How AI Search Is Replacing the 10 Blue Links Model with Andreas Voniatis

Welcome to the B2B Growth Blueprint podcast, where we talk with entrepreneurs, founders, and leaders at B2B services, tech, and SaaS companies to learn the systems and processes they've used to create success for their business. And we'll also learn from marketing and sales experts to provide insights to help you grow your business. I'm your host, Mark Osborne. And the podcast is brought to you by Modern Revenue Strategies and their 10x ROI B2B Growth Guarantee. Schedule a free diagnostic call to learn the fastest path to double your B2B revenues in just 90 days. Hello and welcome to another episode of the B2B Growth Blueprint Podcast. I'm your host, Mark Osborne, and today I'm very excited to welcome my guest, Andreas Vognatas. Now, Andreas is the founder and CEO of Arteos.io, which is an AI search optimization or generative engine optimization agency that helps B2B companies get recommended by ChatGPT, Claude, and other AI platforms with guaranteed results. He's a certified accountant turned data scientist, and Andreas built his first large language model, or LLM as you hear them referenced, from the ground up all the way back in 2019, 3 years before ChatGPT launched. He has really unique expertise in how AI systems process and recommend information. He's the author of Data-Driven SEO with Python and has nearly 25 years of experience in SEO and digital marketing. ChatGPT recently positioned Arteos as the leading generative engine optimization agency in the United States. So Andreas, welcome to the show.

Thank you. Thank you for having me, Mark. I'm looking forward to sharing my knowledge on GEO.

So this is— I get asked about this every day, at least once a day. So I'm really excited to jump into some of these topics and really share a lot of insight with our audience. But before we do, I do want to get a little bit of your background because There's so many people out there slinging snake oil and acting like they know what's happening in the AI space. And I know from some conversations that you and I have, you have some really unique perspective and some background that gives you some expertise. Tell our audience a little bit about that so they can really sort of understand where you're coming from.

100%. So I started in SEO Around 2002. And I did that because I must confess, I heard about affiliate marketing and I wanted to make money in my sleep.

Sure, sure.

And so I thought, okay, well, that's all great, but how do I get the traffic? And then, and that's when I first came across the word SEO.

Sure.

Now, bear in mind, I was just a qualified accountant at the time. I just Got my qualifications, got my certificate and everything. But this was far more intriguing because there's no world championship for accountancy, right? So that's right. But with SEO, it's incredibly competitive, and it really piqued my interest. So I did everything I could, reading online and testing and experimenting what I read, and I found that 90% of what I was reading. was a load of rubbish in SEO.

Yeah.

And so I found that it was actually quite easy to be pretty competent and even good at SEO if you were prepared to do the dirty work of experimentation. Anyway, if we fast forward a little bit, uh, data science was becoming a thing back in 2012, and by that point I was starting to get a bit bored of SEO. So I was thinking I was learning data science to to perhaps become a data consultant as a career or open a data consulting agency. What it actually did was it gave me a newfound appreciation for information retrieval or the science of search and made me a much better SEO. But the other thing it did was it showed me the software engineering side of AI and mathematics, the maths behind AI, and got me into programming. So, and it was then I saw the future of search and where it could go. And that's what led me in 2019 with the help of some ex-Amazon engineers to create my own LLM. Now, I was feeling really pleased with ourselves because of what we achieved. We trained an LLM on 2 million documents so that we could automate the optimization of web pages without having to touch any of them.

Wow.

And then GPT-2 was announced and I thought, well, I don't think I'll be competing with OpenAI, but I see the writing is on the wall for SEO and it will become AI search.

Yeah.

And in 2022, you know, ChatGPT was announced. But by then I had already been harvesting like a mad scientist data daily for, you know, what was then the last 3 years. And I— and that put us in a very strong position to prepare for AI search. And, you know, when you build an LLM from scratch, you know inside out what all of these AI platforms are looking for. What will, you know, what will they reward? What will they not? What will they ignore? And that's what's led me to this point today.

So, so many things to get into there and to dive in, but to sort of set the frame before we get into, you know, what they look for, what they reward, what they ignore. Let's talk a little bit about, you know, the king is dead, long live the king. Uh, and, and this is a thing that happens in, in marketing on a fairly regular cycle. You know, radio is dead, long live TV. There's still radio around today.

Yeah.

TV is dead, you know, long live, uh, YouTube and the internet. Well, TV is still around today, or even before that, it was you know, broadcast is dead, long live cable. And then, you know, cable's dead, long live YouTube. And now YouTube's dead, long live the scroll or whatever it might be. And so there's this sort of chorus, you know, around today of, you know, SEO is dead. It's now generative engine optimization or, you know, however you want to describe, you know, optimizing for the answer engines. And I think that's maybe the wrong question because in the same way Some people still listen to the radio. Some people still watch TV. There will still be people that, you know, continue to use search, but actually they kind of serve different roles. And I think they'll sort of land in different roles. Talk to us about the right way to think about how these generative engines are impacting consumer behavior and the right way marketers should be thinking about that.

So there's a lot to unpack, even from that question right at the end. If we start with SEO, I don't believe SEO is dead. Like the examples you gave, you know, we still send paper posts despite email being the most popular way to communicate. AI is heavily dependent on websites having good SEO. Okay, so I wouldn't say SEO is dead. I would say the SEO industry is on life support. Okay, so there's still a place for the SEO industry to provide support to businesses, and it is vital support, you know, especially if you're an e-commerce store. You know, there's bound to be tons of errors that keep being created as products go out of stock or new things get added. That still needs SEO maintenance. And even if you're B2B, you still need to make sure that all the basics are in place, such as having content that's discoverable and prioritized within the website system so that AI and search engines know what is priority for their users. And the other thing is as well is that AI does use RAG or retrieval augmented generation to, you know, their models are 6 months old, right? So they need— they often do internet searches just to make sure whatever they're giving you is up to date. So they rely heavily on search results for that. And so there's no way that SEO is dead. It's just on life support, or at least the industry's on life support or consulting's on life support, you know. And I say that because You know, if you're a business owner, you probably wouldn't be prepared to spend anywhere near 5-figure sums, much less 4-figure, or at least high 4-figure sums for your SEO consultant like you used to, you know?

Yeah.

So, you know, like everything, we all have to adapt. It's survival of the fittest, not the strongest, but the most agile. So that's my view on that. If I can segue into the other part of your question, Yeah. You know, back in 2000, um, what, you know, I remember talking to someone and, uh, we were talking about how books were being sold on Amazon and, you know, they were like, oh, that will never catch on. I was thinking, I don't know, I'm not sure I agree with that perspective. And look, you know, The internet back then was really the Wild West. You know, if you— I imagine the internet a bit like— have you ever seen the movie Tron?

Oh yeah.

Yeah. I don't want to assume. So Tron, you know, you have like this virtual space and, you know, it feels like the internet is like Tron, but it's been remade. So 25 years ago, the internet was like the search engine would give you highways direct to your website. Okay, or your store, okay, or your office. Uh, it seems like with AI, it's like it's been wiped clean with a few highways, and now what you have is one big closed shopping mall. Okay.

Interesting.

And, and so if your information gets used, you don't get a highway to your store. What you get is a screen with your name on it, as if it's like content sponsored by your brand name or your business name. And that, that's the age we're in. We're not optimizing for clicks anymore. You do get clicks, but what you're really— the real, the new normal, so to speak, is not traffic from search. The new normal is impressions, right? It's about mindshare. It's about people remembering who you are so that if they receive that cold outreach email from you, or you hope that they actually search for you online, either in the mall or on their phone or when they get back home. So when they're on the way home or when they're at home, you just hope that with enough impressions that they will actually search for you. And, you know, people are getting leads from AI, but we're not in the search traffic age anymore. We're in the search impressions age.

So let's kind of illustrate that a little bit further, because I think that's a really valuable sort of insight, which is in the old age, if someone searched for, you know, providers of microphones, I'll just use this as an example since it's here. They would be directed to choose from a list of websites that sold microphones. They would go to that website. But they might not actually be looking to buy a microphone. They just might have a question about it, but they would go to that website and learn about it. They might be looking to buy, and if they were, they could potentially buy it there. Today, they will go to an answer engine with that same question and get told, well, this particular brand of microphone suggests that the answer to your question is X. And so potentially that intent, that buying intent wasn't there in the past and isn't there today. So there's gonna be less site traffic from sort of lower intent. But if that brand of microphone answers the question in a very compelling way or in a way that sort of aggravates the need or sort of generates intent, they will potentially then still remember to go search for that manufacturer's website and then buy that microphone or that adapter or whatever the thing is that they were, you know, sort of stimulated on. Is that the way that people should be thinking about it?

Yeah, I think that's certainly on the buying end. The one thing to remember about AI search is to be recommended, first you must be cited. And therefore, to be cited, you're You need to be authoritative. Okay. To be recommended, you also need to be liked. Okay. So unlike search engines per se, which relied heavily on authority, AI relies on both authority and sentiment. So if people are saying nice things about your company, then that is what will get you recommended. If you're authoritative, you'll get cited and you may get recommended, but I'm in the business of guaranteeing AI visibility, so I don't want to take chances. I want to make sure, A, you're liked, and B, you're authoritative. So in terms of the funnel, getting cited, you know, you may not see as much traffic until the very end because, you know, AI is actually summarizing a lot of the information, not just from your website, but from other websites. But the good thing is, is that I believe it's— although you're not getting the traffic, it's speeding up the buying cycle. And the reason why it's shortening that buying cycle and speeding it up is because your prospects are now saving a lot of time not having to read through those 10 blue links or 20 blue links. AI is just done them the job, right? So it's like, thanks, I've got the information. You've just saved me a few hours, maybe 1 hour. Now I can ask the next question, you know, what can you recommend, right?

So, so interesting. And, and so you, you sort of laid out there what I think are some of the important ingredients, uh, and I think there's some nuance between them, uh, which is, which is worth. So in the old age of search engine optimization, you got ranking through being an authority. And the way that you sort of demonstrated that authority was having the content on your page that other people linked back to. And that's fundamentally how Google's algorithm works. The more people that are relevant and of, of high authority themselves that link back to your content, the higher you rank, you know, for that particular content, whether that's organized by keyword or longer form or sort of semantic indexing and all those other things. But so that's, you know, sort of fundamentally one way of ranking for search engines. For ranking for the answer engines, it's having authority, which there might be a new way of measuring authority, not just backlinks. So I'd be curious to know that, as well as sort of citations, which might be similar to backlinks, but I'm, I'm curious to understand that. And then also a sentiment analysis because just because you have authority on a topic, it may not be content that anyone enjoys consuming, like academic content or things like that that wouldn't show up in an answer engine but might show up in a search engine. Talk to us a little bit about those differences of inputs.

I think the one thing to bear in mind is that The principles of information retrieval— retrieval— the principles of information retrieval are not thrown out of the window just because we have a new technology that's made it much better. AI synthesizes from memory. Now, so when we talk about information retrieval, which is the science of doing a good job as a search engine or an AI, Information retrieval still has the principles of authority, still has principles of making content discoverable and the content itself. Now, the game changers here are that all the principles that have changed are that search engines have a very limited input and a limited output. The limited output is the, uh, Temboo links. The limited input are a bunch of keywords. Okay, AI is— because it's so much more sophisticated, it can, it can have a much wider context window. You can feed it, um, you can feed it code, you can feed it, uh, and, you know, text for an ebook. It can take in, you know, you know, documents, never mind, you know, huge swathes of text. And the output is a lot more— the processing is very sophisticated because it can actually reason to a very high level. And, and it has a model and it has a lot of memory, you know, that that model was trained on. And therefore the output is a lot more sophisticated. So we're moving from keywords here to people. Okay. Because it could— because, you know, you type in a pair of— you type in a search engine like men's jeans, But who's typing men's jeans? Is it someone who wants American Apparel? Is it someone who wants Dolce Gabbana or somewhere in between? We don't know. But with AI, they do know because, you know, people put in a lot more things. They say, oh, you know, it's already learned from your previous queries for a start. You know, it knows what kind of income level or it can infer what kind of income level you are. Where you're going. Yeah. And people would type this thing in, you know, I'm going to a party at— it's going to show my age because I don't know where the coolest parties are. But let's say I'm going to Billionaire Club in Monaco. I need— I've just arrived. My suitcase hasn't landed from LaGuardia. I need to go shopping for a pair of jeans. Make some recommendations, you know. So it's unlike a search engine which has to try and guess, like, you know, even with personalized search, which jeans you might like, it will go to the most authoritative. But, but with, with AI, it will know what kind of brands you're more likely to, to, to, to purchase. So that's why AI is far smarter and it's more people-driven. And what that means for you is if you're, if you're using keyword tools like Semrush, you can totally throw them in the bin because they're not going to help you. What you do need to be doing is using market research planning tools So we've already obviously created our own proprietary platform because we've been data mining the entire internet for the last 6 years. And so we've got a lot of opinions based on who it is that's making those opinions. So that, that is the kind of research that's going to make your content sing in AI. In terms of authority, well, you know, backlinks, you know, there's a whole industry out there that's selling backlinks by authority level. That's not going away. It still works. You know, because SEO, because search engines heavily rely on them. Obviously you've got to be really careful because, you know, you still want high traffic sites, but even that's not a guarantee. What you're looking for are behaviors to make sure that that site is not selling backlinks. It's not obvious, you know, whether there's expertise on weddings and then another expertise on industrial chemicals for glass production. Hello.

Interesting. So you said a bunch of things that are really worth, I think, unpacking, which is moving from keyword research to market research. And that's because rather than looking for keywords, types of audiences look for types of solutions. And so being able to describe your solution the way your target audience describes that solution and through sort of the lens of that persona is more important than the keyword was historically. Am I understanding that correctly?

Yes, yes, exactly. So, you know, there's, you know, keywords hide the person, but with AI, we could get— we could get—

That's really interesting. That's interesting. And you also, you've said something a couple of times that I want to make sure doesn't get buried because in a previous conversation you and I talked about it and found it to be really powerful. Which is, you know, it's sort of thrown around a lot. Well, you know, the ChatGPTs of the world, they only update every 6 months. And so you kind of can't impact your results within less than 6 months. But the difference is, is one of the inputs that they use is the results from search engines. And so as you're doing a good job of optimizing your website for search engines, that are re-indexing, you know, large swaths of the internet on a rolling basis, that rejiggering of your location within search can actually impact your location within the answer engines as that signal of authority and sort of citability.

Yes. So, you know, just because you'll be good at SEO doesn't mean you'll be good at geo. Now, AI may still select your content if it's very well optimized from a search engine standpoint in the absence of content that is not available to AI that meets the AI threshold for what we call in data science information gain.

Right.

Okay. So information gain is how much value does your content add to the field of knowledge on the internet on that thing you sell or have expertise on. So in the absence of that level of content required for consistent AI presence, it will of course select from what it can, what is available, and that will come from SEO content.

So this concept of information gain, is super important. And, and you and I have chatted about this a little bit, and, and you sort of referenced tools a minute ago. And so there are a lot of people out there that are spending a lot of energy and effort spinning up content using AI to generate content for AI to rank. But you just said something that really sort of reveals the fundamental flaw of that, which is If you're asking, you know, the audience to tell you about a topic, you're getting sort of the most watered-down answers and versions of that topic back. So if you're using AI to generate content, there's no information gain that's being then shared in that content that you're putting on that site. And so it's not likely to impact the answer engines. Am I understanding that correctly?

That's absolutely correct because You know, AI, let me ask a business owner this question. You know, what reason would AI have to change the status quo and recommend your business within the new status quo, right? What reason would you give it if you're not giving it anything new that adds value to its models, which are used by its users? Okay, so your content, if you want to change the status quo, you have to come up with something that teaches AI something new. It has to be something, unseen insights that cannot be obtained anywhere else on the internet, because, uh, if it can be obtained, then, well, there's no reason to give you the credit, right? As I said earlier, to be recommended, you must be cited. So It is. So information gain is what we describe in data science as, you know, what is that value that your content gives to the overall field? And this is something that can't be obtained by prompting AI, even if you use DeepMind paying $200 a month. The thing is, if AI sees content that comes from AI, even if it's from a different AI, It's going to be— it's probably got that data itself. So it's going to think, well, if it was a person, it would think, well, you haven't really taught me anything new. What you're trying to do is feed me water, but it came from my own sweat. So I'm not going to drink my own sweat. Sorry. So sorry for that rather visual graphic analogy.

No, it's great.

But I think it's rather fitting. So you really have to give Real food and nutrition that, that, that, that adds value and gives life to AI. It becomes better as a result of your content. That is true information gain, and that is what's going to change the status quo so, so that your business gets recommended.

So to use an old school sort of framework within the SEO space, you kind of competed with tonnage. You know, so tonnage of information about your topic and that was high enough quality that you could get people to link back to it, but wasn't always that it was novel or sort of new information. You could really sort of compete by creating a tonnage of it that was nice enough, maybe in the way it was displayed, or there was some interactive calculator that was a new way of sort of showing it again, not really necessarily a new way as much as just an interesting way. And you're saying that the answer engines reward that novelty. So how does a business that, you know, doesn't compete or that competes in a marketplace where maybe they didn't invent the, you know, the solution that they're selling, how do they create that novelty to then, you know, sort of have that space?

To create that novelty, you need to be doing research. So you would commission research that, A, your buyers would find interesting. Preferably, if there's a bit of a linkage in that, if your buyers find it interesting, then journalists will find it interesting. And therefore, that research has value. And, you know, this is why I say your research for AI content should follow people. Not keywords, because if people are interested, then they're going to be prompting about it. And if they're prompting about it, well, you've you've created the answer that makes your business the answer. So you know there's there's different levels of research. You know the very extreme end, you could be going to Gallup, which is an American market research company. You know, so if you don't have a Tesla or you know an infrastructure like I have, which is incredibly expensive to build, then. You will have to go, you know, that, that makes it extremely cost-effective. You know, like for example, we can ask, we can produce reports that, that, you know, that have over 10 survey-style questions and don't cost anywhere near $10,000 to $30,000 per report, you know. And that's before you hire someone to analyze and go through that data and before you can hand it over to the writer. So technology makes it extremely cost effective if you've got the raw data like we have, or if you're a business, you know, and you're trying to do it on your own, then you would definitely hire some kind of market research company in order to commission research. And then you've got the data in order to produce something novel that AI will like.

That's interesting. Well, and it also, it aligns, I think, with some approaches to the way that really smart people approach the way they build their brand, which is sort of through the idea of kind of creating a category where they have a novel point of view. And to use an example, we'll come back to the microphone. Lots of companies make microphones. On the outside, they seem largely the same.

Right.

But they actually are quite unique in different ways. And so we'll take this one because it's right here. They've obviously chosen to have some form that's important to it. And so if their point of view is we make microphones that look cool and sound cool, and then if they create content around how they make their microphones look cooler by using new, you know, sort of products and materials or referencing retro designs that aren't available from, you know, other manufacturers today or All of the different things that can be unique about their approach to making something that's largely a commodity, that unique point of view can then feed something novel to those engines that then when someone is looking for that type of microphone, they'll get recommended through those answering machines.

Yeah.

Is that right?

Yeah, 100%. I think, yes, depending on how competitive the market is, because if it's incredibly competitive, you can be sure there's going to be a lot of novel— Sure.

It's not really that novel. Yeah.

Well, I mean, it might be novel, but it might not be— it could be the best microphone in the world, but if nobody knows about it, then, you know, if it hasn't got the authority behind it, so, you know, you still need to invest in the authority even after you create this amazing content. The other thing to mention is that, and I think your point about novel, you know, having a novel perspective is incredibly important, which goes beyond AI search because in a world where everybody's niched down, how else can you compete? You have a perspective.

That's right.

And you don't try to win all your customers, but you win the customers that you're gonna get along most with and who are gonna recommend you and make your business highly successful. But going back to AI, AI is ultimately an information engine. So I'm glad you touched on this because we actually do both. We provide the scaled insights, but we also actively encourage our clients to add their novel perspective on the data. So yes, it's done for you, but you get so much more return if You're saying what you think about the data, and it's even better if you disagree with what the internet thinks, right? Because you're showing your thought leadership report is less one-sided and more multi-perspective. And think about it, this is why AI is hooked on community forums like Reddit, X, because it loves the multi-perspective. you know, it's more critical in terms of its evaluation or assessment and the discussion of that topic. But if you can replicate that in your own thought leadership, then you're gonna have a lot of success in GEO.

That's fantastic. So Andreas, you're obviously an expert in a ton of super interesting topics, and I could drill you on it all day, but I gotta be mindful of your time. But before I let you go, like, give us the pitch a little bit. Tell us, you know, who you, who you work with and how you help them at RDOS.io. And tell us a little bit about that.

Yeah, so look, we're helping businesses get chosen by their prospects when they're searching in AI because AI is choosing winners and losers in search. And we're guaranteeing our clients are the winners. Okay. So and we do that by producing content cost-effective. We're doing that by producing content cost-effectively that teaches AI and the world something new. Now, unlike SEO content, which as you've alluded to earlier, you know, is a mixture of desk research and making a better version of what you're currently seeing in the top 10, The kind of content that we're producing is not just appealing to AI, but it's also getting discussion on LinkedIn, right? So this is really important. I'm a terrible salesman, as you can see. I'm more into the science of how it works, but it has to be that way because I want to deliver. So, you know, your content can't just appeal to AI. It has to appeal to people as well. And we promote it using digital PR. and community discussion forum outreach. And bringing it all together, that ensures that our customers are going to become the answer in AI. And that's money-back guaranteed.

That's fantastic. Well, Andreas, what's the best way for folks that are listening and want to get more of this out of you? What's the best way for them to engage with you?

Well, they can reach out to me on LinkedIn personally. I'm not using an assistant to manage my account. Or they could simply come to— And, you know, if they want to get in touch, you know, please, please book a call or send me a message.

Fantastic. Well, Andreas, thank you so much for taking the time to join us today and share these really valuable insights for us.

Thank you for having me. It's a real pleasure, Mark. And love the questions, by the way.

Talk to you soon. Thank you for joining this episode of the B2B Growth Blueprint podcast. We hope you've gained insights to grow your business. If you enjoyed this discussion, please rate, comment, like on your favorite platform, and share to others who will find it useful. Do you know a CEO, founder, or leader at an early-stage B2B services, tech, or SaaS company who would like to be a guest and share their story? Send us a note. Let's schedule a conversation.