One in three are using AI for personal finance, every week

Episode transcript

Amelia (00:16)
Hey there, and welcome to another episode of Fin the Week. I'm Amelia.

Paul Wood (00:20)
I'm Paul.

Pat (00:21)
I'm Pat.

Russell (00:21)
I'm Russell.

Amelia (00:22)
How's everybody doing? Good week? Paul, you've just come back off holiday. You feeling refreshed?

Paul Wood (00:26)
Yeah, very refreshed. I was just saying to everyone that as I was leaving Madeira, where we were, the weather started to turn a bit bad, and we arrived back here and it was glorious, so I think we've brought the weather. So yeah, very refreshed and happy to have been away for a week.

Amelia (00:45)
We all appreciate you bringing the sun back with you.

Paul Wood (00:47)
Yeah, it was the best thing to bring.

Amelia (00:50)
So this week, we're diving into a major piece of research. It's the 2025 Consumer Digital Index from Lloyds Banking Group. It's their 10th annual report and this year it's their generative AI edition. And the headline stat is a big one, but before we get into it, let's chat about everyone here in the group. Have any of you used AI for financial advice this week? And be honest about this.

Paul Wood (01:17)
I haven't, no, but it's one of those things I've kind of… it's always on my mind that if I've got a thing I need to figure out about anything, finance or anything, I would probably turn to AI as a first port of call. Not necessarily to get me the final answer, but to get me in the ballpark. A bit like you would do — I've certainly done it with ailments, and if someone's feeling sick, checking up on that.

So yeah, and I don't use it to sort of replace a doctor, but I use it to maybe point me in the right direction, or possibly panic me.

Pat (01:58)
I've used it for calculating loan repayments. My wife's about to move her salon to a new premises and as part of that there's a business loan involved, and we've been looking at different interest rates and just going into Claude and saying, you know, X thousand over five years with a 7% interest rate compared to a 9% interest rate, what's the monthly repayment? And it'll just calculate it all for you.

Using, you know, amortise the loan repayments out over the fixed term. And it's been really useful for that. I'm constantly in there asking questions about this and that and helping with my personal finances a lot. Yeah, every week.

Russell (02:41)
I haven't recently. I opened a savings account but my wife sorted all of that out, and we did — she compares different providers. I don't know if she used AI. We actually… yeah, I'm not sure, so from a financial perspective that's as close to the situation that I would have been in as a user. I probably would have done if I was comparing like products myself.

Russell (03:10)
But yeah, a bit like Paul, I kind of use it to compare like everything. So for example, if I'm looking for a new pair of running trainers, which is normally quite a lot — very good at comparing things, because it pulls on resource, it pulls on the information it finds. So it's very good at comparing products. Even as far as, like, I've used it for things that seem a bit odd but are actually quite difficult to find online. So like the model number of a gardening tool, comparing against another model number of a slightly different model of gardening tool, to find out that one of them has like a pull-and-hold lever and the other one has like a thumb switch or something. And it's so good at breaking down and comparing products in a table.

Amelia (04:03)
Definitely something people are using. Let's get into this. So Lloyds Banking Group published their 2025 Consumer Digital Index. This year includes one of the largest studies to date on AI in personal finance. And the headline is 28 million UK adults — that's 56% of the adult population — have used AI for personal finance in the past year.

Pat (04:03)
It is very good. Yeah.

Amelia (04:31)
And almost one in three are using it weekly. And here's where it gets interesting, because while people are clearly using AI for money management, there were real questions about trust, accuracy, and what this means for financial brands. Paul, before we get into the details, what was your first reaction when you saw those numbers?

Paul Wood (04:49)
I think my initial reaction is one of… I was surprised at how high the percentage is. I wouldn't have even thought — this might sound crazy, but I wouldn't have even thought that percentage of adults regularly use AI. When you factor in family members and things, if I was to split my wider family in half, I could probably imagine half of them have used it, half of them haven't.

So it surprises me that that amount of people are using it regularly. But then it occurred to me — I was at an event quite a while ago now, and it was discussing independent financial advisors and wealth managers and that whole industry. And it was talking about a lot of people don't use IFAs when they probably should do. So for instance, when they're planning retirement or planning some big life change. And the group was discussing, like, why might that be? And one of the core reasons seemed to be that people just don't think a financial advisor is for them, because the assumption is that you only need a financial advisor if you have money. And the problem most people probably have is that they feel like they don't have money and therefore [don't] go into an advisor.

It's a bit like me going to a hairdresser and asking what I should do with my hair, and they'd look at my balding head and just sort of laugh me out the door — you feel like that might be the same with a financial advisor. So then when I think about it in that context, when you suddenly have this AI tool sitting there ready to use that you can just ask all of the what you might think are silly questions about your own personal situation, and it's not going to judge you, and it will give you answers that get you, like I was sort of saying before, in the right ballpark towards an answer — kind of makes a lot of sense then. And actually, does it open a lot of people up to the concept of, I don't say financial advice, but something that's close to that, so that then they can say, okay, now I know what I need to go and ask a financial advisor.

So yeah, that whole thought process has then made me conclude, maybe I'm not surprised by that number. Maybe people don't feel threatened by it and therefore are using it because it's plugging a gap that was there.

Russell (07:25)
Yeah, I feel like it's kind of like having a conversation with someone who's really intelligent. So why would you not have a conversation with someone who's intelligent about important matters? You know, whether it's finance, whether it's health — it's much better at crunching numbers than me. It's off the chart compared to what I could do in my head, which is that kind of real-time conversation that you're having with someone.

Imagine having a conversation with someone who could calculate numbers to the extent like ChatGPT could in a few seconds. It would just almost be impossible. I don't know if anyone on the planet could do it. And I think that's why it's good for things like… it's good for making objective decisions. So if you gave it some information to use, let's say you gave it your monthly budget breakdown for example, I'd expect it would come up with some quite interesting, good decisions. Even if you didn't have a lot left over, it might help you push the margin on your budgets for example, to the point where you might be able to afford to pay for your flights for your next holiday next year. It would definitely be worth exploring that.

Where it has issues — I've seen someone post about this on LinkedIn. So they were using AI to look at their dividend tax rates and how much they would be paying in tax on their dividends. HMRC introduced a different percentage tax percentage on the day that they were speaking to it, to what it was giving back to them. So they were getting advice on how much tax they were paying, which could be quite substantial. I think the percentage difference was a couple of percent. ChatGPT was kind of like 2% off what they would be paying from that day, because it hadn't pulled in the latest HMRC tax information, to the point where they had to direct it to the exact page and the exact wording and the exact sentence of that update to then get ChatGPT giving them the exact advice. But let's say you were acting on that decision on the day and you didn't know yourself that it changed, and you didn't know where to point the AI, and you made a financial decision off the back of that because you thought you were going to be paying less tax. I think that's where there's a real danger having this conversation with the AI, like it would be for finance or like it would be for health.

Paul Wood (09:59)
Some interesting stuff in the Lloyds report on the idea of trust. I'm just looking at it now. So it says 40% of people trust banks and financial advisors more, whereas only 15% of people trust AI more. But that's still 15% of people trusting AI more than a sort of professional institution. That's quite shocking. And then it says as well, among 25 to 34 year olds, nearly a quarter now trust AI-generated financial advice more than traditional sources. That's crazy to me, and kind of a recipe for disaster potentially.

Pat (10:43)
Is that because it's so hard to access a traditional source these days? Because you've got banks, you know, you've got high street banks shutting down all over the place. Websites are kind of almost universally quite difficult to use, especially if you're younger and you're used to reading technical, financial articles and documentation. You've got your AI, ChatGPT right there, it's super accessible, it speaks to you in…

Pat (11:10)
…in your own language that you're comfortable with, and it gives you largely sound advice. Like, there'll obviously be instances where it gets specific details wrong, like your example, Russell, where they happen to be discussing a tax rate that was changing on that very date. So it might get specific numbers wrong, but largely, if you ask ChatGPT about budgeting recommendations, it will give you a really sound, rounded answer on how to improve your budget. And that's a resource that — if you imagine you're like a 23-year-old straight out of university, you've got loans, you've got rent — like, 10 years ago you'd sort of be on your own, you might ask your parents for some advice and they'd give you some advice. It might be right, might be wrong, might be 30 years out of date, but you probably wouldn't go further than that, right? But now you've got ChatGPT and it would say, "Okay, well, you probably [should] start thinking about your pension." And then that person would be like, "Why would I want a pension at 23?" And it'd explain that. And suddenly you've got loads of young people thinking about pensions when they didn't previously.

So overall, I think it's really useful. Personally, one experience I've had with it, using Claude Code actually, which is an agentic tool — we had this issue where in our (my wife and I) shared bank account, every month we were going like 250 pounds overdrawn, constantly having to top it up. It was quite clear we weren't paying enough in each month, but it was also quite clear we were like spending money in all sorts of random places. So I downloaded the last 12 months of bank statements from our online banking, dumped them in a folder, and just unleashed Claude Code on them and just said, right, help me, we're going over budget every month, what's going on? And it went through every single transaction, reconciled each transaction into a category, and produced an in-depth financial report for me and a budget plan. And it said, right, each week you're spending this much on groceries, which was way more than we thought, because I was just thinking we go to Waitrose every week and spend that, but actually, you know, my wife goes to M&S on Thursday and there's all sorts of other bits we buy. It's just spending this much a month on subscriptions, this much a month on power and electricity, and all of our bills had gone up and we hadn't really been considering that either. And it also asked me along the way kind of how much do you want to save? How much do you want to save each month? How often do you want to go on holiday? And I had this back and forth.

And it came back to me and said, if you want to go on holiday to this higher-end place, then you need to start putting this much into your bank. But if you're happy with just jumping on the ferry and going to France once a year, you can get away with putting this in. So it really helps to sort of relate the amount of money we're putting away each month to real-life outcomes. And the upshot of that was we sort of reluctantly stuck another couple of hundred quid a month in each into our shared account, cancelled a load of subscriptions, reorganised some of our grocery shops, and we're in a way better position as a result.

Pat (14:20)
And I probably could have gone through that process myself, but it would have taken hours and hours that I never would have done, because it's just, you know, your personal life and you've got other stuff to do. AI feels like a priority.

Russell (14:33)
It's amazing, isn't it? It's amazing because… you could kind of — if you can document how you live — I mean, I think we were talking about one of the previous podcasts about how the input is the most important thing to create a good output. If you can document your life daily, whether that's spending habits or travel — if you have that data there, you can essentially use the AI to come up with suggestions to hit targets that you kind of set it, or to come up with its own suggestions. And you just gave me that idea: if you gave it your budget sheet, you could perhaps even ask it how you could become more sustainable as well.

And it might actually come up with some good ideas around, you know, how you could become — I don't know, let's say it'd be like, "Well, you could save some money by not driving to work and cycle to work, because that would be better for the environment and you'd save 600 pounds a year," let's say. I'm sure it would be very good at that. And it'd be interesting if it went through — you could put in your shopping lists, especially people who order online. Imagine how powerful that would be. Most of our monthly expenditure goes on shopping, to be honest.

Could you hook in your Waitrose account through an API and tell it to analyse everything that you're buying and whether you need to buy it, and what the price points would be if you could travel down the road and do this? Could it find those little micro adjustments that would have macro improvement on your budgets if you have all this data that it gives you? That could be quite an incredible exercise actually, and I'm sure that will become easier.

It's just — where does it pull all this information from? I mean, my finances have just been managed in a couple of spreadsheets, really. It certainly crunched that really well, I know it would, but could it go to that micro level where it's analysing the cost of [things], probably, to save you money?

Pat (16:47)
Honestly, if anyone could go to AI and go back and forth with AI to help flesh out your five or 10 year goals — those could be financial or career-based or health-based or whatever — and the AI tool would genuinely put in place a long-term plan for you to help you achieve those goals with things that you can do right now, things that you can do in the next week, and milestones along the way — that kind of goal setting and planning was just never accessible to a large proportion of the population until AI came along. So in that respect, it's a really incredible tool. And it's not really surprising people are using it for financial advice, because most people's future finances feature quite heavily, don't they? Most people say, "I'll have a goal of getting…"

Pat (17:37)
…reaching 50 grand a year or 100 grand a year salary, or "I want to own a house, I want to own an Audi." You know, people have these fairly materialistic goals that are kind of at the top of their list. And genuinely these tools will help you make decisions and guide the way you live your life in order to achieve those goals.

Paul Wood (17:56)
Just playing devil's advocate, because the scenarios you're describing there are quite pragmatic, and you're sort of looking at a smart way to budget and coming up with it. So your intentions are good — like, you've got an idea in mind and you're just trying to organise yourself. What about someone in a different scenario who's perhaps up against it financially, or someone who is looking for the next get-rich-quick scheme? Do you think AI… because the power of an independent financial advisor is that they're trained, they're regulated, they have to explore the market, they have to present stuff clearly, they can't lead you down the garden path, or you know, there'll be consequences. Whereas people just relying on an independent — well, independent in a different sense — AI tool like ChatGPT, that doesn't really answer to anyone currently — is there a risk that even a tiny bit of misinformation about a crypto scheme, for instance, or moving your pension to this thing, could cause havoc? And who, you know, how should that be handled? I don't know if there is an answer, but I think those scenarios are probably playing out.

Pat (19:22)
They can be quite sycophantic, can't they? So you could have a really crap business idea, and you could go to ChatGPT and say, "I've got 10 grand and I could either use it on my family, or I could invest it into this great business idea I've got," and ChatGPT will be like, "Hey, that's an amazing business idea. Let me build you a website for that." And then you just blow your 10 grand on the crap business idea and it's just a complete waste of time.

And then if you do ask it for financial advice, it will always add a disclaimer, you know, you should get professional advice before investing and all that stuff. But it depends on the model you're using. Like Grok, for example, is probably more likely to recommend you invest in crypto than ChatGPT, because it's kind of less bounded. But the risk is definitely there.

If an IFA recommends you invest in a dodgy scheme, they're going to get in a lot of trouble. And so they naturally approach advice with a lot of caution, backed by a lot of expertise and training. But these AI models don't — there's no one really responsible for it. So yeah, that's definitely a risk for sure. In my experience, I haven't really… the models are trained to not do nefarious things like that, but there's always a risk that they will. Especially if you demonstrate that you're keen and you go into that conversation with the context that it's a good idea, the model is more likely to agree with you.

Amelia (21:00)
That's interesting, leads on to… in the Lloyds report, it outlines some of the things that people are most worried about when using AI in this way. 43% very concerned about data, privacy and security. 60% worry about the security of their personal or financial data. 45% concerned about being misled or scammed. And 26% worried about AI bias affecting recommendations. So I mean, there are very valid concerns there, right?

Paul Wood (21:34)
Pat, from your knowledge of how these models process data that you put into them — let's say you use a basic version of ChatGPT and you funnel in, like, a PDF export of your bank statement — is it going to swallow that up and forever be a part of ChatGPT's future models, including your address and your account numbers and all of that stuff?

Pat (22:04)
That's what they'd like. Yeah, that's what they'd like. And actually when OpenAI released ChatGPT, it kicked off this kind of AI revolution. By default, that's exactly what they did. So every conversation you had with ChatGPT at launch would be saved by OpenAI and then used to train future models. And that actually got AI off to a bit of a haphazard start, because straight away companies, you know, blacklisted AI as insecure from a data protection perspective, because immediately they just assumed that by default that's how AI models work, that it would remember everything you discuss with it. But actually they don't work like that at all. That was a feature that OpenAI deliberately introduced. And these days by default, as long as you're on a paid version, they won't store your data and use it to train future models. If you're on the free version, it will ask you for permission. I think some of them will force you to do that. You know, anything that's free isn't free, so you're paying with your data. A lot of the platforms now will kind of try and jostle you into that position where they are keeping your data and using it to train future models. And you have to also remember that all of these companies are based in the US — Anthropic, OpenAI, Microsoft — it's all US-based, all their servers in the US. So every bit of information you put into these chat tools is going across the Atlantic outside the scope of the GDPR. If it's your personal data and you're using it yourself, it's up to you. It's not a massive issue. But if you're doing it in a business context and you're passing client data into these tools and you're not doing it in a GDPR-compliant way, then that's the problem.

So you have to make sure that you're paying for the tool and you've got the right guardrails in place and you've got a data protection agreement. There's lots of things to consider from that perspective. So yeah, data privacy and security is valid. You know, and OpenAI are starting to surface ads now on the free versions of ChatGPT.

Paul Wood (23:56)
Because that sounds like quite a practical challenge, because I would imagine with you having a background in technology, when you got to the scenario of uploading the bank statement information to help you plan your budget, you probably provided that in a kind of quite easy-to-process format, like a CSV that didn't have anything personally identifiable — it was just transactions in a spreadsheet. But to a novice user, if they're chatting to their AI and they're saying, "I need help planning my budget," and it says, "Fire across your bank statements and we'll have a look," I'd imagine most people would probably just download their bank statement and chuck it in, and that would include everything. That's potentially a big issue, isn't it?

Pat (24:53)
Definitely. Honestly, within the next five years, what you'll start to see happen is you'll upload your bank statement, right? And if you're in your overdraft, right, and you upload your bank statement, then immediately you'll start having loan ads following you around the internet. That's exactly what's going to happen, because OpenAI have already started surfacing ads in ChatGPT, and at the moment they're not based on your personal data and they're very subtle and minimal, but the irresistible sort of journey of these big tech platforms is to gradually grow that advertising revenue. And at some point OpenAI are going to start taking your highly private conversations, using that to build profiles of you, and then selling that data to advertisers so they can target things at you.

You know, your bank statement literally tells you everything that's going on in your life. So they can see that you shop at Waitrose, see you've been to the doctor three times in the last month, and suddenly you've got these creepy ads following you around the internet. And that's where data privacy and security becomes like a real challenge.

So I think my advice to anyone is, before you upload any personal data in, get yourself on a paid plan. $20 a month, you'll get better models that are more accurate, you'll have an ad-free experience if you're using ChatGPT, you'll have the option for it not to save your data and use it in future conversations. And just kind of be aware of that really. If you're aware of it and you know how to go into your settings and configure it, it's not a problem.

Paul Wood (26:31)
It feels like a challenge, because quite often the people who will benefit the most from these types of [tools] — the people who have the least access to an IFA are the people who have the least amount of disposable income, I suppose. And they're probably the least likely to have a paid subscription to any of these services as well. They'll probably more [feel] compelled to just use the free ones and see what happens. Yeah, I imagine the regulators are probably having to have a close look at this sort of stuff.

Pat (27:09)
Yeah, definitely. But from an advertiser perspective, the information that people give to their AI tool would be an absolute goldmine. The potential for monetisation is absolutely enormous.

Pat (27:25)
And honestly, I think with the cost of running these models, advertising is probably the only way that these big tech firms are going to actually start to turn a profit. So we'll start to see it emerge. Right now, we're in a stage of the technology where all these big players are fighting for market share. So the tools are kind of largely free and very easy to use, and not tarnished by layers upon layers of banner ads and targeted ads. But the process of enshittification is starting. Have we talked about enshittification before? Yeah, I think we have. It started. OpenAI have started surfacing ads, and they're going to become more common, I think, across other providers as well.

Amelia (28:01)
We have.

Russell (28:11)
It feels like a lot of these platforms start out free because essentially it's the humans who are testing these platforms and using them, so they can evolve them in the right way to the point of monetisation, serious monetisation. We're essentially the guinea pigs, you know, kind of testing all these apps. And I was just thinking about it from a UX perspective as well.

I think they're heavily used for advice at the moment, just heavily used in general, because the conversation to begin with, it feels private. You know, there isn't any ads, it isn't pulling from your data. You feel like you're having a private conversation with this intelligent being. I mean, you know it's artificial. That kind of sycophancy as well, that kind of intrinsic desire to please, you know, is something people like. You don't want to give it your financial information and for it to kind of pass out or to judge you. You're kind of getting that non-judgmental, or fairly non-judgmental, leaning more towards the side of helping you, you know, than not. Which people like. It's human nature to speak to someone and feel like you're being understood, and to feel like they're giving you advice to improve. People want that, and from the user-experience side of things, that works really well with the AI — whether all that advice is sound or not is another [thing], it's something that we've been discussing.

When they start introducing ads, you're going to know that what you're telling it is [being read], and the ads that are popping up, your bank statements are producing loan ads — I think people will automatically start trusting it less. You make a good point, Paul: if you're asking the AI for advice on how to budget, and it's trying to tell you to sign up to a $100-a-month subscription when you're losing $100 a month — how can it provide that advice?

So I think it works really well from a user's perspective at the moment, because it feels private, it feels non-judgmental, and I think people who are in a bad financial situation will probably feel more comfortable — not everybody, but I think some people. And perhaps, you know, I don't know, perhaps that kind of student kind of use case as well may feel a lot more comfortable speaking to ChatGPT quickly, which is free and it's accessible and it's going to give them advice. Whereas they would be, you know, kind of divulging their financial situation to a human, like in a one-to-one office or a video-call kind of scenario where, you know, if they do need some serious help, they might not feel comfortable asking for it.

Amelia (31:13)
For sure, can feel I'm sure very intimidating for that person. So what does all this mean for financial services brands? You know, who controls the message, and are brands optimising their content for AI consumption the same way they optimise for Google, for example?

Paul Wood (31:13)
Yeah, I think there's lots going on, because I suppose first of all, there's a battle between brands to sort of be the ones who are referenced by these AI models. So if you're talking to it about a mortgage or a pension or whatever financial product, you want to be one of the brands that it name-checks as it's going through that. So I think a lot of companies are focusing on that, and the way they're doing that is fundamentally what they've always done with search engine optimisation. It's just a different flavour of that, maybe with a few details.

And then I think — we've discussed before — it kind of ups the ante slightly, because I think AI takes care of (I think I might have mentioned it in a previous episode) the top of the funnel, where you'd normally do your research and you're learning about a particular area. Normally you'd search on Google, then visit a few websites, do your research there, and then you'd move on. Whereas now you would use ChatGPT or one of the other models and do all of your research in that single space.

And so this sort of research paper, almost, would worry me most if I was an IFA, maybe, because I'd be thinking, who's gonna go and use ChatGPT instead of speaking to me? But then I kind of argued myself out of that, because I thought, actually, it probably opens more people up to the concept of financial advice, so that in the long run they're more likely to be clued up and go and speak to a proper advisor once they've got the lay of the land.

I'd be worried if I was, like, a Compare the Market type company that compares products, because arguably these AI agents could potentially do a much better job, because they could understand you to a much deeper level. You could probably tell them all of your information once, and it would probably — I wouldn't be surprised if in the near future there'll be an AI tool that just learns about you and your life and all of your insurance needs, and then takes care of it from that point on and just comes to you. In fact, I think Compare the Market — their own auto-search product — I haven't really looked into what it is, but in my head it's that sort of thing, where you maybe give it your information and it automatically goes and finds your products.

So yeah, I think they're probably worried, because they think AI is going to come and eat their lunch. And then I suppose you've got the big banks and institutions that don't want to be misquoted and misrepresented, and so probably are fearful of AI spouting stuff about them that isn't necessarily correct, or at least is out of date.

So yeah, and I don't really know the answer to any of those things. I think there's probably — I mean, I know people don't want to crack down on what tools are able to say and what they aren't able to say, but I think with something as important as finance, and health as well, having regulators heavily involved and setting the tone early is probably a good thing, because it would allow people to just work on a level playing field. But yeah, that's kind of how I view it. I think a few companies are probably worried about things.

Pat (35:11)
You just think about the concept of a personal AI assistant, which a lot of firms are chasing at the moment. A tool that knows you inside out, has access to your bank statements, knows what ailments you have, knows what your family setup is, your house, your mortgage, knows what your preferences are, like how much you care about the environment and all that kind of stuff. That agent would then be able to recommend the perfect — it would be able to go out onto the internet and find the perfect provider for you in any area, really. So if you're looking for a new car — I've been doing it recently actually, browsing for cars online and saying, "What'd be a good car for me?" And it comes up with three options. They're all perfectly aligned to my values and my budget. And then you can say, "Well, what's the best loan to help me fund this car?" And then go out and find the best provider that's perfectly aligned to me.

And as a result, that kind of research step, like human research step, where I'm going into Google and looking at paid ads and all that, and going to Compare the Market — arguably, if you extrapolate out what's happening, those tools kind of won't be needed anymore, because the AI system will just know what the best platform is.

If you look at that scenario from the context of a business, you've kind of got, you know, your marketing budget is almost a bit pointless now, because that whole connecting your business with your customers is just done by the customer's AI agent. So your marketing needs to be almost, like, focused on all of these little micro niches. So is there going to be a growth in targeting specific micro niches? Are we going to see businesses blossom that focus on, kind of, families in Guernsey with two kids and a car who live in a specific part of the island? A little business that just focuses right on that tiny little niche, and all of their content and products are perfectly aligned to that niche, so that the AI assistants always recommend that firm for those particular people.

I don't know, it's interesting. I sort of thought about this quite a lot a couple of years ago when these tools first started coming online, thinking forward about what a world like that might look like. And I think there'll definitely be a rise in businesses that — already, as a business, it's a wise move to find a niche and target that niche. As an agency, we target financial services because we're good at that, and we've got experience in that, and it makes us a better fit for those businesses compared to an agency that serves all different types of business. But actually, moving forwards, is there going to be value in us targeting, like, two or three very specific micro niches within financial services, because businesses in those niches using their AIs would specifically find their way to us rather than us finding them? I don't know.

Paul Wood (38:15)
That model sounds sort of idyllic in theory, but then it kind of reminds me of the olden days of SEO, where you'd have in the footer of a website you'd have, like, "Cheshire web design company, Manchester web design [company]" — like, they cover every single place in the country — and the ones that would show up wouldn't necessarily be the best companies, they'd just be the best at showing up. And so you wonder if the AI models would be too easily gamed in that approach.

I mean, I don't know, because we've also spoken about the concept of a web for humans and a web for bots and agents. I can see that happening. And the companies that surface up-to-date, easily usable information on, say, mortgage rates and loan rates and blah blah blah will win, simply because they are playing the right game and they're kind of providing the right information. But I do think brand will remain an important part of a company's appeal as well, because you have to sell to the heart as well as the head.

Russell (39:37)
And I think ultimately it would be a human decision on which agency to kind of pull the trigger on. It has to be. Otherwise that could be quite catastrophic. I'm conscious of time, but you did kind of give me an idea, Pat, of there's probably going to be a personal assistant and a business assistant, with two kind of very different objectives — you know, looking for a website, or looking to save money to go on holiday. And it'd be interesting to see kind of where those assistants live.

Because it feels like a very manual process to input this information at the moment. You probably could, and the more technical people, I'm sure, could streamline that, but you have to keep being updated. Who's going to win the race to this personal assistant and this business assistant, like, platform? Will it be someone like Google, for example? You know, at basic level, you've kind of got something like Google Home, which pulls all your devices together in your house and you can control them all from that app. You know, who's going to release that? Who's going to release that app which pulls together your whole life, like, easily into this one space, both on a personal and business level? It's going to be quite interesting, and it's going to be a bit of a race, I think, over the next few years — which platform is going to be able to do that the best.

Paul Wood (40:57)
I really like the idea or the concept of a sort of — I was going to say digital passport, but that will probably trigger people who are worried about ID cards and stuff — but the concept of a sort of a bucket of information about you. So this is me, this is the products I use, this is stuff, you know, just relevant information in a secure way that you can kind of pick up and place in whichever platform you need to use. And then if you don't want to use a platform any longer, you take it out of that and you put it in something else. That idea sounds — that would be life-changing for a lot of people, I think, because then you wouldn't need to keep inputting data into another tool about, "no, I need — this is my car registration number at the moment, this is the date the renewal is," you know, that just boring stuff that you could probably automate all of that just by having a single packet of information that you can just gift to a tool that you need to use.

Russell (41:57)
Where does the package live? I guess that's the challenge, isn't it?

Paul Wood (42:00)
Yeah, but at least — yeah, I think it needs to be independent, doesn't it? It needs to be something like you own and you control where it gets given to. But yeah, who [knows]?

Russell (42:09)
Someone's always trying to make some…

Russell (42:11)
…money somewhere. That's the problem.

Pat (42:13)
And right now there's this proliferation of open Claude-based tools, which are wrappers around Claude Code primarily, which run on your machine and have a memory on your machine and build up a picture of everything, and all the files are stored in your machine. And then they can work on that data and use large language model APIs to be useful and do things for you. That's exactly the kind of ideal setup for a personal AI assistant. But it's quite technical to set up, and so it's not really accessible to the wider market. And I think if someone manages to take that concept and package it up in something where you can just quickly sign up online, then they'll be onto a winner.

Meta in the last 24 hours have released a new model that's focused on providing personal AI services. But honestly, I wouldn't touch them with a barge pole. They're just one of the worst in terms of abusing personal data. But they're going for that market because they can see how lucrative it is. They're going for it hard.

Paul Wood (43:12)
Funnily enough, my holiday reading book was Careless People, the book about [Meta]. Yeah, it's stuff I always suspected, but what a shocking company on the face of things.

Pat (43:21)
It's a good read, yeah, it's a good read, that. It's terrifying at the same time.

Russell (43:32)
Sounds interesting.

Pat (43:32)
Absolutely shocking. Shocking and all. The whole culture of that company is just like Mark Zuckerberg from the top down. And he hasn't got humanity's best interests at heart. I won't say any more than that. Just read the book.

Amelia (43:49)
There we go, Paul. Maybe we need a Fin the Week book club now as well as a podcast, right?

Pat (43:54)
I would recommend reading that for sure if anyone's listening.

Russell (43:54)
That'd be good.

Amelia (43:57)
So a bit of homework for everyone there. Shall we move on to jargon busters?

Paul Wood (44:05)
Let's do it.

Amelia (44:05)
So this is our feature where I've got a list of industry terms, and [we] see if you guys know what it means. I'd say we've had some mixed successes over the week so far. Today's term is ETF.

Paul Wood (44:22)
Exchange-Traded Funds, I think is what it stands for. As for what it actually is, I'm not 100% sure.

Pat (44:31)
I think it's like a collection, it's like a sort of group of investments that are packaged together that a fund administration firm or an investment firm will offer, so that you, as an end user or a company, can invest into that. Yeah, so you kind of like, it's like a…

Paul Wood (44:53)
Yeah, I feel…

Paul Wood (44:56)
…like Russ knows this, because I recall you explaining it to me once and I've probably forgotten it.

Russell (44:59)
Oh god, yeah, I mean, I'm glad I went last, but I couldn't remember exactly what the acronym is. I think you were right. But yeah, so a group of funds. They could be — it could be a sustainable ETF, for example, or a specific, you know, kind of group, technology, something like that, where you just have funds kind of grouped together within one ETF instead of wanting to invest in them separately. I think that's right.

Amelia (45:30)
Yeah, bang…

Pat (45:31)
I feel like we should know this. If we don't, it's embarrassing.

Amelia (45:31)
…on. So.

Pat (45:35)
It sounds like, hopefully, we got it right.

Amelia (45:37)
Absolutely right. So, as I [guessed], exchange-traded funds — a basket of securities that trades on an exchange like a stock.

Paul Wood (45:45)
Yeah. Passed.

Russell (45:46)
Phew.

Amelia (45:46)
Which is pretty much what we said.

Pat (45:47)
Yeah, largely correct.

Amelia (45:49)
Well done, well done everyone. So we'll have more of that next [week]. Will be an interesting chat again, I'm sure. Thank you so much for listening, and yeah, we'll see you next week.