Property AI Report
00:09:50 · Mal McCallion, Matt Goddard
You know, the court cases seem endless and numerous over there as well. So it's a um yeah, it's an interesting way of trying to unsettle, I guess, the the kind of market leaders and seeing uh seeing how they respond. All right, next up um on our property news, we've got some innovation from a major agency. So Strutt & Parker, owned by BNP Paribas, Bank Over in France. Um they have uh made some significant investments, haven't they, Matt, in uh in their tech. [00:10:13] S1: They have indeed, and it's um following on from the sort of things where you've been lobbying the portals to do for a while, and certainly you've introduced with my porter, isn't it, which is the natural language search um approach to it and transforming that search approach on the website. And so it's it's Tamara Couchman who um I worked with when she was back at Carter Jonas, has been sort of spearheading this as part of them. And um it's good to see that that following through because it could be quite easy for for website redesigns just to go with the the standard approach, right? Do you have a chance to look at it?
00:10:43 · Mal McCallion
[00:10:43] S0: The tire self, yes, briefly. Um, yeah, look, it it it looks it looks good. I think that there is a a more I think there's a there's a move from some of the larger agencies towards this much more. I think it's really interesting, you know. Again, if you actually think about the kind of um the structure of these and uh do you just show your own properties? What happens if somebody asks a question and there are no properties? That's probably the biggest thing to consider when you're thinking about putting natural language on your own site. You know, how do you actually help people to find something? You know, so if they've asked for a three-bedroom property, you know, um with C View near a um near a golf course and uh and one of those things is not present, you need to have a really natural kind of off-ramp that says, well, you know, if you actually remove this particular golf course requirement, then there are six here, but there are none that exactly match your match your criteria. Obviously, with the larger agencies, I think it's it's much more straightforward for them to at least have something coming through. So yeah, I think for for smaller agents it it's interesting how you it's really important that you think about how that happens. But I think more and more, as we see with Strutt & Parker, as we're gonna see with that with many, many other agents, this idea of people having to fill in boxes, you know, and again going back to what we were saying earlier on about the rental side of things, you know, actually putting in those ticking boxes, sending something, sitting there, waiting for somebody to come back, all that sort of thing. Those days are, you know, if they're not got already, then they are gonna go pretty fast because people just don't will not accept yeah, just having to wait and having to make things more generic than they need to be. They want hyperpersonal, they want things that are gonna be really right for them. Um so yeah, I think we're gonna see much more of these, and I think it's great. But yes, I think anybody listening just needs to kind of think through the whole journey because it is a it's it's quite a big change.
00:12:22 · Mal McCallion
It feels like it's almost cosmetic, and obviously the right move and others have tried to make it very cosmetic, but actually it's a really deeper, it's it's almost AI first kind of thinking in terms of what you're looking to do with it. But yeah, bravo to them. Um and yes, good to see uh good to see some uh one of the big agents um really embracing it. All right, that is our property news for now. Let's move on to our AI and tech news. All right, so one particular company has kind of been dominating the uh the airwaves, the um the press last week, um, and that is Google and its parent company Alphabet. So there's there's the good, there's the bad. Um let's start with the fact that they've unveiled some new models. Alright, so uh they brought out Gemini for carbon. So um, readers from sorry, readers, listeners from back in the day, I will remember about this time two years ago was it, I think, when Gemini 3 came out, and I particularly um was very, very, very much raving about um how great it was, and you know, thought that basically Google were were now just gonna run away with that, and everybody else was toast. Um, shows how great my you know specific predictive powers are in terms of large language models. But yes, the large language models every Claude and um ChatGPT came back and overtook. And now what we're looking at is Gemini trying to recover. They brought out um Gemini 4 Carbon, they brought out Gemini 4 Argonne a couple of weeks back. Where do you think this is gonna end up?
00:13:31 · Matt Goddard
[00:13:31] S1: I think the first thing is don't bet against Google, right? Um, first and foremost, even with half a billion knocked off their share price, which we'll come back to. Um it's like you you sort of uh you make the point about, or you know, you don't really know what you're talking about, but it it it basically became they weren't there, were they? It's a recap. They were glue on pizzas, they were doing some really bad stuff as Bard, relaunched, and then suddenly Gemini 3 was here, and it's everyone, including yourself, is like, oh my god. Yeah, here we go. But then we start exactly. And so coming into this year, you would say it was a three-horse race, then suddenly Google disappeared. And 11 months on, 10 months on, they were still very, very quiet. They were releasing odds and sods within their sort of distribution, weren't they? Like tweaking things and like releasing flash flash models. Yeah, little bits to the imaging and stuff like that, but nothing. But then but then in the same time, you know, the two-horse race suddenly use Wonderful came along, and okay, it wasn't as good. Meta are back in the ring, and uh Grok suddenly with cursors acquisition and what they're doing, that it's like you wouldn't put them up there, but it's is that four foundational models, and then Google's back. Um and I think the point I was chatting to you about off air is you know, with with Argonne, once you get a model that is punching at the front line, you can then get into RSI and not the tennis elbow that we we talked about, but the recursive self-improvement, and and that model starts hammering it. I think the really important point about carbon, uh Gemini 4 carbon is the fact that this is their elite model, this is their mythos class model, and to that extent, okay, they've announced it, but they're not releasing it because of the inherent dangers and having to go through the American vetting system and all that sort of stuff. But by the benchmarks, and just a little bit of a sideline on the benchmarks, they've had so there's a particular company called Artificial Analysis, who was very much the LLM arena originally. They they would be the ones that everyone put their models on and they got benchmarks across a suite of metrics.
00:15:22 · Matt Goddard, Mal McCallion
They've actually rejigged their benchmarks this week to um sort of make sure that they weren't being gained by agentic AI, because this year is all about agentic coding and and everything else. And it's actually meant the the Western labs have have shot in front of maybe the sort of the Chinese models, but that's not to say that they they won't close that gap pretty quickly. I think we were saying that maybe the Chinese labs were only sort of four or five weeks behind recently. [00:15:46] S0: Yeah. [00:15:46] S1: And so I think that's dropped slightly, but I'm sure it'll catch up. But back to back to Gemini 4 carbon, it's there, it's going through those situations, and by all accounts, it's on a par with with mythos and um astracy.
00:16:00 · Mal McCallion
[00:16:00] S0: That's right. It's significant, isn't it? So and and I think you know, it is a welcome addition. It does keep pushing these things forward more and more different, um, and just like natural competition. The more the more that's in the mix, the harder everybody presses. I believe it's only a year ago. It's like two years ago. This year is absolutely flown. So but yes, that we've been waiting for this for ages. There were rumours in the summer that basically um four was not up to scratch, that it was, you know, there were massive problems with it. So now that it is here, as you say, you know, once it gets released, it's going to be great to kind of play around with it. Um I think you're also right, you know, never count Google out. You know, it's that that they are so huge. And one of the reasons, and let's talk about the kind of financials. Um, so upon announcement of all of this and and and the kind of big um investor calls that they did last week, they they managed to lose half a billion, so 500 million off their um their market cap, which is the second largest fall, I think, ever. Um, and I think NVIDIA only beat it once, and I think that was when um DeepSeek first kind of popped out. Um, so so yeah, it is a huge, huge, huge loss because basically analysts are looking at this, the numbers, and starting to sweat. And and I think that's a really, really interesting story because what they came what came out on the call, um, Sundar Pichai, um, the chief exec basically said, look, AI mode on your Google search results, just not getting the clicks anymore. You know, people aren't buying the blue, you know, the blue ads because nobody's looking at that. But we're answering the question up front, so therefore it's not going through. And it's it's it's this kind of like, you know, are we going to be cannibalized ourselves? Are we going to cannibalize ourselves, or are we going to be cannibalized by somebody else, you know, out there who's going to come along and eat our lunch? That's not mixing way too many disgusting metaphors. Anyway, so yeah, it's it's it is that the analyst now starting to to consider and potentially price in what happens if Google isn't able to do this.