Join Heather Hopkins and Alasdair Walker at Handford Aitkenhead & Walker as they explore the transformative impact of AI on financial advisory firms. Discover how industry leaders are integrating AI into their practices and preparing for the future.

In Episode 3, Heather and Alasdair explore the cutting-edge world of AI with guest speaker Kapil Ram Bhatia from Salesforce. Discover how Salesforce is revolutionising tech for financial advice and wealth management firms, and get an insider’s perspective on the future of AI in the industry.

Watch via:

Transcript — 

Disclaimer: This transcript was produced with the assistance of AI, so it may contain errors or inaccuracies.

Heather 

Heather. Hey, welcome back to nextwealth insights. My name is Heather Hopkins, Founder and Managing Director of nextwealth Ltd, joined by my co host, Alistair Walker, Managing Director of HW&A Chartered Financial Planner. Hello, Alistair.

Alasdair

Hi, Heather. Great to be back for episode three, I just wanted to give a couple of quick reminders to our listeners before we kick off with the episode with a great interview we’ve got lined up. By way of reminder, the series is a companion to our AI Lab membership group, and the whole project wouldn’t be possible without the help of our sponsors and their Aviva, Fidelity, SS&C and Salesforce, and I understand we’ve got a great guest today from Salesforce who’s going to give us some insights into how they’re shaping the future of tech for financial advice and wealth management firms.

Heather 

Fantastic. Yeah, we’re delighted to have Kapil Bhatia with us. He is senior RVP of wealth health, pensions and asset management at Salesforce, UK, I should disclose Alistair that I’m actually I have a small shareholding, really small, but about it’s maybe 15 20, years ago, I had a pitch from somebody when I was living in Boston for Salesforce, for the CRM, and we were using this terrible database to manage our CRM. And so I bought a I wasn’t able to convince the company that we should implement Salesforce, but I thought, this isn’t a bad shout, not a bad idea, so I bought a few shares. So So full disclosure up at the beginning, but Kapil has extensive experience driving digital transformation for financial services through Salesforce suite of products. He’s passionate about helping firms leverage technology to improve client experience, service and operational efficiency, particularly through AI and cloud based platforms. He leads Salesforce, UK’s Enterprise team to support clients across sales, AI, marketing, customer engagement and service, data integration and analytics. And prior to Salesforce, Kapil worked as a management consultant at McKinsey BCG strategy in the US, Middle East, Africa and South Asia. So I’m really looking forward to this conversation. Welcome Kapil.

Kapil 

Thanks, Heather. Lovely to talk to you and Alastair about this very exciting topic. Fantastic.

Heather 

So I think, I mean, I’m a bit familiar. I already mentioned a bit of my background and and interactions with Salesforce, but it, although Salesforce is known as a CRM system by many, and it’s got huge adoption, particularly with larger enterprise clients, but all sorts of clients, it’s, it’s better known as a CRM. But that’s not all you do. And in fact, you know, looking at the revenue mix for the company, it’s become less important. So how, how would you describe Salesforce?

Kapil 

I’m very glad, Heather that you asked that question. Actually, while we are known as CRM, and it’s even our stock taker, CRM, is just 11 to 12% of our revenue, the core sales CRM. We have many other products beyond the core CRM line. So when we began 25 years ago, the vision was to have software on the internet, in the cloud, you know, so to speak. And since then, we’ve grown in the last 25 years from going from zero to 28 billion now, and 80,000 employees, and most of that growth has come by adding products along the customer journey. So we started with Sales Cloud, then we had Service Cloud, then marketing cloud, commerce cloud, so on and so forth. So any touch point of the customer, we had a cloud for that. And as a result, today, like I said, you know, 88% of our revenue comes from outside of CRM. I think just want to also highlight two other salient characteristics of Salesforce as a company. Firstly, you know, we are very kind of a values driven organization, and customer success is one of our top values. And to that and across all our products, we are a very low code, no code, kind of platform, and it’s there in our ethos which means our consumers or our clients can actually customise the product for their own processes and their own needs. And secondly, I think about Salesforce, about 9 – 10, years ago, we began another push, which was to make our products very industry specific. And to that end, we have a product called Financial Services Cloud, which has a data model which is very much oriented to wealth management and asset management, and also has many processes pre built into it. So the idea is that we are able to give our clients products which are A customisable but also have a good balance of out of the box capabilities for them to deploy.

Alasdair  

I think the customisable bit is quite interesting. Because I suspect our listeners will have used Salesforce as end users more often than they might have realised. And what I what I found is in my time, and I maybe pay a little bit more heed to some of this stuff, given my background, but I’ve only ever noticed I’m using Salesforce if something’s gone wrong, so like, if the internet’s dropped out, or I’ve had an error code, then it drops out of the, you know, the intelliflo self service system, which I’m pretty sure was built on Salesforce. For example, I, for my sins, have to interact with the FCA semi regularly on their Salesforce platform. But what’s great about it is, well, I say great. I mean, what’s interesting about it is that you don’t know it’s Salesforce, right? So from a sales perspective, for you guys, that might not be that great, because you use this all the time. You just don’t realise.

Kapil 

Yeah, no, actually, it is quite prevalent, right? Heathrow Airport. If you’re going through the check in points, it’s all Salesforce. If you go to pred get a coffee, you know, behind the scenes you have murals of software working. There’s many, many instances. And also, I think the brand Salesforce, also, you know, categorizes us more into the sales bucket, whereas we are, like you said, you know, very much broader than that.

Heather 

We heard on our first podcast with Pete beardsmore about the partnership between ssnc and Salesforce to be able to deliver SJP’s Advisor Assistant, which is an example closer to home that some of the people listening might be familiar with, and it’s another example of people wouldn’t necessarily know unless they looked behind the scenes that that you were involved in that.

Kapil 

SJP is a big partner for us, and it’s a big reference for us in wealth management, where we are the single pane of glass for them across all their advisor network. And I think Heather, you had written an article many years ago, or couple of years ago about how a typical advisor has to refer to 10 different systems, and what we’ve done, and what’s our overall aim is that all the data doesn’t have to be in Salesforce, but it should be visible through Salesforce, so you have the single pane of glass, and your advisors don’t have to spend too much time searching for data. They’re able to see and get as comprehensive customer view as possible, so that they can spend time and give the best possible advice to their end clients. And also, you know, the data is in one place, or at least visible in one place, so when there is a changeover of advisors, the new advisors can be added and which is a big issue now, you know, with the whole generational wealth transfer and advisors retiring, that’s, that’s, that’s very important. So SJP, you know, they’ve chosen us as the one single pane of customer glass. Sorry for us. We had a rapid launch with them. Nine months we were able to roll it out to 10,000 advisors. That’s also because our system is, like I said, fairly low code, no code, kind of system. And we were able to decommission 17 existing systems which they had, which, of course, also made a big difference to their cost base.

Alasdair

You know, you’re absolutely speaking my language, as somebody who runs a business that’s 46 years old, and we’re dealing with, you know, multiple generations of tech debt. The idea of being able to roll some of their many systems, and we currently have to refer across to some of that’s my own doing, of course, is, yeah, definitely, definitely music to my ears. When we’re interviewing people on this podcast and when we’re talking at the AI Lab events. One of the things we tried to do early on in the conversation is a bit of level setting and an interesting sort of changing beast, actually, in a lot of ways, is the way we’re thinking about and defining AI. So what started out being a question has become a really active learning process for me. And you were at our last event and and I really did think Professor Adrian Hopgood introduced this idea of these sort of quadrants of AI, you know. So you’ve got a spectrum of generalist to specialist, and generative to predictive. So you’ve got these like different areas to consider. I’d be really interested to hear what your thoughts are about defining AI, and how you think about AI, and perhaps how Salesforce thinks about AI as well.

Kapil 

So we also have the same, very similar classification in terms of predictive and generative. Predictive AI is something which we’ve had in the product for a long, long time, almost five, six years ago, when we started releasing predictive scores based on AI for leads in Salesforce and even you might have, you know, might have used it, or it might have, you might have come across it. Generative is relatively recent, and that’s one. Is one angle. The other axis for us is more driven by business value, so you have predictive and generative on the other axis is, what are the use cases? The use cases are divided into two categories, one is on the revenue side, and second is on the cost side. So which are the use cases which help drive productivity, improve efficiency, so on and so forth. They are on the cost side and the use cases on revenue side are more about those which improve your revenue. Maybe identify areas of cross, sell, up, sell, so on and so forth. So when we think about any particular industry, like wealth management, for example, we’ll think about the use cases, and then we’ll plot them into this similar matrix where some are predictive and good for cost. Some are predictive, good for revenue, generative for revenue, typically, typically, the predictive for cost are the ones which get faster adoption, because predictive is, you know, has been around for longer and has less, less risks. And the ones, I would say, on generative drive revenue, you know, broadly, roughly, are the ones which are more further down the line.

Heather 

Yeah, it’s a really helpful framework, I think, and, and I think that discipline of thinking about because it’s really easy to play with these tools. And when you know people are in businesses thinking about, you know, I could do a bit of this, I could do about it that, but really having that structured approach to thinking about, okay, what are the, what are the use cases in my business, and that category of  revenue generative and cost saving, I think, is a really, really useful framework for people.

Alasdair

This comes back to conversations at the last event about profitability and about demanding a return on investment. And I think to further split that between cost and revenue is really interesting, because you can make arguments one or the other, but it is that starting to try and demand some return on investment is a really interesting point as well.

Kapil 

Yeah, what we typically see in the current environment, and which won’t be surprising to either of you, most, most of our clients, want a business case where the cost is neutral, and most of the investment is paid off based on the cost, and then the revenue is the cherry on the cake on top. So when we try to propose an investment, it’s always trying to be cost neutral or even cost positive, because the AI can drive a lot of efficiencies. We have situations where the existing setup may have 1.5 or 1.6 paraplanners per advisor with AI, you could take it down, or you could help them expand with while keeping the back office capacity the same right. So definitely a very useful framework to think about getting ROI and investment.

Heather 

So one of the things that you mentioned to my colleague Julie, when she was doing some interviews about research and due diligence on firms. You know, when they’re looking to work with an AI provider, any kind of tech provider and, and I think, you know, you talked about this, this revenue growth or or cost savings, but the other, I mean, one of the, one of the big considerations, is the cost for change management. And I’d love to hear some more of your thoughts around that, because I think what you what you cited, was that for every dollar or pound that you spend on tech, you have to spend five on people. And I think that’s something that people might underestimate sometimes, how difficult it is to change behavior.

Kapil 

It’s I, to my mind, that’s a very, very big point. You know, we see a lot of projects across the industries, and almost always the difference between successful and, you know, not so successful projects is the emphasis on change management. So much of it is about changing users behaviors and getting the technology and the user to meet halfway that, you know, you just can’t hope to have a successful project without proper change management. And it’s especially important for AI, because within AI, then another aspect of change management becomes important is that there is also AI skill gap. So,  Alistair, you and I were talking about agent force a bit earlier, right to implement any of these AI projects, you know, you need to have folks or people who are skilled at, let’s say, prompt engineering, right. Figuring out what kind of prompts I want to give data and data engineering is also very important. Those are all of that comes under the bucket of, you know, change management and and skill gaps. I think from a, you know, rough, rough, rough metric, I would say technology is 15% of the whole solution. And we get excited by Chat GPT and and Salesforce and our products. But really 15 to 20% is data, and 60, 70% is the change management aspect,

Alasdair

That’s such an interesting point that I don’t think particularly in the circles of sort of techie, geeky people that I find myself talking to, that we don’t consider there’s an anecdote that I’m always reminded of in the situation that was when I joined the business that I now run, there was a member of the team who’d been at the business since its inception, and she’d been, you know, computers have become a thing that people used throughout her career, and this was in the sort of mid 2010s her process for searching the internet was to open her web browser, and that opened up with Yahoo, because she liked reading the news, but she knew Google was better. So she searched for the word Google in the Yahoo search bar, hit enter, clicked the Google link, and then searched for what she wanted. And that was her workflow. That was an adapted workflow. No one had ever looked at that and gone, wow. Like that’s 30 seconds you could have back every time you want to search the internet. And I always try and remind myself of that when I’m thinking about implementing something new, because at the end of the day, the people pushing the buttons aren’t the people who are, you know, deeply immersed in this and understanding this and and that was from a training perspective, particularly when you’re providing the software with all the bells and the whistles on that must be a really interesting challenge for you guys.

Kapil 

You know, the success of the project is as largely, largely correlated to the quality of discovery we do so before we even deploy, or before we think about, okay, what is the solution, having done the discovery understood, understanding the existing user habits, understanding the existing pain points, and then leveraging the most important pain points, or, you Know, with with our technology solution to say, like, if you are, let’s say, inputting this data as an advisor twice in two different systems, and our solution can address the double keying of data as an example, using just that one level, can then help you drive a broader change, and, you get your ultimate business objectives achieved.

Alasdair

Absolutely and I think we need to think about that more, trying to implement things. And I guess sort of thinking about advice firms and wealth management. And you know, that’s a big part of our of our audience. And given that we know that Salesforce are massive in the wealth management space, in the US, and again, sort of behind the scenes, potentially in the UK, a little bit more than sort of front, front of mind. Be really interested to hear how Salesforce is being used by advice firms, what some, some sort of best practices that those in the UK can learn from there.

Kapil 

So, going back to my earlier point to which I made, to Heather, the main point of you know, in wealth management for Salesforce is largely on trying to give the advisor a single pane of glass for  the advisor, so that they can view all the customer data in one place. And then on top of that, what we are doing is we are we have built in the processes inside Salesforce, so that basic processes, whether it’s onboarding, whether it’s querying financial data, or it’s capturing consumers financial goals, or creating, suitability reports, all of that can be done within Salesforce so that it makes their advisors life easier. To us, that’s the ultimate investor, you know, the best deployment of of Salesforce in wealth management.

Heather 

Just recently, there was some really exciting announcements out of Dreamforce, the big Salesforce conference, about agents, and it’s something that we’ve been thinking about in the AI labs group about the role of agents, and there was a number of different agents, and as I’d be really interested to hear your thoughts about how you see that playing out in financial services and wealth management in particular.

Kapil 

The new, new announcement we made was about agent force, which is basically autonomous agents. Yeah, and we are going to release three autonomous agent profiles later in October, going to slowly expand three to almost 100 different profiles, so there could be one for underwriters, and they will definitely be one for advisors. So there will be an advisor autonomous agent who can be trained to do tasks which normally an advisor would do, so, thinking about any of the mundane tasks which an advisor does, which could which the agent could be trained. And of course, you will train the agent in a way, and you can define guardrails. So let’s say when, at a particular point in the process,  if the task becomes beyond the capability of the autonomous agent, then, of course, it will be handed over to the advisor. But we think there’ll be a fair number of existing admin tasks which the autonomous agent could be could be doing themselves. I can describe a bit more about what’s there behind the behind the agent. But essentially, it’s the agent interface, yeah, which we would commonly know, know as a chatbot, right? There’s a reasoning engine, and then there’s an agent builder, which lets the clients, or lets Wealth Management firms, define and build individual agents. So you could have, you know, one agent for the advisor. You could even have one agent for specific tasks to do, back office, report creation and so on and so forth. And we think I would imagine, in the next two years, or, 18 to 24 months, we will start seeing the first roll out of these agents and wealth management being fairly widely adopted,

Alasdair

And just to try and bring that into the practical realm, from sort of a business owner’s perspective, are you saying, you know that effectively I could, I love the idea of agents, because I’m imagining all these people wearing suits and ordering martinis, but the idea being that we might have an email agent that’s just going to receive sort of email correspondence. I might get an email from a client that says, Oh, by the way, we’ve moved we’ve changed address. And so I go, right, I’ll loop you in with your key administrator. They’re going to be in touch and tell you what information the product providers that we deal with on your behalf are going to need to make those changes. And I can see this sort of, there’s a proliferation of work from that very simple request, you know, if you, if you tell me, Oh well, no, actually, no, the agent just does that and works all that out and sends the information out, then you can have a lot of advisors asking you where they can sign, right? Because that’s like half of the bane of our lives, going through our email inbox.

Kapil 

Yeah, no, absolutely, going through, any changes in address, updating that in the record, scheduling meetings with your clients. You know, that’s probably also taking a fair bit of time. You know, accessing calendars, doing all of that autonomously. You know, prospecting clients, like sending emails out, putting clients on a journey in terms of various life events, and sending them emails about those life events, updating financial data, making recommendations and predictions, which is not to advise to the end client, but at least giving you a first draft of what the recommendation could look like. All of that, I would say, is within the realm of possibility over the next few years.

Heather 

It’s really interesting the name agent I was listening to. I’d love to credit it, but I can’t remember which podcast it was, but they were talking about the word agent as being really important, because it because they have agency. It comes from the root agency. So they have agency to act on behalf of the company, right? They have their they or to act, essentially autonomously, as you said, and I think that’s, it’s so exciting, but also absolutely terrifying, because we go from what you were just talking about, about learning how to prompt, you know, prompt engineers, learning how to prompt properly, and the idea that you’d have agents interacting with customers or prioritizing your inbox and and although it’s really exciting you can go down a path that makes that absolutely terrifying, in terms of the risk in the business, how you establish trust, and the sense of confidence with your customers. So, so talk us through how you’re thinking about the sort of, I guess, the parameters that you mentioned, guardrails within which these agents will operate.

Kapil 

Oh, that’s a very good question. Before I answer the question, straight up, I would say like there is, you know, in wealth management, we’ve identified, and some of our clients have identified, 700 odd use cases, right? So there are definitely use cases where these risks are much lower, where you could already get started, yeah? And I think there’s been a fair amount of discussion on what those use cases could be or are. Beyond that, trust is very important for us. It’s actually our absolute number one value, yeah. So in our system, we’ve already built various filters in terms of toxicity and bias. Secondly, as you set up the agent, you get an option to set up the guardrails, which means you can define at which point the agent has to hand over to a human right and our idea is that there’s always a human in the loop. So before agent, or any AI, driven output, interacts with the end customer, there’s always a human who is looking at the output and and is approving that output, at least for the near future. We expect this to be the mode of operation, and this should work. And I think trust and risk definitely very, very important. We’ve already seen some, you know, not so pleasant outcomes with you must have heard about DPD, where they created their own the AI created its own cancelation policy, and with Canada Air and so on and so forth. So we are very. Careful you want to absolutely be avoiding that from our perspective, having the toxicity filters, having the bias filters, and then always having a human in the loop, plus giving you the guardrails to define at what point do you want to hand over from AI to human, will provide, hopefully, the sufficient protection for at least the basic use cases, there’s many to go through.

Alasdair

As as somebody who watched a video over the weekend of a sort of big science communicator on on the internet hacking the phone of another big science communicator as a case study in how easily breakable very secure bits of technology and sort of communication are. I feel like the working assumption has to be, whatever you build in an agent or a chat bot is going to get broken by someone, as you know, somehow. And the question then becomes, how do how do you minimise the damage when it does, rather than assuming that it won’t I mean, is that a fair assumption?

Kapil 

I think with that assumption, then you will be probably working very hard to you know, may end up spending a lot of resource on protecting against a low probability event, but yeah, better better safe than sorry.

Alasdair

On the subject of managing risk and keeping the humans in the loop. One of the questions that I’ve seen come up a few times is is how we manage the human in the loop around that risk. You know, take a AI, advisor, helper that’s going to write a report, and that report is going to be, most of the time, pretty good, but the advisors taking the ultimate responsibility for checking that it’s going to get sent out. Is there anything in managing the risk, or are there any tools that that that Salesforce are thinking about or providing that can help just avoid the situation where the advisor goes, Oh, well, the last five are good, so this one would be good too, so we will click that and off we go.

Kapil 

Ultimately, with the human in the loop, the ultimate responsibility rests,  with the human or with the advisor, right? So there is, there is not any foolproof technology solution to avoid the human error in that. What does happen is that, over time, the LLM of the program learns from the suitability reports or learns from the various whatever the output is, and that keeps improving the accuracy of the program right. And even before you implement, as you think about implementing AI, there will be a minimum data set requirement, which at least ensures that the first version output is of a particular quality, or a particular, you know, calibrated at a particular quality level, but the ultimate responsibility will always be, with the human, at least in this scenario.

Heather 

So we promised on our last episode that we would ask the AI bot to ask a question, and Alistair and I have independently have asked different bots, and the question that that’s come up is quite a good one. It’s how you see the role of the human financial advisor evolving as AI capabilities, like autonomous agents become more prevalent. Now I should add there were five other questions we took the one we liked, we thought was most interesting, but, but I think that question about skills of the future financial planner is really important one. So I’d love to hear your thoughts on that, Kapil.

Kapil 

I think the biggest advantage, or the biggest way they should change, is it should free up more time to actually provide the advice. I’m hoping that AI will be able to reduce the admin load on an advisor will be able to make the amount of time they spend on prospecting and finding customers much less, and hence give them more time in terms of giving advice. And they could, they could spend that additional time on, advising more clients, even, which will hopefully help address the advice gap. But at least in my mind, the future would be that, I don’t know, maybe they’re spending 50% of their time on advice and 50% on the admin and the various tasks, and that ratio becomes 70:30, or 80:20, even, and eventually we can make a dent on the advice cap in the UK. So

Alasdair

That’s a really good point. I think there’s a there’s a bigger question around skills and capabilities, as this stuff embeds across sectors, right? I always hear people say, oh, you know, be really, really great for people, they won’t need to spend all that time on on the kind of repetitive, standardized tasks. And then there’s a bit of my brain that always goes, what about the people that that’s all their job is, I don’t think that’s something we’re going to solve here today, though. We have a set of kind of rapid fire questions that we that we like to ask every podcast guest. So the idea is, you know, give us, give us your top of mind answer to the things we throw at you. I think Heather’s going to kick us off. Yeah.

Heather 

So how can firms get started? We talked about some really big ideas. Some are shorter term, some are long term, but, but how would you recommend firms get started? And we have a rule, it has to be something other than meeting notes.

Kapil 

I think creating internal knowledge articles that should be that could be a very good first point. AI, can you know process your unstructured data even there’s so much data in all the PDFs which you have, the emails and so on so forth. Scraping through that, creating knowledge articles, I think would be. I didn’t give a rapid fire answer. Sorry, that was a bit long.

Alasdair

No, that’s that’s a really interesting one, and not one that I’ve heard before, but one that’s actually been on my mind personally. So selfishly, my might ask you about that afterwards. So my question is, is about one, one application or use of AI in your life could be work, could not be work that’s made a big difference to your day to day.

Kapil 

So I, in our company, I use Slack, and Slack is slack AI has made a big difference, because as soon as we got Slack, the number of amount of time I spent on email has reduced drastically, like, you know, 80-90% and then with Slack AI, you can summarize documents, you can summarize channels, so I don’t feel lost, you know, I can just go last seven days, What did I miss? And it’ll point out the most important things, and if there’s something which it has missed, then hopefully it’ll come back. But so far, nothing has come back. So I Slack and Slack AI is what I like.

Heather 

That’s a good action for me, because I do sometimes get lost in the Slack, but it has cut down my email traffic, definitely. And do you have a podcast or book recommendation for our listeners?

Kapil 

Our chief AI scientist is a lady called Clara. She has a very popular AI podcast. It combines the technical details along with real world examples. And I like that a lot.

Heather 

We’ll link to that in the show notes.

Alasdair

Yeah, absolutely. This is, this is my favorite awkward question to ask. I always reminded of Heather’s cake. What is your biggest AI fail? Well, it doesn’t have to be big, but an AI fail that you can recall.

Kapil 

You know, the one we just referenced earlier, where the AI created its own cancelation policy, and then the company had to refund. I think that was that was very interesting for me.

Heather 

It’s a really good one, and slightly higher stakes than a cake going badly wrong in my house. And I now have a Google doc tracking all of my AI fails. If you want more examples, I’m happy to share them. So you know, just to wrap up the conversation, what one thing should people do tomorrow? You know, if they’re, if they’re just getting started on this journey, what should they do?

Kapil 

So I’m going to give a slightly, you know, deeper or a longer answer than the one thing I think, what I what tends to happen is that people get excited and they, you know, go out and try to build something or, you know, get a chat GPT license and try to do something on that. I would say it’s more important to step back think about your business, and think about the key value drivers in the business, and see which of those would you prioritise first to deploy AI. I think having the business and strategy perspective of why, part of it very clear, is more important than just, you know, trying to go out, get a get a tool and build some chatbot, or whatever else the case may be. Stepping back, thinking about the business, thinking about where it makes sense in terms of impact and feasibility, I think that’s that is very important.

Alasdair

Yeah, that absolutely lines up with the sort of evolving discussion around demanding some return, right? So is this, is this something that’s going to drive, something that drives the value in the business, or is this a shiny thing that I just think is interesting? That’s a question. Sadly, I don’t ask myself nearly often enough.

Kapil 

No, exactly, exactly, because it’s not, you know, the fundamentals of the business are still going to remain the same, right in wealth and I advice, it’s really about understanding where technology can make that quick and fast impact, and very often we don’t see that clarity. And that’s why that’s important.

Heather 

Amazing. Thank you so much. I’ve learned loads. It’s such a privilege to be able to have these conversations with people and really appreciate your time. Kapil, thanks.

Kapil 

Thanks Heather, thanks. Alistair.

Heather 

That was a really, really interesting conversation and a different perspective, I think, than we often hear about advisor tech, because it’s one of the big tech companies or somebody from a big tech firm, and I often hear that big tech isn’t interested in financial advice and financial planning and wealth management because it’s too regional, it’s too compliance heavy or to regulate, I guess so it was great to get that perspective from someone in a big tech firm that is 100% focused on our sector. And I thought that single pane of glass analogy was really interesting because I got a picture of Tom Cruise in the matrix with all that stuff on that single pane, being able to move it around. And maybe that’s what their marketing people were thinking of when they came up with that, with that line.

Alasdair

Yeah, it’s a, it’s a great promise that, as I say, the so much of the tech stack that financial planners use just isn’t currently delivering. And I think I suppose, speaking personally, one of the challenges I find when trying to trying to find it’s effectively trying to find access to the big tech solutions as a small firm. And that’s an that’s a really interesting challenge that I know there are people sort of developing around Salesforce who are trying to deliver that to smaller firms. And I can very much understand that we’re not the the, as a three planner firm, you know, we’re not the primary market, but to be able to benefit from some of that as a secondary consequence, I think, is something that’s that’s really interesting. Also thinking like a big tech company, and applying some, some relatively straightforward questions, is this going to drive return? Is it going to drive it via revenue or via increasing revenue or reducing costs? What are the key business drivers in your planning firm, and how are you going to deploy AI to fix those, helps to cut through a lot of those hard questions to answer that I think we find ourselves asking when we’re trying to review different solutions people are trying to sell us.

Heather 

Yeah, and it’s also a really refreshing approach, isn’t it, because it’s not necessarily putting together the business case that I’m going to I’m going to save X amount by implementing this. It’s helping to prioritize and think about where do I want to try a solution, because it has the potential to drive cost savings, or it has the potential to drive revenue. So, no doubt you have to put together the business case for the investment at some point. But, but just as a, as a thought exercise, creating that quadrant of you know, is it, is it generative or predictive, and what’s it going to deliver to the business and matching those things up, I think just helps people to prioritize where to start and also what to do next once they are on that journey. I thought that was really helpful. Your point about how customizable it is, I think is really, really important and quite different. One of the things I thought he said, I thought was really interesting was about the skills of the financial planner. And, you know, with 50% of advisor time is spent on sort of more administrative tasks, that maybe they could spend more time on client focused tasks, with the tasks that are more valued by clients but also more rewarding for people to do. I thought was really interesting, because actually, we’ve just done a survey, we haven’t published it yet, but about 50% of advisor time is spent on activities that can only be done by a regulated individual, but that’s only the advisors, and then that’s only a small proportion of the staff within that firm. And so the impact on not just how advisors spend their time, but how that firm is structured, could be fundamentally different. One of the things I’ve been trying to do is so it’s not using agents, but it’s that idea of, you have a marketing agent, you have a, you know, HR agent, you have a you know, whatever your agents are. It was actually Johnny recommended the implement AI podcast. And I’ve been listening back to some of the back catalog. And one of the things that they’d suggested was, when you’re prompting, say, I want you to act as my legal assistant. I want you to act as my PA. I want you to act as a strategy consultant. And so telling the AI the job you want them to perform, I think, will help us think about those agents and how you know, Kapil mentioned 700 use cases, but what are those specific agents you could use in your business and starting to treat whatever if you’re using code or chat, GPT or copilot or whatever it is, as you’re prompting, telling them how you want them behave, I think could be really helpful.

Heather 

In terms of a way that people could start to try and develop that thinking. I’m pretty sure that Chat GPT premium, the subscription thing, has got a an assistant builder built into it where you can effectively pre code, that kind of prompt in so we could all have a go at setting ourselves up with some agents for exactly that reason. And I feel like that’d be, that’d be a nice action to come from, you know, from that, that kind of thinking which, which is definitely, again, a further evolution of of what we’re thinking and talking about. The other and this is this is relevant to our to our next guest as well, I think. But the other thing this really ties in with is the idea of multiple different sort of subroutines almost building a bigger picture and being drawn together by something bigger. And in our next episode, we’re going to be featuring Professor Adrian Hopgood, and at our last AI Lab event, he was talking about exactly that. You know, he spent over 30 years as both an academic and a developer of these models, building, albeit not quite in the way we might imagine, ChatGPT, but these very specialist, effectively, agents or subroutines that are controlled by something bigger, and how we can develop that in the in the sort of generative AI era, particularly for the benefit financial services, is going to be a really interesting episode next time as well, I think.

Alasdair

Fantastic, goodness. My brain is going to explode before Christmas. Alistair, this has been great. Thank you to you, Alistair, for joining me as my co host. Thank you to our sponsors, SS&C, Aviva, Fidelity and Salesforce and Kapil is not here anymore, but thanks to him and the team at Salesforce and and last but not least, thank you to our producer, Artemis Ervin, who helps take out all the ums and ahs and all of the talking over one another? Thanks. Artemis, see you next time,

Alasdair

Yeah, thank you as well to you, Heather, see you next time.