Google wants to replace your landing pages

Episode transcript

Amelia (00:30.844)
Hey there, welcome to episode three of Fin the Week. I'm Amelia.

Paul Wood (00:39.142)
I'm Paul.

Russell (00:40.43)
I'm Russell.

Pat (00:41.727)
And I'm Pat.

Amelia (00:43.144)
And here we are rounding off another week. So how is everybody?

Paul Wood (00:48.849)
All good. Yeah. Glad it's Friday.

Russell (00:49.858)
Yeah, good, thanks.

Pat (00:50.131)
Really good, busy.

Russell (00:53.4)
Very busy. Happy Friday everyone.

Amelia (00:55.974)
Always glad it's Friday. Now, we said when we launched the podcast that we had some kind of core subjects that we plan to cover over the coming weeks and months, but that there may be some kind of emergency episodes as and when things crop up. And today is one of those, right? So it concerns something new, and that is this patent that Google has applied for and been granted.

Amelia (01:23.472)
Let's get into it. Paul, do you want to explain what it is?

Paul Wood (01:27.515)
Yeah, so the basic premise of it is that a year ago, in January 2025, Google filed for a patent application, and it described a process that, essentially, following a user's search on its search engine, would assess the top result in the search results.

So it would assess the landing page that the person would arrive at, and apply a scoring system to that landing page. And if the scoring system decides that the page isn't up to scratch — for a number of different reasons; it could be that it's hard to use, or it doesn't present a good level of information — then Google would use an AI-generated landing page in its place. So it would take the information from the core page, but generate a personalised result for the user.

Google was granted this patent in January 2026, so at the moment it's all theoretical. It's not actually in use, but there's a lot of chatter online about it. It's caused a bit of controversy because some people aren't comfortable with that idea.

So I think it's worth discussing at length today because, particularly for financial services firms who have to be very careful about how they present information, that sounds like a risky prospect. So yeah, I think that's what we'll get into today.

Amelia (03:11.688)
So this is, as you say, it's been quite controversial. There's been a lot of opinions, a lot of discussion around this. You know, off the bat, what are we all thinking of this? What are your first kind of thoughts?

Paul Wood (03:25.946)
I mean, my initial reaction was that it's not a huge surprise, because I've heard people at events already sort of asking the question: will websites be a thing in the future? Because in theory, a good AI agent could take data and information from businesses and create personalised experiences for people as they browse the web. So I'm not surprised that Google has already started to think about this.

But it definitely causes some concern, because if we think about it from a financial services perspective, you've got a few things. You've got the regulatory concerns if information is presented even slightly in a misleading way. That's a big problem. You've got the brand concerns — you know, companies spend a lot of time and effort creating a brand, selling a story and a vision. The last thing they're going to want is for an AI agent to come and just tear all of that out and present their own thing. You've got then concerns over things like bias, you know, presenting information with the wrong weighting might be a problem.

I've been watching the Hannah Fry documentary series on AI on the BBC this week, and that addressed some of the bias in the insurance industry where, in the US, health insurance claims — I forget the details exactly of it, but there was bias in there that meant that people of, well, ethnic minorities were losing out on insurance claims just because of an AI system that was biased.

And so you can imagine that a tiny mistake that is given a huge amount of power with an AI system that can just spread it far and wide could be a massive mistake in the long run. But yeah, ultimately, I'm not surprised it's going this way. I think things will go this way. I think personalised experiences will be a thing. I think websites will have to think about how they interact with AI agents.

Paul Wood (05:51.92)
But yeah, I think there are a lot of concerns to figure out before it becomes the norm.

Pat (05:56.416)
I mean, I think Google have been doing it for years already. And so social media companies — I know Google got in quite a lot of hot water about years ago for taking news stories from loads of different publications and just resurfacing it in its own news feed. So platforms like Facebook have been surfacing news stories pulled from all over the internet for years as well, Google for years would surface videos within the search results that you could actually watch outside of the platform from which that video originated. So they've always been pushing the limits of what they can do from a sort of copyright perspective. And I see this as just kind of a natural sort of extension of that, really.

And now they've got these large language model-based tools that can repurpose and rework content so that it's aligned for the end user, it makes sense. I can see why they've applied this patent. I think with all this stuff, the devil is in the detail, and it comes down to how it's surfaced. If you visit the Guernsey Financial Services Commission website, and then Google rewrites a bit of legislation on there, and it's not clear to the end user, and it's missed some subtle nuance about how you should run your fund administration firm — if you then go and set up your fund administration firm and you get it wrong and you end up breaking the law and getting in trouble, that's bad. That's very bad.

But if Google repurposed that content with a clear label, or perhaps repurpose it in a way that it's manually triggered by the end user, then it's more acceptable. So it might be that you go to this page and then a little dialogue would pop up in Chrome that says, you know, "remake this page in a way that I can understand it." And it will just completely remake that page using AI. And that's fine because the user completely understands that AI is reworking the page, and the user in theory should also understand that that might lose some of the nuance of the page as well.

Pat (08:26.181)
Because that's what people are doing day in, day out already.

Paul Wood (08:27.538)
Although I'd say, do they understand? I think we understand. But I'm thinking about if my mum was looking for an insurance policy, she wouldn't understand that the page has been recreated.

Pat (08:33.279)
Big assumption, to be fair.

Russell (08:44.238)
Yeah, I think the same as well. If you have a pop up saying, you know, "improve the understanding of this page to match your personal needs" — people just see AI everywhere at the moment. They would just think it's the actual website itself, kind of changing the presentation of that information, not the browser. So yeah, I could see people getting in quite confused, especially if we're talking about using AI and it perhaps hallucinates, although I know that's improving. If it is something in a regulatory space and there is just a minor hallucination, it gets like a word wrong or a rate or something like that, that's going to be — who's liable for that at that point? Is it Google?

Pat (09:43.09)
Yeah, but then on the flip side, if you think about what this could mean for people with learning difficulties or people with accessibility issues — you know, if you struggle to see and you rely on a screen reader, then when you visit a site that isn't well set up for screen readers, Google could completely reshape that page in a way that makes it screen reader compatible. And suddenly you can consume that content. Similarly, if you struggle to read or write and you've got a reading age of 12, something like that, then this tool could rewrite the page down to a reading level that's aligned with your capabilities. So there's definitely positives as well, but it just really comes down to how Google frame it, I think. How they surface this technology. It just needs to be clear to the end user that it has been rewritten and that there are risks associated with that.

Paul Wood (10:53.369)
Yeah, this is probably worth looking at the kind of journey that the story's gone on, because the news came out earlier, like a few weeks ago, that this patent had been granted. Somebody kind of uncovered it and shared it on LinkedIn. And the initial reaction generally was one of, I suppose, negativity — that this was not a good thing. And so I looked into the patent and did a bit of research, and I don't know a lot about how patents work, but from what I've read, the applications tend to be quite broad. They're written in terms that are as broad as they can get away with so that when they do apply the technology they're protected, and that in actual fact the application might not be as broad as the document suggests, because what the process describes is that essentially they could do this for any page — whether it's in the kind of organic search results that you just get in a normal usage or whether it's a paid advert search result.

But then I've read some follow up articles. So there was one from Search Engine Journal, which was very sort of kind of tried to play down the risk. So they sort of described it as, there's a fair bit of drama about Google's patent, but then they went on to say there are no concrete examples in the patent that are related to editorial content, news sites, academic pages, blogs or informational content. So the clues overwhelmingly point to transactional commerce-oriented pages. And that's true in the patent. All of the examples they give are ones of people searching for things to buy. So it's like shopping sites and things like that, and you can kind of imagine — yeah, okay, if it's used in those cases, that if you're literally just trying to make a transaction to buy a widget, then fine, that works.

But then there's the Forbes article, which was much more on the risk-highlighting side. So they were clear to say that this isn't a feature announcement —

Paul Wood (13:09.125)
This isn't a feature announcement, it's just a patent application, so Google's not actually using this right now. But they did say that they had a theory that this lines up with an earlier announcement for Google's web MCP. And maybe — I don't know a lot about that, but Pat, you were talking about MCPs earlier this week and maybe you can kind of unpack that a little bit. But basically, what I understand is that the way they described it is, it's a protocol that lets websites expose structured functions and information directly to agents.

So I suppose in layman's terms, from my perspective, that's: you build a website in such a way that there's like a human version of it, and then there's a kind of data layer version of it that an AI agent could come and make use of. So it's almost sort of saying — in fact, they used a really good way of describing it: it's a way of taking the component parts of a website and reassembling them in a way that's good for the user.

Yeah, I mean, basically, it seems that if this web MCP concept and this patent kind of combine, what you've got is a model for Google to personalise everything using data. But yeah, I suppose we should —

Paul Wood (14:30.875)
Should we start with MCPs? What does that actually mean?

Pat (14:34.857)
So model context protocol, MCP, is just a protocol that was developed by Anthropic. And it's basically a way to — if you run a platform, let's say Notion or Google Docs or ClickUp or Slack or whatever, the chances are you've got an API which allows other bits of software to interact with your platform, pull down data, make changes without a human being involved. So for example, you might want ClickUp to be able to speak to Slack, or you might want Google Drive to be able to speak to, I don't know, ClickUp or something like that.

In the world of AI, the natural next question is: how can I get my AI agent to speak to the API so that my agent can do stuff in that platform? So I want — let's take Slack as an example — I want my AI agent to be able to go through all of my Slack messages over the last day and send me a daily summary. So as a developer, you can either build your own connection to that API so the agent can work with it and talk to it, which is, it's possible. It's easy to do, to be honest, these days with things like Claude Code. But actually it's easier if you have an MCP. So an MCP is a way to simplify the ability for agents to connect to your platform. So Slack would have their API and then Slack would also develop an MCP.

And then if you're in like Claude or ChatGPT or whatever, you can just install the MCP and then suddenly your agent can speak to Slack and access all of the functionality of the API. And it's really taken off. There are thousands and thousands of them out there. I use MCPs to speak to ClickUp. I use MCPs to speak to FreeScout, our sort of ticketing platform. I've been experimenting with a Google Slides MCP.

In the context of what Google are doing now, the web MCP basically means that you'd be able to install your web MCP in your Claude or whatever, and then suddenly it would be able to browse the web in a way more efficient way. You know, agents can browse the web already, but they're trawling through millions of lines of HTML and CSS, and a lot of it's quite unstructured. With this MCP technology, if you ask your agent to go out and research basketball boots for 12-year-olds, for example, right now it would go out and scour the internet using Google and Bing and go through all of these websites, and it would take quite a long time. With this web MCP, it would kind of take a fraction of the time and use a fraction of the number of tokens — just be way more efficient and way more accurate.

So yeah — kind of framing it in the context of part of this wider system to build out personal AI assistants, I think this patent makes a lot of sense, because in order to have a really powerful AI assistant, you need it to be able to research and access websites in a really efficient and structured way, because that will make the results faster and more accurate. And the first firm that can crack the personal assistant space in a secure and efficient way — so that you can just ask your assistant to recommend and buy a pair of basketball boots for my 12-year-old and then trust that it will get it right — the first firm that can crack that is going to be onto a massive winner. So yeah, it kind of makes a lot of sense in that context.

Amelia (19:04.103)
Paul, you touched upon this — you know, you were saying this was first announced, the general consensus, the general feedback seemed to be of concern perhaps, or, you know, negativity. What do you think it is that is that main concern? Do you think it's the unknown? Do you think it's an element of control — that, you know, maybe this feeling that actually even more so, Google's got the control over your brand, whether that is how it looks, whether that is, you know — what do you think it is?

Paul Wood (19:33.965)
Yeah, I suspect it's a mixture of things. I think it's partly just the kind of default human response to a change, which is that change is bad. I like the way things are. But I also think there's a bit of distrust of Google because this ties back to one of the things this will be most used for, I would imagine, is Google Ads. So people who run search ads, you know, adverts on Google search results — the Google Ads system is notoriously fiddly, and they're forever rolling out new features that are presented as really helpful to you as an advertiser, but in reality tend to waste a lot of budget. And that's quite a cynical view to take, but I've kind of written articles in the past about this — that if you dig deep enough, you can find justification for a lot of the features they roll out. But in actual fact, they're so badly designed, in my opinion, that they make it so easy for advertisers to fail.

It certainly reminds me of, like, in Google Ads, the system that lets them automatically create what they call display ads. So visual adverts that you see on different websites — rather than designing your own ones, what you can do is you can upload a bunch of assets, so some headlines and some images, your logo, and the Google system will create a suite of adverts that will then be used across the web. And I've never seen a good-looking ad that they've created off the back of that. And so I would imagine there are a lot of people thinking, "my goodness, like, I don't want Google doing this with my whole web page." You know, I take a lot of care over my landing page and what it says and how it presents the information. The last thing I want is for Google to come along and tear that up. So I think there's just a level of distrust there, which is backed by years of experience of poorly implemented solutions in Google Ads, which the cynical among us —

Paul Wood (22:00.132)
— would say are there to just squeeze more ad revenue from you. But yeah, and I think the other big thing, which is probably something to ask Russ, is, you know, what does this mean for websites in general? Because Russ, you've been designing websites for your whole kind of career. And kind of a big part of a website is interpreting a brand and showcasing a brand. And what does something like this mean for that whole world? Does it change that whole thing? What's your view on that?

Russell (22:39.181)
Well, I don't think people are going to stop visiting websites and apps tomorrow. But I mean, we've already seen — did you refer to it as the great decoupling, or did I make that up? I think it's where impressions have spiked and website visits have kind of dropped. And it was this kind of trend that was happening in analytics. No one could figure it out to begin with, but then I think it became quite obvious that it's because AI summaries are kind of coming into play and pulling websites into that, and perhaps that information being displayed on search is counting as an impression. And there's a whole kind of shift in the way that people are accessing data now, and it really does depend on the website, for sure.

I think however AI surfaces a website, and even if it kind of reshapes a website for your needs, you may find that the kind of savvy people, the savvy kind of research-minded people, would still like to see an original version of the website, just to be like, okay, I definitely know that these are the people I'm going to do business with, these are the people I'm going to, you know, transfer a lot of funds to, or these are the people I'm going to, like, purchase something from.

But there'll also be another whole segment of society that are going to trust everything AI kind of sends its way and just experience that whole journey which has been shaped by them and believe every single page and every single kind of step in that journey. I actually wrote an article and I see that being kind of for kind of key experiences that we're kind of moving into.

I think one of them is the augmented experience, where you could probably choose what AI is changing on the page, like in the browser. So perhaps you could — what people have been using up until now is kind of plugins, especially in the accessibility space. So you've got kind of plugins to increase font size, improve websites for screen readers and any kind of access need challenges that you might have. And now browsers would kind of do that out of the box using AI. So a plugin is a big thing in the past, but it would be that augmented experience. Then for transactional journeys —

Russell (25:11.503)
— which it sounds like the patent is leaning towards. I think people will never leave an AI agent. You know, like a booking kind of process, for example, and we've seen OpenAI's Operator and Google DeepMind's Project Mariner basically produce these — you know, already they're out there. You know, these are their projects and they have that whole end-to-end process working already. It isn't anything new. Then the research and validation kind of journeys, I think, like I mentioned, people would kind of still visit sites and look, you know, in the more traditional kind of method.

And then I think it's really interesting in the branding space because, as you mentioned, Paul, I think brands spend so long, you know, working on the page layout, a lot of user experience research kind of goes into these pages as well. These companies are investing heavily into user analytics and looking at heat maps and scroll maps and changing small button placement on the page, and like wording on buttons and wording in titles and doing A/B testing and all that side of it. What does all of that mean in a world where, for all they know, a big proportion of their users are now having that page completely reconfigured when they visit it?

So all that investment into UX research — how valuable is that if most users are kind of going there and they're having a completely bespoke experience which the brand hasn't put in place? It definitely raises a lot of interesting debates about what the future is going to look like in that respect.

And then also, I mean, you kind of touched on it, Pat, but the accessibility space — just, it must be so exciting for anyone with access needs. Last year, we worked on a project which actually adapted a website. It was actually a plugin — when the visitor visited the page, we created a Chrome plugin that adapted the website and made it easier to understand for people with cognitive access needs. And it's that kind of plugin, that adaptable interface idea, and projects like that — I can see them all being superseded by this work. We did nine user testing sessions as part of that project, testing with the bookends of people with cognitive access needs. And the actual community was the IDD community, the intellectual and developmental disability community. So we had people with really severe disabilities. There's one person that we tested the website with, he couldn't even read a word on the website. He could only understand what the website was telling him by visuals, what was in the picture.

So for someone at that extreme kind of bookend on the disability community, could you imagine converting a data-heavy or a text-heavy website into something much more visual using images and video? So they could actually understand what their website was about. And that's really important for them in their daily lives. It could be finding out important government information. It could be finding information from their bank. The whole world would be transformed if there was this kind of generative website experience in place based on your personal needs.

So yeah, I think I can see why people are worried, because, you know, it feels like it would be monetised. And I think that's perhaps the scary proposition that's heading our way. But I guess there's a lot of good that is going to come from it as well. And yes, I'm worried about where that leaves services like UX design, because how can we be too precious about, you know, sweating over every small little change to a website if at the end of the day that user is then going to kind of come along and change that? I guess perhaps there's user behaviours we'll have to look at probably changing what we're kind of measuring and where we're investing that time, because there's probably underlying user behaviours that would carry through on the traditional site and the modified site that we can continue to evolve and track. So yeah, we'll have to adapt our processes accordingly.

Pat (30:00.896)
I think — even if Google is transforming your online presence and adapting it to the end user, I still think the age old adage of junk in, junk out will still apply. Having done a lot of work with AI and looking at how it summarises things and crunches things, like, the better quality of the input into these systems, the better quality output as well. So I do think, even if Google's rehashing your content and turning it inside out and presenting it in lots of different ways, if the original source of that content is still very well made and very well produced, it will still produce better outputs as well. And it might be that, actually, if the UX of your original source material is very good out of the box, Google might not need to switch it around. So actually if you get it right straight out of the box and from the get-go, perhaps that means that your website, your brand, will be presented online kind of unedited, or have a higher chance of that happening. I still think it will still be very important to have very high quality source material for sure, in terms of —

Pat (31:28.447)
— the written word, but also the layouts and the UX and accessibility and everything as well.

Amelia (31:34.92)
Yeah, I was going to say, on that note, because I was chatting to some of my friends about this, who work in the wedding industry. They work with a lot of very small suppliers who have, you know, probably very small — probably made their website themselves, perhaps even. For those smaller companies, smaller organisations, is this better for them or is it worse? Are we still going to be penalised because you don't have a big team? You might not have people who know about SEO and all those things — is it gonna work in their favour?

Paul Wood (32:12.138)
Yeah, I think it's on a knife edge. A bit like — I come back to Google Ads as an example. Google Ads is a self-serve advertising platform. So it makes itself available to those small businesses, and it sort of sells itself as a sort of, "you can use this yourself and you can get up and running and be on Google in a day. Just put some money in there and off you go." But it's also built for enterprise level advertisers who are very sophisticated. And as a result, what it's become is this very complex system that, if you do everything right, works brilliantly. But if you just dabble in it, you can blow all of your budget without getting anything out of it.

And I fear that this has the potential to be a similar scenario, where if you're a small business you have to think very carefully about where you spend your time and your money, and therefore you probably build your own website. But what you won't necessarily have is the ability to structure it technically in a clean enough way so that an automated system can interpret it correctly — which is kind of back to the point Pat was talking about, where if you've got good quality coming in, you get good quality coming out. But if you don't necessarily have the technical prowess to organise everything, you might find that working against you. Having said that, I mean, these systems are just getting better and better all the time, so I think there will be solutions. But yeah, I think it presents a challenge. I think this will suit enterprises. It won't necessarily suit smaller businesses.

Pat (34:10.409)
It's similar to SEO Schema. So SEO Schema is structured data that you put in your web pages that is invisible to the end user, but it gives Google's spiders, the crawlers that help to populate its search index — it gives these crawlers context and structure, and tells them that this product is in this category and, you know, is of interest to certain users, and it tells them that this company is called X and their address is here and this is their main phone number. And I think firms that invest in their SEO generally have better SEO Schema because they've paid a developer to set it up properly, and as a result they perform better in search.

And I think the same will be true when it comes to surfacing your website in AI search results, or making sure that your website is interpreted in an elegant way and represented in an elegant way by Google. And that will probably come down to making sure that technically your website is structured in an efficient way. Maybe surfacing your content via an API or via a channel that an MCP will interpret much more easily.

So it might be that more advanced companies that spend a bit more on their site might actually end up surfacing the site via two channels. You've got your human channel and you've got your agent channel. The human channel is with pictures and layouts and your classic user interface. And the agent channel is just JSON. It's just text or Markdown.

In layman's terms, JSON is just a structured way to transfer data, and Markdown is a very basic way of structuring written content. Yeah, it might be if you've got the funds to create a Markdown or JSON version of your site, you'll generally probably perform a little bit better in these kind of AI driven searches, because the agents can consume that content much more easily without having to sift through mountains of JavaScript and HTML in order to get to the actual content underneath.

Russell (36:43.298)
Actually, hearing you speak, I'm thinking — what we'll probably have, when we spend a lot of time putting proposals together for websites and apps almost on a daily basis — I think you're right. What we'll probably be offering is that kind of visual representation of the website, which actually, what we'll probably be saying to clients is, it's going to score highly. So there's less chance of Google manipulating what it's going to look like to the end user, and all of a sudden that becomes a USP, because, you know, over the years we've become quite skilled in, you know, designing and building these websites.

So we would like to think that, if that is the case — if it is a scoring matrix — we'd like to think that we would design something and build something robust enough so it's perceived as they want it to be perceived by the end user, which is quite a strong pitch to someone using, say, I don't know, like, a kind of — it's a bit easier these days, but perhaps not having those skills to kind of create a really solid backend, a really, really, really solid front end, like, kind of, we would, so that kind of level. And then there's the code version of the website, which the agent would use, as that kind of secondary kind of proposition.

So I think, yeah, there's kind of two things there, but it's probably going to change the way that we kind of sell websites as well. And actually, kind of going back to that scoring matrix — what would be quite good is, I don't know if you use kind of Google Workspace, but if you use Google Slides, you've probably seen "beautify this slide" quite a lot kind of pop up at the bottom of the screen when you're creating slides. And to be honest, I've clicked on it before. And it's basically transformed my slide into something which looked like it was straight out of the 1990s and looked awful. Yeah. So there's no guarantee that, you know, how well is this going to work, if that's anything to go by. And also, how do we stop that "beautify my website" or "change the website" popping up for clients if the website is well designed and built enough as well?

Pat (38:45.393)
Uglify. Uglify my slide.

Russell (39:08.459)
So perhaps there's a kind of "test your website to see if Google is going to change it" kind of idea for like a kind of testing kind of process or testing app. Although I know Google keep these algorithms kind of really top secret, and it's quite probably going to be quite, you know, difficult to find out what's going to pull that trigger. But there's, you know, certainly going to be different mechanisms to take into account now as we're designing and building, and also sending websites to clients.

Pat (39:43.262)
I think there's also, like, thinking about it, a massive privacy angle here as well, because right now Google acts as the signposts and they send you off to a website, and all they know is that you've gone to that site. But they don't know what you're doing when you're on that site unless they're tracking you with cookies. But, you know, those third-party tracking cookies are becoming less prevalent these days. But imagine if you've got this Google filter running in Chrome and it's filtering everything you're seeing, and hoovering up every single web page you look at, processing all of that data and crunching it into a new form for you and presenting it back to you. As well as the benefits of having — obviously you've got the benefits of having it presented back to you in a more readable form, but actually, Google are now effectively monitoring every single interaction you have online.

And so suddenly the mining of data, the mining of your online behaviour, becomes absolutely enormous. It's stepped up a grade. And I guess the temptation from Google to start using that data to build out more detailed profiles of you, so they can advertise in a more targeted and effective way, will be incredibly strong, and it would be very tempting to do that.

I think out of all the big tech firms, I'd be very worried if Meta was introducing this technology, because they would just do it hands down without hesitation, irrespective of the law. I think Google are probably a little bit more ethical. They've had their moments over the years. But I think from a privacy angle, I definitely wouldn't be comfortable with them effectively looking at everything I'm looking at and processing it all — going through, you know, presenting all of my emails back to me, presenting all the articles I'm reading back to me, all the stuff I'm buying, all the research I'm doing on various ailments that my family and I have. It's, yeah, like, a privacy issue there.

Paul Wood (42:02.784)
I mean, there's stuff out there that suggests that that's what Chrome is and has always been — a sort of data mining tool. It's like their eyes to the world. But I suppose it's not necessarily done on an individual person basis. It's more so they monitor how the web is used across the board.

Paul Wood (42:31.956)
Yeah, but I think that'll continue, won't it? But I think if I think about this whole concept of a web for humans, or a website for humans, and then a data layer for AI agents — the idea kind of excites me, because as someone who, you know, I work in a technical industry, I get a kick out of having information organised cleanly. You know, if a website has a clear structure and content is organised properly, that means I've done my job properly. And so the idea of having a layer to a website that is just clean, organised information that could then be repurposed, it kind of appeals to me as a concept, because I think you can think of different use cases. Like, you could be in the car driving home and remember that you need to apply for a car insurance renewal or something. And you could then chat to your AI agent in the car and say, "I've just remembered I need to go and get my new car insurance policy. I'm looking for something that does this, this and this. My renewal quote has been this. Can you kind of go and take a look at what's out there?"

And then it could talk to you and say, "I've looked at these companies, this one looks quite good because it fits with what you're looking for. I've used their sort of data layer to understand everything. Here's their current fees and blah, blah, blah." You know, as long as that information is all properly organised, it comes back to that thing I think we might have talked about in a previous episode, of the top of the funnel, the kind of sales and marketing funnel where the top layer is research. AI is increasingly going to sort of take care of that stage for you, or with you, and then you'll kind of interact with a brand later on in the process. And I think this whole movement that the patent describes goes hand in hand with that, and I think that's quite exciting if it's done well.

Amelia (44:49.104)
And, you know, as we said, this is clearly just, you know, an idea at the moment. When and kind of if this comes to fruition, we'll have to see. But if there's kind of a take home to any sort of brands who have heard this and they're kind of sitting there terrified thinking, "I don't know what this is about. I don't know what this means for my brand" — what would you say?

Paul Wood (45:08.048)
I think right now, nothing fundamentally changes, because the rules are: present your content and your information cleanly, well structured, have a website that is kind of well built, doesn't have a lot of stuff in the way, is easy to use. The kind of common rules of user experience and technical kind of health, I suppose. And I think that continues. You know, you just have to do that well. And if you do it well, you're reducing the chances for an AI agent or a bot to come along and misread things and use things in the wrong way. So I suppose that means work with a company like us.

Amelia (45:59.048)
Perfect. Perfect segue there. Shall we have a bit of jargon busters?

Pat (46:03.625)
Quality in, quality out.

Paul Wood (46:10.25)
Yeah, sounds good.

Amelia (46:11.848)
So this is our feature every week where I put you guys to the test, and we choose a term — an industry term from our jargon buster list — to see how well you know these kind of terms. So this week, that term is fallen angel.

Amelia (46:32.956)
What do we reckon? Does anyone have any idea?

Paul Wood (46:38.156)
I have no… I've never heard that as a term. If I had to guess, it would be along the lines of — well, we spoke about unicorns a couple of weeks ago, and they were like highly valued businesses. So a fallen angel, like a former unicorn that has since died.

Amelia (46:39.663)
Lot of blank faces.

Russell (47:05.269)
Yeah, oh dear, I was actually thinking — I can't help thinking, for these jargon terms, that I always revert back to the kind of investment industry, and I always start thinking about stocks whenever I hear something like this, and I'm kind of in the same boat. But I think "angel" as a word —

Russell (47:28.595)
— means good, doesn't it, and pure perhaps, instead of like something valuable. So a fallen angel might be a company that started off good but then does something quite devious, and then their price drops because they've kind of turned to the dark side.

That's what I think.

Pat (47:50.268)
Yeah, I'd have no idea as well. I'd say maybe a firm that perhaps was valued very highly based on its potential and then ultimately completely flopped because its potential was just not true. The one firm that comes to mind is, what was it, Theranos?

Pat (48:17.459)
The blood testing firm that promised that they could detect all these different kinds of cancer through pure blood tests. And it was all a complete web of lies. And then the CEO got thrown in jail and all of the value was just completely lost as a result. Perhaps that, I don't know, as a guess.

Amelia (48:39.432)
I mean, yeah — fallen angels, so it means a bond that was investment grade, but has now been downgraded to junk.

Pat (48:51.135)
Okay, we're on the right tracks.

Russell (48:51.789)
Okay, so we weren't a million miles away. It wasn't so much to do with good and bad then, after all.

Amelia (48:52.134)
There we go. Yeah, along the right lines.

Yeah, exactly. Exactly. But, you know, we'll have more of this next week. And speaking of next week, we're going to be discussing whether AI models are only as good as the data you give them. So Pat, we're hoping to answer the key question: can I literally use AI for anything, or will it fall at the first hurdle with bad data?

Pat (49:24.249)
I'll answer that question now: probably yes. We can debate that in more depth next week, looking forward to it.

Amelia (49:32.38)
So as I say, we'll get into that more.

Paul Wood (49:32.852)
Yeah, it ties in nicely with what we've been discussing today actually, doesn't it?

Amelia (49:38.982)
Yeah, exactly. So we will get more into that next week. But yeah, thank you so much, everybody. Take care and we will see you next week.

Paul Wood (49:49.43)
See ya.

Russell (49:49.613)
Cheers, bye.

Pat (49:50.08)
Cheers guys, bye.