music was done with, oh, man, it starts with an S. That's going blank. I thought it was Suno. No. It's a different one. It's the other one. My phone's over there. Anyway. But yeah, we should definitely let you guys decide so we can keep it, scrap it, start it over. Anyway. Guys, man, welcome back. Episode two. Episode two, AI with Friends, man. This has been exciting. And also, too, we crossed over a hundred listens already on for EP one, man. So definitely thanks for you guys tuning in and listening. So we're going to keep it going and stuff here, man. I am one of your hosts, Marlon Avery and stuff here, man. I'm going to let the fellas introduce themselves as well. Adrian, man, kick it off for us. What up? I got the answer. Stable Diffusion. Ah, it was Sable to Fusion. Yeah. We did Sable to Fusion for the music. Appreciate it. Yep. Yep. Hey, how's everyone doing? I'm Adrian. I'm a head of tech at DistroNation and overall AI enthusiast, nerd, builder, hacker, whole nine yards, making it work. That's it for me. Yep. all right how we go um so hey everyone I'm saku I'm the cloud engineer extraordinaire working over at uh pinterest um you know I'm I'm a ai this enthusiast always looking to learn new things and share all my ancient history around tech with folks and that's why we call you the wise wise. The one he's starting to believe. And I guess I'm Roland Avery, AI practitioner here, AI enthusiast as well. We're in a company called AI Impact where we do training and also, too, we do building consultants as well for AI solutions. And so, yeah. Guys, I'm over here in D.C. D.C. is a vibe. and stuff over here. D.C. has definitely been a vibe, man. It's been welcoming. They got some fun things going on and stuff into the city, too, as well. Adrian, man, you said you're not too far from D.C., too, as well, right? Not too far at all. I'm about three and a half hours away. You was considering, wasn't D.C. one of your cities that you was considering? Oh, yeah. D.C. was definitely one of my cities. Yeah. It's just, you know, it's a lot of driving in D.C. It's a lot more driving in D.C. than it is in the Bay Area. Like, you really... Wait, really? Well, you can't just jump on a train to get where you gotta go, right? Yeah, the metro. It's not popping. It's not a metro, but everyone lives far out. Yeah. you know but yeah I love dc the food the dc go go you know I'm not ready I'm not ready for the dc go go this has been fun I've been up here a few times uh they've always kind of like welcomed me in I've done a few workshops up up here I uh I spoke at a conference yesterday to as well. That was awesome. And so it's been interesting. It's definitely got a great energy. And I already know a bunch of people, not a bunch of people. I know a good amount of people and stuff right here just doing some awesome work, you know, in tech and government and things like that. And so, So, yeah, man, it's been good. But, yeah, man, let's cook it off. Last week, man, we had some great conversations, though. You know, we talked about everything from, you know, AI phone agents and everything to, you know, the California bill and stuff was going on. And so we're there. You know, eventually we'll have discussions around the AI EU Act, too, as well. You know, probably next week and everything. But there's some popular things going on this week, man. So let's dive into it. OpenAI announced... omni one and omni one mini uh for me kind of a shock you know because also we've been kind of like waiting for the anticipation uh gbt five um and so uh you know guys what's your thoughts you know they they uh open the eye announces that basically this is the first step you know to proper reasoning uh within um you know artificial intelligence and lms and so And so I say right now, I think on a PhD research level, um, and it can kind of take a thought process through steps, everything organized says, and kind of, you know, give you a more efficient and proper responses there. Sekou man, kick us off, man. What's your thoughts? Oh, I'm loving it. I'm loving it. I burned through my allotment of requests in like the first. Fifteen minutes before realizing there was like an end to the request. But luckily, they actually open it up so you can, I think, get up to, I think it's like, fifty on preview, if I'm not mistaken. And maybe the same thing for many. But it's way smarter. It is... it can actually give you a lot more substance. But you still kind of get to the same place in some ways. I was looking at a, I think it was a meme of someone doing a whole bunch of skateboarding tricks and eventually hitting their head. It just means you can do more tricks before you hit your head, essentially. Oh, that's great. I'll take that. I'll take that, even if I was skateboarding. Even if I was skateboarding, I would take those odds. A couple more things, and same thing. Same place, though. If I could land a kickflip, yeah, I'll follow you. We pushed out the component of concussion further out, so it gives you a little bit more fun and stuff here. Also, too, I didn't know that it had limits on it. I haven't hit any limits yet on the Chachapati side and everything. I've been messing with it as well. They might have removed them too as well, so who knows. They adjusted it pretty quickly when they saw folks burn through that initial allotment that they gave out. So, yeah. Adrian, what's your thoughts? Well, I mean, you mentioned before that it's comparable to like a PhD, like a PhD professor, someone who holds a PhD there. I have yet to try it, but I would say that what I'm using now is like a teenage intern when it comes to the AI stuff, where Um, I really have to prod a lot, um, and you know, just short of encouraging it, uh, to get, uh, really what I want sometimes. Um, I've been doing a lot of, um, experimental things, uh, with that, like, um, even within the, like my IDE code itself, opening up just a commenting section. and turning that into a chat interface, which I saw used and I had to replicate, which was you just open up a block of comments and then type user colon, your question, and then hit line return, type AI colon, and it'll finish it. And then that'll start a chat interface right there where you could ask more specific questions instead of just relying directly on co-pilot, you know, out of the box. It's like a weird sideways engineering approach. That's kind of cool. You know, here's one of the things I realized. I used it quite a bit and stuff already, but I... I typically don't go watch some of OpenAI's preview videos or video stuff they walk through as they typically do. I'll look through something like the documentation, the research, or the documentation. But I want to go watch some of the videos. And specifically, I was looking for how they was prompting the new model. And one thing I noticed very quickly is that it wasn't the... the kind of the prompting strategy that we're used to. So, you know, things like assigning the roles, giving details. It seemed more like a pseudo code, you know, in a way. And so this is, I mean, this is additional question stuff here. Do you think prompting has shifted already? And we're going to see more so like pseudo code workflow and instructions kind of going into it. And before you answer that, let's take it to the audience, what a pseudo code is, and then, you know, answer the question. Yeah, pseudocode is generally when you're writing the expression of what the code would do in layman's term. You're not actually writing code that would run, but you're giving instructions on the flow that you should take. And maybe to answer the question, I feel like to get the best responses out of chat TPT, Like, I found myself setting schemas so it can give me output in a very particular way and putting certain information in the middle and the first initial prompt. I feel like I'm still kind of doing that. And actually, I find myself actually doing it a little bit more if I want to get the best out of it. But the thinking portion on Omni, where it's actually taking multiple actions, is actually, I would say, it actually feels a little bit less, even less predictable right now because it's so good. Yeah. So it's really interesting. It's like the fact that you ask it a question, you give it all these parameters, it'll see what you said, ask itself some questions, make a decision, maybe ask itself some more questions. And then finally give you a doctored output. I think that changes the game, I think. Interesting. So is the, because I'm sure I'm not the only one who's seen this, that the GitHub repository of awesome prompts, is that just out the window? Is that going to be something that's going to be still usable? I feel like we have to change the strategy a little. I don't think it necessarily be out the window. I think you still may start with the first line, you know, a line or two or anything as the prompt descriptor. And then you're kind of like filling in from there, the steps and like the pseudocode, you know, after that of like what you want, you know, to see if your output, you know, I've, I've tried both like strategies and what I've seen so far is like, you know still kind of like giving it a role of like hey I want you to act uh such and such and then kind of like following what they did in the video it's like you know giving like that pseudocode you know like stuff or thing I've seen good results and stuff for that uh but I mean it's still like llm and everything and so I'm not sure if we've moved away from that um so as of yet but I can definitely see a direction that we eventually will yeah but when it's wrong It's like it's doing flips in cartwheels while it's wrong. Adrian, you made an interesting comment earlier when we was offline and stuff here. You said it feels like Omni is what Slane promised to be. What was the tool of the platform? Auto GPT. And I was talking about Omni in combination with the Replit agents. Replit conversation. Yeah. Sekou turned me on to that. That's my new favorite thing. Oh, so good. Let's dive into this. Sekou, tell us more about what you've been experiencing yourself with the Replit AI agents and things like Cursor as well. So Repl.it uses, well, AI agent is essentially where you have different agents where they are tasked with doing different functions in order to accomplish a single task. So if I say, you know, write me a book, there'll be an agent for writing the actual book that's only focused on writing. It'll be an agent that's focused on making sure that the grammar is correct. An agent that's focused on copyright. All these three agents produce one output. So Repl.it essentially does that, but does it with code. And it's actually pretty dang good. They have a pretty nice interface where you can go in there, you can put in a prompt. And it feels like a new way of coding, in a way. You're not actually coding, because you're giving instructions. But you're massaging what the final output is through prompts. And I've been able to create some very useful things using this. And it's just surprising how easy it is to use. I might have a page worth of conversation to get there. But the amount of code that gets produced, it actually creates files. It will import libraries. It will swap between different languages. If you're doing front-end code or back-end code, it will do the whole thing for you, which is, yeah. That's probably the best implementation I've seen so far has been Repl.it AI. Yeah. Adrian, what's your thoughts on the, you know, I know you say you just you start to get into yourself, but the combination of things, especially we'll be seeing with the replica and also to just like cursor and other ideas and platforms. Yeah, as far as the replica thing, the replica system, I remember being frustrated with auto GPT in that it would just go off the rails quickly, quickly, very quickly. It would go off the rails and do its own thing. Where it feels like that has become a lot more tighter in a system like Replit, like what Replit has. In the IDE side of it, you know, I'm a creature of habit. So I really, I'm still using predominantly VS Code to do A lot of my work, but my experiences with cursor and just how much better and faster that that is. Has been has been great. It's been really great. It's been making my days a lot less frustrating. I think the, the, the leap between GPT three and GPT three point five was the, uh, the embedding of starting to understand, uh, uh, coding. Um, and that kind of like, it was like a baseline interest of reasoning as well, but also too, I think that was like the. the foundation of what we start to see when it comes to things in like workforce and automation, how it can be an assistant, you know, to us. And so you guys, do you think we'll start to see, you know, Replit type like AI agents for other industries and everything? So like for product manager agent, for, you know, UI, for taxes, for just like, you know, entrepreneur assistant, you know, what's your thoughts of that? I mean, this is just kind of like, you know, V one of the world we're about to see. with agents and stuff. What's your guys' thoughts? I would say, yeah, agents, I think that looks like a product. I haven't seen a lot of things that are truly a product. Some things are cool emojis or changes, small little surface area changes, but this is probably one of the most substantial things I've seen. Companies being able to integrate AI agents into their workflow to be able to like augment and work beside an actual engineer. Just like the example I showed you of like writing code through prompts and essentially just doing entire workflow without actually having to write, write the code is, I think that's, I think that's, that's huge. Imagine doing the same thing for a legal document. You know, imagine applying that to, you know, creating banners for a site or something like that. Like, yeah. Adrian? Yeah. Whew. I don't even know where to start. I mean, I feel like what I get the most excited about, yes, agents are the future, but the, you know, splitting up, you know, what you mentioned earlier, Sekou, about you put in a prompt and then you have a front-end agent, a back-end agent, maybe a QA agent as well. I mean, I'm a solo developer. I've been a solo developer for a very long time, and, you know, I'm working on a rather large project now where I'm wishing I had a team. And sometimes I'm reaching those crossroads because it's just so much to do when you're thinking about even a website that doesn't do that much. Of course, I decided to do a streaming music one, but that's my own. ambitions I guess but uh how you know that just changes the game you know I'm from the startup world and um you need to really burn the candle at both ends because a lot of the times just um you know it's a startup it's a small team you know you got to just make it work you get a bootstrap um from one uh milestone to the next so yeah that makes me so With the combination of all of those things, like I said, I think software engineers are typically some of the first ones who kind of get to see the full benefits and stuff of these things. But I think also, like you said, from the legal aspect, from the design aspect, from the QA aspect, I think we'll start to see you know, more agents. And speaking of agents, man, Slack announced that basically they're about to turn it into an agent hub. They said the head of Slack, Denise Dresser, tells TechCrunch that she is shifting the business to a chat platform and into a work operating system, specifically by making Slack a hub for AI applications from Salesforce to Adobe to Anthropic. The company CEO sees Slack as more than a workplace to chat with your coworkers, and they announced several features on Monday for a pricing tier of the national platform announcing Slack AI. So guys, what do you think? Slack is going to get into the aspect. I would say they're going to become an AI agent hub, obviously for several businesses, for several organizations. Agent Man, we'll start with you. What are your thoughts? I think it has the perfect bones to carry that, man. There's already so many extensions you can add into Slack already. It seems like it makes a lot of sense. It makes me, within the corporate environment, like within the job, it makes me wonder about which, you know, I got pump faked on this before the last podcast, Marlon, where I thought that I found some publication about HR being replaced by AI. That right there is making me wonder again if there was an avenue by which HR would be supplemented or kind of like augmented by AI, it would be through a chat interface like that potentially. But also it could be for annoying job surveys as well. You know, it's interesting on the HR side of things. I've learned very quickly, depending on the industry, depending on the company, that the definition of HR within the company can be extremely different per company. And so I... I won't say any names because we're trying to get out of platform or stuff here. But it was a company I worked for for quite some time ago, and I quickly realized that HR was basically the mafia of the company. Yeah, they were there to protect the company and the company, you know, alone. And so when it comes to who they, the intersection of, you know, agents, you know, replacing AR, I mean, HR, they I see some, I see a lot of benefits, I see some things, depending on what the company wants the AR to represent, I see it to be in challenges, and stuff in those ways. But I also see, you know, agents definitely being able to support tasks, you know, things like payroll and, you know, figuring out how to, you know, submit a anonymous, you know, feedback, you know, from, you know, an employee to your manager to, you know, whatever it may be. And so I see some benefits up there in how, you know, these agents definitely can support that. I think also, too, just having the definition of HR. for a company per per organization makes it even more challenging and difficult. Um, so yeah, Seiko, I'm not sure if we're allowed to say, but you used to work for, um, that wasn't select directly. Was it, or was it like they got acquired? No, it was, it was quip. who got rolled into Slack. Well, somewhat. Well, yeah, pretty much got rolled into Slack. Yeah, I think the Slack has a pretty huge advantage with the integrations that it has, the workflow engine. But not just that, but they have access to actual real data. which is the hardest part of some of this. So they're already positioned, I think, to actually kind of win this. It's just really how comfortable they get folks to use their technology. I feel like there are a lot of good benefits to Slack. I think everyone has had the, like, situation where you, like, you're working at a company, you know, you got to, like, ask around, around, like, hey, what does this acronym mean? Or what is, how do you do this? Like, it's your onboarding process, right? And it's usually, like, you know, someone you get sending a message, someone creating a doc that automatically gets, like, outdated the day that you actually write it. or it doesn't exist at all or it doesn't exist at all but there's a not it's a giant knowledge base of information that is inside of slack channels like if you're able to connect that into um like use cases that support that part those parts of the business I think that's I think that's like one of the places that we'll see a lot of good things come out coming out of that other ones like um Working in operations, I think every company I've worked at, we've always had an idea of an incident channel, like a place where you have everyone congregates if something blows up. We all get in one room. We try to keep all of our documents and everything tight inside of there. But you can only keep so much context when it's time to replay how you did things. It's always challenging. Like boom, drop AI bot inside of there, you know, um, correlate that information to a living document, not just things you write to get outdated as soon as you write it. So yes, there's a lot, there's a lot that you can do with that. So I'm actually, I'm super excited, but there's the other side, HR part we were talking about, like, you know, for every good thing you can do. With AI, there's a shadow, right? There's always that shadow. Someone's going to do something that's unethical or just creepy to do to employees. So I'm just waiting to hear the first story, the first lawsuit. And hopefully Slack Salesforce kind of has their act together on how to like mitigate those type things or, you know, make so folks can at least trust the environment. Yeah. I also see this as a play where Slack can become even more competitive with companies like Microsoft Copilot, with tools like Microsoft Copilot, because, you know, guys, we've been having this call now for like, you know, over a year or so. And typically we've done initially on Google. And then, you know, we did it for a little bit for Zoom. And then for whatever reason, some of the most clearest and uh seamless video uh version that we had was discord um and and now you know we're doing our stream yards you know and so integrating something like that I would imagine you know slack would be something similar to discord because then we'll start to see teams that are starting to meet and just simply into slack everything and having using that as their main you know video meeting platform and then we're starting to see integrations of you know ai note taking you know even ai project management taking you know segments of documentation doing uh the retrieval information and sending emails to people who are supposed to be in the meeting but they're out of office you know things like that and so I definitely can see how this can kind of like you know lead them up to a competitor of microsoft copilot as well Oh, yeah. I definitely have. Who can get the most Nvidia GPUs to make that happen? Is that with these Nvidia GPUs, is this a point where it is who has the most money or this is just straight relationships? Who likes who better than to send out these orders? Oh, yeah. I'm sure it's relationship-wise. For sure. Yeah. For sure. Yeah. I think it is. You got to pull up with the leather jacket. You got to join the bicycle club. Reinserting Blade into the Marvel Universe. You know. Roland say here, he says, hey, thanks for describing my monthly issue, constantly rewriting outdated policy docs. Wish I could automate that. Man, Roland, we definitely feel you and understand you too as well. uh you know speaking of uh sake will you just mention the war that's going on uh hopefully nothing lethal uh that's going on but intel inks inks ai chip deal with aws uh they're announcing they're they're trying to get into the space uh you know as well and so Intel says they announced a key customer win and challenges to its Foundry business. And as a chip maker looks to execute a turnaround, Intel is taking steps to transition into chip Foundry division, Intel Foundry into a subsidiary. The CEO, Patrick, announced this stuff on Monday. Says the leadership isn't changing and the subsidiary will remain inside Intel, but Foundry will gain an operating board, including independent directors. And so they said they had announced a deal with AWS that they're moving forward to getting into the chip war, if you will. False. I think that's... Oh, man, that's so interesting. Because AWS... This actually reminds me, we were talking about the... It was Anthropic and AWS Connection last episode. Yeah, and this is actually almost on the same lines because AWS has their own inference and training chips. It was called Trinium and I think it's Infraentia or something like that, a lot of two chips that they have, which has always been interesting to me. And they also have their own Graviton chip, what they make. So I think they're just trying to have options for folks to choose from. Because the NVIDIA stronghold, they have a huge moat right now. They have a huge moat. And one of the biggest challenges, even with their product, you've got to choose between, oh, do I want to build a system here that can do training? Or do I want to build a system that can do inference? Whereas if I got a Naviga GPU, I can do both if I have a good CPU on the same system. And that's pretty huge for folks. You don't have to pay twice for something. So Intel enabling that, I feel like, is a smart move on the Amazon side. But the question is, what does that mean for their other chips that they have, right? Yeah, I have so many questions. I'm like, I was about to say, Adrian, you had a bunch of questions and I had to kind of put a bookmark and everything. So definitely take us away in this next segment. All right. So with this, so with this deal, what exactly is AWS providing? Are they providing the chips to Intel to be rebranded as or to be like a kind of white label as Intel chips? Because as far as I knew, Yeah. It says Intel has signed a deal with AWS to co-develop an AI chip using Intel's eighteen eight chip fabrication process. Intel has also agreed to produce custom Xeon six processors for AWS also as exchange building on existing partnership between the two firms. Sweet. So Intel. did the manufacturing and the design within the same company before. And now they're spinning with the foundry division, they wanna spin off the manufacturing part into its own thing. Which makes sense because I guess wasn't it earlier this year that uh some shareholders sued intel over maybe being not so forthcoming about um earnings and their performance and like how they were doing um you'll enjoy this right here Uh, Gisling, Gislinger, Gelsinger. Wow. The, uh, the CEO said we have trippered. I would deal pipelines at the beginning of the year, uh, uh, describing the AWS deal as a multi-year, multi-year, multi-billion dollar framework that could, that could possibly involve additional chip designs. He added that it demonstrates the continual progress that we are making to build a world class about your business Intel cost cutting and deal making along with a newly awarded a three point five billion dollar contract to to build chips for the Pentagon. Since the company stocks company stopped soaring over a six percent at market close. It's a bright spot for Intel's otherwise, you know, for their fiscal year. Intel posted a four hundred thirty seven million dollar net loss that widened to a one point six billion dollar loss in QQ. It's insane. That's what I'm saying. This this move is for the shareholders. I mean, this has got to get them the warm and fuzzies a little bit more because it was going down. You know, it says it says Intel foundry posted a five point three billion dollar in operating losses in H-I despite a slight year-over-year climb in revenue. Oh, man. I feel like I have so many mixed emotions with Intel because, like, you know the newest chip they came out with, the Lunar Lake, which when Microsoft unveiled all those new co-pilot PCs running Qualcomm and there was no Intel AMD, like, I knew that hurt. That must have hurt tremendously. But their rush to get a chip out that can be a co-pilot chip has led them to use TSMC to fabricate it. So Intel's not even making their own chips at this point in time, which is the strangest thing. They outsourced it to TSMC, the same folks who make the Apple's chips. But I don't know. I think I know that they're trying to turn things around, but it's, I don't know, it's a little scary. It's hard to do both, I would imagine. Both things, like manufacturing and design, nowadays, possibly. I don't have that much knowledge of the landscape there, but I feel like if they would specialize in one the other it would be maybe easier you know easier for him what's the definition of difficult if the contract is three point five billion I mean the contract is three point five billion with um aws you said No, they signed with the Pentagon to make chips for the Pentagon. They got to make the chips. They got to make them then. They got to make some chips. Make some chips and do what we need them to do. That's the hard part. Exactly. It's been a lot of crashing. It's been a lot of issues with the chips too. Yeah. I remember that. There's a chip that they recently came out with, a desktop chip that's just been Feeling pretty bad, which, yeah. I feel like we do need Intel to be pretty strong for the U.S. government, at least. Yeah, I feel like just for really, just like the U.S. period, man, like it's, they put us ahead of the chip race, right? Wasn't that the jumpstart that we needed? And we've been like ahead since then? Or am I making that up? They were. And think about it. Like if, you know, TSMC is Taiwan. It's a Taiwan-based company which has tons of issues with China. We want to at least make it so that the chips are being made here, which they are doing. TSMC has a factory in the US now and all the Apple chips are coming from them, but Um, you know, still not owned by American company. So like, yeah, we, we, we definitely need a strong AMD and Intel out in the U S I think so. And Biddy is based in California. Yep. It's in California. It's a U S company, definitely a U S company, which is awesome. Awesome too, but they're not fabricating in the U S. So all their ships are coming from Taiwan, which hopefully they'll use those factories in the US and Arizona now. So we got a good fan and stuff here, man. Our friend over at Sake. Sake sent us a request last week to talk about, man. How do you think, how will AI automations impact the social media landscape, specifically things like YouTube and YouTube Reels? I mean, YouTube Shorts, sorry. Uh, I think it's doing it already. There's a, there's a giant pipeline of AI generated content that's just being shipped to, uh, the social media. I think right now what we are all kind of doing is like taking long, long stream, long, what's it called? Long form content, slicing it up automatically and then dicing it up through shorts and things like that. Like that is like the first entry into that. But the next one is like, Hey, we don't need to make that long form content. We'll just maybe record ourselves and let something get generated. Like that's the next place we're going. That would be great. That'd be great. But there's, I mean, there's a lot of tools out there to get you going. Like, um, I've played around with one, um, in video, you see in video.io is the website. Um, you can create videos with a text prompts and they have a pretty good free plan. Uh, I played around with that. Um, you know, I, I work in the YouTube space. And the, I try to stay up on what people are doing as far as like generated, uh, really AI generated channels and how they perform. And it's kind of a mixed bag. Um, but the people that, um, got it, got it, you know, they really are getting like probably. Two hundred K views per video on a regular basis. And the, but it's, it's few of those. It's very few. yeah um this is this is marlon a with the practitioner because I I got a bill to learn um and just kind of like start doing things so I didn't talk about this guys but I guess I'll say it now uh I created a tick tock channel uh from a bunch of different youtube videos that is all talking about stories about kobe bryant Okay. I just want to see. What? So many questions. I got so many questions. But this is exciting. I've been wanting to do this, man. I gotta hear more. Yeah, so I saw a couple creators talking about basically how they're creating these faceless TikTok channels and they're just making a lot of money and stuff for it from TikTok's creative program. And I kept seeing it and I found myself one day going down a Kobe Bryant rabbit hole. I would just watch it. I watched like ten, twenty videos. I mean, I got I got to get off of this and everything. And so then I was like, huh, I know for sure, you know, if I'm doing this or there's tons of people doing this or anything. And so I just put together and I use Opus clips or Opus dot pro and stuff with it. I use Opus dot pro. We'll put the link and stuff in our show notes as well later. So I use Opus dot pro earning and created an account credit, credit, try different, uh, videos. Could we get a little bit of editing and everything, but open star pro does a sensational job of taking the video in long form and then creating clips and everything on your behalf. And so then you kind of go in and you edit per clip. Um, and so it was awesome. It was awesome. And so I created the channel last Saturday. I probably have. eight to ten videos over a thousand views um and so you know just talking about just you know kobe you know kobe stories from your favorite athletes to you know former trainers and things like that and so I just scheduled out you know a bunch of videos to be scheduled out for the whole week I just want to see how I perform and as of right now it's performing very interesting and it's kind of annoying because I mean that page is getting more activity to my own personal page But I get it. I get it, though. I think it's things like this where we're seeing AI platforms like Opus or video that AI that will take the continual process of long form to short form and being able to link back. I definitely think that's one of those things that's not only going to only happening, but it continues happening. And then additional to that, I definitely see a world guys have you seen this these uh like these ai instagram models that like there are four pages now I definitely want to receive see a world where now those those ai models in ai creator influence will start to have youtube channels Oh, for sure. And they will start to talk about different topics and everything. I don't think it's going to be anything serious or anything, but I think they'll start to show, like, you know, things of, like, I can definitely see how they're going to start to create, like, affiliates and things like that because basically then somebody like Burberry will create, like, digital clothing, you know, for that AI, you know, generated model that, you know, that person or that person I don't know what to identify it as but that ai you know they could basically start to wear and label that and it will just like start to have the and I've seen I've seen a couple pages now that are um that have like thousands of followers um you know as well and so the transition that I definitely see a world you know where I mean honestly I think did we talk about this before but I also see a world where you're going to start to see on the social media side fully generated ai podcast from topic to automation to using a platform like eleven labs and then the podcast itself is just fully ai generated and guys honestly that'd be that'd be a fun episode for us to uh build you know, a platform or we'll build like a episode like that and stuff here over AI with friends and let it run, you know, for episode and everything and see how everybody feel about it. And we may even like start to create AI versions of ourselves that we invite on as guests as we're having these conversations too. That... yeah you know a couple years ago there was this um say about six years ago there was this instagram model that popped up called lil miguela miguela um who was yeah remember that and then she started to get brand deals and it was a fake model like I don't know if the body looked real but the face was like clearly like cg or something like that but I don't know whatever happened to that, but you know, she ended up getting whatever, whoever ran that ended up getting like brand deals and a lot of publication, uh, beyond behind that. So the, the idea of something like you mentioned before, AI models being like, oh, I want that AI model in my campaign is wild to me is really well, um, you know, but. It's totally possible. A friend of mine was also doing generated videos, generating music on Suno and then creating the videos with it wasn't stable diffusion. It was something else he was using. I'll put those in the notes as soon as I find it. But his intention was to create a character like the study girl. I don't know if you guys seen those videos where you know I don't know if you uh you know like music to study too yeah lo-fi girl yeah yeah yeah yeah his his intention was to create um a character like that um to you know basically be the background to a bunch of generated lo-fi music, but then to also have a story that it kind of follows along. Which I don't know if you ever went into lo-fi girl lore. Trust me, it's deep about just whatever theories about that character and whatever she got going on. You can find many videos about it online. But a lot of that makes that a lot more possible now, where before you would need a graphic designer. to get that done now, you could probably bridge that gap a lot faster. So the AI model is still around. That platform now has two point five million followers and they just posted something yesterday. Oh, wow. Wow. Yes, I never I don't know who runs that page either. I actually saw the founders who created it. They were on A-Sixteen Z podcast. They were talking about it. It was like two years ago or something like that. I'll send it to you guys. It was interesting as well. OK, so here, Sekou, I want to start with you with this one, because this is something that you've been alluding to, you've been talking about for quite some time. So here, I feel like if you don't build it first, I don't know if anybody else will. But this idea of artificial intelligence becoming the second brain for us. And so do you think AI will succeed? Do we think we'll actually see it being the second digital brain? You know, we're starting to see companies spin up, you know, who are trying to take, you know, who are trying to tackle it. You know, Fathom, they just raised seventeen million dollars. You know, we also know she's been trying to do Obsidian. You know, it's been trying to do it, too, as well. Is everyone still around anyway? And so I don't think it is. So there's a few companies been trying to figure it out along with our co-host Sekou. What you guys thought? Yeah, I think it's really interesting where we are right now. Like right now, you can you can create the concept of second brain, which just to give the audience a little bit more detail in this idea is that you have a second brain is like a notepad, another place to store information that relates to your own thoughts that you can either come back to and either reminisce on it. But the AI association to the second brain is when not only can you, uh, record your thoughts but you can be reminded of them and you can also search the thoughts to be able to search your your second brain to be able to pull out relevant content to your life um so yeah I feel like I feel like we're almost we're getting there like right now like there's a lot of there are a lot of hobbyists out there who are building tools to support that I'm doing some of that, doing some of that work of like, you know, back in my, my Evernote days, I used to, I remember having a port between Evernote. I don't know, the using of Evernote or you used to work for Evernote or both? No, I didn't, didn't, didn't work for Evernote. That's actually one document storage place I did not work for. But I was a heavy Evernote user, and I had a lot of information stored in Evernote. I continue to do it throughout my career of just taking those great aha moments, things that maybe I've never implemented, things that I ended up moving to some type of major design from. And then of course, Evernote like deteriorated and died. So I had to move to something else. So I moved to OneNote and that was terrible. Yeah. That's backwards, man. I know, but it was the ability to kind of sort my actual thoughts and categorize them by time. Like it had the best tagging system, right? But it's all connected to being able to search, like just writing things down on a bunch of notebooks that you cannot like you know, pull out the information you need or, you know, search for, when did I think about this thing? Like, are we able to do that? But now I'm using, I have like a little, I have a little pipeline right now. I'm using, using this company's called Pipedream, which is kind of like a Zapier marketplace for integrations, but they do workflows and, And I use my little Apple, um, voice recorder, record my notes, drop it in a little Dropbox, uh, the Dropbox into a Google drive and send it to Whisper. Whisper transcodes it. Embedding. It does some embedding for me, takes that information, sends it over to OpenAI and You know, I have a very nice prompt to kind of like massages the, try to try to get more depth into what I've talked about while also maintaining the actual transcript itself. Now, then I drop it into my notion and now I have it to. kind of pull it up. It tags it for me, which is great. That's also a huge part of like being able to make your, your personal notes searchable is having a good tagging strategy. That's all being done by AI now, which is, which is great. Like if I'm just, if I'm doing this like on a weekend, imagine what like people are doing right now trying to like solve this. So yeah. Yeah. Yeah. Ever since you brought up the second brain thing, Sekou, I've been losing sleep a little bit over it because I'm like, how would I do that? And I know that my brain is weird. You know what I mean? My brain is very weird. I was a hardcore user of Evernote. I remember leaving a job and I was like, hey, look, all the documentation's right here. I have it right here. It's all laid out. And when they receive the documentation, it's like they could maybe distill. I would say maybe thirty percent of it as usable, which is I mean, it wasn't it wasn't bad, but it was just like, you know, I would get a text like, I can't believe you're asking me these questions. It's right in the Evernote. It's right there. I have been over the weekend. I had an idea about what my second brain would be. And the idea I had was a Chrome extension that would allow me to track what I'm researching. And it would have a kind of chat or text interface so that I can, you know, send it off with notes or some kind of like topic of research along with the web page. So I could, you know, sometimes I'm going through the motions trying to get something configured correctly in AWS or something. And then months later, I'm like, I know I've done this before. How did I figure this out? If the very loose idea of a Chrome extension that would allow, you know, allow me to sit in the web page to um an lm model along with you know some other you know supplemental information um it would allow me to be like all right let me just search backwards of this problem that I'm sure I've solved before it would be huge but that that pipe dream to voice to g drive to whisper to notion uh like pipeline you just laid out man that was awesome like that is really cool I think I'm going to give that a shot as well yeah I think um I have high hopes for the second brain world that we live in and also have high concerns um the hope side of things is Second brain is not only an indication of past memory things that we work on, but now we're talking about being able to hand over, integrate these type of life changes, impacts that can help us overcome hurdles in life. So if we hand it off, we can hand it off to a therapist, an AI therapist as well, that can discover something that we're dealing with currently, that something that's in our subconscious that happened to us when we were fifteen years old. You know, I see that as things that can really help us elevate, think through, you know, move forward in life. My concerns and when it comes to AI safety. This is not the it's not definitely not my, my focus. But when it comes to things like this, I definitely have high concerns, because I also understand how it can be manipulated, you know, to as well. And so, you know, even with, you know, Mark Zuckerberg writing me a letter a couple weeks ago, admitting that there was a political campaign that was, you know, kind of strong or Facebook, if you will, to push certain agendas. I think those type of things don't go away. I think it will just start to be manipulated into things and being able to you know, being able to, you know, impact people in ways that are, you know, not fun, friendly, or even ethical. And so, you know, but also too, I think that's the component of any large technology or large chain of technology, people are going to use it for good, people are going to use it for bad. And I don't think you can change that because that is more so human behavior than necessarily technology. And so, you know, how do we stop that? I have no clue. But I do think in this world that the benefits can definitely outweigh, you know, the bad side of it, but the bad side can definitely have high impact. And I hope that we don't see that world. But again, there's individuals out there that has that human behavior and stuff within them. Yeah, absolutely right. Yeah, I think about that too. And it's also why I want the stronghold that NVIDIA has to go away. The fact that we need their giant GPUs to do these things in the cloud is what stops folks from doing that at home, disconnected. I feel like that's such a huge part of this story. The reason why we talk about the fact that we're using these services is because there's no choice. There's no choice to do anything alternative. You can't just have a server that's at a psychologist's office or something like that that is able to provide this for you. At least not yet. Yeah. Not yet, but The pursuit of the runnable smaller language model. There's the Gemini Mini. People could be sooner than we think that we'll get something like that. Yeah. I can't think. I'm going blank. What's the terminology when a company owns too much of a percentage of an industry? Monopoly? Yeah, Monopoly. Everything. In this industry, NVIDIA is definitely a monopoly. Typically, our country has sometimes fought against monopolies or made them force up into smaller companies. We saw that happen with AT&T some years ago. Do we think it's needed? I know this is a conversation that's not even thought about So happening right now, I know some of their competitors and stuff like that. What's the pros and cons of potentially, you know, if they are a true monopoly, you know, what's the pros and cons of like breaking them up into smaller companies if you need to? Well, I feel like there's like a difference between monopoly and being the best. Well, I guess I kind of go ahead and being the best innovator, the best innovator in this field. Right. there are small companies that are building chips that are specifically made to optimize using transformers. That's their only task, these chips. They're not for gaming, not for anything else. They're simply focused for transformers. And if the transformer technology ever went away, they are out of luck. So I feel like there's still competition. That is what the monopoly is when there is Not only there's no competition, but you're making your moat so wide so there would never be competition. I think NVIDIA right now is having a really good two years on this. But I think economics are going to change as soon as the interest rates change. Yep. And I think the same thing. Yep. Same thing, anti-competitive practices. If they start to engage in more of that, I would suspect as long as NVIDIA is doing good for the people that need them, like larger companies, then I think that they're set, but if there, there comes a point where they are shutting down competition that could be cheaper, more affordable, then I think, uh, they'll start to, you know, get a government will get involved in. I'm just remembering something else. I saw someone said that open AI is a wrapper around Azure and Nvidia. Right. Essentially, when you're buying NVIDIA, I mean, buying OpenAI or using their service, you are just giving money directly to these two folks. Right. And then, of course, Azure's platform is like a wrapper around OpenAI, and OpenAI is a wrapper around TSMC. Right. Yeah. You know, so... I think there's definitely a rule. As long as the TSMC is partners of all these different companies, making chips for them, and NVIDIA doesn't own the complete process all the way end to end, I think there's still, you can't call it a monopoly until something like that happens. When they own all components of it, I would be very scared. So from the concept of competition, uh money being poured into this um also too adrian you just mentioned you know even uh you know, the concept of money being expensive right now, you know, what we are starting to see that definitely hasn't slowed some people down. So, for example, Salesforce just announced that they are doubling their AI fund to invest into companies. They initially announced the first launch of the Salesforce AI fund was going to start in June, two thousand twenty three, where a two hundred fifty million dollar fund that increased very quickly to a five hundred million dollar fund. And now they just announced a one billion dollar fund. On top of that, BlackRock and Microsoft reportedly announced that they are focusing to build a thirty billion dollar A.I. focused mega fund. It said the investment powerhouse BlackRock is set to launch a massive A.I. focused fund exceeding thirty billion dollars in collaboration with Microsoft and the Abu Dhabi back investment Alpha MGX. So according to the outlet, the fund among Wall Street largest will focus on creating data centers and funding among infrastructure to support AI. Chip giant Nvidia is also reportedly contributing to the expertise around the around the already energy demands of AI technologies, which are only expected to grow. So there's a lot of money stuff coming in as well. That is a large amount of money. But also, too, I like what Adrian mentioned, money is also expensive right now. And so what you got to tell us? Yeah, OK. The BlackRock side of it seems to be more just what you mentioned earlier, Sekou, the OpenAI being a wrapper around NVIDIA. The BlackRock investment seems to be focused on the infrastructure and the kind of data center side of it. Let me see. I can find it, but that end of it. So, um, making money that way, which I think is smart. I think that's a good way to do it. And it's, it's crazy how people, you know, when you have a lot of money, you know, you could make big moves like that. You can really cut the, you know, um, right into the supply chain of a, uh, of a, uh, industry and, um, start to make money there, which, um, I think is cool. Um, I think it's a little bit interesting. So Abu Dhabi was part of the what investment fund? That wasn't the BlackRock one, was it? Yeah, it says the investment powers BlackHouse set to launch the massive AI-focused fund of thirty billion dollars in collaboration with Microsoft and the Abu Dhabi back investment outfit MGX. Interesting. Yeah, that's interesting. Yeah, I think with... I don't know. I think folks are preparing for these interest rate cuts. This is positioning right now. There is an expectation that when interest rates go down, the one... While it's going down, there's going to be a talent pool that is going to be more available, right? And the startup's going to, you know, the ability for folks to invest is going to go up. We'll keep going up. The amount of IPOs are expected to, you know, like increase over that time. So I think we're actually, I think we've been kind of building up for, a big amount of innovation. The fact that this whole AI revolution and GPTs actually became such a big thing in the middle of a high interest rate environment just means that once money is cheaper, it's going to be crazy. Like it's going to be insane. The amount of like startups and things you're going to see products you go out. Cause right now, like if you, if you don't like have a good balance sheet and things like that, like no one is going to spend any time on your startup. Right. But in an environment where you can take more risk, well, yeah, burn some cash, be innovative, figure out what you're going to do. It's a different world when you don't have that same pressure there, which most of the last, was it the last ten years? We've been in an extremely low interest rate situation. That's one of the reasons why we've had so many startups IPOing and so much money being thrown around Silicon Valley. So if we get back there or even close to that with all the things that just happened, this is going to be insane. Sekou, you've been around the industry for quite some time. Can you compare this to a time period that you think we're about to experience? So there was a time period where we saw a lot of growth within the tech industry or within certain industries across the board. Do you think we're about to experience something similar or can you compare that to a time So think about when some of these large companies came to be. When Google really came to be, it was after the dot-com crash, right? Yeah. Which also started out as a high interest rate situation. And then as folks started losing jobs and they wanted to get the economy started up, they lowered the interest rate. Innovation went up. But if folks that take chances went up, same thing for the housing crash. When the housing crash happened, we went to an extremely low interest rate environment, which is like majority of the large companies that you see will start around there. I feel like we're just going on another cycle. Even though there's no recession, Technically, that was attached to this. We're going through a very similar cycle. I think the COVID was technically our version of a recession in some ways. So what I'm hearing is interest rates go down, lessons go up. Cool. That's a good podcast title. That's a good episode title. Oh, that's hilarious. Yeah, I think it's gonna be I think it's gonna be interesting, you know, literally, because there's gonna be so many different startups. And, you know, so I also think so even if I think hardware also to going to see a skyrocket of opportunity, which typically, you know, you know, those companies are is very difficult to get funding, just because the iteration of process, you know, takes, you know, longer than software. But like you said, when, you know, when the interest rates goes down and people kind of like how a little bit more money play with it, you know, and they also knew, you know, you know, these chips and GPU stuff were needed. I think we're going to see a lot more funding and stuff round into as well. And also too, you know, I think the, the backend of this image, this, this area that we've got to go into, I definitely can see the backend of being this of what we'll start to see the first iterations of in-home robots. because operations and things will be cheaper. And we have a lot more money to store around. And then also too, we already have a knowledge management system with these LLMs that you'd be able to embed and train and kind of get it to understand things more quickly from communications and everything. And then also too, we'll still see even more product iteration of things around like computer vision, where you start to clean the bathroom, be able to, you know, do the dishes and stuff where it may be. So I think on the back end of this, we'll start to see the first, uh, kind of in, in the house robots to kind of like help us with some of these daily tasks. Oh, like, uh, Rosie from, uh, was it Jetsons? Is that it? I'm dating myself now. I know Rosie. I saw somebody put, I think it was Kevin on stage, posted things like, who was your first cartoon crush? And they put Rosie. It's like number seven or something. That's funny. Number seven. Cool, cool, cool. Well, fellas, man, here, man, we've had another phenomenal, successful show. We got to talk about OpenAI announcing OmniOne and OmniOne Mini. Slack is turning into an AI agent hub. Our friend Sank there gave us an awesome topic around how AI impact automations. I'm sorry, how AI automation impact things like social media specifically, YouTube and YouTube Shorts and And so these we're starting to see on TikTok as well. And also too, Sekou's favorite topic, which is how will AI become our second digital brain? And then Intel X, the AI chip deal with Amazon. And also too, there's a lot of money being put up now around AI-focused VC funds. Gus, man, definitely a great episode. We're going to put our handle and stuff here. Again, if you guys want to, definitely like, comment, subscribe. Definitely love to hear more from you as well. Eventually, we're going to put in where you guys can call in to ask you some of those questions as well within the show. Guys, you guys can follow me. See me on all platforms at I am Marlon Avery. You can see it at the bottom of the screen. Those are listeners. It's just I am Marlon, M-A-R-L-O-N, Avery, A-V-E-R-Y. Adrian, what about you, man? Give us some of the final words to the people. Cool. Infamous Adrian on Twitter. You can find me at www.adriangreen.xyz as well. Yeah. And you got me, Saku. You can find me on Twitch. You can find me on TikTok as SakuTheWiseOne. Not the non-wise one out there. You know, it's funny. You know what's funny? I think it's going to get to a point of like episode eleven, twenty two, seventy three where this can be kind of part comedy show. Yeah. Well, I hope AI isn't a joke by the time we get to episode seven or four. Or AI is telling our jokes. That's good. We'll have our digital twins on here. I know, right? Little Miguel up here. yo that should be a goal when they go digital we should invite her as the first AI guest yeah that's really good That's really good. I hope we play this clip back. Let's play this clip back right before we actually have Lil Miguel on. That'd be great. That's going to be the intro to the episode, and then we're going to have Lil Miguel on. And I think she's twenty-one years old forever. So am I. I'm also twenty-one years forever. All right, guys. Well, definitely, man. Definitely appreciate you guys listening, tuning in. Definitely like, comment, subscribe. Give us a listen and stuff. Give us a follow-up too as well, guys. As always, man, we appreciate y'all, and we are out. All right. Peace. All right.

People on this episode

Head Shot

Marlon Avery

Host
Head Shot

Adrian Green

Co-host
Head Shot

Sekou Doumbouya

Co-host