Episode Transcript
[00:00:00] Speaker A: Like, I lived not far from there in a temporary place when we were moving. So we had a condo.
We had, like, an early condo in Soma, one of the first condo buildings, so maybe 2004.
And then we lived for like, a while.
[00:00:14] Speaker B: Too long.
[00:00:16] Speaker A: We lived there for, like, I think, like, 10 years. We were like, ah, we'll. We'll live here for a couple years. It's kind of cool. And then I was like, I had enough of Soma. And then we had a temporary place in Pack Heights, like, not far from where you are. So there for like, maybe two or three months. Like, my wife did not like that place. And then we moved to the East Bay in Berkeley. And I loved living in the East Bay, and I love living in Berkeley. So how long you been in San Francisco for now?
[00:00:43] Speaker C: Oh, very recent. So I just came from London.
Yeah. In September, I made the move. I went briefly to New York for the Advertising Week, and then. Yeah, now in San Francisco.
[00:00:55] Speaker A: Yeah.
[00:00:56] Speaker B: It's funny.
[00:00:57] Speaker A: Like, we could have done this later, but I set the time for London time, so that's fine. We'll do it. We'll do it early.
[00:01:04] Speaker C: It's all right.
[00:01:04] Speaker B: So we're live streaming a little bit on a different. Different time today, but exactly.
[00:01:12] Speaker A: All right, get started.
[00:01:14] Speaker B: Yeah.
[00:01:15] Speaker A: You want to share the link? Yeah, I'm.
[00:01:18] Speaker B: I'm looking for it.
Let's.
There we go.
[00:01:24] Speaker A: Okay.
[00:01:27] Speaker B: Shall we get started?
[00:01:30] Speaker C: Yeah, let's do it.
All right.
[00:01:36] Speaker A: Well, welcome to the Greg and Paul Show. I'm Gregory.
[00:01:40] Speaker B: Hey, everyone, I'm Paul.
[00:01:41] Speaker A: And we always break down latest SaaS, startups, AI, whatever the Internet happens to be debating this week, and boy, were there lots of debates. I can't wait to talk about some of the cluely stuff.
We always have lots of dumb takes. We have some smart ones. We try to aim for smart ones, but unfortunately, we usually make a lot of dumb ones.
And then of course, we love memes way too much. I have a couple memes I hope we get to. To share at the end. That's our most popular short, you know, is the AI meme where the guy ate the poisonous mushroom because AI told him it was safe.
[00:02:13] Speaker B: That was the last number one.
[00:02:16] Speaker A: Number one. That's our most popular one.
[00:02:20] Speaker B: That's awesome. So we. We stream on Life on X and LinkedIn every Friday, typically noon Pacific time, but today is a little bit earlier. And then we post our videos straight to YouTube. We're also live or. Sorry, we're also up on Spotify and Apple podcast now. So finally, Lisa, take Us on the, on the go.
[00:02:43] Speaker A: We've been announcing that for like weeks. We're like, we'll do it in five weeks, 20 weeks in, like Paul did we get on Spotify. So we're finally, we're finally on Spotify. No one listens to it on Spotify. But that, that's okay. Eventually someone will listen to it over there.
[00:02:58] Speaker B: No, people, people listen. Has seen us on Apple podcast.
Yeah, yeah, totally. It's a thing.
[00:03:08] Speaker C: Cool.
[00:03:08] Speaker A: Well, so we have a guest, Andrew Tortella. He's a guy that I met on LinkedIn. So I remember how we met you. There was something maybe that I post the, like the LinkedIn point system. There's some funny posts I did on, on LinkedIn that you commented on and I replied out all the comments usually. And so I looked at your profile and I was like, dude, this guy has an LLM ad network. Like, I remember just writing, like, we have to talk. Like, I've worked at embarrassingly for ad tech companies and worked in big advertising. In fact, like with Vibersass. I'm trying to get away from doing ad tech to some degree, but I'm passionate about it and I know a lot about it and I really think that the area that you're focused on is, is one that is exciting and innovative and interesting. Something I think our audience wants to hear about. So here, let me, let me do my little official bio. I wrote you a nice bio. So we'll give you a formal, formal intro because you can tell, like, this show is like incredibly buttoned up and we're really going to challenge. Yeah, CNBC for their formal takes.
[00:04:19] Speaker C: Okay.
[00:04:20] Speaker A: Andrea Tortella is the co founder and CEO of Thread, an AI native advertising platform that enables brands to place ads directly within LLMs and AI applications. Based now in San Francisco, he's pioneering advertising in the age of AI, making it possible for companies to reach users when they're actively engaging with chatbots. Andrew previously worked at Perplexity, I think, I think I've heard of them.
Is an Aspen Institute fellow and holds a degree in economics and business from ucl. Welcome to the show.
[00:04:58] Speaker C: Thank you so much. Excited to be here. Excited to chat today and unpack everything that's happening in AI and advertising.
[00:05:05] Speaker A: Yeah, no, super exciting. So let's just jump in.
And the first question that I want to toss out there. Everyone knows that OpenAI's ad platform is eminent, or I believe it's eminent. Like, it has to happen.
So you're someone who's at the intersection of this innovation and technology in advertising and in LLMs. And I wanted to get your perspective on what OpenAI is working on and is it going to be some kind of Google AdSense moment when this launches? Like that product was an enormous success for Google in terms of, let's say, revenue, but it also was an enormous success for marketers. It opened the door for them to buy inexpensive keywords that really drove, I think, the expansion of many, many Internet success stories from very early days of its inception. So what do you think?
[00:06:10] Speaker C: I think that we'll have the same, we'll see the same thing emulated with OpenAI in just a much bigger scale, much bigger scope.
I think OpenAI will be the biggest advertising business in world history we've seen so far. And I think in the same way will benefit in a big part all the advertisers, especially the entry level, less sophisticated advertisers because of the nature of ChatGPT with the long tail prompts and conversations, right, you're not constrained to only a few keywords, but actually where you have very long prompts and you're able to target very long tail conversations, very random stuff. And so that enables the flourishing of small businesses that are very, very niche, that are able to capture a lot of value and to reach for the customers, people that they want to serve. And so absolutely, I think as much as they're, there are some challenges for OpenAI to start and you know, an ad business, they will do it. They, they have the means and the talent to do it and this will happen. And I think it's, it's happening very shortly.
You know, nobody knows the, the exact time frame. I think that also fluctuates. They might not even know themselves just because of, of market dynamics, but it will happen very soon, right? And marketers need to be ready. And so we operate in a, in an adjacent space where we're doing the same thing, but for the independent AI ecosystem, keeping it vibrant. And for sure there's ChatGPT, but there's other players outside of them. And so we serve players that don't have hundreds of millions of users, but that individually have millions of users. And so aggregating those guys and then helping them monetize while they build a great product. But I'm calling here from Silicon Valley where the Valley wants to save the world with their AI. Now they can give it away for free or they can make it cheaper and subsidize their experience. And so I think in that extent that will benefit us, right? Actually the day that OpenAI launches ads. I think that will be the best thing for us where every marketer in the world, every agency in the world will have a mandate to advertise within LLMs. And, and so just like today, you have the War Gardens and the Open web and I think that will be the same thing for us.
[00:08:37] Speaker A: Yeah, I agree. I have so many thoughts, but I'm going to refrain from speaking too much, at least here in this part. I want to hear hear from you. I think our audience want to hear from you too. So, so I guess the next I think obvious question that most people would have, anyone I had had when we first talked is like, tell us a little bit more about like how it works, user experience, like how do you put the ads into a chat experience like some people are for? I actually look pretty closely at their perplexity experience. So there's some understanding out there of like how it works, but I think most people aren't familiar with it at all. So yeah, explain us like how it works.
[00:09:14] Speaker C: Absolutely. I mean we're still so early, right? Then how it works is, is not defined and set in stone. This is just how it works now and how things are, are being experimented at this stage. And so that's why I think OpenAI is, is not in a position where it's time sensitive for them to launch ads. You know, this is going to be a business at scale, this is going to be the most scalable business. And so they want to make sure that when they do it, they're, they're certain of what they're doing. And so that's why us as a small startup were able to put in the reps, we're able to be battle tested and experiment and see what's working. And so advertising in chat is all about making the experience as native as possible, as endemic as possible and so exploring different formats within the chat to make it as native as possible. And so what we've done so far and what seems to continue is having sponsor messages where the native LLM experience is chat to chat. And so the ad is going to be a chat, right? It's text to text. Google is text as well. And so that's the ad formats that we have now where once in a while I'm having a conversation with LLM and it's just going to inject a sponsor message. Right. What is also interesting is to look at how opinionated some people are on X or online about how ads are polluting the experience and ads are horrible and they're destroying the world but actually, definitely advertising has some challenges and that's why as a startup, this is our duty to actually improve it. And so to build an experience that is not only complementary but actually additive, where the ads are so good that they bring value to you. Where actually I prefer the experience with the ads without the ads. And so this is something that already exists today. For example, even Sam Altamin himself mentioned that Instagram ads are amazing. And so that's really what we, what we aspire to do. And people hate ads because they're, they're not great ads. Right. And we think that we can do a good job at doing great ads. Essentially what people are doing when they're having LLM conversation, that they have a problem.
Right. And so we just show them a solution. Right. And there's lots of things to, to figure out, but I think intrinsically we are positioned really well where, you know, from a privacy standpoint, it's extreme privacy friendly. Right. You don't need to have behavioral targeting and know everything about the person and older family, but now you just have the context from the conversation and you don't really care who the person is. Right.
And, and also from the targeting perspective. Right. That, that helps a lot as well.
So yeah, it's the early days, we're still trying to figure it out.
The trust component as well is very important, especially for ChatGPT, but for us as well. Right. Where people have very high trust in LLMs and so you don't want to break that trust. And so it seems that people don't necessarily make the connection that the ad can be separate from the LLM output, which is the case for us today. Right. The sponsor message is just a second separate follow up message to the actual conversation.
And so it's clearly labeled as an ad and it's not meant to deceive people. Right. We're not trying to force feed people with ads, but actually just trying to show them something useful that they might like.
[00:12:41] Speaker A: Okay, so, so let me just make sure I understand.
Like you're in. You type a question into chatbot and then is that answer says like sponsored answer or is it like gives you an answer and then it has like a sponsored one included, like second one. So it's like, so there's a sponsored like section or portion and that's how you transition between the context. Right. Because like the context might not match perfectly. Right. Would depend on how many advertisers you have. Yes, a little, I know a little bit about advertising, so, so that's interesting. So I Get a response. And then it's like, there's sponsored. There's a sponsored section. And then can you like, chat with that section? How does that work?
[00:13:21] Speaker C: Yeah, you can chat with. With the whole thing, right? So it doesn't. If you chat.
Depends on what you chat with, right? So you chat. The chat is. The experience is the same, right? The idea is that nothing changes, right? It doesn't even feel so much like an ad. Right? The best ads don't look like ads, they don't feel like ads. And so that's the same thing for us. Whereas you just keep chatting, right, about the ad, not about the ad. You know, whatever you want people chat about.
[00:13:50] Speaker A: So the first thing I think of when I hear this is like, how do you enable the advertiser on the creative side so that it gives good answers?
[00:14:02] Speaker C: We automate all of that with AI. We. We think that, you know, there's a very big arbitrage of information that can be done in real time, right? We have the conversation, we have the brand knowledge. And so what we do is that every single creative is done in real time based on the conversation. That is what's going to deliver the alpha, that's going to deliver the performance, that's going to deliver the outcomes. And so we do let advertisers come up with their own creatives, but actually the AI will come up with the best creative. Just because it's so contextual. Just because I have a problem, I have a question, I have a prompt. The ad will reply to that. Exactly right now. And so if you have your own selected list, then it's not going to be as contextual, it's not going to feel as personal, it's not going to hit people as hard. But actually it's not something off putting that I've seen from brands, right? You would expect maybe with brand safety. I'm scared that I'm going to leave AI, come up with the creatives. But the creative generation is very strictly done based on the campaign parameters and the input that the brand gives you, right? And so if they tell me, hey, here's who I am, here's what I want to sell, I want to promote this in this way, then obviously we give you a preview of what the ads are going to look like, so then you can adjust and then you can launch your campaign, and actually, brands are quite open to testing that. And I think there's been very few brands, at least with us, that say, oh, I'm so scared that AI is going to generate some text that's so random because that's not the case.
And so it's not really an issue technically. Therefore practically for the brands, they're fine because it's just text. Right. Whether if we would be doing images, I think that would be a different story.
[00:15:54] Speaker A: Fascinating. So is there an opportunity for like prompt native advertising agencies that would emerge and be people who would, you know, enable and engage?
[00:16:02] Speaker C: I would hope so.
[00:16:03] Speaker A: Do the creative development for this.
[00:16:05] Speaker C: I would hope so. I would want to start one myself. I think I'm just too constrained. But I would want to invest in companies and agencies. So if anyone out there is starting one, hit me up because yes, I would hope so and I would want to.
[00:16:20] Speaker A: Yeah, it's like the, that's where my head goes, right. It's like, oh, you have to create this experience and it's a whole new way of thinking about it and there need to be guardrails and parameters and I don't know, like first thing I thought was like coming up with like a funny or interesting like Persona. Right. Like the brand. In my opinion, if I chatted with it, I'd wanted to take on what I imagined that brand would be like. Right. So brands have like a rich history of like developing themselves and have like a really specific tone, style. They obviously lend themselves this like a luxury brand. I can imagine how I would want that to chat with me versus like, you know, if I'm having a conversation with, with a more challenger brand, like a Nike or Liquid death. Right. I can imagine them having like some really off the wall type of like approach to how they want their chatbot to respond.
[00:17:15] Speaker C: Absolutely. And I think that's so fun and I think that's why this is so exciting for brands to be able to explore this new channel. I think it's a new world, a new rather that is opening up. And I think this channel is so unique and has so many intrinsic values that are, that are valuable for advertiser. Right. I think this is the first time ever where brands can talk to their customers one on one at scale. Right. When I see a billboard in the street for Nike shoes, I can't talk to the billboard. If I see an ad for those same Nike shoes on Google or on Instagram, I can't talk to the ad. Right. I can't ask how light are the shoes or other shoes waterproof or do you have the same shoes in red or complaining even about these shoes.
So you can't even say that you like them. And so in the LLM you can, right? Where you can ask, how light are the shoes, waterproof, who's the founder of the company, how much they cost, when are, are the next shoes dropping? And so what is so incredible is that now you have attribution from this information, right? This is happening in your brain. But now on the lm, you can transcribe what's happening and you can transcribe your thoughts. And this is all information that goes to the brand and is insanely valuable. Like you're literally seeing what your customers are thinking about what their problems are and how you can better serve them. Right now, you know, oh, people are complaining about my shoes, actually I should change the color. I should. They, they think they're not comfortable. Right. And this is really mind blowing. Just from a measurement perspective, but also from a targeting capabilities perspective, right? There's a lot of things that can only be extracted from the semantics. For sure. Maybe Google and Meta, they know everything about you, right? But the targeting capabilities, there aren't ways to target language per se, or there aren't ways to target stories per se. Right. Or moments of people's lives that are shared or even more subtle things like personality, for example. Right. You might find proxies to, to do that across the walled gardens and the rest of advertising. But you can say, hey, I want to target people with a big ego. Right? But from the conversation, right? And so this might be valuable for.
[00:19:37] Speaker A: No, no comment there.
[00:19:39] Speaker C: Exactly. But like this is a, this is like a real thing that we had come up and brands, you know, suggest that where, if I'm a financial institution, then I know that, you know, people with the big ego, they're going to invest more and that's going to be more valuable.
Right? And so that's kind of mind blowing.
[00:19:58] Speaker A: Awesome. All right, so let's, let's switch to some of the mechanics. I mean, I think our audience, mostly professionals, marketing professionals, professional audience in the sense that like, I think they have a lot of questions on the mechanics of this. So people that are like marketers who are familiar with typical advertising measurements, how is this stuff measured? How is it bought? Right? Like in, in advertising you have, you know, cpm, right? You have the cost per thousand, which is how most advertising online is bought or through like a click. So I'm curious, like, what are your thoughts on or how do you guys like measure it and how do you sell it based on existing advertising metrics? So those carry over or so we.
[00:20:41] Speaker C: Build the end to end ad tech infrastructure where we connect directly with the AI, native publishers, the LLMs they take our API, we read all these messages and once in a while we inject a sponsor message for the advertiser. And then for the advertiser, we have our own dedicated media buying platform where they can go, they can launch a campaign with no minimum ad spend in seconds. With a single prompt, they can say, launch a campaign for the Gregory and Paul show for $10 launch, and the campaign will be live. We actually applied for the Guinness World Record for the fastest campaign in the world. Fastest marketing campaign. From idea to real time bidding, right? Where you go from an idea to actually the ad being served in real time, Real time bidding.
Just because it's so fast, right? We don't make the creative, that is what takes the most resources and time. And so now, literally in seconds, you're going to be live, your ads are going to be live. And so this is something interesting that I think I can bring up to your audiences where I think time is something that is very undervalued within a marketer's marketing mix, right? You think about the cost, you think about the audience and the reach and the scale. But actually, if you think about what's the ideal, what does the ideal marketer look like, what does the AI native marketing marketer look like? And so just like in startups, speed is the moat in marketers, right? You want to market at the speed of thought, you want to market at the speed of news, you want to market at the speed of culture. And so if you have a channel that is intrinsically very long and time consuming and you know, is going to take a lot of resources, then that's suboptimal, at least for testing, for experimentation. And so I would just encourage all of your audiences and all marketers actually to, to make this mental experiment. How long does it take me to go from an idea right now? I just came up with the idea, I want to be live now, right? How long is it going to take me? And so that's going to take me minutes, hours, days, weeks, right? And so we allow you to, to do that in seconds, right? And so that's why our channel is intrinsically powerful for speed, for experimentation.
How you do the media buying is so essentially through our platform, right? You can launch a campaign either with a prompt that's like the vibe marketing way, or. Or for more sophisticated marketers, we also have more granular targeting where they can target a multitude of factors and targeting capabilities.
How are you thinking about measurement, attribution, even monetization is that I think with this new channel comes new measurement and new attribution.
But at the same time, we're also a business that needs to operate now and you can't make something that is too new for people, that is too foreign. And so we still use the same standard metrics. Right now we're in a CPC instead of a CPM just because we're intrinsically very performant channel and so we have a lot of clicks, we're confident in the clicks.
It also align incentives as well with all the stakeholders, right? We're now not incentivized to spam people with as many ads as possible. On a cpm, right? On a cpm, the more ads I show, the more money I make, right?
Whereas this is also suboptimal for all the stakeholders. For the users, they're going to see ads that they don't actually want. Whereas now on a cpc, why would I waste an impression to someone that I know is not going to click on my ads, right? And then for the advertiser, it de risks it for them because they're thinking, oh, this is a new channel, I don't know if I want to spend. Whereas now you only pay for the performance, right? This is my unity economics. I know I'm willing to pay this for click. If you're doing a good job, then I'll pay you. If you don't do a good job, then I literally pay you zero.
This is what's happening today and then all the standard metrics. But I think again, within these LLMs you have this raw material, the raw material of the conversation that you can then process. And so this is what we're thinking about and looking at, processing all this raw material where we have all these conversations, these semantics and all this contextual data metadata to extract the value from.
And so I was actually speaking to Brian o' Kelly who is advertising industry pioneer, and we're thinking about new monetization where actually the click might not be the best way to monetize the AI today. This is what we use. And people, people are still clicking, but if you just look at the total amount of clicks that are happening in the world and in the advertising industry, if you have an industry that based on clicks and people are clicking less, right, Just because they're not clicking blue links anymore, they're chatting, you're just gonna, it's just a race to the bottom, right? The clicks are disappearing, there's going to be fewer and fewer clicks. And so this already something that has happened dramatically and is going to get even worse. And so Even for our model, right, we cannot sustain the click. And so this, I think is needs to be something that, you know, is transcends our company, that needs to be adopted by the whole industry where, okay, what is the next monetization model? What is the next metric? What is the next unit of measure? And so what BOK proposed was cost per turn, right? Which was. It needs to be something that is quantifiable. And so cost per turn is how many messages do you have after the ad, right, because you can reply to the ad. And so actually what that means is that what this assumes is that actually it might be more valuable for the advertiser to have the user stay in the chat, be engaged, rather than be redirected to a website where they're going to be lost. Right?
And so this is one way, I think it has limitations, but at the same time I think it's, it's really smart. And so I think now we're in a transition period instantly transition period. You either need to stick to one side or slowly adopt a hybrid version. And so we're so early that we're just going to keep it simple, start with the cost per click. But I don't think that will last.
And so we're very conscious about that. And so now proactively thinking about what's something that advertisers are happy with, that we're happy with, that makes sense for everybody, that's easy to understand as well.
I don't know if you all have any thoughts on that, but how would you make something that is easy to understand? Cost per turn is simple, right? You have a reply. How many replies? Okay, they reply 1, 2, 3, 4 times. Okay, you'll pay based on this.
Another metrics that we have internally that we're super excited with but would be harder to standardize across the industry is something we call internally attention shift.
Attention is a super buzzword in the advertising, but specifically attention shift, right?
And what that means is that, you know, we come from a very naive background, right? I just finished university and so we're jumping in this new space where we don't know anything, we don't know anyone, you know, we're so we need to, you know, think first principles about everything and just even thinking about what is advertising in the first place, what is our role?
You know, if we want to survive here, what are we even doing? What can we try to do? So advertising comes from the Latin adver tire, which means redirect the attention, right? And so that is what advertising does they redirect the attention, in this case redirecting the attention from the publisher, who owns the attention of the user to the advertiser. That's our job, to redirect the attention from the publisher to the advertiser. And so we need to facilitate this transaction as much as possible, make it as smooth as possible and redirect it as much as possible. And so then you can think about, okay, well that's our job, we need to do that as much as possible. How do we come up with a metric to measure that? And so if you apply this to an LLM conversation, you're going to have the conversation before an ad and you're going to have the conversation after an ad.
Then you can look at the user attention before an ad and then after an ad you turn this into vectors. And so you have topic of the conversation before an ad, topic of the conversation after the ad, and then topic of the ad. If I'm talking about fitness and then I have an ad about protein after the ad, do I talk a bit more about protein or do I talk more about fitness? And so here you can look at the attention shift and to us this is mind blowing and we think that this is one of the most truthful metrics. I think all these metrics, the cpc, cpm, all these things, they're a bit BS metrics, They're just proxies for reality, they're just proxies for this attention shift, they're just proxy, it's for just customer acquisition. What does advertising do? In big part it delivers customer acquisition. It depends where in the funnel you are, but just generally just acquiring a customer. And so the ultimate metric that you want to have is just percentage is just take rate on the customer acquisition, take rate on the roas. And so I think we're just going upstream now where we can get closer to reality and find more ground truth in those metrics.
[00:30:11] Speaker B: That's very cool.
[00:30:13] Speaker A: Yeah, a lot to unpack there.
So I think. Go ahead, you'll go, Paul, I've talked a lot.
[00:30:18] Speaker B: That's awesome. So do you have like, are you guys thinking ways of just measuring the difference of the conversation?
[00:30:26] Speaker C: Maybe like that's what we're already doing. Okay, okay, got it right now. So all our advertisers, they see the attention shift and so we have it both as a, as a graph and we have it as a percentage. And so you see basically, okay, here's the topic of the conversation. Then the ad happens here and then you have a spike of attention and Then you want to see, okay, how, how much you can keep, how much.
[00:30:48] Speaker B: You can keep it and like how maybe like it's how much of a difference in the conversation between the before and.
[00:30:55] Speaker C: Exactly, exactly. So you want the shift.
[00:30:57] Speaker B: You want to, right.
[00:30:58] Speaker C: You want to spike it as much as possible. Okay, here's, I'm talking about fitness. Boom. I showed the ad about protein. Now how much do I talk about protein, right. And do I stay high or then do I fall back and talk about fitness again? And so that is like, that is mind blowing, guys. Like there isn't.
[00:31:14] Speaker B: It's very cool. You could be, it's amazing. You could even like get them to shift into a complete separate topic. Like let's say just randomly travel, right. For whatever the reason, you talk about fitness and then you interject the ad about traveling and they shift the conversation. Traveling.
That will be completely mind blowing.
[00:31:33] Speaker C: Yeah.
[00:31:36] Speaker A: Wow, it's, there's so much.
Well, the first thing I was thinking about when you guys were talking was like what you're just, what you're describing is like how magazines used to work and what, like what like the old school advertisers like Ogly believed and then they, you'd like read them. You'd read the entire magazine and you would generally like read the ads and the ads would be in between the articles.
[00:31:59] Speaker C: Right?
[00:32:00] Speaker A: That's what we described. So like now you're doing it in an environment where it's dynamic. You can measure it. There's a whole bunch of like um, it's instant almost. Right. And, and, and I, I, I like to. The Internet tends to be nonlinear. I think magazines, even books, I kind of like not that excited about books anymore. I, I'm not excited about like a linear experience. Like I have to read it in order and it's really long and I, I, I'm busy. I have a lot of things to do and I don't have the kind of time like I used to as a kid read like four books a week or something. So I love this whole like non linear methodology that let's say chatbots ultimately lend themselves to.
And then inserting the ads into it makes total sense. Right. And you can shift it. I also like the idea of like an engagement measurement rather than a click. I think clicks have, they're so good incentives too. There's also some bad incentives with clicks and I can see that like a metric maybe, maybe there's options to choose both.
[00:32:57] Speaker C: Right.
[00:32:57] Speaker A: I want to measure on it per click or on a measure on an engagement.
[00:32:59] Speaker C: You want to make the most clickbaity ads. Right? And so people just click actually, but then they leave. Right. And so that, that's what you could, you, you could do. Right, Right.
[00:33:07] Speaker A: It has, it's, it's, it's, you know, like you've seen how like platforms like LinkedIn stuff have reduced the ability for people to share links because they want people to stay on the platform. Right. So there's, there's, there's, there's strengths and weaknesses to both. But I like the engagement one because like I'd see how that becomes popular brand advertisers.
So maybe I'm not selling a product that's easily bought online. Like, you know, traditionally like beer or something like that, like alcohol is a good category where you don't, I mean now they have more online sales, but it's not really how they sell it.
[00:33:41] Speaker C: Right?
[00:33:41] Speaker A: Yeah, they want to have an engaging experience. Or cars is another really, probably a better example. You really just don't buy a car online. You do research. They need to advertise it. So I'm like a conversation with like, you know, Toyota or Subaru or Bentley. It's kind of interesting. Very cool.
[00:33:58] Speaker C: Yeah.
I want to double tap on the linearity because I think there's a, you can have the other take where actually there is a lot of linearity or at least linearity of attention within the LLM just because you are, you have a lot less content switching, right? If you think about social where you're just doom scrolling and you have a constant back and forth between content which is not the same content, right. Or sure, you have the algorithm, but you're, you're intellectually not stimulated, right? You're just doom scrolling. And this is why, this is the. Our channel is so amazing, is that people are very intellectually stimulated, right? They're having like deep conversation where we have some conversation of hundred plus turns, 100 plus metrics. So 50 conversations from plus 50 replies. And so that's deep guys. That's like deep intellectually stimulated conversations. And so if you think about from the advertiser perspective and the pure attention plus the attention shift, that is amazing. Right? The user is a lot more intellectually stimulated and so the quality of the user that you can acquire, the quality of the user that you can talk to is a lot higher because they're a lot more engaged, they, they're a lot more knowledgeable, they get educated. Right. And so this is also what advertising is. It is, you know, share ideas that matter to people who care and educating your users and telling them, well, actually in this case the protein, you actually need it and you understand how your body processes it. And so now it's not like I'm just scrolling and see, oh, damn protein. I heard about this. Click buy. Now I understand why do I need it, right? The cart size is going to be bigger, I'm going to buy it again. And, and so generally the better customer, better user, higher lifetime value because they understand, they're more educated. And so that's the whole point of the LLM.
[00:35:54] Speaker B: Okay, so I have a follow up question to all of this. Tell me if you can't say so, is companies like OpenAI looking at all of this and then training this knowledge back into the LLM?
Okay, what I mean is right now LLMs are very one directional, right? It's very, you ask a question, it answers the question. Very rarely would it do the adjacent kind of conversations. For example, like if it's talking about fitness, very rarely would it be like, hey, you might need to travel here, for example.
But in the future, if you're having a real conversation with like a fitness trainer, maybe that is a recommendation, right? They'll be like, hey, what you actually need is to visit this local gym or go here, there's a conference that you might be interested attending next.
[00:36:54] Speaker C: So proactiveness, essentially.
[00:36:56] Speaker B: A proactive, exactly. Or here's a, here's a piece of article or a conference video that, that might be interested to you from a year ago.
So like what you're doing is kind of doing this, right? Like you're advertising and shifting the conversation to something relevant.
Do you think that this will eventually get trained into the model itself?
[00:37:20] Speaker C: Yeah, I think, you know, it's interesting what we do in terms of the attention shift and that you want to think about how you want to minimize the attention shift at the same time, right? You want to show an ad product or service that is as close to the topic and so you can minimize the absolute attention shift. But you want to have the highest relative attention shift. Right?
[00:37:40] Speaker B: Exactly.
[00:37:41] Speaker C: Close.
But in absolute terms. But in relative terms, you want to have the biggest shift in terms of what OpenAI does. They're training on everything, right? Like every conversation that you put in OpenAI, I think if you don't have like the dark mode or incognito mode, they train on everything. So they're going to train on everything.
But that is what Meta does today, where they know what you want even before you know that you even needed this. Right? And you didn't even know that this existed.
So yeah, I see no reason why they wouldn't do the same to an even bigger extent. And I think OpenAI is going to have a much easier life than metadata because I think it's at least before it was quite challenging to extract a lot of high intent between images of cats. Right. But now when you have somebody literally telling you what they want and writing paragraph long messages, literally typing their whole lives, now you can have much better understanding and world model of the person of their intent rather than just extracting that from. From a few comments or, or a few Instagram posts.
Very cool.
[00:38:57] Speaker A: Well, should we maybe do some news articles? You want to join us with some of our weekly tech wrap up?
[00:39:05] Speaker C: Yeah. Okay.
[00:39:07] Speaker A: Paul, should we do this this Data Centers in Outer space one.
[00:39:17] Speaker C: Start.
Oh.
[00:39:23] Speaker A: Paul, are you there?
[00:39:24] Speaker B: No, I'm here.
[00:39:25] Speaker C: Okay.
[00:39:26] Speaker A: Wasn't sure we had a. We had a technical difficulty last time so. Okay, so this one made the news. Oh whoops. Made the news rounds. But Google at least they claim they're going to be launching data centers in outer space, which sounds wild.
And according to this the ability to operate in the, I don't know, outer space, it's like the stratosphere or something, right? It's not technically outer space but Anyway it gives them 24 access to solar power and they claim somehow I find this drive terribly but like somehow by doing it this way it's affordable and so they're going to put these data centers into, into outer space, which is awesome. Like I would love to see like a real commercial space startup or, or I guess not sort of initiative happen.
So they say this will launch by 2027. So what do you guys think about this?
[00:40:34] Speaker C: Well, there's a startup called StarCloud that is already doing that and I believe that just last week they did their first mission on, on SpaceX. So I think it's still early and they're, they're definitely an ascent player, a challenger. But yeah, we should.
[00:40:50] Speaker A: All they're doing, they have, they have. The data centers are running already in space.
[00:40:55] Speaker C: No, they're not running but they sent the first.
[00:40:57] Speaker A: Oh they're gonna like be a competitor or similar type of service.
[00:41:01] Speaker C: Yeah. So. So they, they did their first mission where they sent their first satellites to space. Right. So I'm not sure if you know the GPUs are going to be melting, you know, this week, but they sent their first capsule already.
[00:41:15] Speaker A: Yeah, there's some obvious advantages. Right. But like is it is the ability to like do the data transfer. I don't understand actually that piece like it's like that's going to work. Like how does that. Do you have any insight into that, Paul? Do you understand that piece? Like they can get the data back and forth in a timely manner.
[00:41:33] Speaker B: This is data.
I think they're going to use this for training, right? Mostly. So like small computations, not really life.
[00:41:42] Speaker A: I mean where they just upload the Dropbox.
How do they get works?
[00:41:50] Speaker C: I think that there are technologies where you can, you can get the, the input. Right.
[00:41:59] Speaker A: I, I did actually meet a guy in Seattle who works on some initiative that I don't believe is top secret.
[00:42:04] Speaker C: But.
[00:42:05] Speaker A: And like I'm not, I'm not a. I'm not like a hardware space engineer. So I know they have like large base stations, so there are systems they put on the ground that communicate with satellites and do the, do the data transfer. But still seems like it'd be a lot of data. So I don't know.
[00:42:19] Speaker B: I guess, yeah, maybe you basically just have to. What they're doing is they're going to send a bunch of data up and then the output is very small and it's not live. Right. So it's tpus tensor processing units.
So what's interesting here is that Google is predicting that the, the most it's going to be energy. That's the most rare resource going forward because Microsoft did this whole entire project under the sea where they thought that, you know, using the cooling with the water is more, is more important.
But Google actually thought that using solar power in space is more of a important initiative.
Lower cost, lower cost in terms of energy.
So that's interesting.
[00:43:13] Speaker A: Did you say something?
[00:43:14] Speaker C: Sorry? Yeah, I was saying n. It's pretty cold out there, right? I think it's colder in space under the ground. Right?
[00:43:21] Speaker A: Yeah. There's some obvious advantages. Right. Well, interesting. Like I mean Google has always been doing these crazy moonshots and what's, what's interesting in retrospect, like they were just kind of early. Like a lot of those things work. They were like, they made that mass investment drones and now drones are, I mean their drones are dominating the Ukraine war. Right. So drones are a reality. They were right about Waymo. It took a long time.
They put a ton of money like I don't know how much they spent on it, but like they've had, they've had 25 years of research, but it works.
So this thing like I, I don't know if this will generation of like people or initiatives that are doing this are the ones that are going to work. But I could see, like long term that this makes sense. Right. I think you identified the two issues. So, like access to power, which is better in space. Right. They were very clear in the thing. Like 24 hour access.
There's no cloudy days, all that stuff. And then cooling. Right. Like it's freezing up there, so you don't have to pay for cooling costs.
[00:44:23] Speaker B: Yeah, yeah.
[00:44:24] Speaker C: All right.
[00:44:24] Speaker A: Should we move on to Tim Draper?
I like this one. Tim Draper agrees with me.
Open AI. He called them the AOL of AI. I think I called them the Yahoo. I still prefer my analogy. I still think Yahoo is a better, A better analogy for what, for what they're doing. But I thought, this is interesting that he kind of called it in a similar way that, that I did. Right. That OpenAI. For me, the parallels that they offer, everything. They're trying to be everything to everyone, which both AOL and Yahoo did.
Because the Internet was small, there wasn't a lot available.
And they're very successful. You know, like Yahoo and AOL both were like massive businesses and they did, they did extremely well. So it was interesting to see him, him post this.
And he, he, yeah, he, he put some. He was made the. He made that massive bitcoin call. Right. And he got, he got roasted for that one point. I think he had a $250,000 per coin price tag, but it's over 100. So.
Any thoughts on Tim Drapers?
[00:45:32] Speaker B: I totally agree.
I. I can even prove that I agree with the June 29th tweet.
All right.
[00:45:40] Speaker A: Get that up there.
[00:45:42] Speaker B: Over here. Let me, let me. Sure.
You know, June 29th, this is exactly what I tweeted.
[00:45:50] Speaker A: Oh, you did?
[00:45:52] Speaker B: Prediction right there.
[00:45:54] Speaker C: That's incredible.
[00:45:56] Speaker B: So that's pretty good. I told you.
[00:45:59] Speaker A: Pretty good.
[00:46:01] Speaker C: Pay.
[00:46:01] Speaker A: Pay your royalties.
[00:46:04] Speaker B: I, I think so. Like with Open AI, I think they're early to the game. We really don't know what AI looks like in 18 five years, let's say.
[00:46:14] Speaker A: Yeah, it's like first generation. It's a, it's an easy call to make.
[00:46:18] Speaker C: Right.
[00:46:19] Speaker A: It's not like.
[00:46:20] Speaker B: Exactly.
I think they did really well in terms of distributing AI to the masses.
[00:46:27] Speaker A: All right, I got another OpenAI one here. This was cool. So I've written a lot about this. There's been a lot of information going around, but this was pretty cool about the 1 million customers working with OpenAI. Yeah, that's a big number.
[00:46:46] Speaker B: They're very good at. Exactly. They're very good at getting in front of people.
[00:46:51] Speaker A: So I, I doesn't it doesn't go into the details here. Like, like how exactly how they calculate this. Like, it can't all be API access, but it's, it's, it's a massive number. It's very impressive.
It's much larger than early Yahoo or AOL numbers.
You guys have any thoughts on this?
All right, all right, this is going. We got, we got some. Good. So we get. The thing we really want to talk about is all of this cluely drama that's been going down on, on Twitter this week.
[00:47:26] Speaker B: These guys are very good at staying in the spa life for some reason.
[00:47:30] Speaker A: Oh, totally.
[00:47:32] Speaker B: I almost forgot about them.
[00:47:36] Speaker A: I think the. Well, so. Okay, so let me see if I can summarize it.
So a bunch of stuff has happened in the past, like, week or so. So Clulee relaunched and rebranded as a note taker. So they were the, the cheating app that allowed you to, like, cheat everything. And that was the positioning. And they did well with that. They had all of those fantastic viral videos and the rage dating and all of that stuff. Built a massive brand very quickly.
And then, you know, they, they made some, some public statements about trying to hit a million dollars in ARR.
And so with the pivot, and then let's say, going dark on some of those communications, the haters came out in full force critical of, of Cluley, asking him, like, why they haven't been saying much about revenue growth, why they've been pivoting. And so there's been a whole bunch of.
Of drama on Twitter. Gary Tan got involved. There's a whole bunch of things.
[00:48:43] Speaker C: So.
[00:48:43] Speaker A: So this was the TechCrunch article, but let me see if I can find. Which is wild because, like, people being really critical of, of TechCrunch and then the writer even. So this is Roy's explanation about the pivot, which I thought he did a good job, you know, following up to like the, the feature that he got here on, on the Technology Brothers podcast.
[00:49:07] Speaker C: But.
[00:49:10] Speaker A: What do you guys make of all this?
[00:49:13] Speaker C: I don't know that it's such a big pivot, to be honest.
Right.
I mean, the cheating eating was pretty much taking notes before as well. Right. So I think they were a feature to a note taker. Now they're more of a note taker. I don't think that it's product wise, such a big pivot. That's why they were able to, it seems to me, to do the pivot so fast. Right. It's not a hard pivot, so it might be somewhat of A micro pivot. But I think it could be to some extent, who knows, Just another marketing stunt where they say they're pivoting, but actually they're not pivoting. I don't know that from a branding perspective, that deprecates the brand even more because it makes you look on the back foot. But I guess just do whatever it takes.
It doesn't seem such a major, substantial pivot, at least if you just look at pack pleats from the product and feature standpoint.
[00:50:14] Speaker B: Completely agree. And also they're going into even more crowded market.
Right. Like, I can't really see what their differentiator are from all of these other note takers.
[00:50:26] Speaker A: All right, so here's what I wrote over the weekend about the Cluley pivot. And I want to start off saying, like, I think they're amazing and I think what Roy's done is fantastic. Like, I'm totally aligned with like their ability to generate top of funnel. And I totally agree with them. But what I think the issue was is that the positioning is really challenging. When you go from an app that's kind of like a pirate app, it's like cheat on everything. They took this very kind of challenger brand, or they were trying to be very irreverent and edgy, which I'm supportive of. That's great. I think in an early stage company you should do those types of things. But then they pivoted into like a note taker. And I think the problem is that in that space, trust is really important, Privacy is really important. And so they have a brand that doesn't necessarily establish them as like people that you trust. Like, I can't see a healthcare company wanting to use this as a note taker because of the, the brand.
[00:51:28] Speaker C: Right?
[00:51:29] Speaker A: And so that's what I think. That's what I think is the challenge here. And I still, like, we were joking earlier. Like, I still think there's a 15 to 20% chance it can work. They just need to like, change the positioning. But I like this in the sense that it illustrates how important your positioning is. Right. In some ways you could say, hey, product, we could just kind of rejigger here. We have a new product or it's servicing a new market. But the positioning of how people think about you in terms of a brand is much more difficult to do. And there's some people who may like, never get over it. They're just like, why you guys post all those rage bait and did all these like goofy viral videos. Like, I just don't trust you with.
[00:52:06] Speaker B: My trust is a thing. Market differentiation is a thing.
Isn't there one that, like VCs use granola.
Right.
It's a media note taker.
[00:52:22] Speaker A: Yeah. I think that was all the comments.
[00:52:23] Speaker B: For like, VCS go to granola.
[00:52:26] Speaker A: Yeah. So. So what I. Like I said, I think it's 85. It's not going to work. Like you're saying, Paul, you even thought that they might just get acquired.
[00:52:34] Speaker C: Do you think that they do it? They did a disservice themselves by pivoting and they should have stayed because they. The best competition is not competing with anyone. Right. And so if you are the only. It seems that they had no competition. Right. Truly define its own category. Whereas now they go in another category where exactly.
[00:52:54] Speaker B: But they raised a ton of money.
[00:52:56] Speaker C: Right.
[00:52:58] Speaker A: I. I think you're really. So I think you're totally onto something. I think like, that would maybe be.
My marketing recommendation is like, yeah, but. Or I think they should have considered that more. So, like, we. What is our brand and what are we known for today?
And then is there a way we can like, use that and either increase the scope of the product, like reach more people and keep the brand. I think pivoting the brand, you know, I don't want to go as far as say it's a mistake, but it's. It's very, very difficult. I think more difficult than they anticipated. And then they went to a market that's very crowded.
[00:53:32] Speaker C: Yeah.
[00:53:32] Speaker A: There's a lot of competitors.
[00:53:34] Speaker B: Competitors. It causes brand confusion, people.
[00:53:37] Speaker C: And it's not like they. I mean, you say they raise a ton of money. It's not like it is. They raised their $15 million series or so, but I think Granola has raised more. And then all the other companies that have been in this space, I think they have raised more. And so they haven't. In relative terms, they haven't raised much after they raised too little.
[00:53:56] Speaker A: Correct.
[00:53:57] Speaker B: Yeah.
[00:53:58] Speaker A: So I think if they could have stayed some kind of like, pirate app or whatever, whatever that was, I think there was ways to do more with that and just continue to sell to the same people. I'm gonna get really boring. Like, there's a Boston Consulting, like a BCG Squares thing that talks about this, like, like new markets, new products, and like, you can talk about which sector they go in. And so if you have a new product and you're going after a new market, that's the hardest one. The easiest expansion is generally, like, you have an audience, you have a group of people, you sell to and just sell them new products. That's always what. What I always recommend that's the easiest and a lot of startups I think get this wrong because they're very usually product centric and focused on creating great product which is great like and they tend to be technical. They're good at building things.
[00:54:47] Speaker B: They don't.
[00:54:47] Speaker C: They.
[00:54:48] Speaker A: They underestimate the challenges of like selling into a new.
Into a new audience or new. New segment.
[00:54:57] Speaker C: Absolutely.
[00:54:59] Speaker A: All right, should move on. What we got five minutes.
[00:55:08] Speaker B: What is this one?
[00:55:09] Speaker C: Paul?
We can't hear you.
[00:55:16] Speaker B: This one's on Icon, the famous AI ad maker which followed the Cluly playbook has started turning into a UGC creative agency and they're desperate to add new revenues. I thought this is interesting. I thought I would like to get your take on this since it's slightly relevant.
[00:55:39] Speaker C: I have heard that Carter rumors for a little bit of time already. I think this is something that is industry wide in their space in the AI ad generator. And so I think that is not intrinsically their fault. That is just where the tech is at where they're not as lucky as us where just an LLM call to make a sponsor message but they actually need to combine a bunch of different technologies together to make great ads. And brand safety is very important there and to do a good job is very hard. And so.
[00:56:11] Speaker B: Right.
[00:56:12] Speaker C: It's just still not there yet to do that at scale.
They don't control the whole supply chain like we do. Right. We have that end to end and so we can do that. They don't own Meta. Right. And so if Meta were to do that I think that would be a different story for them as an player to do that. That's, that's even more challenging. And so I think it, it does make sense. Right. Maybe it seems desperate just because they were in like this hyper growth mode and then now they're plateauing. They need to do this kind of business model pivot but I think they're. They're just a tech enabled agency and so I think that is gonna basically merge like the tech side, the agency side. And I think agencies need to become more like and tech companies need to become a bit more like agencies and. And this you know may is the testament of harmonization of, of the two spaces.
[00:57:09] Speaker A: Yeah, I think you didn't know about the fact that like the creative development piece particular a third party it provides like, like nominal performance increase increase and you know, increase.
So like every time I've seen all of these tools that do like Dynamic, creative and all this stuff like ever gets very excited about it. But the reality is that like the juice isn't worth the squeeze. They're a ton of work to get them to like work appropriately and they don't necessarily result in better performance.
The messaging matters a lot and I think like Facebook in particular proved that like your message matters, but the production value, the quality, quality of the ad, the way it's produced, like it provides some, some value, but it's not huge. And so all these ad builders and ad creative things and art, like, it's just, there's a lot of effort and then not a lot of value in terms of like driving ad performance. And I think it's always been, always been the case.
[00:58:09] Speaker B: Yeah.
[00:58:10] Speaker A: All right, should we do one more and then we'll do some memes?
So I got this post and actually I think you might be really the right person to like weigh on it. So Stripe, they shared this strike data that was super interesting. So I'll just read some of it.
We're noticing a Trend@ stripe. US startups are pulling ahead of their peers. These charts show average revenue growth for software startups in each location.
[00:58:38] Speaker C: U.
[00:58:38] Speaker A: S startups typically show growth somewhat faster than those elsewhere elsewhere. However, since mid 2023, US companies have accelerated a lot.
Interestingly, at least according to him, this is not because of AI or it's not just because of AI startups. Server knows that AI startups are scaling really quickly. But he said if you strip that out, you still see US startups are growing fast. I'll show the chart.
And so this is the US versus UK versus eu and so US is the line that's really scaling. I think they break out the AI piece here because the argument if you see this chart is like, oh, it's just AI and cursor. And these startups are scaling really quickly, which is true. So you remove that, you can see, yes, US startups are scaling really fast. But here's the line here without the AI startups.
[00:59:28] Speaker B: So it's the same.
[00:59:30] Speaker A: It's still scaling much faster than other than startups from other regions.
[00:59:36] Speaker B: Okay, so I actually have a question here. Doesn't this also mean that AI is not as big as it seems?
[00:59:45] Speaker C: I think it does say that. I think this chart makes sense to me to the extent that non AI startups are essentially in established businesses with established revenue models with established customers. And so the customer base here is already there and you use the AI to just have cost optimization and just other optimization that just increases the Margins, Whereas the AI startups, they're just going in a new market and so you're still trying to find product market fit. And so obviously you have some breakouts, but it's still a new market that still needs to reach maturity. Non AI startups, they already have a business that is set up, they know what they're getting into. And so if you are just a service business, even a product business, doesn't matter what business, but you're an established market now with AI you can just sell more, faster, cheaper, so you're going to grow faster than your AI peers just because you have a market that is already here.
[01:00:39] Speaker A: So the part that I highlight here is where he gives his hypothesis. So you're dead on with that for sure. But he says that US startups, at least as their hypothesis, even if they're not AI companies, are adopting new technologies.
He calls out not just AI but stablecoins and that they're adopting them faster than companies in other regions and that's why they have started to accelerate relative to startups in the UK or in the, or in the eu.
[01:01:09] Speaker C: Well, I think that's because they, they've been pushing stable coins. I think they're definitely biased, a little.
[01:01:15] Speaker B: Bit biased, yeah, I agree with that.
[01:01:19] Speaker A: As well, but I still think it's an interesting chart. I thought that like they have really broad based data from a wide range of companies and so I still think there's some validity to this.
[01:01:31] Speaker B: It's. Yeah, for sure.
[01:01:35] Speaker A: All right, should we do some memes and then we'll wrap up?
[01:01:37] Speaker B: Let's do it.
[01:01:38] Speaker A: Okay, so I got a couple memes.
Let's see. This, this is the one. This one did this. Yeah, this did a couple thousand views. So this was on the Mandami stuff. Some of you caught this one, but yeah, my wife laughed really hard when she saw this.
So she says, yeah, I know, I get it, it's ironic. I voted for socialism but I only dance date finance bros. Just sue me.
[01:02:07] Speaker B: Very New York.
[01:02:08] Speaker A: I like this one.
And then, and then I got one more. Okay, I got one more. Montgomery. This, this one, this one I have to explain for young kids.
Live footage from downtown Manhattan near Goldman Sachs hq.
[01:02:22] Speaker B: Yeah, this.
[01:02:24] Speaker A: Yeah, you have to explain this photo. So this photo, this photo is taken at the end of the Vietnam War when the communist guerrillas invaded the south. And it's of the U. S. Embassy when they were evacuating.
[01:02:37] Speaker C: Oh, right, and everybody was climbing the wall, right, yeah, yeah, of course.
[01:02:41] Speaker A: Evacuating people to the ships out in the ocean they they flew flew so many helicopters it's actually a very serious photo. They flew so many helicopters that they had to start knocking the helicopters off the boats.
[01:02:50] Speaker C: Yeah. Ocean to get more space.
[01:02:54] Speaker A: Yeah.
[01:02:54] Speaker C: In the planes as well.
[01:02:56] Speaker A: Correct. So they could evacuate Americans as the communist guerrillas invaded. So this there is some comments in this thread that just go on and on. We're like like providing live reports as the communist guerrillas invaders Brooklyn and take over Central park and we have certain areas cordoned off so that Goldman Sachs executives can escape the airlift as a socialistic over New York.
[01:03:23] Speaker C: Yeah.
[01:03:24] Speaker A: All right enough of that.
Well dude, thanks for coming on excellent I'm sure we could talk for hours about that was really interesting and I wish you the best luck we'll catch up when I'm I'll be in San Francisco in December.
[01:03:43] Speaker C: Perfect.
[01:03:44] Speaker B: How can people follow along hey you.
[01:03:47] Speaker C: Should go on thread AI t h r a d AI or just follow me on LinkedIn. Andrea Tortella.
[01:03:56] Speaker B: Awesome.
[01:03:57] Speaker C: So great to chat.
