Help with Your AEO: Boosting Brand Visibility in the Age of AI with Waikay – 299

I’m very excited listeners. We have ah a very special guest, but I have to give you a word of caution. I don’t think you’re going get any, not anything. You won’t get as much out of this unless you actually go to our YouTube channel and and look at his presentation, his product. I’m very excited about the future.

We’re in a world of possibly transitioning from SEO to what we call AEO, which is artificial intelligence. uh engine optimization so i’m just going to get right into uh into the introduction here and uh and tell you a little bit about guests so again i’m excited and honored to introduce dixon jones he’s the ceo and co-founder of inlinks an award-winning seo software company recognized for its innovative use of entity-based optimization and natural language processing. Dixon is a leading figure in the internet marketing and SEO community.

Oh, let me stop you there. Let me stop you there. It’s a lovely way to introduce me, but i yeah i i’m i it’s fine. I yeah I used to run Majestic. though I used to be marketing director of Majestic.com. I’ve been in the industry for 30 years. And let’s not bore everybody with all the details. ah appreciate everything.

Hey, I tell you what, us us nerds us nerds we will certainly appreciate this. So I’m going to just continue here, despite I appreciate your appreciate your humbleness, but this is these are some serious creds. So um now I’ll explain this to my wife, and she’ll roll her eyes, but but everybody else that knows that’s in the know in the community, it’s like they say, when you know, you know, um they need to kind of hear your your youre your accomplishments because it and think it’s really important.

So… um he’s He’s received numerous accolades, including lifetime achievement award in search and has been named search personality of the year and an outstanding technology individual of the year.

Throughout his career, Dixon has founded several successful ventures, including ah Initiative Unlimited in 1989 and Receptional Limited in 1999. One of the UK’s first search consultancies, he played a pivotal role as marketing director and later global brand ambassador at Majestic.com, also MajesticSEO.com, where he helped the company earn two Queen’s awards, for innovation and export. He’s also been involved in DHJ Ventures and serves on the board of Zen Intelligence. Dixon is a regular keynote speaker, and this is where we get nerdy, at major industry events like Brighton SEO, PubCon, SES, and SMX, and is widely recognized for advancing the understanding of entity SEO and search marketing.

Most recently, and this is really important, Dixon has launched Waikay, short for What AI Knows About You, a groundbreaking tool from InLinks that reveals how major AI models like ChatGBT, Gemini, Claude, and Perplexity perceive and represent your brand.

Waikay enables businesses to analyze their brand’s AI footprint, fact check AI generated information, and benchmark their reputation against competitors helping brands proactively manage their digital presence in the age of driven search.

Again, I’m very excited. And Dixon, if we could just jump into your product and we can show everybody that’s involved in SEO, how ah you’re innovating and breaking new territory here.

We sure can. And thank you very much for all of that. Anyway, thank you so much for the the introductions, Bob. ah Honestly, that just went on for… um um’m ah I’m getting old, really.

But the Queen’s Awards, I mean, I know you guys are in the States and I’m in the UK, but that meant I got to go and see the Queen at Buckingham Palace twice. So those ones, I like care. you know they they they were They were probably…probably the highlight and uh and i didn’t do that all on my own so um but thanks for the introductions uh and yeah i’ll just dive into way k i’ll try not to spend too long on way k guys and we can talk about it and have questions afterwards um and uh yeah as you say it stands for what ai knows about you the reason that we built it is because um we were looking at uh the various, there’s not that many tools yet that are properly trying to help us to influence ah SEO, ah SEO, if you like, within LLMs, within ChatGPT, within Gemini, within Claude, et cetera. And the ones that are there seem to be thinking about it in a way that isn’t going to scale. It’s kind of trying to trying to take rank check-in and just dump it into into um AI tracking.

And we found that that didn’t work. So we’ve come up with a different approach, which really looks at the way in which the LLMs are looking at knowledge. And so we’re looking at yeah what AI knows about you, where AI is getting that knowledge. um Can you fact check that knowledge? And and then you know how can you fill in the gaps and the knowledge and and influence the that the knowledge graph or the the information eventually.

So there is, by the way, a free account. So you can you can get a free free version if you go to slash free. But it’s at wayk.io. Impossible to pronounce, impossible to spell, and impossible to see it because it’s not even the.com domain. But the fact that it stands for what AI knows about you means that you can kind of get there pretty pretty quickly.

um And what I was going to do was jump in and just do Intrycks as a project. And I just i did it a little bit before. So if it if it goes wrong, then i I’m going to go back to this one. But let’s go and try it with Intrycks. And and what what we’re going to do with any um come with any brand, what we’re trying to do is um

Is is that spelled right? No, it’s not. Yeah, it’s it’s a little different. Yeah, Intrycks. I got it wrong. Sorry. Speaking pronouncing words and names, it’s like, if way to is bad, this is 3X. It’s very difficult for people. Fair enough.

Okay, so so so I’ve got it right now. So you’ve got to choose a market because we do a bunch of Google search lookups and things, and you can’t do that without specifying a market, but that’s fine. You’re in the United States. We know it’s fine.

So what we what we’re trying to do is find out what all the AI knows about Intrycks in the it in kind of context to your competitors. We do need to choose a couple of competitors.

These ones look a little bit too generic. So I was going to go with woollymammoth.com, or have you got a couple of couple of locals that you that you might want to compete with?

Not offhand, and maybe that’s because I… I stick my head in the sand too much. So I’ll tell i tell you, let me just dive back in just for the sake of time here. So so you choose two and then you build a project. But I’ve built the project anyway.

And I chose themightymo.com and woollymammothdesign.com. So they’re both… say that they’re based in in in Minnesota. so So I pulled those out.

the first thing that the system does is it asks Sonar, ChatGPT, Gemini and Claude, what do you know about Intrycks.com? What do you know about WoollyMammoth.com? We build a knowledge graph of concepts that are on your website so we’ve crawled your website and we’ve crawled the other two websites as well and we’ve understood all the underlying concepts that those websites are talking about we then have asked the llms what do they think the website is about so now we’ve got two data sets of information because we pull those down into the underlying ah We call them entities, but topics is a better word, really, of of ideas.

And so we’ve got this map of things that you’ve talked about on your website and they the um things that the LLM has said you’re you’re about. So if we have a look at something with Xero on here.

So I’ve got to explain TD and g GR. So training data, TD stands for training data and GR stands for grounded data. So ah anybody that’s been playing with the with with AI, um if you’re using Google Gemini, then the chances are you’re using it in AI overviews on Google. And so what it’s doing is it’s blending and approve a pure LLM with training data that it got from a long time ago potentially and blending it with RAG, retrieve search basic search results. It’s finding finding using the training data to get some information about where they need to look and then it’s getting search results to help augment that data on the fly.

Not everything does does use um training data or grounded data. So grounded data means it’s got a search lookup. So that’s why all three of you have got zeros when it comes to Google Gemini as a score.

And if I click on it, you can see the actual response. It said, I do not have access to real-time information. That’s because it can’t do a search lookup, including websites like Intrycks. So it doesn’t know that much about you. So it’s given a score of zero for you and your competitors. But in the real life situation, Gemini is very good at pulling in its own search results, at which point you do great. And if I click on here,

It says some interesting stuff. It says you’re a digital design company. It’s got a whole bunch of different things here about who you are, et cetera. And all that is is put in and Now we know that you’re that that it’s got quite a lot of information about you.

What’s interesting, what I found quite interesting in here is ah that… um You’ve got to a bunch of facts here.

And ah okay I saw earlier on something about Sweden. So one of the search engines is is fighting it’s finding you and thinking you’re a Swedish company. I can’t find out which one it was now. did to do Yes, here we go.

So ChatGPT. has said Inlix is a Swedish digital marketing agency specializing in SEO services. This is interesting and I think shows exactly why companies need to pay attention to this stuff now because the machines are hallucinating and they’re getting things wrong.

And to say that Inlix is a Swedish digital marketing agency is clearly incorrect. You just got look at the website and that’s not correct. So one of the things that um we do at WakeA is make it very, very easy for you to start checking all the facts that are spouted out by all of the AI systems.

By the way, you’re not alone. InLinks, by the way, it said InLinks was a rank checking service and we’re not. um And I did did one on Majestic and it said Majestic was dissolved in 2021. And that’s because a UK company called Majestic Wine had dissolved and it and it got confused.

So this fact checking… of of of AI is an important thing for brands. And the bigger the brand, the more important it is, but even small brands should be checking this stuff.

And so what we’ve done is we’ve turned all of the output from ah from from asking the LLMs things and turned them into individual facts that you can very quickly go and see.

Right. Intrycks is a digital marketing agency and website specializing SEO. Check. That’s fine. Fine off the list. That’s great. Vast majority of these is going to be fine. But occasionally you get something that’s that’s incorrect, like the Swedish digital marketing company and flat check GPT saying it’s wrong. So we can flag that so you can easily flag ideas.

And now these reports The idea is that you run them over time so you can run them monthly or weekly or daily or whatever. um So when you do run them again, then we’ll anything you haven’t flagged will just wipe because otherwise it just becomes too much information.

But but the the the fact checking is the first important thing. The second important thing is these numbers. These numbers give you an idea of how accurate the information is in different search engines. Both for you and for your competitors. and now we’ve got some concept of competing brands. um we We can also, all of these all of these things kind of come back with a bunch of um sources. So for example, there wasn’t one in here, but let’s go to sonar.com.

So here they’re saying all of the all the yeah URLs that they’re seeing in this response um that they’re citing as ah as relevant. So we can then have a look at all of the…pages that that that are on your so in your reports and they’re not just necessarily your own website, there might be other websites as well, but we can check to make sure that they’re all there. They’re actually redirecting with 301 redirects, which ah some of your users might may not know. i don’t know if they know about 301s or ones and not, but let’s say they do.

I’m assuming. But if there’s 404s in there, then that’s a clear signal that you’ve got to change anything. um and and And again, with your competitors here, we can go and start seeing which pages are mentioned in competitors’ um citations.

So, for example, we’ve got Capterra here with Willie Mammoth Design mentioned on Capterra.com. That’s a signal that says, hey, well, um the LLMs are using this Capterra stuff, that’s very easy. You could go and get yourself on Capterra and maybe you could also next time be seen as a, ah and use that to to get mentioned. So you can start to see some places where you can get your stuff talked about to interrupt the LLMs for next time around. That’s exciting. Yeah, that’s that’s crazy.

So it’s cool. So now this topic stuff I’m going to move go move on from because it gets a lot more interesting when I start to create a report around a concept. So right now, that was a brand overview report, which you could get for free, actually, you can just close sign up and get a free one.

But where it really gets interesting, I think, and now I’m working in real time here, is when we start to say, right, what do we want to be better at? So, you know, this is kind of like, you know, if you think about keyword, you know, getting to number one for a particular search phrase in SEO, what you really want is when when you ask the LLMs, you know, what’s the best search engine optimization company or web design company, then you want, you know, intrigues to come up. That would be the end point of the thing, which I think is, you know, some years off because we needed to learn a lot about how LLMs working.

But if I choose web design, It’ll create reports saying, OK, what does Intrycks and those two competitors, ah what do the LLMs know about Intrycks in the context of web design?

So now we’re not just looking at your brand. We’re looking at something much more contextually accurate or something much more of a business focus. um that make sense?

It does. It does. Yeah. Right. Okay. So again, we’re we’re going out asking them to to to give their responses and we’ll have a similar looking screen at the start. But so my extra stuff is going to turn up, which is going to be very, very important. um At the same time, we’re um augmenting the knowledge graph that we’ve already built about your website. So any web pages that we’ve seen in other reports are We’ll go and crawl those in the background and make sure we understand those so we have a better view of you um as we as we go on. So the more reporting you do, the more accurate we have an understanding of your actual um actual focus. So while that’s going, let me show you this mention thing as well. Because this mention thing, all of the yeah all of the um yeah URLs that have been mentioned so far, um in those lookups that we’ve already done and now mentioned here.

So this is a good way to go and see you know the capped errors and stuff of that. And you can break it down by domains. You can say, look, that’s interesting. Other domains, WPA engine has been, WordPress.org has been mentioned a couple of times can see the Mighty Moe has got a profile on WordPress.org. That might be an interesting thing if you’re ah a WordPress shop. Other domains here, you get to see the Capterra stuff. But now you start to also get an idea of market share from these numbers. Market share in the context of LLMs. So LLMs are mentioning you 18 times, Willie Mamma 13, Mighty Mo eight times. That to me is not a bad um representation of how well you’re relatively represented in the LLMs. um So, right. So now we’ve done this ah this web design.

Again, we’ve got scores now, scores we can dive into. and ah One of the interesting things now is that I don’t if you saw on the last one, I went quite quickly, but the Mighty Mo wasn’t understood in ChatGPT for its brand.

They had a zero and they also had a zero for Woolly Mammoth design, I think. But in the context of web design, The LLMs now understand which which Mighty Mo they’re talking about and which woolly mammoth they’re talking about. And they’re not talking about just general woolly mammoths.

They’re talking about this one. and So they get a much ah much better idea of these businesses in the context of the products they’re selling. And that’s natural, right? you know Hopefully, if we did did you know what’s in tricks.com in the context of London, you wouldn’t do so well.

But because it’s in the context of web design, it’s got a pretty good result. Again, not in the original training data from from from ah Gemini, but 82 out 100. That’s a pretty good score. so we know we know what it does.

And we know. And did you did you take over Fargo Web Design kind Company at some point or is that an SEO page? There we so ah So we we can we can see bunch of facts here, um which ah which hopefully are are right, but you can flag them again if they’re wrong.

But the the great thing here, and I’ll stop talking in just a second once I’ve shown you this button here. This button here is the action plan.

And this action plan, I think is where we really can um start the process of actually optimizing within LLMs. So what this is doing is recommending things like topic gaps. And what it’s doing is some really fine-tuned ideas for how you might be better represented within the context of of web design.

And the report is not that long. This this is a report, that’s a whole lot, that that that’s it. And they’re really strong bullet points. So firstly, go on to strengths, Then we’ve got Intrycks a strong location-based SEO and web design.

You have strong location-based SEO and web design pages. um You’ve got multiple city-specific pages like Fargo. You’ve got diverse service offerings, technical SEO content, etc. So these are your strengths.

This is Mighty Mo’s strengths. They’ve got a good court for portfolio and showcase collect selection ah section. And then Willie Mammoth, a really good about section so we can understand where they came from, et cetera.

But now we get onto the content gaps and we can see a bunch of structural gaps, thematic gaps, what we consider critical content gaps and then significant content gaps and then and under mentioned um topics.

And these, I think, are really interesting, but we can tighten these down even more. and go to recommendations. And this comes to six things that our tool says that the LLMs think is well you you would need to need to get better represented in LLMs for the context of web design.

Six ideas, that’s it. Six suggestions, only one sentence long each. So for example, So around Minneapolis Web Design Service, we want a a dedicated Minneapolis Web Design Service page similar to your existing city page. So you’ve got city pages.

They’ve surfaced. The Fargo one did. But for one some reason or another, the Minneapolis one didn’t. know if that’s correct. I hope it’s ah it’s it’s insightful. um And then develop a comprehensive portfolio page ah because because um ah mighty Mighty Mo or whatever it was um created, ah has has got a good um showcase page.
Also, it gets very in detail on existing pages. So on your web design page, it says, ex exist enhance that existing web design page to include detailed information about the design process, including some research methodical…method but yeah methodology and revision policies. I’m talking very quickly um and expand the current web design ah content to emphasize the approach. So these two things, not as urgent, but we think they’re interesting.

And then a couple of structural improvements. um a bit more clarity in in your packages and offerings and they integrate some case studies into the service pages.

So these are not a large number of things, but I think that they’re pretty um powerful. And then we reduce that even more into here, recommended steps. Do this, do that and do that.

And then if you wanted to go and also get the chances to get listed, then go and have a look at getting yourself into WP Engines, you know, company agency directory or something as well. So ah that is it in a nutshell.

Did it make sense? And was that interesting? Oh, very interesting. I mean, it it’s, you know, I think we’re going to talk a bit here about where where things are going yeah in terms of search and search engines and um how that transition is taking place, what the timeline is for the that transition.

One of the questions I have is, is you know, what level brand or what, how relevant is this to, let’s say local, let’s say the plumber or the, or the bakery or whatever the case is versus. i like the resume yeah I think if you’ve got content and competitors and you’re proud of your business, then it, then it’s useful. I mean, this thing only costs, you know, $20 a year. So it’s not as if we’re, you know, spending a, sorry, $20 a month, you know or starts at a month, you know and that’s,

OK, that’s you know only eight reports a month. But, you know, a small business can do, you know, you know, they they sell bread and they sell cakes. They want to do that rap report every every three weeks or you know well every month. So and then then see their brand.

But what you you you definitely want to know. if the LLMs are hallucinating because um they can pick up data from anywhere.

It’s not just your website. One of the things with a big company that we found was Microsoft, for example, are blocking Gemini from crawling their content. They’ve made a policy decision as a company that they don’t want to educate Gemini with and make it easy for them to to to to um know everything about Microsoft.

The irony is that Gemini comes out with perfectly good information about Microsoft, just what everybody else in the world says about them. So they may not come off better because they’re no longer in control of their narrative because they’ve actively said, I’m not having i’m not having Google on my yeah on my on my site, thank you very much.

And I think that that was a well-meaning decision that probably wasn’t thought through ah all all the way and and maybe something that they want to to revise.

It’s also a problem that Google’s gonna have to think about when ChatGPT and Bing starts you know coming around Google’s content as well. But if you if you can’t control your narrative, then um you can expect a lot more hallucinations.

Yeah, yeah. They’re almost these LNNs hello LLMs are almost like nation states, right? I mean, they’re going to get, I mean, it’s going to get interesting for sure.

And what’s great about your tools is is it it just helps us create that clarity as best we can as small entities to and say, okay, this is who we really are. And yeah you don’t have to be big do ah to to to to to need it. um it It may be that you’re not looking for web design. You’re looking around web design in the context of Minneapolis or something like that. you know And so you you’re you’re further refining the topic that you’re actually searching about. You could have typed in any word there to build that report.

We suggest entities and subjects, proper subjects that we can relate back to Wikipedia concepts, for want of a better word, you know um ah but encyclopedia ideas.

I think one of the problems that SEOs naturally fall into is that and particularly small businesses may fall into is that they come up with a quite a long string of something that’s very narrow and unique to them so yeah they’re ranked number one for it but um nobody’s actually necessarily typing that in but that’s what it means you know that’s what it means you know you you don’t necessarily want to do web design for somebody in in uh los angeles because because you know you would rather meet them face to face because that’s what we do in Minneapolis you know and so you see I’m a man of the world I’m a man of the world and so

So you don’t want to have to always say Minneapolis if your only customer is supposed to be a Minneapolis. The LLM’s got to understand that. You know, it’s got to kind of get, you know, Minneapolis, Fargo, you know, but not Los Angeles.

And I think that that is an interesting place for us to think, you know, thinking in terms of what knowledge do you want these LLMs to have and what do you not want them to have, actually?

We’ve been playing with the idea that in the future, our about pages may not just say what we do, but they might also say what we don’t do so that we can inform the LLMs to get get things clear so they can yeah see exactly what ah what they are and aren’t doing. let me just turn that TV behind me off.

Yeah, no that’s interesting. And oddly enough, we do you know we do work in Australia, Germany, did some work in Ireland.

And ah in ah you know for a ah Yahoo Midwesterner, it’s kind of weird that we’re doing you know, business all over. But I mean, the, yeah I’m sitting in the middle of the middle of a shire, a shire town in shire village in, in, in, in London talking to a guy in Minneapolis. It’s a weird world now and that’s yeah I mean, the world is, is flat, so to speak. And, and, uh, you know, we employ people all over the the world and because they’re talented and, yeah and and very good at what they do. And and I never would have thought that 30 years ago when when I started various businesses that, you know, my dad’s basement that we’d be talking about, you know, these these things. And and so let’s look yeah if I can bring it back to, you know, what you see the future You know, as as i you know I did a little research on Waike and tried to grasp as much understanding of how this is going to impact things. And I’m very excited. I mean, today, i don’t think any of us are going to go out. They could. But I mean, you know, and make our fortune in a AEO, if you will. But how do you see how do you see those search engines shifting? Like, you know, Google is yeah know is is what the largest search engine at this point YouTube’s the second.

Yeah. um which is And then all the large language models are are gaining traction. What do you see going on there for a timeline and and what do you see? Well, I think I think over that there’s a lot a lot of blending already.

um yeah so so So what happened, so that so the history as I saw it, the first time I really started paying attention to this was at a conference in Brighton, only you know about about a month after ChatGPT launch. What was that? twenty Was that 22 or 23? 22, yeah.

Yeah, not so far ago anyway. And it was about a month before. And I was actually sitting next to John Muller in this conference. and that’s and And this guy talked about you know LLMs and showed us all the output of LLMs.

Bloody hell, that’s ah that’s impressive. um and and ah And as we were walking out, said to John, so why don’t you take those 10 blue lines and just make a story out of them and do that instead of search results with all the information but in a story?

And he said, oh, I don’t think that’ll happen for a long, long time. I think though you know we’re very traditional search engines and stuff look look forward two years and and and and then they’ve got these AI overviews that are pretty much, they’re not taking the same 10 blue lines, they’re taking different different data.

But I think what is really interesting in in that interim is that The the chat GPT when it came out and Gemini’s training data when it came out was fine as far as it went, but immediately it was static data.

So the data, unless it starts getting updated, um only knows what it knew at the point of launch. So as soon as you change a president, then, you know, life changes, you know, so so so that information starts to degrade in usefulness over time.

Also, when when these systems initially came out, nobody had any concept of trying to manipulate that background data. So so they they could take all the books in the in the British Library and photocopy them or whatever they were going to do to get their data, proprietary data, public information data, take it all, and all the people that were trying to do things there were not necessarily trying to influence the the AI. So was…

ah less likely to make hallucin make factual errors because no one had tried to put anything in there to make it a factual error. But at the same time, it did did seem to hallucinate because it didn’t understand in that training data that this was an authoritative piece of content on that subject and this one wasn’t.

So it’s it’s going to random pages that that may or may not be appropriate. um and may not be the best source of information. I think what has happened is that they’re seeing that now.

And this is why RAG, retrieval augmented generative search, um it has has already become a big thing. So the LLMs would like to give you a straight answer if they could.

But if they’re not sure, then they’ve got the opportunity to go on to Bing or go on to Google or go on to Duck, duck, go or whatever and pull back some or ah some search results to hone in and crawl those bits on the fly to augment the information and give you better data.

That gives them ah the search engines have spent all that time working out the difference between a good page and a bad page. So they’ve got this kind of concept of authority in there that the LLNs don’t have. So that’s reduced some hallucinations, introduced some other problems along the way.

It’s already moved very, very fast. So when I went to Brighton two weeks ago and we had the Waco stand there, we just sort of launched the action plans that week. um The stand was really, really busy and they were all saying our customers are asking about.

Yeah. AI and, yeah you know they they they they they there is you know, the problem with SEOs is, you know, we like we we we look at bright, shiny, learn twinkly things and since you come along, everybody wants to go there. you know, well, something new has come along.

We know it’s fundamental. We know it’s going to change our our world. And if it hasn’t already, it’s changing changing your world very, very quickly, isn’t it And and um so it’s it’s it’s what all the customers are saying they want to to focus on now. You know, um if we haven’t fixed SEO now, well, it’s going to change very, very quickly now. So let’s just focus on the next thing and and and and go there because ah I think as as human beings,

We are already starting, a lot of us, a lot of us in this industry, and I’d imagine a lot of people just just who who, you know, do not do any tech stuff at all, are starting to press things into their computers. You know, I saw Facebook um has come out with a llama, but you can ask llama. So basically, if you’ve got Facebook, you can now ask llama or anything in Facebook, for example. Well, that’s…that’s an AI doing its thing. So so this is ah this is this is coming to you know a phone near you, whether you like it or not, and you won’t even know that you’ve changed your habits.

So of course, businesses need to stay on top of this because You do not want your you know your bad reputation you know a bad reputation to be to to be propagated.

The LLMs learn in a very similar way to humans in that we just listen to the wisdom. ah The wisdom of crowds can go terribly wrong. We all know that. And and so so you don’t want a person who is perfectly interested in buying your product to ask a machine that in between misinterprets the product and then gives a different representation of yourself.

That could go terribly wrong and it’s something easy to track right now. And oftentimes it’s easy to fix right now. I said that InLinks was seen as a rank tracking service. We tracked that back to the original LLM output that that that had said it.

It turned out to be a website that I had never heard of that had done a mostly positive review of InLinks. um And so I asked my my my VA to reach out to them, see if they wanted to.

But they been put this vote this false feature in there. She reached out to them. They immediately changed it. um They wanted accurate information. And it was the easiest fix in that I’d ever had, actually. And job done. Some of the fixes are not going to be that easy, you know, but. ah but some of them are. It’s already here. I think it’s already here.

Yeah. So to to your point, you know, on some of the, it’s about protecting the knowledge base, right? So the, or, or, or affecting it or or making sure that the knowledge base is pure.

I have a couple of small businesses and we created a um ah phone voice AI. So when you call after hours, it can navigate and give you pricing and set up appointments and everything else.

And that was a struggle. It took probably three, four months because… you know, we just chose the the website as the general knowledge base, which was a huge mistake because you, you know, 600 pages, there’s, there’s blogs, articles and stuff that the AI get got confused and, you know, pulled information from that versus a very contained knowledge base. So, you know, this is our opportunity with your tool to, to go in manage our knowledge base and, and, and where it’s hallucinating, like you said, go in, change it, change what you can, because

you know, the struggle is going to be other people’s, you know, websites and articles that it’s, you know, you can’t necessarily control that, but you can do what you can to, to affect that. So, and to your point, you know, this, I came across to you based on just by happenstance, listening to another podcast, I think it was the edge of the web.

Um, oh yeah yeah and, uh, and basically, um, you know, did some research. And in that time, while I was doing that, a client reached out a medical clinic and said, hey, what can you tell us about ai managing AI information and da da da da da it it was just perfect timing.

But to your point, this is going to go crazy because I’m going hand this podcast out to a bunch of my constituents and they’ll have you on their podcast and we all need to get educated on this as quickly as we can.

I’m delighted i because scott I think Before we can make too many recommendations as to how to how to manipulate it, we’ve got to first or influence it. I should say influence, not manipulation night, you know, ah so changing manipulation.

SEM SEO. Anyway, so I. and you know before we can absolutely come up with all the things that we know about seo seo we know that the h tags are important we know that the structure is important you know internal you know the link internal linking and external links you know in and out are important we know citations you know so we we’ve got a pretty good idea of things that we have to do because you know People like you and I have been in the industry a long time and and we’ve learned and we’ve learned by trial and error. And, you know, but this LLM stuff is really new. there The LLMs themselves are learning in a manner which is very hard for any human to ah to to um comprehend.

But we can still start to record information and understand patterns and behaviors. So we’ve seen some really interesting things. We’ve seen, for example, some of the some the because we’re making a lot of API calls and LLM lookup calls now and some of the times they they they show their reasoning. So they’re showing their reasoning and logic and some of the things.

What we’ve seen is someone thinking in a different language. And that that I found very interesting um because when I think about it, thinking in Chinese characters is a lot more efficient than thinking in English because there may be more characters, but they mean more concepts. And so they can actually get to a a reasoned point in ah in in a different way. But it’s not always English. I’ve seen Spanish in there as well. as ah Sorry, it’s not always Chinese. you know I’ve seen different ah different thought processes and they’ve just jumped into thinking or showing their thinking in a different language.

And that’s kind of weird. Maybe it’s because we’re outside the US s and we’re sort of, you know, my business partner’s French and stuff. So, you know, we kind of see stuff, but I don’t know why I saw Spanish or or Chinese, but I’ve definitely seen these in in the thought processes.

So they’re thinking and in weird ways, but When, and so e I apologize to anyone that’s listening only, because I’m gonna use my hands now, but when an LLM is asked the question, you know, what is, you know, give me 10 good web design companies in Minnesota, then it’s gonna, it’ll start by taking that in, turning it into its underlying constituencies. What are we talking we’re talking about? about design and Minnesota, really, and good, you know or or best, or some concept of best. Those three things. and so However you phrase the question, it can break it down to that concept, at which point it can start by saying, well, okay, let’s go and look up a directory of companies in Minnesota and see which one’s web design companies, for example.

Or it could say, right, let’s go and have a look at web design companies and then see which ones are based in Minnesota. Or it could say, you know what, let’s go and have a look at the concept of web design and what it means to be good at web design and come up with things like UX and these kind of concepts and stuff and then see which ones of those ideas then are are are mapped onto the web design companies I’ve seen in Minnesota.

So it’s got these different thought processes and they they run parallel thought processes just like a human does, ah well, a little bit like a human does. And then as I go to the next layer, the next layer, the next layer, they’re kind of looking at more and more information.

And we as businesses and as optimization experts have the opportunity at every layer, as long as we know the sorts of things they might be talking about, which is what ah hopefully wakek is trying to do in the topic app stuff.

We can say, you know what, the reason you need to talk about this page over this concept over here is because at least one of the LLM processes with one of the competitors, it went through that process and was talking about this concept.

And it and and it it seemed important to the LLM on this occasion. So why don’t you interrupt it and write some content about this? Because it’s part of the thought process. And then when it gets to what’s the best, my hope is that eventually we’ll say it will say, well, I’ve kept on coming to this brand as I’ve been doing my thinking process. I’m going to put them on the list.

Yeah, it’s… it’s I just think it’s it’s one of these things that probably I’m guessing a year from now, there’s going to be somebody in my company that this is all they do.

They monitor this for a client or are all our clients. And it’s part of our- I think that’s going to be the step number one. You’ve got to monitor it first and you can’t you can’t make realistic ideas of what to do um until you monitor it and and and see what’s consistently coming out as incorrect in the LLMs or where it’s making mistakes or seeing where there’s clear opportunities. The LLM has not seen this part of your service offerings. How can we make it more obvious to the LLM for next time?

And then there’s a whole new thing about that that training data, for example. So Gemini, yeah the example we go, none of us, none of the three were in the Gemini training data. If I went to, I did one yesterday for Nike, Adidas and Fila.

don’t know if you know Fila, but anyway, it’s three footwear companies. um and And Nike and Adidas were in Gemini’s training data, but Fila wasn’t, for example.

And that’s ah they’re a big company and they’ve been around a long time. So it’s not it’ it’s it’s everyone’s problem. But eventually, Google is going to want to update its training data. So we need to be thinking about that now as optimization people, because um back in the day, Google used to update its index once a month.

And we all called it the Google dance. And we just waited for it to happen. And it took a week to do. And we could see all the rankings going up and down, which way is why it’s called the Google dance. and and so we we were always had two sets of you two opportunities to optimize one got to get indexed to got to get to the top of the search results. This getting indexed part, if you aren’t indexed, you’re never going to get to the top of the search results.

And right that is already the way with training data versus rag retrieval ah search search based grounded data. And, um and clearly, if you’re If you ask a an AI, you know, what is seven plus two, it doesn’t need to do a search lookup. You know, it knows the answer and it doesn’t have to go there.

The more complicated the query, the more chance it’s going to need to go to ah search lookup. But if you’re already in the training data, then you’ve got much more chance of being represented in the search look up some way, shape or form. And so it it gives you a double a double shot at the cherry, I think, you know, two two shots of the I’m mixing my metaphors, but shots of the prize. Can you can you speak to the training data piece? So I’m i’m a little confused on how, how well that you and And I think people are going to want to know, you know, obviously, you know, it seems like people are are getting into the main, the various like large language models fairly easy, but the training date data seems to be a trickier deal, obviously for an entity such as Fila, which is huge.

I mean, how would you, if they were your client, how would you go about saying this is what we need to help get you into training data? So I’m unable to say for sure this is 100% the way to do things because we don’t know whether ChatGPT or Gemini or when ChatGPT and Gemini and Claude are going to completely give you a new set of training. It’s it’s it’s like they’re to have to go in physically putting new hard drives with the new data on the systems.

But what we can do to help that along and what we’re doing in the Waikay products and what I think other people should be doing as well um is asking lots of questions around the topics that you’re talking about. So what what what your product is. So if you’re a bakery, ask about, you know, cooking bread and baking Victoria sponges and, you know, uh, bakewell tarts or whatever, find out all of the places that they’re, they’re saying they’re getting their their data from their sources from that gives you a really good smattering of places that they think, um, that that, that, that, that they want to read to be able to, to, to get data and, if i was If I was running this training data versus grounded data approach, I would be saying, well, I’ve got to get new yeah URLs to check my put to put in the training data next time around.

A good place to go will be the ones I had to get to but didn’t have um in when And I had to do a search lookup because search lookups costs an LLM a lot of money. And I know this because we have to pay for all the APIs and doing a Google training data API lookup costs nothing compared to a Google RAG, you know, grounded data lookup. So it’s much more expensive to do.

It’s much more resourcing expensive and and pounds and better dollars and cents expensive as well. So if they can of if they can, the more they know from the training data, the long-term, the cheaper the product is for them to supply. So it makes sense for them to start learning from where they had holes before, basically.

And so that’s a good data source to look at. Even if you don’t succeed there, it’s a good data source to look at because the search results are bringing that information back. So you still want to be in there even if you don’t get into the training data.

So i I can’t at the moment say, hey, ChatGPT updated its training data yesterday. I think we will get there. I think ah Dan Petrovic from ah Dejan SEO is running a tool called AIRank, I think it’s called, um which is trying to take all the data and then, you know he’s hoping to get some signals of when when when everything might be changing. So when they might might come out with new models. I think we’ll try and get to a point where we’re trying to do that as well. for sure, but but for sure They can’t continue to not update their training data. Some people are good training. I think ChatGPT is augmenting its data because we’re starting to see, you know, a week or two later, the the numbers on ChatGPT were saying zero and then they suddenly drop and come up.

So so they’re they’re um they they’re updating their data as they go. Gemini seems slower to update that, but it doesn’t need to because it’s got it’s got Google search results to back it up.

But I can’t tell you how every LLM is going to work because everybody’s innovating so fast. But to say it’s always done this way is a is a yeah is a tempting fate because, you know, as soon as that happens, someone will prove you wrong.

So we’re we’re getting towards the towards the end here, but I wanted you to touch on if if there there is a topic in terms of where do you see your product evolving? What features are you hoping to, know, are you working on? This a relatively new product, right? It’s only been out for a period of time.

You know, where do you where do you where do you folks? So the the things that we we were we were taking some feedback from from ah from having a standard brightness. One of the things although I didn’t show it um here because it just takes a couple of minutes.

um we we we actually build this knowledge graph of your website when we’re starting a project. And everyone was saying, I want to see that. And it cause didnt didn’t appear in the end product. It was just something to keep you entertained whilst we were building the thing. So that’s coming out actually next week. so So the reason I didn’t show the topic stuff right at the bottom is that’s going to turn into a nice little bubble graph of topics that your that represents your brand and things that your competitors are talking about that you’re not and stuff like that.

So a nice little thing there. So that that’s coming out. um obviously you know we wed we We try not to launch on a Friday because you know just in case it goes down over the weekend. you know and we’ve had requests for sentiment analysis. They they they like the the the fact checking and stuff, but you know they’d like to know with whether the the facts that you know about it are saying good things about your brand or bad things about your brand.

we’re not so sure how useful that will be in practice but if customers are asking for it we’ve got to react with that so so that’s ah that’s an important thing um I think that those are the two urgent things but I think we have to acknowledge that the technology is changing so fast and and of course when I started in SEO, you know, AltaVista was a big player in town and Yahoo was chasing them. And and so ah well the world tends to contract into bigger organizations at some point in time. so

You can see the two mammoths are, you know, ChatGPT and Google Gemini. And then there’s Claude and Perplexity and stuff. But then, you know, Grok’s got its thing out there. There’s Deep Seat with the Chinese. says There’s Llama in Facebook and stuff. They’re all playing the game. And and and they’re going to be trying to eat each other up and spit them out because they’re going to have those those big fights because there’s…big money in here, but there’s also big costs to the losers you know because they’re putting huge amounts into these things. so um So I think just measuring and measuring and measuring as much as we can, but but but making sure that the action points make sense to a business in human terms is where we really want to focus. And and so so the action plan in the one i example I showed you, um you know hopefully these are um on point.

So and not overly, you know, assumptive, you know, so or presumptive. ah If we go too far, and then I think ah we’ll we’ll come a cropper, as we say in England. The other developments are going to be we want most people on multi-user access. So they’ve got to ask that already. Large organizations want credit facilities. it’s I hate that. I hate large organizations.

Get your credit card out like all the rest of us. You know, I’m happy you can cancel this stuff at any time if you don’t like it. But do not make me take 45 days to take your billions, Mr. right eos you know whatever just get your card out you make us do you do it you can do it as well that’s my right well and and sorry i’ve got got on a little personal rant there you know no no and i as a business owner i’ve i’ve had large organizations do that to me and i i’m like fine it’s you know 10 more yeah you know it’s it’s interest at the end of the day it’s a money game and yeah we’ll play the game but your product is now 10 more so

I’d love to pay six months in arrears with Netflix subscription. Great. No problem. Yeah. my And I think my wife would do. So but ultimately, um so that that’s exciting. I mean, I really, i really see this as a ah for agencies. You know, for individual businesses, for sure. Right. And to your point, try it. in And if, you know, if you don’t, you’re not getting anything out of it after a few months. and If you don’t like it, you’ve spent 20 bucks, you know, that’s, you know, where you’re not, right you know, and if you do like it.

you know you yeah you you can you can you can just leave it there tracking and it will build up a bit of picture over over time um and and then when you’ve got time to really pay attention to it or somebody comes along and says you know hey, you know I’ve just seen this company down the road who’s gone bust because you know everybody thought that they’d closed down because that’s what it said in the LLM.

That’s when you want to pick up and say i need to make sure that’s not happening all on my side. Yeah. Well, and I see for me, I see this as, you know, I own a couple of small businesses and obviously I have the agency and I, you know, I’ll talk to my team and and like we’re going to do this and we’re going to start to present it to as part of our core reporting to every month or quarter, depending on the client. and and saying, okay, we started doing this um and then just see their reaction and then you know it’s it’s it’s a feature, you know it’s all labor. So if we have to, you know, yeah adjust things. We, we charge accordingly, but it, that was another thing that we, yeah yeah, you talked about the roadmap. Where are we going with that? One of the other things that we were asked to do was to, you know, create those reports out in a way that they could put into look a studio or into some, into it some kind of thing. So we’re going to create the ability for you to have one of those, you know, Google docs links. So you could just give the link straight to the customer so they can see their action plan and, and see what they’re going to do. you know Right. That would be fabulous.

And, and, and like, Most of our reports we give customers, you know they do better with graphs and big pictures, and that’s true for me, you know yeah so versus spreadsheets and minutiae, so to speak.

Yeah, yeah, yeah. yeah you got to be people People need to know the why as well as the what. You know you can’t just give them people a list of things to do. I mean, how many of us in the SEO industry have written these audit reports, given to a customer, and then six months later, not a single thing has been looked at, done, or possibly even read? in ah um yeah And to i’t I don’t want that to be the case with our products.

So let’s let’s the last couple steps, let’s review pricing, if that’s all right, and then how we get how we get in touch with you. ah And then all the yeah URLs and everything that’s relevant.

Yeah. So the way the pricing works, as say, there is a free version. It’ll just give you a brand report and it won’t give you that action plan. But, you know, it’ll get you up and running and see the feel of it see just how easy it um Upgrading is all done online. It’s literally press the button there that’s in front of you. And if you want to cancel, that’s just there as well. I learned from InLinks. InLinks, we said, contact us if you want to cancel.

That really annoys customers. We programmed it now. It’s kind of like, I wish we’d. And it annoys me as well, because every time somebody cancels, i have it hits me in the heart. So I just want, if we people don’t like it, press the button and then off they go.

But so but the twenty dollars will give you eight reports a month. Now, a report is either one report eight times or um or or, you know, different topic reports less frequently.

So um so if you’re going to try and do this report every day, so some rank checking people will check their rankings every single day, you know. Well, that’s going to be quite expensive because every time you’ve run read one of these reports, we end up having to make 15, know, five. We’ve got five search it five engines, three it’s three companies, 15 lookups, all costing money. So basically that’s going to come down to the bottom line.

So we’ve got packages where you can upgrade you know from $20 month to and what’s the next one up? month or a month or whatever. And then, ah you know,

So you a hundred ninety nine dollars a month and you get more reports for them but um But essentially, each report, if you think about it, is going to cost two to two and a half dollars, depending on how much you’re spending.

And if you don’t want to upgrade that subscription service, you can just go in and buy 50 reports and then use them as you go. So that’s kind of we’ve tried to make it really easy for not not just small companies, but small tasters, you know people that that you know you could be a ah massive organization, but still the only thing you do is shipping containers. you know So you could be MERS shipping containers and and and all you want to know is shipping containers. you know and you know and And that’s your your your your single topic, so your brand.

and And that’s fine. um ah But but i don’t i didn’t want to ah make it just because you’re a big company, you’ve got to give us lots of lots of money, or just because you’re a small company, you can’t go and use the system. The cost, the actual cost is in doing those API lookups and creating those reports. So let’s make it so that, yes, we get a little subscription, but they the more you use it, the more you’ll pay.

And then you can go and easily change those reports from monthly to weekly to daily if you want. So if you suddenly decide, you know what, we now need to be really well known for shipping containers because it would appear that this other shipping company is, is is is is you know, taking over the world, um then you can just And do it for a day or, you know, go for it go for a few days ah on a daily update, see if you can make ah make a material change and then put it back down to monthly. So you’ve kind of got some information there.

So you can spend as much as little as you want. If you’re an agency, there’s no limit to the number of projects in there. So you can just put in a couple of things. for each of your clients and then say, look, we think you need a lot more than these, but you know here’s a couple to keep you going. and know well we’ll we’ll We’ll invest $5 a month for each of you you know to ah to show you a few things.

And once we’ve got that that free link that you can give to people, that would be nice because they don’t have to try and log in or anything. um ah but so But yeah, I think my hope is that companies will at least generally have four or five concepts that they want to want to start tracking. And then hopefully they want to do that either monthly or weekly.

But that means that they’ll they’re going to start to grow as they have more trust in the reporting. ah report Yeah, I really like your pricing. It’s fair. Obviously, that you know, it’s scalable. You can scale it up or down, right?

You know, and and depending on your season of ah business. yeah you know You know the downside? there’s We’ve got a competitor out there called, hang on, sorry, Bluetech. I mean, sticking things on walls.

So we’ve got a competitor out there. who’s going going for the enterprise market, shall we say, you know, and we think we’ve got a better product. I won’t mention them because it’s, you know, but so ah we we we think our approach is better.

We’ve actually put in a patent for this for this ah this scoring mechanism. We think it’s quite clever, the idea of blending out, building knowledge graphs. um And, you know, they you know They’re one of those companies where the pricing is so much you can’t put it on the website. You’ve got to contact us to even get the price. and and And that’s going to happen quite a bit. And and I think that’s that’s an insult to innovators. I think that that’s a – I don’t like it. But what it means is when a large organization comes and sees our pricing starts at $20, they immediately think, oh, these guys don’t, you know, that’s a cheap product. That’s a, you know, well, they’re going to lose to the innovators at that point, you know, the ones that do that. know Not all of them, of course, we’ve already got some big, big brands in in in the system. So I’m quite confident that but we’ll get over that hurdle. But as soon as you make a ah ah cheap entry price, you suddenly have all these brands who, you know, know um you know, who think it must be, you know, it must be not for them. So i’ve got that battle to move with. yeah Yeah, but I think you’ll overcome it because it’s, you know, again, more exposure can give you. um And again, it’s fair.

I mean, it’s logical. I guess that’s the phrase I want to use. It’s a logical price. It’s logical price. yeah Yeah. Right. You know, it’s just like, you know, no. do you Do you get a great discount for using more? Maybe a little, but but the reality is its your cost is fixed. It’s not like you know the API cost is gonna drop tremendously. You know oh yeah it’s not it might go up and the downside is if if check gpt put their prices up like you know 200 i might have to pass that on to the customer because i actually have to pay that price but you know right but yeah somebody’s telling yeah it so so’s got to pay for the nuclear nuclear power plants that drive these yeah data centers and everything else you know it’s just the fact the way it is you know so Yeah, that’s how it is. Yeah, yeah. so Well, this is exciting. we’ll put a We’ll put a link to your product. Again, I mean, you can… you can I’ll say it over the… over the for For the podcasters, it’s it’s it’s remember, it’s WhatAI Knows About You, which is why it’s called wayk.io. It stands for WhatAI Knows About You. It’s.io.

And the LLMs know who we are now. So, you know, if you type in wayk to chat QBT, it’ll probably get there. i Yeah, that’s side of it. Maybe not the training data of Gemini, though. not for a while no i saw that was it’s pretty interesting well uh i really appreciate your time dixon and and uh hopefully down the road we can we can have you back it you know when as yeah Again, there’s going to be evolution of your product. You’re going features.

I’m sure listeners will love to and viewers will love to see that. and um and And again, i’ll I’ll spread the word to to other people in the industry. Hopefully you can get get on their various podcasts and and get this… ah Because it’s a huge value to our clients. And at the end of the day, that’s the business we’re in. We hope so. Yeah, and we we really want to be the market leader in this now. And that’s another reason for going with a low starting point. Because we you know we know that there’s a there’s a big theres big land grab out there for you know tools around this kind of thing.

We don’t think that many people have got the ability to add a value layer in the product. They just kind of just um asking API, but they don’t have something like our entity knowledge graph system that that that comes that came out of in links, which is at the core of the the methodology.

So, you know, but time will tell if I’m not here in a year, I failed. yes Well, I don’t think that’s going the case. So, I mean, youre you know, too much. And and and i I know it’s a need. I know it’s a need because our clients are telling us that.

I mean, yeah we wouldn’t be here in this interview if, you know, You know, nobody said anything to me clients that were saying anything, andd be like, know, that’s nice, but they’re asking for it. And yeah if they they need it, then obviously we’re probably a year behind, you know, so to speak.

So, um but we’ll, we’ll catch up. Thanks very much for your time, Bob. I really appreciate it. And thanks to your listeners for staying this long. Thanks again, Dixon. We’ll see you now.