Exploring Gen AI's Transformative Impact on Financial Advice and Compliance - Fintech Special

Prepare to have your mind expanded as we journey through the dynamic world where artificial intelligence reshapes the financial services landscape. Joined by industry visionaries Robert Harradine from Ammonite , Craig Barraclough from Finovation and Joseph Twigg from Aveni . 

We dissect how AI is not just revolutionising finance but becoming an indispensable co-pilot for advisers, enhancing their productivity and ensuring stringent compliance. 

Craig's expertise shines as he walks us through the transformative effects on back and middle office functions, while Joseph gives us a glimpse into AI's potential to bridge the advice gap, allaying fears that technology might eclipse human jobs.

Venture further with us as we tackle the complexities of AI in finance, a realm where data is king and the construction of advanced language models like GPT-4 is akin to building digital empires. 

The conversation turns technical yet remains accessible, exploring the critical importance of clean data and the challenges that smaller businesses face in this data-driven world. We confront the phenomenon of 'hallucinations' in AI, the need for domain-specific models in a tightly-regulated industry, and the significance of a robust data strategy that turns potential into power.

As we wrap up this thought-provoking episode, the focus shifts to the profound ethical and practical questions that come with integrating AI into established companies. 

From retrofitting cutting-edge AI into legacy systems to grappling with who bears responsibility for AI's outcomes, we navigate the murky waters that modern companies must sail through. 

And we don't stop there—our gaze stretches to the horizon, contemplating the ethics of neural implants and AI's role in the future of human communication.

 Our  gratitude goes out to our guests for their insightful contributions that have made this episode an expedition into the heart of where technology meets empathy and ethics.

Begin your financial planning career journey today

Whether you are looking to become a paraplanner, administrator, mortgage and protection adviser or financial planner, the Financial Planner Life Academy is for you. 

With limited entry-level job roles, giving yourself the best financial planning career education, will not only kick start your financial planning journey with relevant qualifications and skills, but it’ll also help you achieve success much faster.&nbs

Be sure to follow financial planner life on YouTube for extra content about a career within Financial Planning HIT THAT SUBSCRIBE BUTTON!

If you're looking to start your career in Financial Planning, check out the Financial Planner Life Academy here

Reach out to Sam@financialplannerlife.com in regards to sponsorship, partnerships, videography or career development.

Transcript

Speaker 1: 0:00

Today we are back with a fintech episode of the Financial Planner Live podcast. Robert Harodyne from Ammonite joins us with Joseph Twig from Aveni and Craig Paracloth from Finnovation. We talk about tech stacks. We talk about whether administration and power planners should be worried about their jobs when it comes to AI. What is new in Gen AI? Is it just another AI hype circle? We get deep on that. We also look at the risks and what we should be aware of when it comes to AI. It's a fantastic episode for anybody thinking about building a business and implementing AI into the process. Welcome to the Financial Planner Live podcast. Again. We're back with a tech special episode. Robert Harodyne joins me again from Ammonite. He's got two guests today to talk about AI. Rob over to you.

Speaker 2: 1:00

Great, Fantastic to be here today. Thanks, Sam. Yes, we're going to have about 45 minutes of enlightening chat around the very hot topic of Genitive AI and data. We're joined today by two very esteemed guests. We've got Craig Paracloth from Finnovation and Joseph Twig from aveniai. You're both founders of those businesses. Great to have you on the show. Hi guys, Hi Rob.

Speaker 3: 1:35

Hi, sam, thanks for having us on.

Speaker 4: 1:37

Yeah, thanks for having us on, Rob Sam.

Speaker 1: 1:39

No worries, thanks for coming on. I'm looking forward to sitting back and just listening because, I'll be completely honest with you, I don't have a clue, but I'm really interested. I love chat GPT, by the way. That's a game changer, so if it's better than that, then I'm all it is.

Speaker 2: 1:57

Yeah, I mean, it's a topic, Sam, isn't it, that everybody hears about but doesn't necessarily understand it, so I think this is a good opportunity for us to talk to experts in this particular field, and an open pandora's box, really. So, without any further ado, Craig, do you just want to give just a couple of minutes on your history, your company and Joseph likewise as well, just to give a bit of context.

Speaker 3: 2:24

Sure, yeah. So I run and founded Finnovation Consulting kind of middle of last year. The previous 12 years I'd been CTO, one of the exec directors at Sandringham Financial Partners, the advice network. I decided to start Finnovation after leaving Sandringham to really try and help advice businesses, just to be better. I think we have a fantastic profession but I think we can do things better and I think technologies are a stumbling point for a lot of businesses. I think there's an expertise gap there for a lot of advice businesses. So that's where Finnovation comes in to try and reach that expertise gap really.

Speaker 2: 3:03

Fantastic and working with businesses of all varying different shapes and sizes, Craig, I suppose.

Speaker 3: 3:09

Absolutely. Anybody that needs help, anybody that can help me at this profession better, anybody Rob Good stuff.

Speaker 4: 3:16

And Joseph, yeah, so Joseph Twig, ceo of AVENI. Aveni is around four, four and a half years old in earnest. We're an AI business. We've developed AI products that focus on productivity and compliance, so the whole idea is an AI co-pilot for financial advisors. I'm sure everyone will be coming increasingly familiar with that phrase and concept around AI co-pilots.

Speaker 4: 3:51

Yeah, I think the technology has moved at such a pace over the last 12 to 18 months. I think chat, gpt, took the world by storm and the last year or so there's been people scratching their heads figuring out what impact is it going to actually have? And, although it looks great, what can it actually do that you can rely on? And I think, looking over the next five to 10 years, the impact of large language models and generative AI is going to be astounding. It will change the way everybody works, transform productivity. I still think for financial advice, humans are an essential part of the story looking someone in the eye, building that trust. But everything that goes on around meeting a client can be automated now and I think it's super exciting. And I think the next number of years is going to see some long-standing challenges solved, such as access to advice, cost of advice, advice, gap, things of that nature.

Speaker 3: 5:05

AI will fix everything, sam, that's the answer Come on, get on board, Craig.

Speaker 1: 5:11

Absolutely everything it will fix. I like the sound of that.

Speaker 2: 5:13

That's good, so Craig Joseph just mentioned there, the last 12-18 months has accelerated at a very fast pace. It's become mainstream. Everyone's talking about it now, but do you want to maybe just take the clock back three, four, five, even 10 years around this type of technology, ai and just give it a little bit to its history? Why is this a hype cycle we're going through with AI? Is this a hype cycle we're going through with AI again, or is this here to stay now?

Speaker 3: 5:49

Yeah, I mean, look, ai is everywhere at the moment. I mean everywhere you turn, somebody's talking about AI. I don't think we've seen a hype for a technology like this ever.

Speaker 3: 6:00

I don't think so. When we're talking about AI, we're talking really about machines that have been programmed to learn and make decisions like human beings. The idea is that we can create machines, intelligent machines, that can carry out tasks that would usually require human intelligence, so things like driving cars, writing poems, translating languages and so forth. They're powered by what called algorithms. So a lot of people hear about algorithms. These are basically just sets of instructions that tell the machines how to learn, make decisions and carry out specific tasks. Ai is nothing new. It's been around for about 70, 70 years. It really started coming to the fore kind of in the 90s with likes of DeepMind, and obviously it's accelerated over the kind of plus 10 years. We've used AI in financial services for probably around a decade or so, mainly in back office and middle office tools, but, as Joseph said, that is kind of changing with with generative AI. I think at the moment what we've got is what we call narrow AI. That's what it's referred to. Basically, narrow AI is machines and systems that are designed for a specific task. So, for example, chat GPT some favorite tool is there for content creation. Yeah, chat GPT cannot drive a car, for example. But what we also will hear about are some terms that are theoretical at the moment. So, for example, ai, so artificial general intelligence. So artificial general intelligence is an AI system that can do multiple, very different tasks, very much like a human being, so we can drive a car, we can write a poem, we can translate a language. Agi is kind of seen as what will power the robots in the future, so that robots will be able to be out and about amongst us carrying out various bits of tasks. The other one you might hear about is artificial superintelligence. So this is when AI surpasses our human capabilities, becomes kind of sentient, can think and learn for itself. Basically, whatever we're scared about, we're usually to refer to as the singularity In terms of use. We use AI every day. It's all around us. It tells us what to watch, what to listen to, what to buy, which routes to take on Google Maps. It powers our social media timelines, it filters our emails. It's already creating new drugs, running warehouses, monitoring and managing utility networks. So AI is here and has been for quite a long time. I think the difference is what we're now seeing and why there's so much hype.

Speaker 3: 9:07

Is Generative AI or Gen AI, as it's commonly referred to. So I think Gen AI is powered by, as Joseph said, large language models and I think Joseph is probably slightly better placed to explain LLMs than I am but basically, gen AI allows systems to create new and novel content such as text, images, music and so forth. So tools like chatGPT, microsoft's core pilot for text generation, and Dali and mid-journey for images. I think there's a lot of hype around AI and Gen AI at the minute, as we've mentioned, and it's because Gen AI can help with so many varying, wide tasks. It can enhance and automate so many different things and, secondly, it's easy to use. We can all use it.

Speaker 3: 10:03

It's been democratized and proliferated at a speed I don't think we've ever seen in terms of technology take-up. I think chatGPT, when it went live last year, had, I think, a million users within five days. It took Facebook and 10 months, I think, and Twitter two years to reach that same number and two months. Chatgpt had a million, a hundred million users. You won't have to look at the app stores at the moment to see the number of Gen AI-powered apps within the app stores. So there is a huge amount of hype, I think.

Speaker 3: 10:41

In terms of the hype cycle, I think we're probably we are at that peak of the inflated expectations. As I jumped earlier, we expected to fix everything for us. Whether it will fall into the trough of disillusionment remains to be seen. I think it may or may not. I think it does have some challenges, especially Gen AI. We've seen the New York Times suing Microsoft and OpenAI for copyright infringements for the data that they've used to train chatGPT, so I think we'll probably see some more of that. But in terms of, is AI here? Yes, it is. Is it here to stay? Yes, it is. Is it hype? No, it's not. It's definitely something for the future.

Speaker 2: 11:29

And thanks for that, craig. And in terms of the financial services industry, you know we're heavily regulated, right? So it's in some ways the wealth management industry has been slated for being slow in its advancements around technology, so do you think this is gonna be the catalyst Maybe, joseph, you could answer this one and do you think the regulator is gonna help the regulator, or is this gonna, are they gonna be more in relation to this advancement?

Speaker 4: 12:06

I think the regulator is broadly taking a very positive stance towards the adoption of this technology. We, as a company, met with the regulator a few months back and they were incredibly bullish and incredibly positive. Effectively, you know, there to summarize that conversation, they were saying let's not let regulation get in the way of innovation, which is very sort of promising to hear. I think there's obviously. I think, while we're adopting the technology in a co-pilot setting where people retain the risk themselves as individuals, I think that's fine. I think when you start to see AI advice delivered to the masses without human oversight, that's when the regulator is gonna get very, very excited.

Speaker 4: 13:03

But I think you know, if you think about industries like financial advice, there have been productivity challenges for a generation. You know advisors do spend still spend 30, 40, 50% of their time on relatively low value admin and what Generative AI has done has provided the opportunity to automate that low value admin. So to write your emails, to write your report, to administer your CRM system, you know to take the customer voice content from the conversation, populate that in your suitability report and you know, really chip away at this. Three or four hours of low value admin. You'll turn that into 10, 15, 20 minutes and what that does is allows advisors to see more clients. So if I'm dealing with three, four hours low value admin a day, I have a hard limit of two clients I can see a day. If that's taking 15 to 20 minutes, I can maybe see three, four, five clients in a day. So it can transform the economics of the industry whilst keeping the human right at the center of the advice process.

Speaker 2: 14:31

That's interesting to hear your feedback from talking to the regulator about this, so, and it's great to hear that they're completely on board with it right. So, and in fact you know, I think, in terms of being able to see more customers, do you think that's gonna have? Do you know I think this is the silver bullet for the advice gap in that regards, and would you think it'll be more of the higher net worth clients that traditional IFA would, wealth manager, would go after, or do you think it can actually start to play into if those efficiencies are there, it can start to play into the advice gap and solving that.

Speaker 4: 15:04

Yeah, I think it'll start to chip away. But if you look at the numbers, the market is truly underserved. You know doubling the number of clients an advisor can see on an annual basis is not gonna make much headway into the. You know 12 or 13 million people that have perceived to be in the advice gap. You know you need to combine AI first, digital first advice and guidance strategies alongside very, very efficient human centered advice to really make headway in that space.

Speaker 3: 15:43

I think it's gonna sorry, I think Joseph's exactly, you know, right, nearly on the head. It will be part of the solution but it won't be the. You know it's not the silver bullet but it allows us to, as Joseph says, you know, buy back some of that admin time to be able to free up time, to be more efficient. But it also will help, I think, support clients and consumers much better, especially, you know, around consumer duty. There's gonna be, you know, that ability to the ability to chat with documents, for example. You know you will receive your document from it, from your advisor. You know your annual review pack or whatever. There'll be a capability at some point to be able to chat with that document.

Speaker 3: 17:16

So you can tell me what this means for my retirement. Can I still retire at this age? Would that? Client understanding, consumer understanding is going to be hyper-personalized so that we can, we no longer need to kind of work to a lowest common denominator, we can work to the individual clients you know themselves. So there's efficiencies across the advice process, right across the advice process. But I think the advice gap itself is gonna require efficiencies within the business by appreciation of value from both advisors and clients, education and guidance, you know, and an effort, a much bigger effort, I think, than we're seeing potential with the industry at the minute.

Speaker 2: 18:02

Sammy, you're gonna jump in?

Speaker 1: 18:04

Yeah, talking to the documents, that was really cool. Like literally I could get like a document sent to me. I'm not quite sure what it means and I can go. Can you just explain what this part means for me and how that impacts my retirement whatever, and it will talk back to me. Personalize based on what I'm asking at a document.

Speaker 3: 18:21

Yeah, that's what we're asking. Imagine Chad GPT now, sam, yeah, just ask you a question. So I've, you know I've got. Could you explain this paragraph to him? Could you explain you know the attitude to risk? Why is it important to me?

Speaker 4: 18:34

Wow, that's amazing. It's just clippy, it's clippy.

Speaker 3: 18:41

I've got fantastic team background of clippy that I love to use.

Speaker 2: 18:46

Can I oh go on Sam.

Speaker 1: 18:50

No, carry on Rob.

Speaker 2: 18:52

I was just gonna say. I mean, one of the things we've taken really seriously with all the tech we're building at Ammonite is is the data right? So if you, you can't really harness these AI tools unless the data is clean while structured right. So just sort of hopping back to your point around hyper personalization, the it's gonna be incredible to see how that is gonna advance, but the data needs to be in great shape, right? So I don't know if either of you wanna talk around the large language model, what that is, how you create one, how you train it, how you put the guardrails in place and how you structure the data within it to be able to do those types of things like being able to talk to a document and maybe just to throw in for good measure, like some natural language programming, processing those sorts of things as well on top of that in relation to the documents. So a few different areas, but maybe we could look under the bonnet of some of that. I think that's you, joseph.

Speaker 4: 20:01

Yeah, so I mean, a large language model is effectively a mathematical model that predicts the next word. So if you think a sentence that's very familiar, something like you know, the cat sat on the and then you've got a space, matt would be highly likely to be the next word based on all of the data that the model has been trained on. So there'd be a very high score. Toilet would probably be very low and effectively. If you take that concept and magnify it over everything that's ever been written, every single book, every single website, gpt-4 was trained on literally trillions of parameters, each one of those little scores being a parameter. So large language models are effectively big, large statistical models that predict the next word.

Speaker 4: 21:18

I wasn't sure about the question with regard to all the data that's. Large language models, it's important to remember, are discrete units, so when it's trained, you ask it a question, it considers the question and gives a response. It doesn't retain any of the data. It doesn't learn by itself. People think that that is the case because of all the press with chat GPT. Chat GPT was designed to do exactly that. It was released for user testing and they explicitly said any question you ask, we're going to keep the question and we're going to monitor the response so we can learn and retrain models. So large language models themselves don't present a material data risk. All they do present is a real risk of hallucinations, very plausible but inaccurate responses and things of that nature.

Speaker 2: 22:24

So the hallucinations, joseph? This is another term that some of the viewers may not or may have come across, but it is commonplace. You can't necessarily trust the technology 100%. It might be the cat sat on the toilet and that could be potentially quite dangerous as well, especially in a regulated field. 100%.

Speaker 4: 22:49

And I think that's why these big horizontal models that are very general and large will only be used for certain use cases, and those use cases will always be human first, co-pilot type use cases. If you want to start using generative AI for AI first use cases in a heavily regulated setting, you're probably going to have to train your own large language model, a vertically aligned large language model that's probably much smaller than GPT so 10, 20 billion parameters but designed specifically to create the outputs that you need and that solves quite a lot of challenges around transparency. What was the model trained on? You understand the data source, and that's something that we have really looking at now. So how do you like? This concept of vertical AI is going to be a very hot topic over the next year or two, and so how can you create large language models for very specific purposes, for very specific industries? That, where hallucinations have reduced, explainability, has increased transparency, is there just so you can start using them across multiple other use cases.

Speaker 3: 24:16

And that's the challenge, right, that's the challenge with these large, the GPT type models. Is there not domain specific, as we talk about and Joseph talking about, I think?

Speaker 2: 24:29

we talk about generative AI.

Speaker 3: 24:31

Here as well. There are other types of AI that requires data, so machine learning, so, for example, and I think that is where the data is key.

Speaker 3: 24:42

And especially in financial services. Those larger businesses and larger advice businesses that have larger datasets will have an advantage that they can train their own language models because they've got enough data to make the model worthwhile. A smaller advice business might not be able to train their own model because they just don't have the dataset that's required. And look, we know data is a problem in our industry. We know it's been historically. It's been siloed, fragmented, lack of standards, incredibly difficult to get sometimes. We know that's a challenge. We know people are working on that and we are getting better with that.

Speaker 3: 25:25

But data and having a single source of truth within your business is going to be rightly important as we move forward in this new world and data is going to become more of a commodity For businesses.

Speaker 3: 25:41

Businesses need to start thinking about their data strategy where their data is, how it's stored, where it comes from, how it's used. Because in this world of consolidation, especially where traditionally businesses are being valued by their assets and the management, going forward businesses will be valued based on their data and their systems. That's where the intrinsic value will be in a lot of businesses. So having a data strategy, I think, is key and we're going to see a lot of benefits around natural language processing and the models that Joseph's talking about and the type of work that Avenia are doing. They will help with that client interaction and that front end and front office interaction piece and I think, as we move into the longer term, is where we'll start to see AI and systems that can provide financial planning and financial advice and, as Joseph says, that's when the regulation is definitely going to start coming into it and it is a risk for those that are trying to innovate staying within the regulation.

Speaker 2: 26:59

That's really interesting. Thanks, craig, I think to sort of add at this point no, not at this point.

Speaker 1: 27:06

I'm kind of down my depth, I think, so I'm just going to shut up for once and sit back and listen.

Speaker 3: 27:15

We'll come on to it, Sam. We'll come on to work for us to explain. That's the no, I think.

Speaker 1: 27:20

I don't think they're. I see the whole AI situation and we'll come on to it obviously. I do see the whole AI situation freeing people up from mundane, boring tasks and improving humans' lives. That's the way I view it and that's the way I see it. I don't see it as this kind of massively negative disruption. If anything, it frees us up to do the things we want to do more of.

Speaker 1: 27:45

I'm interested in the scalability of AI. So when you talk about how many, you could actually use your question about the benefits of it right. And so if you're able to see more clients, does that mean you can make more money? Or are we talking about reducing the cost for advice, for instance, and giving accessibility to more people? So those are the kind of things that I don't mind understanding which is more of a positive impact in society to be able to help people with their finances. So could they access somebody such as Alan Smith that I spoke to? He deals with quite high net worth individuals and we talked about how many people follow you, how many people are aware of you, would love to talk to you, engage with you and use your services, but they can't afford it.

Speaker 1: 28:25

Is there an ability to create a masterclass type subscription to somebody who's a genius investment manager through AI capabilities and I think that's quite interesting, you know and at the same time, can they go through the consumer some form of due diligence where they have to kind of understand what it is they are actually doing to be able to access the types of investment strategies. You've got to jump through some hoops to be able to do that, and that's a kind of, again, safeguarding and protecting. It's not sort of saying that you can't have access to this. You just need to understand how this shit works and we're here to protect you.

Speaker 1: 29:01

So I almost quite interested in the level of protection to consumers. I think AI is going to add some real value to that. The other area, I suppose, would be things like you send off that data off into the off into some kind of server, right, you know we don't know where that is and it lives in the server. Are you telling me that? Like it's pretty basic? Is it like blockchain type stuff? It's massively secure. So you know, all of a sudden you've got advisors playing around with AI at the moment chucking a load of data into something and, you know, breaking data protection acts and all that kind of stuff I don't know, those are the types of things I think are a bit out there with it.

Speaker 3: 29:34

I think the first point in terms of that. You know that productivity there is. I mean, whether you use your time to make more, you know to make more money or reduce the cost to services. That's a commercial question. But you know there is this idea and I'm sure Joseph talked about this about digital avatars. You know, having digital twins of the advisors that can service, you know, any number of clients at a time so you can be working, you know, 10x times faster. I think what you were talking about in terms of the client interaction will definitely start seeing things like that. We'll definitely start seeing, you know, the financial health checks that are automated and much more kind of interactive, you know, with a chat bot and much more empathetic than just filling in, you know, certain questions and then that spits out a result. We'll see it'll be more of an experience for the client and then that will point the client on a journey you know the right journey for them back to that hyper personalisation. We can point the clients in the right way.

Speaker 1: 30:41

So just to get a little bit more questions. I know you probably want to push on with the next one, but when you talked about that hyper personalisation and the avatar, so it might write in thinking with Gen AI, you would be able to have a video version of me on a screen that's like a zoom, for instance, and the consumer can ask me questions on video and I'll respond as if I'm like talking to them. Potentially, yes, how far away is something like that?

Speaker 3: 31:12

Well, the way that the pace of change, I would imagine not very long. The question I would have to. That is is about adoption, and will clients want that? So would you as a client, if you think about this, if I'm your advisor, would you want to talk to a digital version of me specifically, or would that be a little bit weird?

Speaker 1: 31:35

I might do it. I might do it. Sorry, sir.

Speaker 3: 31:39

Would you prefer to speak to something that you know is definitely an avatar, or would you like to prefer to speak to a digital version of a human being? I think once we get into digital versions of human being and we've seen deep fake videos if you Google it you can see Morgan Freeman giving financial advice to somebody you can do that, but how comfortable will clients be? A lot of AI technology is going to come down to how willing we as clients and we as to adopt the technology If you grow up that you know no different.

Speaker 1: 32:16

I think if someone's in a crisis situation, they want to know an answer to it quite quickly. I think I'm really interested in that kind of I'm interested in that when it comes to things like suicide prevention, so the ability to actually have that empathetic conversation where it's designed and created to be able to ask the questions that would stop somebody from going down the route of killing themselves, and that I think is really interesting. However, then there comes into the ethical conversation, I suppose, about what is right thing to say and what's the wrong thing to say, and could you end up actually killing that person? So it's kind of it's an interesting topic. I'm just interested. It came up in a recent meeting that I was in and I thought it was interesting. I was thinking because they were saying what's someone saying? It's not far away and I was like wonder how far away it is. I just wondered if you knew.

Speaker 3: 32:57

I mean there's in terms of empathy and it's interesting you're talking about that. I'd recommend looking at PI from inflection AI. So PI's been developed specifically to be empathetic and to provide empathetic responses to conversations. And I found myself the other day. I went on to see it and it asked me how my day was before I knew it. I was having a chat with it about post-it notes and all sorts of various weird things and it was incredible to think that this was a machine that we were talking to. The empathy there was amazing, to be honest.

Speaker 1: 33:36

So I would check out PI and also accessibility to people that perhaps don't communicate very well and aren't very confident and all of a sudden they can have a conversation at the pace that they want to have it without feeling intimidated. So I think there is that kind of element of neurodiversity within it as well, can it kind of? Is it like an accessibility option? I don't know, it sounds cool to me.

Speaker 3: 33:57

This is the thing about AI the possibilities for AI are unbounded. I think it's the first technology really where there is no ceiling at the minute for what AI can do. Especially with the pace of change, especially when you link that with quantum computing and the advances that that's bringing, what will be able to be achieved will be phenomenal. What we can achieve right now is phenomenal. It's what we as human beings will accept in terms of AI. That's going to kind of that's going to do the pace of change. That's where we'll have to decide what we want to accept.

Speaker 2: 34:46

Look, guys, just want to jump in just to the time of things. So one of the burning issues around all of this is obviously job displacement and technology taking people's jobs. So, yeah, I just want to touch on that, and we've we were all used to using technology in our jobs now and love our iPhones or whatever it is. So is this a bit of technology, for the first time, that we need to be really wary of as a civilization Like Joseph? What's your of? Any is essentially automating human processes that are done, like I say, by humans. So what's the kind of reaction you're getting to in wealth management businesses of deploying this technology and what's your sort of short to longer term view of the job situation relating to it?

Speaker 4: 35:37

So it's a really interesting question and you get genuinely held opinions on both sides of the argument here, one being, you know, as with every other technology revolution that's gone before you know, new jobs emerge that replace the jobs that have been automated away, and there are numerous examples. The other side of the equation here is this is the first time both output and the experience of dealing with a human can effectively be replicated by a machine. And so you know, when you step into this world of automating white collar work, where do those people go? No point in going to learn how to write code, because Gen AI can write code pretty well already, and that's going to get better and better and better. So you know there is there is a real sort of a question mark around that, and you know you really step into a completely different realm of conversation here. You know big sort of geopolitical risks and sort of societal level considerations, which is not really my bag.

Speaker 4: 37:04

But if you take a specific industry like financial advice, I think if I was a, if I was a government concerned about job displacement and had some power over how this technology was adopted and rolled out across my country, I would look at underserved markets that have persistent low productivity, because effectively, by improving the productivity, you're going to fill the service gap, and financial advice in particular falls firmly into that space.

Speaker 4: 37:44

So I think there is more than more. There's 12 million additional people that need advice in this country and probably would benefit from some form of advice. There's a lot to go after there with additional automation, so I'm not too worried about that in the short term. In the longer term I mean already actually you could probably deliver AI first financial advice as good as, if not better than, a human. The technology is here already. To Craig's point, it becomes an adoption question. You know, if you think about Robo Advice, you're like well, great, now I can get this advice line completely end-to-end automated. It didn't really work. I mean, it worked for certain demographics that wanted to invest money in a certain way.

Speaker 2: 38:44

And Craig I mean in terms of the job displacement side for you thoughts on that.

Speaker 3: 38:50

Well, I think it's not a simple question, is it? I think you know, with the rate of change and the time scales, that you know with time scales changing, because time scales have changed.

Speaker 3: 39:00

You know we used to say long-term with 25 to 30 years. You know, long-term now is probably 10 to 15 years with the rate of, with the pace of change and everything on technology that we're seeing. So I think, in terms of jobs you know we've heard this it's not going to be AI that's going to take our jobs in the short term. It's going to be somebody using AI that's going to take a job, to do the job better and more efficiently and more empathetically.

Speaker 3: 39:27

I think you know, like Joseph says we're going to see new roles and new skill sets developing. So, you know, the AI engineers, prompt engineers you know someone, someone who know what a prompt engineer is. It never existed two years ago, really, you know. But if you look at job bores, now prompt engineers are, you know, everywhere it's called AI explainers. These are people who explain the outputs of AI and there's people who actually monitor monitor the AI for their ethics and, you know, make sure they're doing the same things. So we'll see new skill sets, you know, coming out. We'll also see new sectors, I think, appearing and new companies appearing. I think it's predicted that the some of the biggest companies in 10 years time in the world will be companies that don't even exist today.

Speaker 4: 40:14

Yeah.

Speaker 3: 40:15

You know the PWC have predicted that AI will pump in some like $15 trillion into the global economy.

Speaker 2: 40:22

You know, so we're going to Microsoft's doing most of that, I think, at the moment. Well, yeah, yeah exactly.

Speaker 3: 40:28

But you know and I know, as Josie said, it's predicted to affect white collar and you know, and knowledge workers. But I think in the you know in the short term, so you know, five to seven years, I think advisors and planners remain crucial, you know, to the advice process for explaining complex financial situations, to explaining the AI, you know, insights and providing that human touch to, frankly, you know, ever better educated clients.

Speaker 2: 40:59

I think also Craig on this, like for me on the job displacement bit.

Speaker 2: 41:03

There's always going to be an element of you know, within financial advice where, from a paraplinus perspective, there's the real technical, difficult questions and using that experience and knowledge from all the study, that actually a large language model is going to struggle to be able to pull together what's in the paraplinus brain you know Exactly, and from previous experience as well.

Speaker 2: 41:30

So I think, in terms of administrative tasks that are repetitive yeah, sure, fine, I completely get that. But in terms of being highly skilled in your job, it's going to take quite some time for AI to be able to get anywhere near that, from my personal opinion and I think on the other end. So going from like a very technical person to then a very good, gregarious, sales relationship type person With financial advice, it's always going to be an element of trust involved and you want people want a relationship with a human at the end of the day, when it comes to their money. So I think there are some protective shields around job roles within this industry and, whilst tech is going to improve everyone's lives the businesses and the customers I think there are areas where there's like a shield that is somewhat protected from AI. I think you're right, robin.

Speaker 3: 42:38

I think and this comes back to time scales Longer term is much harder to predict. I think in shorter term, especially for power planners. There's lots of people who have said that power planners are the first one in line for role replacement with AI.

Speaker 2: 42:57

I kind of disagree.

Speaker 3: 42:58

It's like you said, rob. These are highly skilled, knowledgeable people. What we'll see is them using AI to enhance those skills and enhance their capabilities. That's what we need to do. I would encourage everybody to embrace AI tools because it will just make us better. It will make us better at doing everything we're doing and early adopters you need to get in there. There'll be benefits for early adopters, but everybody needs to start embracing tools and I just encourage everybody to get out there.

Speaker 3: 43:34

I think longer term, I think, is much harder to predict. I have an opinion that I think more and more people will move towards AI-driven financial planning as we get further out. We've said financial planning is about trust and empathy. We've already established that AI can be empathetic, so it then comes down to trust and I think that is on us and how we trust. When we trust the machines, so to speak. There's people that still won't use ATM, so we're quite a long way away from trusting machines without all our finances. But I think as we grow more comfortable in AI in other sectors say for health, for example, health care it will change people's behavior and their openness. I think to AI financial planning. But, as I said, that's longer term and it's very difficult to predict.

Speaker 3: 44:32

We were talking earlier, obviously, about Mr Musk and Neural Lincoln saying that they've implanted a brain machine interface into a human being for the first time In 10 years' time. We could all have brain implants. We could all be part of one big neural network brain and have access to all of the information that we ever need. It'd be like in the matrix, where he knows Kung Fu. It will be like well I know, sustainable, tax, efficient investing. We don't know. We don't know what the future and this is the beauty and the risk a bit, but I would encourage everybody to embrace the tools. There's loads of resources out there to get to understand AI, from the very basic stuff right through to the company. If you want to work as an engineer out there, but understand how AI can help you in your role now and your business and that's what everybody needs to be doing at the moment, but do it responsibly.

Speaker 2: 45:31

And what about yourself, Like in terms of the feedback you're getting? You're talking to a lot of power planners and advisors.

Speaker 1: 45:41

What's the sense of it? Not many Well I'm not speaking to lots that are starting to use it and implement it. I did have Alan Smith on from Capital Asset Management and he is looking at AI and implementing it in the business. They ran through like a model of ABC financial planning and put in all these kind of situations and it took the profitability of the business from 250,000 to 370,000 just by like running a model. So he's really impressed by that and I thought that was pretty cool that he was doing that. Anyway, I think one of the questions that he had actually, I think, was just what pretty much most people want to know right now because it's quite overwhelming. Ai, it's quite like well, where the hell do you start? What's the quickest win in the short term for a financial planning firm right now if they implement AI?

Speaker 4: 46:32

What's the quickest win? Well, I mean, I don't want to talk shop here.

Speaker 3: 46:38

But you are.

Speaker 4: 46:39

I'm going to. That was a nice plug, very fair, the perfect tee up there. I mean, in all seriousness, right now, within five minutes, you can Assign an event in your VC so the bot turns up to your VC every time. You can integrate your calendar, if it's a Google calendar, and from that point on the bot turns up to your millions, captures the audio, transcribes it summarizes the output, writes your ongoing correspondence, writes your suitability report, writes a briefing note for paraplunders all within a minute of your meeting finishing. So that's a relatively low risk, high impact. Where to start and it is just a start. I mean it genuinely is just a start. But you need to have that foundational pipeline that captures the data in place, otherwise you obviously can't use it. So that's a good, definitely a good place to start.

Speaker 1: 47:55

Yeah, do you have much sort of feedback on how much money that or time that equates to? Does anyone actually kind of gone through and thought you know, that's actually saved me this much time this year and this is what it's equated to in respect of, I don't know, productivity or?

Speaker 4: 48:13

Yeah, yeah, we've got quite a bit of feedback and a growing list of feedback and I mean, if you're going to take the sort of average of time saved per case, you're probably looking at the region of an hour and a half. So if there's a, you know, an hour or 90 minute meeting, that used to take two, two and a half hours to sort of write up the CRM, write up the output, now it's taking 30 minutes because it's all written for you and your job is to read, review, amend rather than read through your notes type, read through your notes type. You know we're just getting you sort of 80, 90% of the way there, but that saves a huge amount of time.

Speaker 1: 49:03

So you said does teeing you up? Is that something that you've developed for financial planners at the moment?

Speaker 4: 49:09

It is. So that product is out there. It's been used by a whole bunch of financial planners, financial advisors. It's continually improving, but I think just the base, sort of off the shelf version at the moment now is got to a really, really decent quality. So, yeah, it's out there. It's been used. What's it called?

Speaker 1: 49:35

Aveni, Aveni, Assist. Oh, that's the actual right. I thought it's like a product of yours or something like a different, like no okay, so that's the Aveni Assist one.

Speaker 4: 49:42

Okay, cool, yeah, and AveniDetect is compliance.

Speaker 1: 49:47

Is it affordable for like individuals? So, like you know, your one man ban would be like, yeah, I'll buy that or is it like quite expensive? Is it only really for large firms with numerous amounts of advisors? What's the crack with that?

Speaker 4: 49:57

It's definitely affordable. So this is a good question, right? So this when you're talking about AI, most of the time you're talking about a slightly different pricing approach to pricing. So the cost for an AI provider is based on consumption. So how much do you use the platform? You use it more. It costs us more. It's not like a traditional piece of software that you build, sell and you know that's that you get an update in a year or whatever it is. So if someone was going to use this eight hours a day you know 20 working days a month it would be quite expensive. It'd be, you know, 250, 300 pound. That's just not the way that that advisors work. You know it's six, eight, 10 hours a week and it's probably in the region of 80 to 100, 120 pound a month. If you can double the number of clients, you can see you're talking about return on investment there, you know, which is pretty significant.

Speaker 1: 51:14

Sounds like every advisor should be using it. Well, this should be. How do we make that happen? You and me, we're making every single advisor use it. What do I get?

Speaker 4: 51:27

We might need to take this offline.

Speaker 3: 51:32

There are technologies out there that are, I tend to call them, no brains, and I think this you know of any assist and there are others available on the market.

Speaker 3: 51:45

But that type of technology to me just make perfect sense. You know it is a, it is a cumbersome and, depending on how your business currently works, right, you know you might be quite efficient at that piece. But, you know, depending on how your business runs, could save you between I think George has been quite conservative in saying that you know 90 minutes. I think it could be, you know even more than that. So, depending on how you value your time, that should return. You know that should return an investment per client you're getting back to. You know, even if it's two hours, you know there's 300 quid there If you, you know price himself, right.

Speaker 3: 52:21

So these tools have been used at the front end to onboard clients and so you know it's where it's where they can be really efficient at the minute and where we're seeing lots of new fintech coming to the market around that, that area in terms of onboarding, transcription and meetings. And you know that's where the real you know pain is at the moment and but and that's the real kind of area that is looking at the moment, the back end. I think we'll come as we, you know, as we move forward, but the client experience is where I is really being leveraged at the minute.

Speaker 2: 52:53

And, having having been an advisor myself, you know and I know that a lot of advisors feel like this that you know what they really enjoy and cherish is spending time with clients and that's what the client values.

Speaker 2: 53:03

You know writing up your notes and punching it into a CRM system and you know typing up emails and things like that. That that's the laborious stuff that advisors don't really treasure and they want to be there with the clients, talking to them and enjoying their company and providing value. So it's a dream, a bit of tech really, in that regard, and from the regulators perspective as well. You're capturing every single soft fact. So you know the you can go back and analyze what was said in the meeting and be able to really think about those soft facts and help translate that into the suitability and support and sculpt the advice accordingly. So it is taking like essentially the bits of the job that the financial advisor doesn't really like doing and allows them to spend more time doing the bit they do like doing and that the customer actually values. So you know it does feel like a win-win from that perspective.

Speaker 3: 54:07

It has so many benefits right through the business because that's single source of data and that's the client. You know the client is the single source. So you're capturing everything and, as you say, rob, you know sometimes advisors might not capture everything. They capture what they think is relevant and so forth. You're capturing all that information. That information is then fed in. You know the likes of it. Any others will will transcribe that will analyze. It can populate CRMs. It can populate reports. It can populate response emails. It can populate you know compliance checklists. It filters through the business and builds the efficiencies right through the business. So it's not just at that, the point of interaction, where you're getting time back or you're getting benefit. You are actually getting benefit right through the advice. You know the advice process in the business and you're reducing your risks and you're increasing your efficiencies exponentially. It's just, it's a win-win. It's a win-win.

Speaker 1: 55:06

Why would? Why would love to see right. Let's say, for example, today, tomorrow I wanted to set up a financial advice business. The police are out there. If I was to set up a financial advice business tomorrow, I would love to know what the best tech stack is. If I was to start one tomorrow and I was implementing AI, I was to implement tech. I don't answer this question right now. Maybe this is something, rob, we can do.

Speaker 2: 55:31

You've got Greg Baraclough from Finnovation on the call right there. This is it. Maybe he will tell you right now he's probably on the spot.

Speaker 1: 55:40

Yeah, what is like the best, like literally the best thing. So if I was like I've got this brilliant idea for a financial advice business Rob knows him a bit like that, because he introduced me to the chat from Monzo Bank and I started pitching him about a financial advice business, so it's like what is the best tech stack? You know what is it that? Would you know how much it costs and what is it and what would be on it? So there's.

Speaker 3: 56:01

FinTechs out there sitting there going, say my name, say my name, say my name. So what I will say is there is a vast array of FinTech out there available to firms. It's it's comes down to, it will come down to business vision and operating models, what you, what you're trying to achieve. But for me, if I wanted to build a business now, I would want a single communications platform that all my client interactions go through, whether that's email, text, video call, voice call, whatever I'm capturing that. I'm capturing all that interaction through one platform. But there are capabilities that can do that. That gives me that single source of truth. That gives me all that information. You know, whether it's you know they've mentioned that they, you know the, you know the mum's cat's not very well, or they're going on holiday to, you know mall deeds, or they do whatever they're doing. It's capturing everything.

Speaker 3: 57:02

That then is filtered through transcription software and capability, like of any that assist in the detect. That then populates the CRM and integrated CRM. So your client record, you know, is all there. It feeds all that into into an automated financial planning tool which can then do goal setting, you know, risk profiling and all that good stuff that then provides you with a AI generated recommendation. The advisor can then take that. You know, check it, sanity, check it, everything else. All the information is then fed through to, you know, the compliance team. There's a compliance function which basically becomes almost a tick box exercise because you've had you've had all the touch points and all the information. You've checked. Vulnerability automatically checks by vulnerability, some fantastic tools around vulnerability at the minute, which I think is going to be really key. As we come, you know, consumer duty starts to bite and then you come, you know you then back into the human touch.

Speaker 3: 58:02

So then, it goes to the advisor. The advisor goes out and you know, and talks to the client and then you're into the realm of servicing, you know and adding value. So you then into client portals, your regular client communication communications all automated and personalized for that, you know, for that client. So there are the tools there to do it, which tool is dependent on preference and so forth. But for a growing a business from the ground up today, there are some fantastic opportunities. There are still some challenges as soon as we start dealing in provider land, there are challenges that come with that. But there are some fantastic providers out there now who are really working well in terms of integration. So it's a fantastic time. It's a fantastic time to be in financial services, I think, at the minute.

Speaker 2: 58:54

And Craig it's an interesting question from Sam and you said if I was starting a new business today, the tech will look different to. I've had a business for the last five, 10 years and I have my existing tech stack Because if you've got new weight with a new company, it's a fresh slate in terms of the way in which you're capturing that data and you can actually run the process and it's coherent. So it's coherent for every single piece of advice, for every advisor in the business to run that process, whereas if you've got existing legacy processes, data kept in, perhaps older ways of storing data, deeply nested data structures, these sorts of things then the tech can't necessarily pull the data like it would for the new customer and the way that was captured. So you start to get into this position where there are new businesses coming up but there's a lot of existing ones and they're grappling with this problem really in terms of they want the new shiny tech but they're struggling to get the produce straight.

Speaker 3: 1:00:17

This is where understanding the problem that you're trying to solve with AI comes in, and of any of the prime example that transcribing meeting notes collected that information is a problem we've all got. You've said it, Rob. You added it when you were an advisor. So there's a tool there that everybody can use and utilize. But, yes, you'll have existing tech stacks and processes and ways of working and everything else that you need to be aware of. And that's when you need to take a holistic look at your business and what you're trying to achieve and finding the right tool and the right way of doing it. And it may be. It might not be ripping everything out and putting all brand new in. That's going to be very painful. It might be just identifying a tool that can help you address a specific need and a specific area, like of any assistant detectors.

Speaker 2: 1:01:10

Yeah, 100% Conscious, we've run over guys. Is there anything else that anybody wants to add at all?

Speaker 3: 1:01:21

I just there was just one thing I wanted to talk about, because we talk about a lot of the risks and we could sit here and talk about the existential risks and the desert island question for hours. He said that's a probably a bit more sit-round in the pub chat, but I think one of the things for me we need to be thinking about is responsibility. I think Joseph touched on it earlier when we were talking about the FCA and how the FCA find at the moment where there's still the human in the loop and there's somebody overseeing these things. But at some point we have to talk about responsibility. I guess it's a bit of a wider question about AI ethics.

Speaker 3: 1:02:01

So if a self-driving car runs somebody over, who's responsible? Is the car, the driver to reseller, the manufacturer, the software engineers, the algorithm, whatever? But in terms of us and our profession for SM and CR, who's responsible for the outputs of AI? Who's responsible if there is a hallucination? And how can they gain comfort that the outputs are biased, accurate, in the interest of all best parties? Because I think where we are at the minute, that's probably possible. But some AI is a black boxes Data goes in, something happens, data comes out and we can't always tell why or how. So that explainability that Joseph was talking about earlier and that transparency, I think will be key. So, if anybody's thinking about AI tools, we need to think about who's responsible, because I think at some point the tools will be doing things and doing tasks in a way that we don't understand, and at that point, how do we do? How are we responsible for those tools? So I think that's a question that will come more to the fore as we move into this new AI driven world.

Speaker 2: 1:03:19

That's a whole other can of worms. I think Perhaps another podcast on the ethics of AI awaits.

Speaker 3: 1:03:27

Well, I'm being implants by then, rob. We'll be fine, we'll just talk to each other over there. I hope the wavelengths will be fine.

Speaker 2: 1:03:33

By the neural link. Ok, well, look guys. Thank you very much for coming on, sam, for hosting the Financial Planet of Life. It's been a very, very interesting discussion and, yeah, I've really enjoyed it. Yeah, thanks, robin.

Speaker 3: 1:03:50

Yeah, it's been fantastic. Thanks guys.