[00:00:02–00:00:28]
Good morning everyone. Thank you for coming — so many of you, we don't even have room for it. It's crazy to see, but thank you for coming here this beautiful morning. If it's your first time, then welcome to 1508. My name is Mathias, I work as a transition designer here at 1508. I'm joined by Nina.

[00:00:31–00:00:53]
Hi, I'm Nina. I work as a marketing manager here at 1508. For those who don't know, 1508 is a digital design agency. We have existed since the 15th of August, year 2000, and we have for now almost 26 years done digital experiences and digital design, which has been many things throughout the time.

[00:00:57–00:01:26]
So what we see and say that we do is: we create responsible digital experiences. That is the latest and greatest version of what we are doing here — what we see is necessary to do and be a part of the world with. So we can kind of break that into two different things. First and foremost there's the responsible part. What does it mean to do responsible design?

[00:01:26–00:01:54]
The way we see it, there are three primary things that we look for when we talk about responsible design. We talk about sustainability — so how can we have the least impact on the planet with the digital solutions we create. We talk about inclusivity — so how can we have the highest impact for the people that we design with and for. And in the end, accessibility — so how can we make sure that the things we create are accessible for everybody who needs to use them.

[00:01:54–00:02:25]
The other part of this equation is digital. And what is digital — in the year 2000 when we started there was something, and in the year 2026 which we're in now, it's something completely different. From where we used to talk about websites and apps and digital brands and communications, we can now see that there is another player on the digital landscape that has come in and kind of taken over.

[00:02:26–00:02:59]
We can see now that 50% of consumers use AI-powered search today. This is a report from McKinsey from last year — lord knows what the number is this year — but we can see that the way we interact with digital products and digital services is changing drastically. This number is just going to go up unexpectedly. So what does this mean for the way that we design services, products, websites, all these different things? A whole lot, is what I'm suggesting at least.

[00:03:08–00:03:47]
So what does this mean from a digital perspective? We can see now that there has begun to come APIs and applications that are coming into the agent experience directly — so you can shop through the agents now, you don't even have to go into a website. We can also see that they can do orders for you and you're just met with the final "pay now" button. And here we have an agent as well looking for job candidates, because you want to hire somebody — it can actually find you the most suited candidates.

[00:03:47–00:04:18]
So there's a lot of things that it can go in and do now. It can take in data and process it and give you the correct information that you need to use. So what does that mean for the website — which is where you used to go in and get all of your information, and you had to sit with it yourself and figure out what you need to know and how you can put it together yourself?

[00:04:01–00:04:39]
What we see or imagine is that we're going to have two versions of websites. We're going to have one which is built for the machine — which is the agentic experience — which is structure and data and clarity. So all the things that you want the agent to be able to go in and pull out to its own thing, where you can then read through the summary of all these different things.

[00:04:28–00:05:09]
What does that then leave for the website — or for your digital experience that you have control of yourself? Well, this is where we then want to see if we can create an elevated experience. We want to create something that is meaningful, something that feels human, that tells stories, that is beautiful. It's going to be the super-branded experience where your brand really can stand out from the rest of the herd, because everything is now super automated and easy over in the agent experience. So what is it then your website needs to be? Well, it needs to be your business card — of how you perceive yourself, how you want to be perceived, and how you want to have your look and feel.

[00:05:12–00:05:57]
It also means there's a loss of control of how you are portrayed, which is scary sometimes. Maybe this is — if you go into ChatGPT and write "what is transition design," which is one of our solutions or core value propositions here, you can see that we are quoted three times, but we are also put together with two models — one of which is from Carnegie Mellon — which is not our model. So it's up to the user to decipher what is what, who owns what part. It's just to say: we no longer have control over how we are portrayed in the context of a lot of different things, compared to how we portray ourselves in this more playful, branded experience.

[00:06:00–00:06:30]
So it's just to say: what do you then need to have, if you have these two alternatives? What do you want to look like in the agent, where you want to give the best information for it to have an authentic, truthful representation of you? And what do you then want to look like if you can just be the most true to yourself and your brand?

[00:06:19–00:06:51]
So we also propose these two different ways of looking at the user journey. We have the highway — which is the super-fast agentic experience — where I go in and I write "what is a design agency in Copenhagen that does human-centred design?" You get this answer. Or you take the scenic route, which is our own website where we can tell our stories, show our cases, and all these different things.

[00:06:52–00:07:28]
So what do they then do, and how can you approach these two different ones? We often use this analogy — this is a hashtag, if nobody has seen one before — and this is how most digital platforms up until the year 2025 have existed or functioned. This is how it is to go into borger.dk and order a new passport or something. You are just exposed to everything at once, and then you have to figure out where the hell you are going now.

[00:07:17–00:07:50]
And what we can see with the agentic experience or AI is that we can actually just give you the needle instead, because we know you, we know who you are and all these different things, instead of you having to navigate through all these different user journeys. And normally when as UX designers we try to design something, we try to funnel something where we can in the end give you exactly the information you want. But here what we can kind of see is that you can just get it straight ahead, because it can remove all of the clutter for you.

[00:08:00–00:08:25]
So what does that then leave for the scenic route? To toot our own horn — this is some work we have done for Stray, which is a venture capital fund in New York that exists in the finance industry. And we all know what finance institutions look like from a branding perspective — maybe a bit conservative, maybe a bit boring, you could say.

[00:08:26–00:09:00]
And what Stray has then decided to be is quite the opposite — to be much more playful and use this whole digital experience and digital presence as a business card showing their value. They want to build responsible futures and have meaningful investments. So what we have created with them is a brand that is super playful, super showing of what their values are. And what we see — or hope, maybe — is that a lot of brands are going to lean even more into stuff like this, where they are daring to be super playful or super strong in their brands, because all of the boring stuff you can kind of do with the agentic experience and have all of that information quickly through there.

[00:09:08–00:09:50]
So what is it that we need to do from your perspective? We want to really show some personality and make you stand out from the rest. When we talk about brands and user experiences in general, we here at 1508 talk a lot about the peak-end rule. It is regarding all experiences — both if you're going to the hospital or you're going in to order a new passport on borger.dk. There are two things in human psychology that we most commonly remember: which is when the experience peaked — that is both in terms of peak down or peak up, so it's also a super horrible experience, but also a super good experience — and how things end.

[00:10:00–00:10:40]
So what we always try to really use our energy on is these two different things. So where can we see the biggest user problem or biggest hurdle, and how can we elevate that specific point the most? Or where is the biggest opportunity for creating something super immersive and branded? And in the same vein, how does the user experience end — what is the final thing, how are they going to remember getting out of our experience?

[00:10:22–00:10:48]
And this still holds, even though there's an agentic protocol that can do all of these different things. We're still trying to figure out: so in the user experience, where they maybe get some information from there — what is it then that we have as the peak, and how do we have this branded communication as well?

[00:10:41–00:11:07]
So — responsible and digital. We do responsible digital design. Can they coexist? It feels like there are so many people in here this morning. I just want to have a little hand in the air — are you concerned about AI?

[00:11:03–00:11:45]
Not so shabby. Because of course, when we talk about responsible and digital, there's a lot of things that go on with AI which can scare you a bit — or at least be something you need to consider. So if we talk about sustainability, of course there is the massive energy consumption that all of these data centres are going to require. We can see in Australia that they are lacking water supply, because all of these data centres just suck all of this water out of what the people need to have.

[00:11:45–00:12:24]
So sustainability-wise, not so great. Inclusivity — we can still see that AI targets specific ethnic groups in facial recognition and all these different things. So in terms of inclusivity, there are still a lot of issues with AI and it actually being equal to all people and all races and globally. And accessibility — there is a generation right now who is maybe already challenged by the digital space. It's a hot topic in Denmark these weeks about Rejsekortet going fully digital, and the elderly are scared because how are they going to get around anymore.

[00:12:28–00:13:00]
That's going to be a physical card, don't worry, but it's just to say: when we go more and more digital, and now we go even hyper-digital with AI, what is that going to do for our elderly? What is it going to do for our people who are not so technically advanced? So are we actually going to be more accessible with AI? Of course, this is also going to be an engine into conversational UI and all these different things, so it's going to be much easier to have a conversation and therefore maybe easier for the elderly. But it's at least a concern to have about whether it is actually making it easier for some groups to engage with your services.

[00:13:08–00:13:44]
Yes, and in terms of using AI to make a service quicker and better and enhance its value, we can see Duolingo — which I imagine a lot of people know — has now replaced a lot of its workers with AI, because it can go in and automate the things and save up a lot of time, and therefore be used by more people and save a lot of money and have a higher profit, and all these great things which we love to see.

[00:13:44–00:14:21]
What we then see here from this Reddit user is: "I've noticed an increase of AI — a lot of translation stories are not helpful, there's obviously less humanity involved, it's disheartening." So by actually trying to optimise their experience with AI, they've actually lessened the value of their product and therefore lessened the cost and the value of their brand.

[00:14:06–00:14:45]
Just to say: make sure that you don't remove all the people, because a 70% decrease in stock is at least something. So besides doing all of your boring emails and Excel sheets and all the things you don't want to do anymore, it seems like the thing that AI is really good at is amplifying the existing patterns we have in the world — it's amplifying our inequality, it's amplifying our energy consumption, it's amplifying all of these different things, which is making us stuck in a world where there's a lot of things going on which we need to address quickly.

[00:14:48–00:15:22]
And it also makes us stuck in this — this is the scale of growth since the 60s — and we can see that we have doubled our economy every 24 years, which is super nice, and that's what we're always talking about: we need to go bigger and bigger and sell more, and AI can make us even more growth and sell even more stuff. What does that actually mean? On the other hand, it means that our energy consumption has gone drastically up throughout those same years, and we can see our natural material extraction is also going up.

[00:15:22–00:16:02]
So there is a correlation between the amount of money we use and the amount of energy consumption and material and all these different things we use. So if we can optimise everything with AI and we can sell even more and we can become richer, we are also messing up the planet even more. And how is it going with that old planet of ours? Back in 2009 it was a great time — I just finished primary school — and we only had three planetary boundaries crossed off. And here, last year, we crossed the seventh. Yay. And we're going towards the ninth.

[00:16:00–00:16:32]
So it's just to say there is a lot of things going on in the world, and a lot of things we need to address, and we need to make sure that we do. And I'm not quite certain that AI is the catalyst for this change that needs to happen. I didn't mean to be so negative — it just kind of happened. It's just to say there are a lot of considerations to have when you talk about these topics, and there are a lot of things that we kind of forget when we just type in "can you make me a cool meal plan" or something like this.

[00:16:40–00:17:10]
For me at least — and this is maybe also just a designer praising designers — it matters even more than ever that we are designing with intention, we are designing with authenticity, we're designing with humans, we're actually creating something that has meaning, that has value, that is trying to address all of these wicked problems we have and all of these things we want to achieve. I think there are a lot of people in here who come from organisations that have big vision strategies where they say they want to create meaningful change and sustainable societies and all these different things. So it's just about making sure that we are actually living up to all of those purposes that we are saying that we do.

[00:17:26–00:17:59]
So what can we use AI for? Like I said: reduce complexity for users. So this motorway, highway, needle-in-the-haystack thing — we can give you exactly what you need in the instance, and we can actually then maybe reduce the time you spend going on borger.dk, and therefore also the time you spend doing useless stuff. We can have generative user journeys, which is super cool, because if I am a person suffering from dyslexia, ADHD, or whatever it is, I can go in and have a very specific user journey that is generated to me and my constraints, instead of seeing the same as everybody else.

[00:18:17–00:18:52]
So this thing of actually having the possibility of generating things based on a context is important in terms of accessibility as well. Hopefully all of this automation — and I don't have to do all of my transcriptions from my user interviews — means I can free up time to stay with the problem and come up with creative solutions. Because I hope that it doesn't mean that we have more time to create more solutions or more AI slop and all these different things people talk about, but actually saying: can we spend even more time on looking at the problem, understanding what we want to do and create, and be in that space for longer?

[00:18:50–00:19:24]
And this also kind of ties up with the last one, which is vibe coding and all these different things. It means that from back in the day, where we spent months on building something that was functional, now we can do it in days. And instead of testing with flat wireframes, we can actually test with real solutions that we can give to people, meaning that I can test for actual behaviour and not you being like "yeah, I think I could use that in my everyday life." So this thing of actually testing with real human behaviour is going to be a super important superpower in terms of doing UX and user-centred design.

[00:19:29–00:19:44]
But all of this is well and good — but if you cannot get into the chat, does it even matter? So how do you get into the chat, Nina?

[00:19:38–00:20:17]
Yeah, how do you do that? I'll try to answer that for you. So we are seeing a new reality, and that comes with new challenges. And as Mathias stated, half of the consumers already use AI-powered search tools. We also see that the majority of Google searches today end in zero clicks. That means when a user is asking something on Google, in 60% of the incidents they won't even click on to the website. And that's even though Google came out just recently and stated that they've seen a 17% increase in Google searches last year. So we're searching more, but we don't click as much.

[00:20:18–00:20:49]
We've also seen on a general basis an almost one-third drop in CTR, and we've experienced this too. Last year in May, when the Google AI Overview came out, we saw a 34.5% drop in CTR to our website. So we are facing a new challenge, and this challenge to resolve requires some new demands and new techniques.

[00:20:49–00:21:33]
And these new techniques have a lot of different names already — the way we can see it, it's new, it has all of these names and, to make it even worse, all these abbreviations as well. But most people call this new discipline GEO — Generative Engine Optimization — so we will do that too, to make it easy. So the difference between SEO as we know it from before and GEO is that in SEO it was about being found on the search engines like Google and Bing — if anybody was ever using Bing. And in GEO, it's about being synthesised by the LLMs — being mentioned, being sourced in ChatGPT, in Gemini, Perplexity, wherever the users are using LLMs to find answers.

[00:21:38–00:22:23]
So to understand how to be found on these LLMs, we need to understand how the AI is retrieving answers for the users. Because as Mathias stated, there can be a gap in how you want to portray yourself and how the LLM is portraying you. So to understand this and to understand how to be found, we need to understand the process. So when a user is asking a question, the LLM starts by scraping the content — articles, social media, everywhere your brand is present. Then it goes through this semantic parsing, asking "what is this really about?" — trying to decipher the user's input and making it into something the machine can actually understand.

[00:22:25–00:23:01]
From there it's trying to understand the meaning, the context, the relationship between the user's input and what it can actually find on the internet. From there it's doing this entity extraction — it's looking at who, what, when, where, what is this about — trying to identify organisations and people, and trying to find out how all these data points actually correlate to each other. So if your product data, your service data, or your PR narrative and the content on your website don't align and don't correlate, you might be misinterpreted by the LLM and you won't move on to the next step, which is the credibility evaluation.

[00:23:06–00:23:47]
Because here it's looking at: is this source credible? Is this source credible enough for me to use the information to put out to the user? It's looking at the experience of your content, the expertise, the authority, and the trust. So if you pass this evaluation, then you can move on to the next step, which is being sourced as the winner — or winners, because it's not only one source that is moving on to becoming the final answer to the user. Often the LLM is taking multiple sources in, as we also saw in Mathias's example before with the transition design prompt, where we were mentioned alongside Carnegie Mellon.

[00:23:54–00:24:33]
So from here you will be synthesised into one single answer to the user. This is the seven-stage process that we go through when we are asking a question to the LLM. So how to optimise for this and move on — the answer is here in step number four: the credibility evaluation. So as I said, we're looking at four signals. It's called E-E-A-T and it stands for Experience, Expertise, Authority, and Trust. And this is not a new concept — we know this from SEO, so it's not new. But in SEO we were using it to rank on Google, and in GEO we're using it to be evaluated as credible.

[00:24:50–00:25:37]
So if you want to implement this on your website or your content in general, you can for example use case studies and examples, day-in-the-life or behind-the-scenes content, detailed product or experience reviews — for experience. Content for the expertise: detailed author bios, who is actually writing this piece of content, and about-us pages that are really detailed with everything you can imagine to tell the LLM about your company. As well as content by industry experts or professionals — so if you have some professionals in your industry, have them make an interview with you, or have them make a guest blog post. This is something that can really boost your expertise. Of course, how-to guides — how are we using you as a brand, or your product or your services?

[00:25:38–00:26:15]
For the authority: external mentions and validation. This is also something we know from SEO with link building and backlinks, but here it can also just be mentions of your brand through different platforms where you're present with your name. Also deep topical clusters — if you can really deep-dive into the area that you are working within and have deep topical clusters on your website or wherever you are present, this is also something that establishes authoritativeness.

[00:26:16–00:26:54]
And of course trust: transparent policies — I don't think I need to mention why that is important. Up-to-date content — make sure that your dates are visible on your blog posts or your publications. And of course reviews — this is social proof, something that the user is using but also the LLM is using. So if you lack these signals in your content — not just on your website but wherever you're present — you might not move on from the credibility evaluation, and you might not be synthesised in the answer. So this here is really important.

[00:27:00–00:27:43]
So the reality is this, as I said: a strong SEO foundation supports LLM optimization, but new tactics are needed to ensure visibility in AI-generated responses. Because these are not entirely new tactics — it's not like a whole new thing, GEO. It's on top of SEO, and this is important to remember. So we have created a framework to work with both the old SEO tactics and the new GEO tactics. The framework is called CLEAR, and it stands for Context, Layout, Expertise, Attribution, and Reach.

[00:27:43–00:28:11]
So let's have a look at this CLEAR framework for the SEO tactics. These are still super important, highly relevant, and this is the foundation for your GEO work — really, really important. In SEO we're looking at focusing on keywords, individual page-level optimisation, we're looking at backlinks, domain authority, link building, and optimising for silo channels — where we're looking at how to optimise for Google, for YouTube, for LinkedIn. And of course we're looking at rankings, clicks, scroll depth, and so on.

[00:28:19–00:28:53]
Some people call these vanity metrics, but they are still relevant of course, because Google is still one of the main players regarding search. So in SEO, you could be link number five on the search result page and still have a fairly good chance of being clicked, based on your name or your meta title or meta description. And that's fine if that's what you want to do. But as I said, new tactics are required to be visible in half of the incidents where the users are searching.

[00:28:57–00:29:30]
So the GEO tactics are needed, and here we are focusing on the context, the user intent, the semantic clarity, and we focus on entity relationships. And of course E-E-A-T is a really big player here. We're also focusing on a cross-platform presence, meaning we're not optimising for silo channels but we're optimising for all platforms where we are present — both earned and owned media. And you need to tell the same story. That's what I said before with the PR narrative and the data points — they need to correlate and align. And then we're looking at source links and mentions of course in the LLM platforms.

[00:29:49–00:30:22]
So if we are looking at the CLEAR framework, then we are optimising for the context with these two tops: the layout for page-level semantic clarity, expertise for backlinks and sourced attribution, and the attribution here for rankings and traffic and citation attribution, and reach for silo channels and cross-platform channels. So you need to optimise for both. And as I said before — I want to state it again — this is not an either-or, it's both. So if you already have your SEO in place, then you are a long step of the way. Congratulations.

[00:30:31–00:31:08]
So let's have an example, because I mentioned entities a couple of times. So The Little Mermaid is a good example of something that the AI actually sees as a complex web of facts. The Little Mermaid — she is an entity, she is a node of meaning. And this is also step number three in the AI-generated answer we saw before: the entity extraction. So The Little Mermaid has a literary origin, she is a Danish landmark, she's been exposed to vandalism, she's a pop culture icon, she's a Disney movie, and she's a cultural diplomacy symbol.

[00:31:08–00:31:45]
So this is a concept that the AI can reason about. And in the SEO world she was just a keyword, but in the GEO world she's an entity — a cluster of facts within each node. So when following the CLEAR framework, we don't want to try to own all of the clusters — the whole cluster — we don't want to do that. But we optimise to the context that matches our users' intent. So let's say we are a travel agency or tourist company — then we want to optimise for, for example, the "Danish landmark" node within this entity.

[00:31:54–00:32:22]
So then we have to make sure that we are optimising — this is our context — and then we have to optimise for the layout and the expertise and the attribution and the reach as well. So by using the CLEAR framework, we're optimising both SEO and GEO to become the single source of truth for this node within this entity. I hope you're with me.

[00:32:23–00:32:57]
So does this even work, Nina? I'm super glad you asked — let's have a look. Because if we're taking an example of our Morgenbooster — the one you're at now, if you're in doubt — this could be considered our entity. And each topic of our Morgenbooster is one of the nodes. So before, we were looking at the Morgenbooster as an event with a landing page and a video — a Vimeo video, actually. But now we're looking at it as a content engine, where we are doing a transcript on a landing page, a LinkedIn article, a blog post, a social post, a presentation video, and a YouTube video.

[00:33:04–00:33:44]
And one entity is becoming seven assets — seven content assets. So it's about being present where your users are, but also where the LLMs are searching or crawling. And I know for example that YouTube is one of the main channels that are used in the LLMs. That's why we are of course on YouTube, where we have optimised the videos with chapters and descriptions and so on. So we're telling the LLM — all of these places — about our entity and about our node. For example today, which is GEO, or "website in an AI world."

[00:33:45–00:34:25]
But we have also optimised our website on-page, where we have created topical subpages with links and TLDRs for each of our service points — like transition design, no-code/low-code, vibe coding, systems thinking, and so on. So we have created these deep topical pages. Then we have created citable text passages — these small answer nuggets within each page that are easy, retrievable text passages that the AI can just pull out and serve to the user.

[00:34:26–00:35:00]
We have also created headings that are similar to the Google AI Overview. So if you have tried to search in the Google AI Overview, you've probably seen that it comes forward with key principles or key facts. So we have created the same headings to just tell the LLM: this is a key principle, you can pull it out to the user easily. Of course, glossaries and FAQ modules are also implemented on the site, because this is structured and clear facts that are just easy, retrievable information for the LLM to use for the user.

[00:35:02–00:35:36]
On the technical side of things, we have updated our schema types on all of our pages and modules with FAQs, FA Events, organisation and about-us schemas, because this is a language that is easy for the LLMs to understand — this is the language they read. So of course we have to optimise for that. Then we have also optimised our robots.txt with an LLMs.txt file, because this is just a file telling the LLMs how to crawl a site.

[00:35:38–00:36:13]
So I promised you I would say if this was working — and it is. We've seen a 50% increase in traffic from ChatGPT after we implemented these techniques. And we have also seen — as we saw from Mathias — a lot of mentions and source links in all of the platforms: Gemini, Perplexity, ChatGPT as well. And all of these results we actually started to see in just 14 days after we made the implementation. So it's going a lot faster than with SEO. So this is really good news.

[00:36:20–00:37:05]
So where does that leave the website and the role of the website? Wrapping up what Mathias said and what you have just been told by me: your website is going to be the single source of truth for your entity within the nodes that you are working in. It's going to be your content epicenter — from where your content originates, from where it springs out to all other platforms. It's also going to be the updated knowledge hub, where both users can go in and validate the information that they are getting from the LLM, but also where the LLM can go in and look at new, relevant information to retrieve to the user.

[00:37:11–00:37:42]
Of course, it's a place where you can establish a connection and a relationship with the user by creating branded experiences — you must not forget about the users. And of course it needs to be optimised for this dual layer: SEO and GEO. So the role of the website has not disappeared, but the job description has changed a little.

[00:37:38–00:38:42]
So before we move on to the questions, I just want to make a quick shout-out for our course, class, and network where we are going to look at how to design for complex challenges. We start on March 3rd, so we're going to look at how to tap into your organisation's complex challenges, and if you want to work with them through business design, future thinking, and prototyping, you can in this course. Of course, if you feel inspired today and you want to share this with your organisation or your team, we can come out and do a little quick inspirational session with you — it could be about the topic of today, but also like accessibility, branding, and greenwashing, or whatever is currently on our minds. And of course if you want to get a GEO audit, we can also find that out — so you can book an audit of your AI visibility on your website. So if you're interested in any of this, you can just get a hold of Mathias or me.

[00:38:55–00:39:33]
Any questions? Yes, over here. It's maybe a very loaded question, but — so you're basically saying AI is destroying the planet, it's amplifying inequality, and then in return we get like slightly better prototypes and more time to create things. Yes. I mean, I think if I ask you: do you think that's a good return on investment for the planet? I know the answer. But what do you think — what is the counter, what do we do about it?

[00:39:31–00:40:24]
That's a good question. I think what we can see from how far it has gotten — with 50% search through AI just last year — is that it is an unstoppable force right now. And then it's figuring out how we best exist within that. I don't have the answers necessarily, but it is of course a super big, complex situation. It is coming, and like now when you Google, you cannot deselect it — the AI option is just going to be there. So we also see — I don't know if any of you have seen — there's this called the "Max Gifta" movement, which is this new campaign about people trying to get out of American technology and technology in general.

[00:40:15–00:40:57]
So maybe that's something that's a response to all of these considerations — can we actually go out of Meta and LinkedIn and also ChatGPT and all these different things? Is it actually possible to go out and have a tech-less life? Maybe we're going to see more of that. But there's always a reaction to when something becomes large — there's also people pulling in the other direction. So right now we don't know what's going to happen, and it happens so fast as Nina says. So I don't know where we are in six months, but at least for now we can see that it's just powering through, with a lot of people maybe not considering a lot of these things because they are super hyped about all of the possibilities.

[00:41:01–00:41:36]
And I see a lot of people doing a lot of cool stuff with the vibe coding and all these things in the optimising space. I just don't necessarily see yet that it's creating a super good impact for planet or people yet — besides just optimising your daily tasks. And if there's somebody who does something that actually switches that around, and we can talk about AI for good — because the impact of AI is less than the good output it gives to somebody — then we'll see. But for now I don't see it.

[00:41:42–00:42:35]
Yes, the AI has been quite sparing when it comes to ads and commercials on the channels, because it's quite new technology. But every day you read that somebody is selling out. And when it comes to Google's AI and all the AI behind there, it's quite clear that somebody is paying money to get into the consideration of Google's AI machine. How do you see the whole ad monster coming in the future to affect all this? Because you can be very good at structure and hit the right words, but suddenly somebody comes and puts a layer on top of that that's called "pay me some money."

[00:42:39–00:43:20]
Yeah, so they are already starting to implement ads. We've seen it in ChatGPT, where they started to roll it out in America, where the ads are placed by the user's intent. So if you search something about Mexico, you will get an ad for salsa, for example. But it is based right now on impressions, which is really clever by ChatGPT, because they need money right now, and the more impressions they can get the more money they will get. So it's not based on clicks. But it is definitely something that is really starting to roll out, and it will also take over — I'm sure about that, because we've seen it with Google. So of course this will happen as well.

[00:43:27–00:44:07]
Why not make your own agentic UI version of your website rather than relying on optimising for ChatGPT? Because it's not impossible. But you're right — like ChatGPT is the new Facebook, more or less, for everything. So where are we in terms of how easy it is to just make your website into basically a chat version? I think it depends on the industry, because I don't think you can — it's not a one-size-fits-all solution with ChatGPT. I'm sure that websites for e-commerce — if you have a shoe store online, for example — the website will probably be obsolete in the future. You won't have a need of it, because you just go in to your agent and you shop the latest pair of shoes that you had like half a year ago, and it probably goes by automation.

[00:44:31–00:45:09]
Exactly, yeah. It already is, isn't it? It already is, yes. But if you have a brand — as we saw with the Stray case — it's really important that you can make this branded experience for the user. So I think it's depending on the industry. If you can just put a layer that is just authentic instead of the website — but that's also a thing of being where the users are. So if people are using Claude or ChatGPT or whatever, then you need to be accessible there instead of just creating your own and hoping for people to download that, I guess.

[00:45:10–00:45:44]
Thank you. Thank you for your presentations. And can you share, or do you know of any classic pitfalls that the rest of us can avoid when we try to take shortcuts in creating GEO or whatever — like you also have the concept as the epicenter, and we are prone to making shortcuts. So have you seen anything so far that's like "don't do that"?

[00:45:35–00:46:14]
Yeah, so a lot of people lack the basic foundation of SEO, and that just needs to be in place before you start doing the GEO work. Because that is the foundation, and a lot of the techniques from SEO are also the same techniques used to be AI-visible. So you need to have that in place first, and then you can start doing the GEO tactics. And it's not that much on top, actually. But you need to have the SEO basis. Yeah, so that would probably be a large pitfall.

[00:46:17–00:46:55]
Yeah. When you implemented the strategy for the GEO, did you go and compare different models, different services? We have actually vibe-coded a tool ourselves to try to see how our presence is on the different LLMs. And of course it's hard right now, and a lot of tools are showing up around the landscape in regards to measuring and tracking your presence. We have not made a full report on this, but it's clear that we are showing up a lot more than we did before in the different LLMs.

[00:47:06–00:47:39]
So yeah, and we saw that with the examples as well. And we've seen this 50% increase in traffic from ChatGPT. So I think that's a high number, and that is super relevant traffic, because it's stated that when people actually click — when they're not among the 60% that don't click — they are about 4.4 times more likely to convert. So that's a higher number, so it's really important to get this traffic from the LLMs as well.

[00:47:42–00:48:33]
We've all been hearing about our channel presence for about 10 years, way before AI even. But we've also been hearing about the website as an epicenter or as the content hub. What I've experienced is that in these 10 years, all the resources have been drawn from the website creators and out to the other channels. So do you still believe that the website needs to be a content hub? Because that's not what I'm experiencing. Yeah, I'm thinking the website as the validation content hub. So this has to be the place where you can actually validate the information that is being placed all around on all other platforms.

[00:48:28–00:49:12]
When a governmental institution is posting something on social media platforms, it has the same authority as on the website, even though it's not real estate. So how is the website more important? How do you think that all content should be born on the website for distribution? Why not just distribute on the channel? Yeah, because we're seeing so much AI slop right now and AI-generated content, so the user has to be able to go somewhere to check that information. And one place could be the website as a main authority for the brand. And you have to be able to trust that what you see on the website is actually the truth. So that would be the place to validate the content that you see.

[00:49:14–00:49:30]
So so — the LLM is validating the website higher than validating the Facebook page? Not necessarily, but this is both for user and LLM. So it has to be a content epicenter and an updated knowledge hub for both the user and the LLM.

[00:49:35–00:50:26]
Yes. I come from an organisation that has a lot of factual PDF reports available — we have somewhere north of 1,000 of them. Can you explain how PDFs are accessible to the LLMs? Yeah, so PDFs are downloadable content and the LLMs can read that as well. Of course it will be a little harder for them, because they have to click the button and download it. So if you can have the content live digitally — not as a PDF but as written text on your website — then it will be way easier to pass the credibility evaluation. Much easier to pass the evaluation, yeah.

[00:50:33–00:51:17]
Yes, I'll take a chance and complicate the role of the website a little bit more. Yes, please do. But I think we're a little bit worried about forgetting one very, very important role — and that's the people, the conversion. The website, especially for B2B, your SaaS companies, it has to get people through the door. So I would say: how do you make room for that in your framework, so you don't end up creating something that looks great as a brand expression, robots can interpret it, but nobody wants to read it, nobody converts? There's still a lot of craft in the copy, in the conversions, the hierarchy, all these different things. Is that covered in some of the other things you have, or where does that fit?

[00:51:27–00:52:04]
Yeah, definitely it's important that you don't lose the UX and UI elements. And of course I should say copy is still super, super relevant, and this is also important — and you need to implement that as well. These tools, as we saw in the GEO framework — the CLEAR framework — are for being AI-visible. So when we actually get the click in the 40% of the incidents, you will have to implement UX, you will have to do the great copy, all of the things you did before, still to make the conversions. At least that is how it is for now.

[00:52:04–00:52:58]
And then I think it's about the industry. So if you have an e-commerce business and a Shopify business, then probably the website won't be relevant anymore, so you need to figure out how to be present on other platforms to actually drive the interest. And then when you have the interest of the buyer, making sure that you are visible in the LLM to get the purchases. Yeah, but I think as you said, there's a lot of — if you are a membership organisation like a pension fund or a union, you still need to have people go through a self-service flow and all these different things, so there's a lot of functionality we just can't remove or put into the AI, because there's of course GDPR and personal information and all those different things.

[00:52:52–00:53:37]
But there is the thing of — if I can get aware of all of my rights as someone applying for a job, and I'm kind of confused about all these rules, I can search for that. So what is then the experience I want to have at my union when I have to go and register? I applied for a job and I don't want to get flooded with rules and all these boring things, because I've had that in the other thing and I know what to do. What is then the super-branded, superhuman experience? I think we can still talk about that, even though we of course need to live up to our standard of having a functional user experience. But we can play a bit around with how human we can be when we have all of the boring stuff hidden in another digital platform.

[00:53:39–00:53:45]
Yeah, yes. I think we are out of time, but thank you so much for coming — and have a nice day.