AI search is rewriting the rules - are you showing up where it matters?

Our prediction is that all search will be AI search. So we're already losing the traditional SERP and I think Google is here to stay. What were the kind of top tips that you could recommend to anyone working in the US native retail space? We do way more with local AI search than we do in other industries basically. But we always recommend our users to do this to focus on top of the funnel content so that you're showing up in the beginning of the AI search journey. You launched in 2025, you've already tracked over a hundred million citations. So what we are focused on at first always is to set a baseline. Hi and welcome to this week's episode of the DriveTime podcast. It's a special on the road edition at Brighton SEO. I'm really excited that I've got Hez to group with me from Promptwatch. Hez is the CEO and also one of the keynotes because at the conference, I've managed to peel him away from the event to come and have a chat with us. Thank you so much for joining me, Hez. Yeah, thanks for having me. Great. So just a little bit about you because I know Promptwatch only actually started a year ago and in that year you've already seen pretty fantastic traction. So Brighton SEO for anyone that doesn't know what this conference is all about, it's everything to do with how people find your product or your website and how you manage search engine optimizations. So how search engines will find your business. But we've seen in the last particularly 12 to 18 months, two years AI massively transforming how consumers are behaving online. But first of all, tell me a little bit about you and how you came to set up Promptwatch in the first place. Please. Yeah, of course. So I'm Geyser Gøth and the CEO co-founder of Promptwatch. I have a back for time 27. I'm from Amsterdam and that's also where our offices are. I have a background in finance. So I have a master's in finance. I was an investment banker for a while for about three months. I decided it was not for me. So I started to work on startups, AI, etc. That was really what was happening at that moment. And at some point I met my co-founder class. He's really the tech guy of the duo basically. We both had our hands free at some point. The company got acquired. So we both were free to move basically. And we saw GGPT happening. So class is a lot of followers on Twitter. And he also has a website that has a lot of organic reach basically. So he was getting screenshots of, "Hey, I'm seeing your website in the answer of GGPT." And that got us thinking, like, "Okay, how does this work? How can we influence this?" So in March 2025, we launched the first version of Promptwatch basically. And we have been building, iterating, also learning ourselves from there basically. So yeah, super cool to be on the right that we have been on up until now. So fundamentally, Promptwatch, coin what you do, what your business actually is designed to do. Yeah, we help brands to show up in AI search. And this can be GGPT, but also AI mode, AI overviews, perplexity, all the different AI models and all the different ways of search that are out there basically. So people now, we've seen consumers becoming not lazier, but I suppose expectations change because the way that we search changes, we went from typing something onto a browser to talking to a device in the corner to now having a whole conversation with GGPT, but you're right, it's how do you show up in search? When we started the podcast about a year ago, we were just starting to get our brains around AI. I was anyway, it was one of those things, whereas like, so you're telling me you've got to make sure your website optimizes for AI search. Of course, that wasn't something I think a lot of brands were particularly good at. When you're looking at a business, what time it took me through the steps that you take because we know that you have to tag your pages, write your descriptions correctly. So how are you understanding and working with AI companies to help your clients perform in the best possible way? I think an important concept to understand in the difference between traditional search and AI search is that it's basically a zero click experience. So the way you measure success is completely different and that's also changes how you can know what to do basically. Are you showing up for the prompts that are important to you? Are your pages being hit by the agents of the different AI models? So these LLMs, they send out agents to fetch the information from the websites. And I think what we are focusing on and optimization sends right now is one content structure. So make sure that it's optimized for an LLM to process, not necessarily for a human. And two is technical excellence basically. So in order to be crawled by an AI agent, your page needs to be super fast because they're calling the whole web the whole time. They don't have a lot of time because they need to get the answer back into the conversation with the user. So those are the two verticals basically that we focus on the most. So content structure and technical excellence basically that's what we see is important. So what would slow a website down because I like speed of because that's something I suppose even as humans you get quite impatient and you browse away from something or you click on but actually every robot is doing it if an AI agent is doing it, there's no kind of room for error. So how do you how do you go about tracking the speed of your site or is that what you're doing? You're literally measuring speed of returns. Yeah, so basically what a user of prompt watch does is they upload their site map and then we start to imitate an AI crawler basically. So we start to index all their pages and then we can learn okay how fast is the page? What's the problem here? And that's where we start to give optimization recommendations. Hey, you should remove this image because it's making it slow, et cetera. So we basically imitate an AI agent and that way we can learn okay if this agent would crawl this page would it go correct or would it go wrong, et cetera. So that's how we try to figure out what's going on the page. Yeah, so from a year ago launching and then obviously now speaking at brightness, what do you think have been the biggest shifts in that year? I think a couple things. I think at first everybody was just about prompt tracking and prompt tracking is super important. So you want to determine okay, these are the prompts or the questions that are important to my business and how am I showing up and what can I do to improve? I think the transformation that we have made is really focusing more on the website itself. So more integrating with the website of the customer because that's where you can find the real data. Also what I'm I talked about yesterday and what I'm hearing here is that CDN data is going to be what what's going to drive success. How you can actually measure real AI search traffic on your website. So prompt tracking is really synthetic data. So if this prompt would be used, what would then be probably be the answer and how can I influence it and CDN data. So from from Akamai or for sale or Cloudflare, those types of providers, you can actually start to measure okay, am I really successful in AI search or not. So I think the development we're seeing is focusing on synthetic data to real website level data basically. Just tell me what CDN data stands for. I don't think the delivery network. Thank you. Okay. I think a lot of the people that listen to this podcast are primarily people that work in automotive and they're trying to improve their understanding of the tech that's going to make a difference to their business. So content delivery networks, there's so many and we are full of acronyms. There are millions of them. So just coming back about your keynote then, I know you did one yet, you could have a run yesterday. What was the kind of key theme of your talk here? Content delivery networks. Yeah. Yeah. Yeah. I'm stupid again, but the key of the primary thing was that we really want to understand or start understanding how to optimize for AI search. So just like you mentioned, you can optimize sticks, etc. But we measure millions and millions of data points every day. So we can start to dissect or reverse engineer. Okay. What are these AI models actually referencing? And those are the things I just mentioned are basically, I can go even more technical around that. But those are the main things we found basically. So the talk was really about okay, nice to track prompts, but you really need to understand how an AI agent processes information. And then you can have take the right steps to optimize for it basically. And the other thing that we hear, obviously we had SEO, but then we've now got GEO, which is generative and just as you just explain the concept of that and how that works and how we can start to help dealerships think more about the way they structure their sites or. I think the main goal of GEO is supposed to SEO is really showing up in the answer. You could call it being top of mind of the LLM basically. So that when the AI is trying to come up with an answer that it finds your information first. So it's largely related to doing SEO. It's only percent the same. The end result is different. And the person or robot that you're optimizing for needs different ways to optimize for it. And then the ultimate goal is that I ask a question about a car. What's the best sports car I can find in this in this location, for example. And then you want to make sure that you're showing up on top of the answer that you're top of mind of the LLM. And that's a whole different process than making sure that your web page is number three or number four, etc. So I think that's the techniques are similar, but the end result is really different and it's also quite impactful because a syrup has 10 pages. And then AI answer most of the time just mentions three options. And the user most of the time goes with one or two asks the follow-up question and you're completely out of the equation. So it's a little bit more high stakes, I would say. I've never actually, I was at your right, say people would normally get a search result and then start to scroll through it. And you don't get any of that in a language model search. You mentioned the Chat Gbt, but there are a number we can work with. The Claude, you can see Gemini or Copilot. How the companies that you work with are they across the kind of the spectrum, are you not, or any specific verticals that you've been quite successful in? So we, the interesting thing about our business, of course, as we can help any business worldwide, who is serious about its online presence. What we do see is that the most traction is coming from businesses with a product or service that requires a lot of research upfront. We work a lot with banks, insurance companies, but also the automotive sector. If you start your user journey, you're not going to scroll through all these 10 pages anymore because, mentally, that's too large of a load if you have CHEDGPT, for example. That's really what I see in our. If I look at our customer data, it's primarily a longer buying journey with complex information that needs to be processed by the user. That's where the most traction is coming from, and that's also where you see the most impact on traditional SEO, basically. Yeah. I think one of the. I've worked on a smative for most of my career, and we've always looked at that kind of customer journey piece and how you understand what a consumer is looking for and what their dwell time might be, and then what their comparison sort of basket looks like when they're comparing different makes, models, variants, that kind of thing. We've also seen the shift towards EV. There's an awful lot of knowledge that is poorly presented when it comes to things like cost of ownership, part of our charge, all of that. In the automotive sector specifically, are you seeing. I think we would see often. You would have multiple tabs open, and then they would switch to mobile search or e-tracking things like use by device. I want you to have your accounts up when you go to any device that you're using. When you think about it, you'd like zero click environment. How much of an impact has AI had on the way that consumers are shopping, particularly in those longer purchase journeys? The thing with AI is search at the moment that it's a black box. We try to gather as much information as we can. We do the prompt tracking, we do the CD analytics, we aggregate all the data across our customers, but we can't actually see what the user doing inside the platform. I think the main thing that has changed is that the decision making around buying something is way more influenced in the search experience. You're asking questions, you're really debating with your AI model what the best option is. I think in that sense, the traffic really changed in the way that it's attributable. You can't really. This is a click from JGPT. You can see that on your website, most people do something in JGPT. They come to a conclusion, use Google to find the page they want to see, etc. What we see at the moment is that it's harder to attribute things. Lots of things happen in that black box of the JGPT interface or the cloud interface or whatever you're using at the moment. It's a closed environment. You can't really see. I suppose this year you can see the image of the page collection. You can see how many times people have clicked on certain things to get to the end result. Actually with this, it's much more difficult to track it. It's much more difficult. The best proxy we have at the moment is the JGPT citation bot. That's the bot we see on your website when someone's having a conversation with JGPT. It uses your website in real time in the answer. We can see that in the ContentLivery network, data-based. That's the best way we have right now to measure if your brand is showing up, if your content is used. Other than that, it's still. It's too hard to know what's going on and how these user journeys are evolving. We'll get back to our conversation in just a second. But first, I need to tell you about KeeLit Vega. It's the new predictive intelligence solution from KeeLie. Powered by ThoughtSport, Vega offers dealership next-gen business intelligence. It unifies your data into one live view and connects your entire ecosystem, including your DMS and applications so you can access insights in real time and use them to make bold decisions. If you've ever used ChatGPT, Vega will feel instantly familiar, allowing you to ask questions in your own language to get the answers you need instantly. Best of all, Vega allows you to stop reacting and stop predicting. It gives you the clarity you need to turn insight into action and make big moves before everyone else. Don't forget to get in touch if you want to know more. We can make sure you're fully informed. Back to the main episode. Did the research before we met actually, one of the things that said about you was, you launched in 2025, you've already tracked over 100 million citations. I was going to say, talk to me about that. By the way, if anyone, there is a right next door to the conference so you might hear some background noise. But that was the, I suppose that was the, so the citations is tracking when one of your whatever is your tracking is mentioned in a ChatGPT or LLM search return. Is that right? So we measured in two ways with prompt tracking and that's where the 100 million number comes from. So we fire a problem to the model, see what comes back and analyze the sources that it has used and that say citation. And that's also where you want to show up as a brand or as a website. And then on the other hand, we use the CDN data to measure in real time. Okay, is this happening on the website? Are your, is your content being, being used? So the 100 million citations we measured is really the synthetic testing that we do. Okay, if we fire this prompt, what's the response and what kind of information does the AI model use to come up with the answer, basically? And then the CDN is really to verify, okay, is this also actually happening and what scale is it happening at the moment? I think brightness here is one of those events where you've got a lot of knowledge experts in the room who are their job is just about CEO or optimizing their customer's sites for the search. How ready is the industry or how, because the tech, it's moving so fast, isn't it? How do we have a sure how confident, how ready to have the audience felt to you this year? Yeah, this is my third time here. So last year in April, I was here for the first time just with a backpack shaking some hands and getting a feel for the industry, et cetera. Everybody was super skeptical about AI search back then. It was not going to happen. It's not important, et cetera. Second time, it was already changing a little bit. So the early adapters were already okay. We need to do something about this. This is the third time around for us and it's really, I think the industry has accepted that okay, search has changed and it's not going to revert back to traditional search anymore. So I think there's way more practical insights and people are really working with this and on it at the moment. I think there's really a mind shift change. But it's important to keep track of the latest developments. What's the best model to focus on right now? What are the best techniques out there? Yeah, I think it's quite hard work to keep up basically. I think this is a run for the expert. This is a whole conference full of experts. How on earth can retailers in the car industry be ready for or keep on top of all of that? New kind of way of working and keeping on top of consumers because it's quite challenging. Yeah, it is. Yeah, it is. Yeah, I can't change that. It's quite a challenge. Yeah. So what thing about retailers and how they, because obviously what we see in the automated space is consumers are looking for either the vehicle that they want to buy or they're looking for a service appointment or they're looking at there's a number of different things that they can interact with when it comes to working with the car retail industry or the automotive industry as a whole. What should they be thinking about as they're upgrading their websites or bringing new content or how they saw some new stock? What were the kind of top tips that you could recommend to anyone working in the automotive retail space? I think it's very important to think about where in the funnel you want to show up. So the best, what we always recommend are users to do is to focus on top of the funnel content so that you're showing up in the beginning of the AI search journey and also bottom up the funnel contents. You need to make sure that if someone's looking for a local service appointment and they're ready to convert basically that you show up there. So those would be the main, strategically the main two points and then just get the right content on your website specifically for that AI model. Get the blogs written, do it with AI or do it completely yourself, but focus on top of the funnel, bottom of the funnel and then get the right content on your website. I think that's the most concrete way to think about it and to start working on this right away. Is there a way of building in it because we're looking now at how do you build the agent to do the thing that you can't do yourself because if you're not good at it, could you literally set up the agent to say, "Hey, tell me what are people searching for?" Can you do that? We don't believe that's how it works, unfortunately. Would it be great? It would be super, but I don't think that works. So do you think you're not going to do it prompt-watch and what you're focusing on now? So you mentioned you've been here, you came here this your third time, so you've only been coming here for a year. That's every six months of event. What's the big focus for 2020? For us, it's not only providing insights, but also making it easier and easier to actually do the optimisation. So we're really focused on content creation, content optimisation in real time. So that's really for us as a company. It's really around, "Okay, we can now show quite well how you're showing up when you're showing up the CDN, the prompt tracking, etc." For us, it's 2027. We'll really be about, "Okay, we need to build a platform with which you can easily do the optimisation. So create the content that you need, etc." That's really the main focus. And we're lucky because obviously you're working directly with us. So one of the things that we do is build websites for dealers. And that's just one area of how people will find you. So we talk about acquisition of leads and how we optimise that kind of lead acquisition for our retailers and then how we then help them to convert those leads. So I think I obviously prompt-watch is about finding you first and then it's about how do you then convert. So do you see AI doing obviously we see that in chat how you can start to nurture when the human is busy, the chat can be doing the AI. AI chat can be doing a lot of the work for you. How into the funnel are you with the work that you're doing? Is it very much front end finding mean in the original search? Or is it, are you able to contract through and conversion? - We're really focused on brand visibility in search and make sure that you're showing up when the question is asked. And in doing that, you have all these later on effects because you're also showing up in other interactions with AI, but our focus as prompt watch is really around making sure that brands can control their presence in AI search, making sure that the brand is named, but also that, for example, the information is correct because that's also a large part of what you need to control with AI space. Basically. We're really on that part of the funnel. - And then thinking about the traditional search engines that we, as consumers, have all been using for years, is there a massive risk that people just don't want to go to Google anymore? Or what's your kind of prediction there? - Our prediction is that all search will be AI search. I think Google's going nowhere. The only thing that will change is the way that we interact with Google basically. So I think if we look at our data sets right now, it's JetGPT Cloud in Google that's gonna drive search in the future. So the monopoly is gone, but Google is not gonna disappear in any way, but it's the way we search is just with AI and not scrolling through SERPs anymore. - And it doesn't mean we have to do it. It's just search and Google gives you the AI response already doesn't it? So it tells you get the AI preamble, which is great because it just means you then haven't got a scroll. - Yeah. - It's very dangerous for those smaller businesses. How on earth they're gonna be found? - Yeah. - What's the kind of advice for those smaller companies that are not gonna maybe appear in the top three? What can they start to do? - I think they can. And I think the key right now is to start right away. There's a large window of opportunity. Your competitors aren't doing anything at the moment. So we see, for example, with more local businesses that if they really try their best at showing up in AI, right now there's a huge window of opportunity, but you need to start today or yesterday basically to be able to get that opportunity. - Yeah, absolutely. And then thinking about if we would fast forward 12 months and you're, we meet again in 12 months, what do you, what's your kind of prediction of way you think this is gonna end up for how fast you think it's gonna evolve? Could you do like a year ago, people were quite skeptical and now it's like, there's a room full of people that was there filling the hallway and even the corridors with startups. There's an awful lot. How do you know who to work with? Now how do you find supplies that are gonna help you? 'Cause there's just so much. It's really hard, it's no what good looks like and who do you guys you place your chips? - Yeah, in the world of AI software, it's hard to determine, okay, is this a reliable company or not? I think it's a little bit speaking for our own of course, but I think it's important that you need to look at, okay, what do they say around the industry that they're building the software for? In general, every AI startup is quite new and fresh because they're solving a problem. It couldn't be solved before. But I think it's really about, okay, how do they present the solution? Do they present it as a one click fix? I think that's quite the red flag most of the time. If the realistic about what the AI software that they provide can do, then I think that's a good sign, okay, this is a reliable partner I can work with. But sometimes, I wouldn't say competitors, but AI startup stands to say, okay, you just click once and it's a ready solution. That's most of the time, not the time. - Not okay. - It can help a lot to use AI, but it's not a silver bullet, basically. - Yeah, and then people haven't done really rewrite their websites to be able to appear in the way that they should fit an AI driven search. We have to rip it up and start again, because obviously, - No, I think it's more about optimizing first. So don't burn your chips behind you, but really just, yeah, it's really about doing incremental steps first and then you can pick it up from there, basically. - So when you obviously work with us, and when we brought you into try and make sure that we can help our customers perform better, was anything in automotive surprising or was it better than you thought, worse than you thought it would be? - I think adaption is a little bit slow to be honest, but what I think is interesting about Automotive is the locality of it. So we do way more with local AI search than we do in other industries, basically. And it also provides a large opportunity to, as a brand or as a car, an leadership to optimize. So I think what's interesting about the automotive market is it's way more localized and local AI search. Nobody is really on top of that right now. So that's a great opportunity. - A big opportunity, 'cause a lot of my quite small businesses aren't they? Family-owned. - Yeah. - But obviously geographically, you look that far away from us anyway, but I think in the UK market, where I spend most of my time, people don't wanna have to travel too far, necessarily, particularly to get a vehicle or to go and test drive something. So often it's like trying to find something within a reasonable distance. I think about how people are searching, are they searching in their local area? So actually the opportunities for local businesses are, so don't give up yet, you're not behind the curve. There's still plenty of opportunities to do something and optimize where you live, for customers, where you're based. - Yeah, definitely. So the search behaviors, of course, still the same, the way it's searched has changed. - Has changed. - Still looking for a local card dealership, et cetera. I just use a different wording, and it's more in depth, et cetera, but these questions are still asked. So the opportunity is definitely still there. - That's brilliant. Listen, thank you so much for taking time out. I know it's a busy conference, and I found you downstairs earlier on. Thank you for the coffee. The people that you all, 'cause it's a wrap-up day today, are you guys all stand busy with the questions that you're getting on the stand? Are they thought you were expecting people that you're speaking to confident, totally learning? - Some people are super advanced in what they're already doing in AI search, and others are just completely, they need to start doing something today, help me please, basically. That's the range we're seeing at the booth today. - Yeah. - So yeah, that's, first, that's great to see. - That is really good. So I'm really gonna be busy, and thank you so much for coming to talk to us on the podcast today. Be good to at some point to get you on a fall in the studio, 'cause we are literally talking to the back-offers here at Brightness here, and we've got some space, but it's been great listening to you, and I'm really, I might air even knowledge around AI search, is still very much a steep learning curve. - I hope to help. - Thank you. - I hope to help. - I think we always learn something, every day's a school day in technology, and I think the fact that it's rapidly advancing, just keeps us always on our toes, so it's a brilliant, thanks so much. He's the group. How do I just say, I just want to introduce you properly. He's group, you mentioned, 'cause I see his, he's, and there was a dirt in the middle. - Yeah, no, it's a very Dutch name, so it's heist, but-- - Heist, a group, okay. - Yeah, no, that's perfect. - Okay, great. - I'll say the outro again. Thanks again, so much to you and my heist, a group. It was a real pleasure, and good luck with the rest of the conference. - Yeah, thank you. - Drive time. Brought to you by Keyloop, creating technology that transforms automotive experiences.