From SEO to AEO: How To Win In The Age Of AI with Kasim Aslam and Joe Polish

What I'm up here to tell you is I think Google's gonna win. I think Gemini is gonna be the most powerful AI tool in existence. Google owns all the most important websites on the planet. If you think about it, Google basically captured the collective consciousness of humanity and they put it on a hard drive. They took every piece of information every human has ever created since the advent of the internet and before, cuz that was all digitized too. And then they stored it for itself for a someday use. And so I think in, in a world where we're watching a space race, Google has the fuel fuel necessary to go to space, and everybody else right now is actually borrowing from Google's fuel.

Yeah, how can you be a domino for someone else? The Joe Polish Show. Any problem in the world can be solved with the right genius network. So let me, let me introduce my friend, um, Qasim Aslam. Okay, and that's pronounced like awesome, Qasim. Okay, and I still mess it up every once in a while. He's a college dropout. See what I was telling you? We're not We're not really reaching for the, you know, a lot of, lot of skill here. He's a college dropout turned powerhouse entrepreneur, author, and proud Front Row Dad. He built and sold the world's leading Google ad agency in 2022, and he now heads ParetoTalent.com, a high-performance recruiting agency for executive assistants. Uh, he's also my partner in SeedOil.com. And, uh, you know, I'm gonna ask him a few questions, uh, about some different things, and I think you might learn a, a thing or two. So give it up. From my good friend, Qasim. What's up, brother?

Thank you.

Do you have a preference of seats?

I like that one.

Okay, go ahead. Yeah, go ahead. And so, oh, you know what I need to get from Eunice? Do you mind bringing me the phone that I had earlier? Because I just realized I don't have— I put something in ChatGPT based on that I want— thank you— that I wanted to ask you. So I'll come to that. So You had— what did I not say about you that people need to say? Just rattle off a little bit about, uh, what you, what you do now, what you used to do with your agency. And we're going to talk a little bit about, uh, AEO versus SEO and different stuff like that. And then we're going to do some Q&A with the audience because I know once you get talking, people are going to become intrigued.

Uh, I built the number one ranked Google Ads agency in the world. I had that business for about 15 years. And the reason that's relevant is because I spent $100 million a year of other people's money. Thousands of clients went through my shop, so I got to see a diversity of industry, which was helpful. And the reason that's helpful is because Google, Scott Galloway has this really amazing keynote, a lot of you've probably seen it, where he talks about how, I forget the other connections he makes, but he says something like Amazon is love, Apple is sex, Google is God. The reason Google is God is because on an anthropological scale, the entity to whom people pose their questions used to be God. So you're a shepherd in a field 2,000 years ago and your child is sick and you have no other option but to drop to your knees and look to the sky and say, "Why?

How?

Can you please?" Etc. Flash forward, and this isn't a usurpation of a deity, but, but I think it's really profound to think that that relationship has actually been usurped by Google, or it had been. And so Google is the closest thing we've ever gotten to something like, uh, like intent alchemy. It was the first time ever that human intent started to be distilled in a way that could be measured and managed. And that's different. Google knows what you ask it. Facebook knows what you tell it. And it just— it led to kind of this ongoing, ever-present obsession with the data, with just analyzing the data. I have an entity now that— here's the thing, I don't have an offer. Evan's been making fun of me for the last 24 hours because I don't—

gently guiding.

Yeah, gently guiding. There's nothing gentle. I don't have anything to sell on this business line yet. I have 6 full-time employees that have done nothing but research how LLMs source their citations because I'm interested in that. And this is why you don't give poor people money, because we will spend it in fun and interesting ways. Uh, but we've, we've re— I have more data on this topic than any human alive, I think, except Moniek Ban, who owns Search Atlas, who's also a buddy. But his— the way he's acquiring the data too, I don't agree with his thesis. Generally brilliant, brilliant guy, by the way, and I'd love for him to be able to combat that because I'm sure he's smarter than I am. Uh, but I think that this is, it's critically important to not just business, but also humanity. Because there's a lot of directions we can go, and I promise not to go in all of them, but the one I would offer just as an example is the way that LLMs are currently delivering information is very, and everybody here has experienced this themselves. As a matter of fact, I was talking to John, who was like, oh yeah, I know that, and I've been saying that. LLMs have inbuilt confirmation bias. To a degree that's horrifying. So if you're pro-vax and I'm anti-vax and I'm asking LLMs about vaccines and you're asking LLMs about vaccines and you're looking for peer-reviewed scientific literature, you will get exactly what you're looking for and I will get exactly what I'm looking for. And there's something about— we already live in a world that has the propensity for echo chamber, and now we just weaponized this trillion-dollar, you know, nuclear technology to just send people as far and fast as they possibly can in the direction they're already heading. And there's something about that that's terrifying to me. And there's also a way that I'm going to use it to make tons and tons of money. So I've got like the evil capitalist in me. And then also the— there's just something about it that I feel we need to have these conversations and put those guardrails up. How did I do there, Joe? I ran—

No, well, you know, no, let's just say, OK, what about the non-evil capitalists, right? No, I think that was great. Can you just, at the most basic level, define LLMs, large language models? What the hell does that actually mean? How does it apply to everyone here?

Yeah, so first of all, it's actually, it's inadequate nomenclature. And I've tried to, 'cause words are so important. Semantics are so important. The only thing that separates all of us from any industry is semantic architecture. 'Cause like real estate guys scare the daylights outta me and M&A guys scare the daylights outta me 'cause they say shit like EBITDA and I just go full on like 10th grade dropout. But as soon as you can flatten semantics, I feel like it puts everything kind of on an even playing field. So LLM technically is large language model. And a large language model is a mechanism within which we've put, if you've ever done the research on how they train AI, it's really interesting. There's nothing that's been coded. We've been taught to believe that anything that happens on the computer has been coded by a human. Instead, what they did, and I'm gonna put this in just the, I'm gonna do my best to make this as accessible as I had to make it for myself, I guess is what I'm saying. So I had to dumb this all the way down. But they took a computer and they said— they gave it 100,000 pictures and they said, this is a man, this is a woman, this is a man, this is a woman, this is a man, this is a woman, this is a man, this is a woman. And then they started to ask the computer back what's a man and what's a woman. But they don't know why the computer was able to identify. They don't know high cheekbones, low brow, Adam's apple, like they don't know what about the pattern recognition enabled the computer to know this is a man, this is a woman. And then you do that on every level of analysis and you cascade it across every data repository you can possibly get your hands on. And that ends up being an LLM. An LLM is, it's a pattern recognition mechanism. Now, LLM isn't the right word for all AI. Not all AI is LLM and not all LLM is AI. And people Nobody's really clear on which ones are and which ones aren't. So we're saying LLM, but that unto itself is a problem. What I like saying right now is AI engine, because not all of these are trained on language, which is really, really important. Some of them are trained on models per Simon's Genius, and some of them trained on images. And so we're kind of painting with this broad brush, but the AI engines are pattern recognition mechanisms. And the reason that that's so important is because the propensity for, Confirmation bias is built into that framing. Pattern recognition means compliance with the established understanding of how something works. So where it's basically peer-reviewed, we all agree that this is what something is or what it looks like, and then the computer goes, oh, you all agree, that must be true. It reminds me of the— Einstein had this really great quote. There was 100 scientists that wrote a treatise against whatever he was saying with I think the theory of relativity. And he said, well, if I was wrong, it wouldn't have taken 100, it would have just taken 1. And it's a reference to the fact that the hierarchical structure in terms of information is actually very often inversed in terms of who believes what and who's actually right. And the AI engines are built for confirmation from the aggregate. I'll stop there.

Well, so let's go back to our trip. Whitney, Pardis, we're all fine in Texas, and you gave the whole spiel about Google and how they're the most powerful. And then we'll come back to some of this. I want— because I thought that was such an incredible insight, and we talked a little bit about that at a previous Genius Network meeting, and I just would like everyone to hear that. Yeah, that'll give some contextual understanding. And we'll talk about SEO and we'll talk about how to use all of this in order to position your business, what to pay attention to, what not to pay attention to, and get some predictions from you about where things are going. Not like Nostradamus predictions, but just by stuff that you're looking at every freaking day.

Just out of curiosity, if you're using AI on a daily basis, who prefers ChatGPT to all else? It's about? About? 30% of the room. Okay, who would prefer Claude to all else?

Grok?

Perplexity? Okay, Gemini?

One.

Gemini wins. I'm so long on Google and I'll tell you why. Now here's the thing, nobody raised their hand when we said Gemini because when you use Gemini, you feel like you're getting inadequate responses. When you use Claude, Perplexity, ChatGPT, you're getting mature results. And the reason you're getting mature results, and I can prove this from a data-driven perspective, proves a dangerous word. I can show you correlation isn't causation, but I can show you why I believe this to be true. All of the other AI tools are just Googling something faster for you. So if you ask ChatGPT or Claude or Perplexity a question, they're going into Google's very, very, very mature forest because Google Search is the most complex search graph that's ever existed, and they're mining the forest and they're finding the value. And sometimes there's some distillation of the value. Sometimes there's some amalgamation of the value. Sometimes there's amplification of the value, but it's Google's mature forest from whence the value comes. And then they bring it back to you. When you ask Google that same question, it's not going into its very mature forest. Google is delivering a sapling. And so you look at the Gemini results and you think like, God, this is garbage in comparison. And it's because Google And this is the craziest part. Google is not using Google's own search graph in its AI-driven results, which is insane. It's insane. So here's how we know this. I have 6 employees, 1 supervisor, and 5 technicians. Their job every single day, it's a soul-sucking job, it's horrible. That's how I sell it, by the way. It's how I hire them, is they have to ask 20 questions within the confines of 10 categories. And the categories are the ones that you would guess at. So it's things like local brick and mortar, SaaS, professional services, info marketing, et cetera, et cetera, et cetera. 20 queries and every query is asked of 3 LLMs. And the number 3 has more to do with the data crunching tools that we use and the inability to expand it any further. So you take 20 questions. So like one question would be the best boutique coffee shop in Kingman, Arizona is the example that I use because the data is, I think, the most clear. And we use hyper-specific phrases like that. You want qualifiers. So like best, closest, nearest. Top, best boutique is hyper-specific. And you want that because you want to get rid of the noise. You want to get through to signal. So those qualifications and kind of commercial intent tend to be really important. I also like smaller niche industries like Kingman, Arizona doesn't have a lot of boutique coffee shops. Manhattan does. So if I get a result from an LLM in Manhattan, it's harder for me to understand how that hierarchy might be structured, where in Kingman it's a lot easier. So that's one query. That query is asked of 3 separate LLMs. And then he's at— and then that person's tasked with asking another query and then another query and another query. And they're all based on that same category. So it might be best boutique coffee shop in Kingman, Arizona. And then it might be non-GMO coffee shops in Kingman, Arizona. And then it might be seed oil-free coffee shops in Kingman, Arizona, and so on and so forth. And we do that 20 times across 3 LLMs. And then the LLM delivers an answer or multiple answers. But generally speaking, it delivers an answer. But the answer is— I'm less concerned with what it also delivers is a collection of citations anywhere between 1 and 10 sometimes more, but 10 tends to be the, the max for the most part. Citations, and the citations are what the, the AI engine has told us it used to come up with that query. So the question we have to ask ourselves is why? Why that citation? What made you think that was relevant? So my team takes every single individual citation and they map it. I can show you actually, would y'all mind, let's throw up slide 10. What do I point at to keep it going? There we go. Oh, we'll go back.

So, and if you know, we put it on this one behind us too if you guys want, so it's easy.

Oh, I can see it there. Um, perfect. So just a quick soundbite, a precursor. This is a HubSpot, and this is something that HubSpot shared. Organic traffic has dropped. It's a precipitous cliff, and it's actually gotten worse. This is, I think, a year-old data. This says 60%, it's now 65% of all searches results in zero click. So that's where we are. It's already clickmageddon is what people are calling it. Like you're not getting the traffic to your website, which has a whole bunch of other implications too. This is what I wanna show you. This is what our research looks like. So those are the 3 LLMs in this particular example, and you could swap those 3 out for any. Grok is the outlier by the way, because it's the only AI tool outside of Gemini that really has its own data repository, but there are some flaws to it. Meta's data repository, Meta goes out more than it goes in, interestingly. And so we take every single LLM, we ask the same query of the LLM, and then we track all the URLs. And then we look at the URLs along the level of analysis that you would assume based off of conventional wisdom. So we're looking at things like key phrase permeation, backlinks, page speed, schema markup. There's like 50-some-odd data points. And so if you take 1 query, 3 LLMs, 10 citations per LLM, 50 data points per citation, multiplied 20 times, multiplied by 5 employees, multiplied by every business day for the last year, that's how much data I have on this. So it's not nothing, I guess is what I'm saying. It's supported by some level of, you know, active buttressing. And what's really interesting too is a lot of the stronger assumptions we made are only proving to be more true. So we think we've kind of caught the direction with which this is going. Can y'all go? There we go. So this is what this really looks like in real time. And this is the part that's really interesting. And I usually do a great big grand reveal. Here's ChatGPT as an example, best boutique coffee shop in Kingman, Arizona. Those are the citations. These are the citation prowess, let's say. Those are the things that we would be analyzing from an SEO perspective. There's a reason this is massively flawed, by the way, and I only figured it out half— we're like 9 months into it. Again, more data analysis, page speed, schema markup. And then here's query 2, where to find specialty coffee in Kingman, Arizona. So we're getting really granular and really specific. We can kill this now, kill the slide. So this is the part that's really interesting is when we get these citations, one of the things that we check is in, if you take the rank order that the LLM delivers a citation and the chronological rank shows some level of prowess, we check that against the Google search results. And so if it ranks number 1 in the LLM, where does it rank in Google? And maybe it ranks 1 or 2 or 3 or 4 or 5 or 10. ChatGPT from memory has somewhere between a 40 to 60% overlap rate with Google. Call it 50% on average. Perplexity has a 70 to 80% overlap rate with Google. Claude's somewhere in the middle. Guess what Gemini's overlap rate with Google is? In certain categories, it's zero. It's zero. And the highest it gets is 25%, which means, going back to the original thesis, Google is not using Google's own search graph. Google is building its own knowledge graph from the ground up because Google's the only company on the planet that has the data to do that. Amazon could do it with AWS, but it would be a gross violation of their terms of service. They can't use that data. Apple has some proxy data. They have a more valuable window, but they've never captured it. They took a really weird stance, by the way. The iOS 14 thing was an interesting direction for them to go, and I think they have to regret it. I don't know if that was a miscalculation or if they're smarter than me and there's a switcheroo coming, but Google's the only company on the planet that has the data repository that would be necessary in order to train its own AI engine to this degree. And so the reason that Gemini, and Google has a propensity for staying back and letting the other pioneers take the arrows. They did it, there was a congressional hearing about Facebook because Facebook had 55,000 demographic and psychographic profiling factors on people and they allowed advertisers to use those profiling factors. So what do we have? That actually triggered GDPR, triggered privacy concerns, triggered the geriatric morons in Congress, you know, coming at Zuckerberg asking him what is the internet? When that happened, Facebook had 55,000, Google had 72 million demographic and psychographic profiling factors. But they didn't lift the hood and let people see that. So Google has a storied history of having all of these tools that they don't release to the public. They wait. Look what they did. What's the image generation thing? Banana Panda something, Banana, Nano Banana. All of a sudden, Google goes from not being on the image, the AI image map at all, to releasing what ostensibly is, from the people that I know, the greatest of everything that exists. And it happens overnight without any warning.

They wait.

They wait to see what happens with legislation. They wait to see what happens to public opinion. And then as soon as they can tell the direct, Google's run by attorneys, as soon as they can tell where the wind is blowing, then they release. And you'll notice that Nano Banana came out almost immediately after legislative precedent was established as far as what happens with these images. So Google, and here's the craziest part, when you look at ChatGPT, Claude, Perplexity, they're drawing from a well that Google owns the access to. It's a flawed— well, we know SEO is flawed. Any SEO who tells you that SEO is about quality content is absolutely full of shit. SEO is about link building. You buy links, you rank. That's a fact. It's the reason SEO.co and The HOTH and PosiRank exist. There's lip service given to the content, and you do need some content in order to rank, obviously, but that's not what gets the, the, the top rank. Google search is flawed, and Google knows that, so Google abandoned Google Search Graph is building its own knowledge graph from the ground up. And the day it wants to, it can rip the rug out from every single other LLM. Now, how long does it take them to do that? Will they do that? Who knows? But I'm so long— I've only ever lost money in the stock market, by the way. My thesis has always been buy high, sell low, cry. And if anybody wanted a stock tip from me, dude, I'm deep, deep, deep in Google because I really believe in the data they have and their ability to execute upon it. I just think it's going to be really extraordinary. And so I really like— the other thing that's really interesting about my children, Chromebooks, they're 8 and 10. And I was just thinking through like their life from a vocational perspective. And I'm like, there's nothing left to do in 10 years. What am I going to teach them? And I decided 2 skills. One, learn AI, obviously. And even that is it. I think that skill is, what would you call it? It's almost transitory. Is that a word? Transitionatory? Transitory. Yeah. So it's a transitory skill that's not going to be applicable for very long, but it's important for them to know. The second one I actually think is authority building. So I bought my kids Chromebooks and I said, you can only use this to learn AI and build authority. So they're shooting YouTube videos for other little kids, teaching them how to use AI. Here's why this is relevant. Buy them Chromebooks, fire up Chromebook, help them create Gmail accounts. Guess what Google's default search engine is on Chrome? Guess quick like bunnies.

Chrome.

That's the browser. The search engine is Gemini. Gemini is not a search engine. Google killed— it took Google Search and shot it in the back of the head. It still accounts for 85% of Google's revenue. The fact that, you know, the third most valuable company on the planet took its highest performing asset and no longer makes it the default on all of its new devices and instead delivers Gemini. The way to think about these things, this is how I've explained it to myself, Search is an interstellar space station— or no, is a, is a, excuse me, is a spaceship. Okay? So when you're searching, you're on your planet and there's galaxies and universes and all sorts of things out. And if you want to search and you're like, like right now I'm shopping for a car. So if I want to buy a car, I might have 20 levels of analysis and you don't realize that's how complex commerce is, but it is. There's 20 levels of analysis potentially. There's things like, you know, maintenance and fuel efficiency. And let's say you have a special needs child or I'm about to buy new land or whatever. There's 20 levels of analysis.. So I choose one level of analysis and then I hop into my international spaceship and then I say, "Okay, I want to learn more about vehicles for children with special needs." Then it brings me to the galaxy that has that information and then I have to go from planet to planet to planet and sometimes moon to moon to moon and I get to mine that information to my heart's content and then I capture the information that I want and I do the amalgamation, I do the distillation, I do the amplification. And then I bring that information back home. And then I say, OK, now I want to learn about fuel efficiency. And then I shoot off again into this galaxy. And so on and so forth. And on and on and on it goes. And that's how search works. It's a dumb mechanism. AI is a little Android that you say, hey, I want to learn about cars. These are my concerns.

Go.

And it goes to the galaxies and the moons and the universes for you. And it understands the dependency and the interdependency sometimes better than you do. And then it brings you back. The information, and this is relevant because the LLMs or the AI engines are pulling from repositories of content and everybody right now is fishing in an ocean that anytime Google wants to, they can nuke, which is amazing. It's amazing. And the implications are staggering. So I think what we all need to do is really diversify as we go for the traffic game. And traffic is the most valuable facet I think of the business endeavor. All the big companies are— Amazon is a traffic company. Meta is a traffic company. Google is a traffic company. If you own the eyeballs, you own the business. And so I think we need to A, diversify, and then B, I think everybody in this room needs to do what that man does. I think authority building is the most important. And I'm not good at this, by the way. I've got nothing to sell here. Like what Evan Carmichael does. Look at, you know, everybody's like, Alex Hormozi made $105 million in a day. No, he didn't. He made $105 million in 7 years. But he turned himself into his own traffic repository, and those traffic repositories are— they're unbelievably potent. The value in them, it's a well from which you can always pull. And, and we're all used to being on this, this strange merry-go-round where you have to pay every single time you pass go, and that's, that's— it's a horrible framework for business. And so I think that the world of AEO is going to change that in a lot of ways. Um, but I, I have some, I have some theories on how we can take advantage of that and make some money.

Okay. So, well, let's get into that. Uh, does anyone have a question for him? Is anyone lost in any way, shape, or form? Throw him a mic because I want to make sure that you're following this.

I'm lost.

Okay. Sorry, Beth.

Yeah.

Um, but well, what we're saying is, okay, tell me diversify, diversify. Don't just focus on ChatGPT. Okay? So if you're using ChatGPT and you think, I want my business to be mentioned by ChatGPT, you actually don't. You want to make sure— and this is, this is a new paradigm switch too, because we're so used to single platform optimization. Everybody in here has their one thing. I was just talking to somebody about this. They're like, man, I nailed Meta ads and Meta ads are great for my business. And then somebody else might be like, I do Google search. And even, even within the confines of that paradigm, Google isn't the only search engine, right? There's Google and Yahoo and Lycos and Ask and etc., etc., etc. What we need to do right now, and this won't be true, I expect consolidation because that's the way technology works. What we need to do right now is we need to have multiple horses out on the track. You need to have your ChatGPT track, your Claude, your Groq, your Perplexity, etc. If you want to play this game, this is a game of land staking. This is taking the beachhead. This is internet circa 1997 when somebody got to go in and buy domain names like business.com for $12 and then sell them for $40 million or whatever it went to. Or circa 1800s when some idiot just got to get off the boat, plant a flag in 1,500 acres of the most beautiful Oregon territory and say, this is mine. We can right now capture this intellectual property in a way. When the LLMs deliver a result, the thing that's really interesting is there's a first mover advantage and there's Because confirmation bias is built into the mechanism, once you capture it, it feels like the fact that you were the first justifies you being the first and so will continue you being the first. But the LLMs operate on levels of granularity that are a little bit different than Google does. So we can talk about that in a minute. But what I'm really saying is be solution agnostic. Do not— there are no sacred cows. And we need to really kind of try to do things that allow You need a foot on the boat and a foot on the dock, and you probably need 6 or 7 or 8 feet 'cause we don't necessarily know who the winner is. What I'm up here to tell you is I think Google's gonna win. I think Gemini is gonna be the most powerful AI tool in existence. Google owns all the most important websites on the planet. YouTube specifically being, I think, the most important content repository, second only to a dark horse that I'll mention in a moment. How did I do there, Babs?

Pretty good.

Okay.

Pretty good.

Thank you.

So I really want to be on Team Gemini.

Yeah.

I want to be on Team— because Gemini is Google and Google is YouTube. And I really want to be there.

Turn lights up.

From my own anecdotal experience, I get better results with GPT so far. From an LLM arena perspective, they're always back and forth on who's winning. From a talent perspective, I think OpenAI is getting more of the missionaries, of people who want to change the world. With AI and my, my thesis on why Google is so far ahead right now and like launching Nanobanan and all these other tools out of nowhere. They've been doing this for years. They had their own LLM like ready to go way sooner than anybody else. OpenAI just released it. And so they had to, so all the things that have been coming out are not that they've developing amazing things that they, they already had a lot of these things and now they're just releasing them. And so like, where do we go from here? And can Google stay ahead of the curve? I don't know. I don't, I personally, I don't think they're bringing in the top talent to stay ahead of the curve. And I think GPT is gonna, or OpenAI is gonna, is gonna be the better solution. I, I, I hope that's not true because I love YouTube.

No, I think it absolutely could be. What I will say is you mentioned Google has a cache of amazing AI tools that they have yet to reveal.

You threw that in there. Cache is not a word. Unless you're in Australia.

Where's Simon? Thank you.

So the question I wanted to ask, so there's something you said, we can disagree and go about, I mean, I would love for you to be right, but we can't really test that and we'll have a conversation in 5 years and figure it out. One thing you said though that I didn't understand, and I'm trying to, you said that Google, Gemini is the only one building their knowledge graph from the ground up.

Correct.

But you said Google is also the one that has all the data. So, like, it has the least amount of overlap. So if you're building from the ground up, why does it matter if you have overlap or you have all the data?

Well, so Google's not using Google's own search graph. Google's not a fetch. Google's a capture. So what Google does when it crawls a website is it actually takes all that information and it caches it.

No?

Just like caches. All right.

Now he's doing it on purpose.

Yeah. Google takes that information and it stores it for its own purposes. And it does it on an iterative level that's actually overwhelming. Like the amount of information that Google captures is insane and nobody else is from a search, all, not, I can't say all, every search engine I've ever been exposed to from an internal perspective was a fetch, not a catch, okay?

But why is that an advantage? Like why couldn't OpenAI or somebody else do that too?

They can, but you'd need, so Google has been doing it for 20 years. It has the computing power to do it. Google, by the way, y'all know Google, broke quantum, announced it very quietly, and then has never said an effing thing since. So OpenAI is data thirsty and low power. They went and asked Congress for $14 trillion because they don't have the computing power to do what they need to do now. For them to go do what Google has done over the last 20 years, I think is, it's not an insurmountable problem, but it's a really significantly sized problem. Google has captured all of this data and LLMs, because of the way LLMs are trained, remember the kind of weird sort of proxy training that I tried to explain earlier. That's how they still work. And Google is the entity that has that information. AWS hosts something like 30% of the front-facing web, so Amazon could do it because they actually have this information on their servers. But if you read Amazon's terms of service, they cannot access that information, generally speaking, unless there's a specialized clause. That might change too. We're keeping a very close eye on that. So Google's the only entity that has the information. And can train its own AI tool from the ground up where every other AI tool is still to this day, ChatGPT, OpenAI, ChatGPT is not storing, it's fetching. And it's true, the only thing it is doing is it's tracking updates, which requires storing the same way that cookies are stored on your browser. It's like a temporary storage, but it doesn't have the data. And the data, the content is, that's where the value is. That's where you would go If you think about it, Google basically captured the collective consciousness of humanity and they put it on a hard drive. They took every piece of information every human has ever created since the advent of the internet and before, 'cause that was all digitized too, and then they stored it for itself for a someday use. And so I think in, in a world where we're watching a space race, Google has the fuel necessary to go to space and everybody else right now is actually borrowing from Google's fuel. But in an interesting way, they're borrowing within a framing that is massively flawed because Google SEO, there's no— nobody here would— I think everybody would at least agree with the fact that just because you rank one on Google doesn't mean that you're the best at anything. Like if you search for a chiropractor on Google, it doesn't mean that's the best chiropractor, right? That's the person who gamed the mechanism. We've gotten used to that. But that's massively flawed. And so the other LLMs relying on Google ranking means that they're actually climbing a ladder propped against a wall that we know for a fact Google's going to take a wrecking ball to someday. And Google having the data they do have allows them to rebuild the knowledge graph and rebuild it. This is what's really fun is from any frame in any paradigm they so choose. So Google PageRank was based off of something called Google ranking was based off of something called PageRank, and it's tongue-in-cheek because the guy's name was Larry Page, and it's a peer review system. It's inbound links equal value, but it's a meritocratic structure where the higher the inbound links, the higher the value of the inbound links, the more value you got, and those links were ranked meritocratically, and so it was kind of this meritocratic web. But you could reframe that and you could say something like peer review based off of, um, you know, Doug Brackmann wrote a book, has a doctorate, has however many readers he has. So if Doug Brackmann goes in and says— and likes a comment, and then I come in and like a comment, and Google could decide, no, the doctorate author, his like is worth 100,000 more points than mine is, and completely reframe the paradigm. And so the content paradigm— Google can own and reframe the content paradigm All the other LLMs, if they wanted to reframe the content paradigm, they first have to use Google Search Graph because that's where we're going to fetch. And then they have to overlay their paradigm on top of it or use the paradigm to fetch. But even that is inadequate because you have to go through Google's Hall of Mirrors. And that example I showed you, by the way, the most important website on the internet right now, if you really want to invest in something, get an EA or that stoner nephew in your Ant's basement to go crack Reddit. Profound did a study of millions and millions and millions of Google-driven citations, and Reddit was cited 1 out of 5 times. 1 out of 5 times. 21% of the time Reddit's cited, 18.8% of the time it's YouTube. But YouTube's 19 times the size of Reddit. Here's why they like Reddit. If you really think through Reddit, Reddit is the, I shouldn't say only, I have a bad habit of hyperbole also. I'm trying to do better at that. Reddit is the only site that I know of, en masse, that allows for peer-reviewed content and peer-reviewed content creators. So when somebody goes in and they start talking about vehicles, and I'm talking about cars, and I don't know anything about cars, and Keith talks about cars, and he knows everything about cars, the fact that Keith has been upvoted on other car threads allows that equity to be brought over. And if you think of it from an LLM's perspective, of course they'd love that because they need to know who to listen to. All of humanity is not created equal. And so something like Reddit gives the LLM the opportunity to say like, oh, you are the expert on this topic, content that comes from you, here's what's really interesting. And this is what my data doesn't show. And it's so frustrating 'cause I don't even know how to capture it. My friend John Moran is the greatest Google Ads mind in the world. Everybody who knows Google Ads knows John Moran. I was searching for something in ChatGPT and I found an article. I was searching on the topic of Google Ads and I found an article from Search Engine Journal that cited John Moran. So Search Engine Journal comes up and ChatGPT is the citation. But when I went into the article, 'cause I was interested, most people do not pursue citations. I go into the article 'cause I'm interested and I realize like, oh, ChatGPT doesn't think Search Engine Journal is the authority. ChatGPT thinks John Moran's the authority, and Search Engine Journal just happened to be the network that cited John Moran. This is why this is relevant. We have an inbuilt bias towards URLs. If a URL shows up, we assume the LLM thinks it's the URL that ranked. But no, the LLM knows because John Moran's famous on Reddit. Anybody who cites John Moran on this topic, I'm gonna put their content up. I'm solution agnostic in terms of where that comes from. LLMs like people over brands 100,000 times to 1. An LLM is going to want Joe Polish, not Genius Network, because a person can get a doctorate, a person can publish a book, a person can have a community, and these brands are very transient, as we've seen. So it puts us in this place where, going back to me saying that everything— everybody should do what you do. You, you have to, if you want to play this game, plant a flag as you, as a person, because the LLMs are going to want to draw from the authority that a human being establishes.. And that authority should be on something like a niche and granular level. I've put myself at a massive disadvantage because I own a staffing agency for executive assistants out of Latin America. I also own seedoil.com with Joe. I also own techstack.com, syndications.com, aeo.co. Like, I'm buying all this shit because I went and made an exit and did everything they told us not to do because I didn't have any of these guys. And there's no place for me to establish myself as an authority within the confines of the paradigm that we're talking about now. And it's massive, massive, massive disadvantage. You kind of have to pick. Now you can diversify, but the diversification also needs to be hierarchically structured too. But if you can establish yourself as an authority, go crack the code on Reddit, create multimodal content, you can own industries. And the industries that you can own, that's the thing I'm most excited about. Evan was talking to me about like, why don't you just offer this as a service? I don't want to offer it as a service. I don't want anybody to know I know how to do this. I want to go capture as much market share as I possibly can. 'Cause there are some industries right now where if I own, you know, think about it as being the neck. Of the head that is the LLM. If I own the proxy through which all this traffic goes, I think there are billion-dollar opportunities. There's hundreds of billions of dollars spent on SEO that's just got wiped off the plate. And when, so that search now goes to an LLM result, but the LLM results don't rank content the way that search engines do. It's entirely different. And so everybody here has the chance to do what you could have done in 1992 when we were just key phrase stuffing and capturing market share. And those people have been, The category kings ever since, you get to do that now, but you get to do it from the LLM stance.

Mm-hmm.

I think this is the biggest entrepreneurial opportunity in the 21st century as it pertains to every individual business.

Mm-hmm.

Because most opportunities are, oh, we have to go do that thing. What I'm saying is this is the thing you can do in your business. And I, I, I don't know of anything else that would have, it's, it's, you know, you talk about leverage, it's the little hinge that swings the big door. So crack Reddit. YouTube is the greatest content. The reason YouTube is so good is because it's the most repurposable content. Long-form YouTube can be turned into podcasts and reels and transcriptions for blogs and carousels and infographics and on and on and on and on. And LLMs love multimodal content. They love knowing, oh, you have a blog here and a tweet there. And they love comments too. They love engagement. It's all these things that don't scale. It's all these things that don't scale. You have to do all the things that don't scale. You actually have to reply to people on Facebook. It's obnoxious. But being able to do that lets them know, oh, this person actually knows what they're talking about. And if you do that, you own that industry.

Don't miss another episode of the Joe Polish Show. Subscribe today at thejoepolishshow.com. If you'd like access to the show notes to take action on what was discussed, please visit thejoepolishshow.com/50.