Property AI Report 091 - Silver Lake Sells Family Silver, OpenRent Open Sold & Bots Beat Bolt

Property AI Report

This week covers major moves in property, a fresh wave of AI product changes, and a few strange but telling signals about where the industry is heading. Mal McCallion and Matt Goddard break down what the OpenRent and HomeTrack deals could mean for the market, then move into OpenAI’s ad-supported free tier, computer history, model selection, AI spend, and the growing role of robots and music creators in the AI conversation.

Key topics

• In this episode, Mal McCallion and Matt Goddard discuss CVC’s investment in OpenRent, and what it signals about private equity confidence in lettings platforms and ancillary revenue potential.
• They explore how OpenRent’s model sits in the middle of the Renters’ Rights Act compliance burden, and why that may push more landlords toward full management.
• They look at the tension between OpenRent and Rightmove, especially the earlier dispute over listing fees and market access.
• They cover the sale of HomeTrack by Silver Lake to Providence, and what that could mean for the future of the wider Zoopla / Houseful ecosystem.
• They unpack the proposed mansion tax on homes above £2 million, including likely valuation challenges around land, flats, leasehold, and ultimate ownership.
• They discuss OpenAI adding ads to the free tier in 31 countries, and the tradeoff between lower access costs, better model access, and trust in answer quality.
• They compare OpenAI’s computer history feature with Microsoft Recall, focusing on productivity, task summaries, and automation, but also the privacy concerns that come with screen and keystroke tracking.
• They break down model selection and cost optimisation, including when to use expensive models versus cheaper ones, and why “faster” can actually be cheaper in real-world workflows.
• They react to a ranking of model usage across different contexts, with a recurring theme of choosing the right model for the right task rather than defaulting to the most powerful option.
• They close with the rise of AI in music, including Dr. Dre’s positive view of AI as a creative tool and Will.i.am’s push for “proof of life” in digital music creation.
• They finish with the widening gap in corporate AI spending, highlighting how the top 1% of companies are investing far more per employee than the next tier.
• In Weird AI of the Week, they note that a humanoid robot has run 100 metres faster than Usain Bolt’s record, and also share examples of robots smashing into walls and jumping over two metres.

Timestamps

00:00 - Weekly intro and what’s on the agenda
00:18 - OpenRent gets CVC investment and why it matters
01:04 - Renters’ Rights Act pressure and new landlord behaviour
02:06 - OpenRent’s dependence on Rightmove and future negotiation leverage
04:19 - HomeTrack sold by Silver Lake to Providence
05:13 - Why HomeTrack is the crown jewel in the Zoopla group
07:15 - The proposed £2 million mansion tax and valuation headaches
09:34 - OpenAI adds ads to the free tier in 31 countries
11:20 - Whether ads will pollute trust in AI answers
13:12 - OpenAI’s computer history feature and the Recall comparison
15:06 - Why privacy concerns fade when productivity wins
17:26 - Using computer history to automate workflows and drafts
18:49 - Ranking AI models by use case and enterprise adoption
20:10 - Why the right model matters for cost and speed
21:06 - Agent workflows, token usage, and false economy with cheaper models
22:27 - The “experienced plumber” analogy for expensive models
23:30 - Dr. Dre, sampling, and AI as a creative tool
25:13 - Will.i.am, proof of life, and digital fingerprints in music
27:40 - Corporate AI spending: the top 1% versus everyone else
29:57 - Weird AI of the Week and the robot that beat Usain Bolt
32:23 - Meta’s AI tools and whether it’s too late to catch up
34:38 - Closing thoughts and sign-off

Transcript

Copy Transcript Copied

SPEAKER_00 0:00

Hi everybody and welcome to the Property AI Reports. This is your weekly digest of all things property and AI. Brought to you by me, Mal McAlian, and my good friend and co-host Matt Goddard. Matt, how are you today?

SPEAKER_01 0:11

I'm great. Thanks, Mal. The uh the Premier League's back. Both teams in blue won. Mal, let's go.

SPEAKER_00 0:18

Yeah, really, really exciting weekend. Oh yes, let us, as you say, crack on with our property news. First up, this week, OpenRent um has been um significantly invested in by CVC, one of the big private equity firms. The initial shareholders, the original founders who set out back in 2012, still very much involved. Um it says so yeah, it'd be quite an interesting move. Um, just for those that um uh aren't as familiar with the business model. They essentially charge uh Open Rent charge landlords a reasonably small fee per month to advertise their properties on the likes of uh Right Move and other portals. So um, yeah, pretty much a um uh a kind of vote of confidence, certainly in that business and and perhaps in the sector at large. Matt, what do you make of it?

SPEAKER_01 0:58

It's a really interesting, isn't it? Especially after the Renters' Rights Act coming in and people maybe wondering how open rent's model was going to deal with all that extra legislation and compliance burden. I certainly think from a market point of view, uh the fact that they're investing in it, it does follow private equity investing heavily in the likes of Lohmann, Dexter's, uh, LRG, etc., doesn't it? Yeah. It's the first time. And an interesting one as well for me, because open rent is a bit of um the an unknown, or certainly they don't um put themselves out there a lot in sort of the circles that other letting agents move in, like Adam Hislop and Darius Bradbury are very, very much sort of the the founders, but also sort of stay stay clear of a a lot of the um companies that we sort of talk to from both agency and technology point of view, right? So um I'm sure CBC have seen some opportunities to both grow open rent market share, but probably seeing a lot of ancillary income streams that they can maybe add to.

SPEAKER_00 1:56

Yeah, and and uh like you say, it's um how whether the business model is is going to evolve, you know, are they looking at how they can leverage their relationships with landlords to sell them more products and services? You know, how how does it actually play out with the likes of Right Move? So, you know, they may you may remember last year there was um quite a lot of controversy. I think the the open rent uh contract with Right Move came up. Right move tried to charge them a significant jump uh in an increase in in their monthly fee. And I think open rent then threatened to come off. And in the end, it seems that RightMove blinked and said, actually, you know, there's 20%, I think they've got something like 27% of the market in terms of rentals. So if they were to come off RightMove, that would be a massive hit to Right Move's claims to cover the market. When that renewal comes up again, and you know, maybe maybe it's every couple of years. Um it's gonna be interesting to see how that negotiation goes now that uh it's pretty sure that they've got a bunch more cash. And you know, open rent kind of needs the access via Right Move to all of the for their landlords to to all of the eyeballs as well. So um, yeah, it's it's I suspect obviously that that this has been done with a view to you know 5xing the uh the value of the business over the next sort of three to five years. So it's going to be interesting to see how aggressive they play in the market. And obviously for lettings agents out there, that's that could be really quite interesting, also with with Dwelly kind of coming and hoovering up loads of uh lettings agents as well, right?

SPEAKER_01 3:11

Yeah, and indeed. I did I did wonder about Dwell. I was surprised that you found an angle for right move in there as well. But it's um well, it was very much uh a talk of the town, wasn't it, at the time? And um as I say, coming back to the Rentlers' Rights Act, uh the sort of the idea that there's all this extra legislation, and really open rent is seen as a landlord advertising direct to portals, and that was the biggest issue, wasn't it, with WrightMick? So um a lot of agents I speak to are actually finding landlords um sort of upgrading to man full management to deal with that extra compliance burden. And um maybe they are looking at dwelling and thinking, is there some tech in there that open rent can can utilize and will they start doing a dwelling buying other rent books and natural letting agents up? It'd be interesting to keep an eye on it, won't it?

SPEAKER_00 3:54

Yeah, 100%. All right. Uh next up, um also in the acquisitions sphere, uh Home Track. Um, once part of the Zupla property group slash um houseful, has been um acquired by uh Providence, another one of the private equity firms from Silverlake. So Silverlake, um, you will recall back in 2018 acquired the whole of the Zupla property group, um, including the likes of Home Track and Zupra itself and USwitch and what became Alto, uh the business for all the property uh software. It acquired that for 2.2 billion. Now it seems to be um essentially selling off the family silver um and home track probably being the most valuable of those um particular assets at this point in time. So yeah, Matt, another big move and again another kind of big investment from private equity. What's your take?

SPEAKER_01 4:41

It's it's another interesting one, isn't it, following on from the open rent and the the PE movements around it. But I think with Zupla, we've known for a while it was getting sort of divvied up into the three three sectors around sort of CRM um, the the portals and the the utilities, and um the fact they're selling off Home Track, as you say, is the prizess asset, isn't it? And I think the quote from Charlie Bryant was um quite an interesting one saying that the Mark's an exciting new chapter for Home Trap. Sorry, not Trap, maybe that's one market leading platform. Um but I think Home Trap was pretty much the market leading platform when they first bought it, wasn't it? And the uh I think it was the one that a lot of people went to for the data.

SPEAKER_00 5:18

Yeah, and and still still do. So, you know, it I think that there's some crazy stat like, you know, that uh 80-90% of transactions have some touch of the data that that this particular business uses. Obviously, they've got a lot of deals with lenders and things like that to to kind of automate the valuation of those those products and that the sorry the properties so they can sell the products in. So yeah, it's it it was always the kind of crown jewels, always the kind of uh the bit that was pumping out money. I think more I think the last stats that were available back in 2023 showed that their profit margins were up at like about 40-45%, whereas Zupla um was down at sort of 20. So uh so yeah, very much um the the the kind of um uh as as you say, the sort of crown jewels. I do think, you know, it kind of indicates that perhaps Silver Legs patience is is starting to to kind of come to an end, and perhaps we all we will see um Alto and uh Zupla itself um being being sold. City AM's most recent kind of uh quote was something like um 500 million pounds, the valuation that they've got for for Zupla. But to this point nobody seems to be willing to to kind of pay that much, right? So yeah, going to be interesting to see whether this is the beginning of the end of Silver Lake's involvement with with that group, with those entities. But yes, I'm sure there'll be um lots more to talk about about it in the coming weeks as we see what happens to the rest of the uh kind of stable. All right. Next up, um we want to talk about mansion tax. So the story um over the weekend from um a couple of broadsheets was around this this two million pound mansion tax and how it's going to be policed, basically. So we are going to have the valuation office people coming round um to your mansions to see whether they think it's worth two million quid or not. Feels a little bit intrusive. Um I'm sure that there will be a lot of complaints from a lot of very noisy people who have very loud megaphones via the press and others, um, that this is um very um very much an invasion of privacy and so on. But yeah, Matt, what's your thoughts?

SPEAKER_01 7:08

I just I wonder if there'll be tags in specific people as part of this, you know, people who get what is it, five five million uh uh bonuses or or gifts as part of the not uh campaign tracks. No, so I guess things have got to start somewhere, right? And um uh in the same way as if you just suddenly start saying every property that was a certain size and uh a certain um uh style is uh in falling into a ban, then there'd be the counter-argument, wouldn't it? That it needs to be uh more controlled and a better better view of it. But um it's gonna cause noise as you say, and it'd be interesting to see how it falls out. Um you expecting a knock on the door?

SPEAKER_00 7:43

No, not this week. So uh I I I think you're right. I think it's gonna be quite interesting to see how they try and deal with some things like land, you know, because is is it just the house or is it the value of the overall plot? What happens with flats, because you know, leasehold and you know, actually, you know, how are they gonna find the people that actually the ultimate beneficiaries of owners of particular properties and and so on? So it's um yeah, as you say, they've got to start somewhere, but it kind of feels like there is gonna be a lot of of noise. I think there's a consultation that starts next year, but the the date they've penciled in is April 2028 for this to actually come in. But yes, there's gonna be um a lot more on this um as we go through the next year for sure. Um, all right, that is our property news for this week. Let's get involved with our AI and tech news. Um, all right, Matt, plenty, plenty, plenty to be talking about um in the tech space this week. We have spoken before about ads coming into um OpenAI. Um so if you are on the um the cheaper tier or the free tier, then you will now find that there are ants mixed in with your results, with your answers. If you're in one of 31 countries that they are trialing this with right now. Um so, Matt, it's it's kind of finally landed. I think that that it was started a couple of years ago when they uh when they announced that this was going to be part of their plan. What what do you think?

SPEAKER_01 9:02

I know um I I called you the merchant of doom off air, didn't I, because of the fact that you'd been saying this is going to affect your results and you need to question the things that are being returned. And I I genuinely am interested to see how ChatGPT, OpenAI stay clear of being influenced in the in the responses. I've not actually seen any of the mock-ups of how it looks because I I use the page here and it doesn't tend to get screenshotted in in the releases. But uh I I know it's taken what uh longer to come out over here in Europe rather than um obviously in the US, where it's been been for a while because of our our data policy. So um I think you know if people are using ChatGPT for free and then you get used to those ads. And if you if you don't want to be seeing the ads, then you can pay for the upgrade. And hopefully everyone who listens to this or most people are actually paying for uh some form of subscription so that they can use the you know turn off using my data type settings that we've touched on in the past. But um, what about you, Mal, uh Merchant of Doom? Thanks.

SPEAKER_00 9:57

Um so yeah, I just it just feels a little bit kind of difficult to to have that line. So actually, even in the results that we get in our paid for versions, you know, if I don't know, um McDonald's is like advertising heavily on ChatGPT on the free tier, does that mean that it is going to get a little bit polluted in some some of the results? And that there is this kind of ongoing, you know, the the narrative around OpenAI is trying to get its IPO together and is racing with Anthropic to do the same. Anthropic has specifically ruled out ads at any point, which is what they will do until they then decide to take ads because it's easy and it's a lot of money. Um so yeah, gonna be interesting to see how um how much they push um that, how hard Anthropic come back at them, and or whether Anthropic just actually kind of keep it a bit quiet this time because they want to do their IPO, they want to keep their powder dry and and you know have a story for their investors that, well, maybe we will go down the ads route in due course, we're not really sure. And but what what ads are providing for people who are on the free tier is the ability to get really much better intelligence. So even if you are on the free tier now, you're getting 5.6, JGVT 5.6, um, because you know, instead of you not instead of that being served for free, you know, and and and every query actually costs money and costs um a compute, she's somebody is now paying for it, which is the the advertiser. So you you you are upgraded to to one of the stronger models. But yeah, still um, as you say, Matt, you always say is is pay 20 quid a month just because then it secures your your data and you make sure that that your data doesn't go off and get trained with anybody else. But yes, I don't think this will be the end of that particular story. It will obviously spread to to other countries, um, including ours, over time. And yeah, yeah, I just yeah, I don't know if it's going to shape people's faith in the answers that they get. And that I think is probably my my real concern is around actually how do we know in the end whether these answers are right or not. Um so yeah, we will uh we will see how it all goes. So on uh OpenAI, just because we were going to try and mention last week, um, computer history. This is um certainly one that I know you've been following around how um OpenAI is is sort of um giving you the opportunity to track what you're doing and then replicate it, right?

SPEAKER_01 12:10

Very much. And then just on this one, um it's it's another one that's only been released on Mac OS at the moment, so you'll be able to use it on your your Apple. Now you're free of the the Windows Microsoft of uh approach that was coming in for you for you guys. But uh it's it's a case of it's similar to uh is a follow-on from Microsoft Recall, which if you recall from uh a year or so ago had um massive kickback because it basically records everything you're doing on your screen and takes regular screenshots to to help it learn how how you work. This my um computer history from Chat GPT does a similar thing, but the idea is that it can then start suggesting to you what you've um it could be doing for you um from a genetic point of view, but also helping you sort of say, you know, what have I done today and getting a summary of the tasks that you've completed, also what have I started and not finished, which I'm sure anyone who uses codecs or or co-works quite regularly finds themselves in, you know, which thread did I need to follow on from. And um just generally sort of sat there as a way of helping even you teach teach it how to use new skills as well. And and this is something we touched on Grokbot last week, and and one of the big features of Grokbot is being able to say, uh look, watch me do something and then pick it up and and carry on from here on in. And that's similar with this computer history, you'll be able to get it to teach you, um, and even at the end of the day, review what you've done and suggest new skills and automation. So it's one I'm quite interested in, um, whether I go out and buy a Mac to utilize it or um what about yourself, Mal?

SPEAKER_00 13:39

What I do think is is really interesting is the kind of changing narrative around this sort of stuff, and it's it's always the way, right? Something happens, so you know, the recall from uh Microsoft happened, and everyone went, Oh my god, this is so intrusive and invasive and like privacy, and you know, how can it watch my keystrokes and you know, record all this stuff and take screenshots? And then, you know, fast forward a year and a bit later, and everyone's like, Yeah, it seems pretty good. Seems like just the thing that I really need in my life right now. So, you know, it's it's with privacy, and it's happened with social media, it's happened with all these things that you know previously people were up in arms and pearl clutching, going, Oh my god, this is like so bad. And then suddenly it's like, yeah, it seems all right. And I think this is also true if you look at um things like um recording meetings. You know, again, when that kind of first started, everyone was like, Oh, I'm not sure I like all these things in here, but now it's like, yeah, really helpful.

SPEAKER_01 14:25

So actually, do you think with that one that's done full circle though? Where people started saying, Yeah, these are good, and now a lot of meetings I joined, people just have it locked out completely. So, do you think there's maybe that's more than I'm dealing with enterprise style uh customers? But the bots on meetings specifically seem to be fading away. But I do I do get what you mean. It's like people initially were like, I'm not gonna save my car details in Google Chrome, and then suddenly it's like you you've got Shopify stepping in and and settling up bills for you. It's like just that. Well, you know, you feel comfortable with them um and it just becomes part and parcel.

SPEAKER_00 14:57

Yeah, it is interesting, except to your point about you know the kind of um enterprise people wanting uh bots in there and not. The ones where where I've had people kind of kicking them out, it's more for a hassle than anything else, really. It's just like, you know, actually not wanting and and and and that they're paying. It's like retail and various others that are really quite intrusive. Whereas, you know, there's a lot of them on Zoom, for example, it just it just records or Teams just records and it's it just happens. And almost almost without you really noticing. And I I don't know the last time I've asked anybody when whether they're recording it or not, because it's just kind of there. So yeah, look, I don't doubt that there are um yeah, privacy concerns um still around um some of the this stuff, but I I kind of think it's that horse has sort of bolted a bit, right? I I think people are kind of just going, well, you know, actually we do need to make these notes, and you know, if it's like yeah, if it is a particular um platform, then you know, at least I trust it and and all that sort of stuff. So yeah, no, it's um I and this particular um uh this particular uh profile computer history, again, you know, haven't tried it yet, um, will do. Um but the idea that essentially it can replicate my keystrokes and just do things much, much, much faster is is certainly attractive. We've been looking at a lot of productivity here and it kind of bleeds into the next story because we're going to talk about models, um, where how can we essentially max out productivity? You know, what what are the tools? How can we plug in as many APIs as possible into a CRM, our hand built CRM, um, that then actually means that you know, when I when I open up my laptop in the morning, it's got drafts of all of my um emails from overnight ready to go. Um, you know, it's also got um, you know, all of the the bank details in, so we can see the bank account right off the bat. You can see you know how many support tickets you've got and all that sort of stuff and all in one place. So yeah, um I I will give this a go and uh we'll let everybody know how it uh how it kind of pans out. But yeah, I think it's another very interesting moment where what is convenient overtakes what was concerning, you know, not that many months ago. And on that note about models, um, you were reading something very um interesting last week about um how different models are being uh being utilized, right?

SPEAKER_01 17:06

That's it. It's um it's like uh basically it's a a bit of a viral um news article or uh started as a tweet where it was um ranking the top 15 models from sort of A star, as we would call it, and um you did O levels, didn't you? But GCSEs and A levels.

SPEAKER_00 17:20

No, did GCSEs in the number base these days.

SPEAKER_01 17:23

So for S plus with Fable V up at the top, right the way down to E with sort of Deep Seek 4. And it created a lot of conversation, mainly because of the fact that it's just a follow-on from what we talked about over recent few weeks of where people are using them in certain certain situations. So enterprise, for example, Fable 5, absolutely flying by enterprise because of the Fable V 30-day retention policy. Enterprise just aren't touching it, and those that have access to Glasswing are using Mythos anyway. And then the sort of things that we talked about, and I know you've been looking a lot on, is when to use the open weights, what scenarios to use them for. And actually, um there was talk this morning that Hugging Face may well be getting acquired, and and they do a lot of this sort of ranking on what models to use um in terms of the application, and um it's sort of seeing what the uh where people can do this and quite a lot of the noise and conversation that spins out of it, so when to use what, what not to use, what are the concerns about it. I know you've you spent a lot of time on this recently, haven't you, Mal, in terms of what models you use.

SPEAKER_00 18:24

Yeah. Yeah, you're right. It's becoming increasingly kind of sophisticated in terms of you know the the the right model for the right job. Uh, because you can absolutely burn through zillions of tokens and a lot of cash, fable fiving everything. But if it's just like a kind of daily task that goes and summarizes, you know, um your emails or something like that, then that's that's not a Fable Five task. So how do you deploy Deep Seek, for example, which is like you know, 0.006P or whatever, um, rather than you know, Fable Five costing you about a quid for the same job. So um yeah, I've been spending a lot of time actually over the last week um again looking at this kind of from a productivity and also cost-effective viewpoint. You know, d here, what what we're doing, yeah, we're spending you know hundreds of pounds a week, um, if not more um weeks, um, on building and and and trialing all these different models. So so so we're trying switching out from Opus V, which obviously is one of the more expensive models, um, and putting it down to to haiku again, just to quickly recap, I've I've got eight AI AI agents, all with different skill sets, all with different names that I basically run under OpenClaw, so I can talk to them all and they can talk to each other. So so I've I've been changing each of them into different models just to see what they do and and how that impacts the costs. So one of the things I did was last week I I cleared down everybody from Opus onto Sonnet, and what I saw was actually just the token usage rocketed. So they had to use more and more tokens to do the same thing that Opus was doing. In the end, it cost me about the same, but the time it took was way, way, way longer. So actually, that was kind of a really interesting but kind of false economy. So put many of them back onto Opus and now have got CERN of them, you know. So the the the executive assistant is doing is is on one particular model, whereas the IT director is on a different one. And so, yeah, that there's lots of really interesting kind of nuance to all of these different uh things right now. And you know, obviously, you know, we we will share as as much as we discover as as we build. But yes, what I I think my key take out from last week is actually some of the more expensive models are really, really valuable and that that level of intelligence is really, really helpful because if you are time constrained as we all are, actually getting it to do it fast is is is a is a win for the same sort of money, basically.

SPEAKER_01 20:31

This is similar to the analogy. So that this is the the old analogy of the the old analogy of the really experienced plumber or architect that you get get round to your property depending on what you're doing, and they they might stand there, scratch the chin, and solve something in a in a few seconds or a few minutes because they've seen a huge wealth of these things before, versus the person who spends hours trying to work out where the leaks coming from, um and they might be cheaper per hour, but they spend more hours fixing it.

SPEAKER_00 20:58

Great analogy. It's exactly like that. And actually, I I put um a number of them down to the very basic model on ChatGPT Luna and basically just it just started making stuff up, told me very confidently that it'd done stuff, and then I kind of got one of the opus models to check. It's like, nah, I haven't done anything. So, yes, need to be careful about that. Your plumbers coming round and just sitting having a cup of tea, essentially. Right, that is uh all about the models. We wanted to show just how cool we are on the property AI report, didn't we, Matt, by uh talking about some some music, some artists from um back in the day who are very focused on the AI world. Matt, come on.

SPEAKER_01 21:34

So well, it's Dr. Dre was the one that I pulled out, but then you you obviously pitched in with Will I Am as well, who I believe is part of the. So cutting edge.

SPEAKER_00 21:42

Look at us.

SPEAKER_01 21:42

But joking aside, um with the the Dr. Dre one, um obviously Dr. Dre, well, way back in in the days you put it, was part of MWA, and they they didn't write their own music, but they did in terms of utilising samples, and there's huge um sort of uh backlash in the music. Music industry when sampling first started, which is a it's ironic in a way because from very early on, a lot of songs that sounded similar to other songs and and using chord progressions were sort of um ripping off tracks, and you know, Oasis who I I dearly love, you know, Noel Gallagher is definitely lifted from a lot of other people and sometimes been pulled up on it and other times maybe got away with it to it to an extent. But um, no, there's always been this trend of learning standing on the shoulders of giants, right? And learning from other people, whether it's sampling and wasn't that one of their uh wasn't one of their albums. Singular is unfortunately misquoted, so the shoulder of the giant. But um Dr. Dre's come out and said actually, uh in a New York Times article, I think it was, that he he thinks AI is great. He thinks there's a lot of producers and artists who are actually using AI already without fessing up to it, uh closet AI utilisation. And it's not far removed from sort of knowing what you were doing when you were getting samples and picking those out. And okay, you know, when they were doing it, they're finding records in old on in old shops and listening to them and just picking out the bars and slowing down and speeding up. But being able to sort of unleash your creativity using AI is he thinks is really, really good and really exciting. And um Will A M is actually coming in from a different angle, isn't he? Because he's looking at his um wanting to have this proof of life, and again, he's still pro AI for music, but he wants to have almost a digital fingerprint on music that's created as a proof of life, so you can tell if Mao's actually drums um that that track or whether he's he's programmed it or whether he's he's created it via AI, and um that in itself, because he is a bit of a uh an interesting character, Will I am, isn't he? Yeah. That in itself is like this proof of life. If you I don't know, drum a bar and then just loop it, is that proof of life gonna cover the original bar? Or does it get pulled out because of the fact you use it for the rest of the track? I I think it's great, obviously, it's straight into my wheelhouse with music and and AI, but um, I think it's great to see how these other industries react. And you regularly talk about how the railroads followed the canals and how um you know the the films followed the stage shows. I think this is similar for music, and it's good to see people who are absolutely pro it rather than just being doom mongers, to be like the phrase of it.

SPEAKER_00 24:13

I agree, and and you know, so to Will I am actually is he's one of the investors in Udio, you know, which is um one of the kind of music AI businesses and and has been long been a kind of champion for that digital um education um around sort of what what um indeed what technology in general can do for music and how it again democratizes it and enables even idiots like me to be able to kind of make music and things like that. So yeah, really, really good to see. And yes, you know, I'm sure that as you say, that maybe perhaps the younger artists aren't quite as confident, haven't you know, done it all yet, so therefore won't necessarily come forward and say that they're using it. But it's it's good to see some of the uh yes, some of the the the older, uh the older guard actually coming out and saying that this is uh this is a good thing for for music rather than something to be to be concerned about. Um all right. Um final story is around um how much people are paying for AI on a month-by-month basis. And again, Matt, this is something you've um uh picked up um over the last uh few days. How different the top 1% in America of businesses are deploying AI versus you know the next 10% and then then underneath, right?

SPEAKER_01 25:14

That's it. And it's it comes off the back of maybe four or five months ago where everyone was token maxing and those leaderboards of how quickly people are gonna burn through their tokens. And I think that was very much a case of getting people on board to using AI in their work. And I I still know developers who I come across who are like, oh no, I'm better than AI, I'm gonna crack on and do this myself. So like token maxing was a phase, and it's amazing how quickly these phases pass, isn't it?

SPEAKER_00 25:38

It really is, yes.

SPEAKER_01 25:39

It's like a nanosecond. Yeah, exactly. It was May, and we're now at the end of August, and already that's the thing of well, why would you even have that?

SPEAKER_00 25:46

Yeah.

SPEAKER_01 25:46

That's on our sort of you know conversational side of things, right? I generally do go into companies who are tech companies who are like, no, no, we're not going to utilise that. And I guess I'm lucky enough to see there's the broad gamut of uh companies are doing it, and Uber as well. They they burned through their AI budget by April because of the fact that the employees would absolutely hammer it. And I thought what was quite stark here is corporate America's top 1% are spending $7,400 per employee per month on AI. So that was figures from July. Whereas the top 10% are only roughly $650. So it's really sort of stark contrast between that top 1% and the remaining 9%, which I expect there's some in that remaining 9% that are really low, yeah, and dragging that 650 average down.

SPEAKER_00 26:32

Yeah, I would imagine it's yeah, almost quite single figures, single double figures, um, a bit below that. But yeah, it is a really interesting uh kind of contrast because this per employee. So even though they are like massive, massive top 1%, you know, they're spending millions and millions on helping their employees to use it. So I guess the kind of the big headline from this is if it's good enough for for those businesses, then you know, not necessarily to the tune of 7,500 or whatever it is per per employee, but certainly, you know, perhaps you should be looking at the the thousand um you know per month just just to see what happens, just to see how you can use it and what what you can possibly deploy this stuff for. But yeah, it's just uh it is fascinating to me that actually it there is just that stark divide um and and the fact that the top 1% see the value in it already and are really, really going for it, and perhaps you know we'll pull further away as a result. Right, that is our AI and tech news for this week, and we are going to take a little wander through our occasional series, Weird AI of the week. Right, um, this is my favourite story this week because the bots have beat Bolt. Um so we have a robot that has run the hundred meters faster than Hussein Bolt did. Um Bolt's record um has stood since and was it Beijing or was it Tokyo? I can't remember. Yeah, it's it's it's um uh seven nine, I think, seconds is the fastest time anybody's ever run uh the hundred meters. Um this bot did it in 9.58. So yeah, really, really quite um quite a moment, right?

SPEAKER_01 28:01

Incredible, isn't it? And these are very specifically these humanoid robots, right? So they they are running it. It's not that they've got like a wheel at the back that they're leaning on and getting pushed along. But I know you also really like the other part of this weird AI of the week, which was bots running into walls, right?

SPEAKER_00 28:18

Yes, so it's and it's the robotics championship, the second um annual robotics championships over in uh in China. So plenty of very, very impressive um Chinese bots on display, but also some really, really good, fun ones where they run so fast that they actually end up sort of running into walls and blowing themselves up, things like that. Another kind of big milestone this week, and again, if you can find the video, it's it's um it's it's incredible. Those that subscribe to the Property Air Report newsletter, well, I've got it on Sunday, um, but it's a video of a robot jumping from standing star, jumping over two meters high. Right? It's it's it's insane. And kind of like you think, wow, it can really jump over high walls now then. So that prediction's out. But yeah, it's um it's it's phenomenal um what these things are doing. If you also look back a year at the first robots championships and you see people like humans running alongside roads, you know, kind of um uh basically um uh moving the robots themselves remotely as they run along, and then they're all kind of falling over and clanking out. But now the change in a year is insane. So if we look at what's gonna be happening in a year's time, it's gonna be even more insane. But uh yes, um, those it's fun and games. Um the robots are uh doing some some superhuman things already. So I think it's just gonna get um better slash worse depending on your perspective. All right, um, that was our weird AI of the week. Now, tool of the week. Matt, this is Meta. It is, and it's Meta on a Mac. Yes. Sorry that all Macs get all the uh all the fun stuff first. Um but yes, Meta's um AI wants you to talk to your apps or wants to talk to your apps on um on on your behalf. So yeah, this um again, Meta hasn't really been at the cutting edge until quite recently, right? Their superintelligence unit has uh built out some some some pretty impressive stuff. You know, again, Grok has has has kind of come to the party as well, so there's some some outstanding developments taking place right now. Yes, what's your kind of um feeling around the kind of uptake with with Meta and is it too late for them really? And I think it's too late.

SPEAKER_01 30:16

Um I think it's dangerous to cancel anyone or count anyone out given what we saw happen to Gemini uh a year ago and then what managed to happen in to Gemini in the last few weeks. And as you touched on, you know, even a couple of weeks ago or months ago, yeah, let's go months, um, you would have thought it was a two-horse race, and then and then Grops come well back in. And um, like with Grokbot, it seems like people are targeting very much sort of a a more personal or or even sort of uh more more um mass market approach to to AI usage. I think meta is it's targeting the enterprise, but it's bringing things in that people might use anyway as part of their sort of wider meta usage. But the the voice is certainly something you've talked a lot about over recent years, isn't it? You you basically drive around the country talking to AI. I I struggle with it just because of the sort of nature of the working environment I'm in, not constantly able to talk out out loud for these things. But meta-targeting voice and new spark certainly has been quite impressively received, so worth having look. It's just I find myself now happy that there's only two I'm really using because I need to remember which one's got what context and where I'm working on things.

SPEAKER_00 31:23

Yeah, so so this this is but presumably much more for those that are using the nice of Facebook um ads and um you know Instagram and all that sort of stuff as part of their daily lives, WhatsApp and so on. Um, but uh yes, I I do almost always struggle with just this trust idea. It comes down to, you know, bless him, Zuckerberg and you know, actually what's he gonna be doing with data? Even if he says all my data's safe, it's like, is it uh but anyway, it's out there. Uh go play, see see what you think um of it. But uh but yes, I think it's gonna be one that uh I'm not necessarily gonna dive into too fast. Right, that is all we've got time for this week on the Property AI report. Thank you for listening in. We will be back again next week with plenty more um on the uh the property AI paired results. So, yes, looking forward to catching up again next week. Uh, we'll see you in seven days' time.

SPEAKER_01 32:10

Take care, everyone.