Artificial intelligence in wealthtech: Delivering personalization at scale

Gavin Spitzner, president at Wealth Consulting Partners; Andrew Smith Lewis, special advisor at CAIS; Koley Corte, senior vice president and global head of business transformation at AllianceBernstein

Transcription:

Sharon French: (00:08)
I am going to reintroduce, myself and Tiffin because the room is at least 50% more people, if not more. so my name is Sharon French. I serve on the leadership group at, Tiffin BA by the way, stands for technology in finance. The firm is based in Boulder, Colorado, and at the core of what Tiffin does is combined in AI, which is a good reason why we sponsor this particular track and investment intelligence to change the user experience within wealth and asset management or investment management to provide better individual outcomes. And you heard a lot from the panel before us talk about things like democratization of access to investments around how digitalization and actionable insights for the advisor provides them more value. Certainly allows them to scale, to develop certain individual custom plans, as Sam said and around this whole, issue around continuous dynamic planning. This next panel in particular is around it's called, delivering personalization at scale, which is very near and dear to our hearts, we've off this AI track. We've got three sessions. This session, we have a short break and then a third session. So make sure to come back for that one to complete your whole experience here today, so our moderator Gabby. Okay president of wealth consulting, partners Andrew is a special advisor at case and Coley. Hi Coley is a senior vice president and global head of business transformation at Alliance Bernstein. And I happen to know that Alliance Bernstein is doing really cool stuff, transforming how they think of their go-to-market strategy. So with that please enjoy the session and I'll be back up to introduce the third AI panel. Thank you.

Gabby: (02:30)
Thank you very much. Great to be with you all great to be back in person at invest. I think I've been at every invest since the beginning in 2015, couple virtual in the interim, but wonderful to be with all of you today. So I'm Gavin Spitzner after a career in wealth management, asset management, and then in wealth tech, I've been consulting for the past seven years on all the stuff that we're talking about today and tomorrow in the wealth tech space, more importantly, I'm joined for our conversation about AI and wealth management, and how we're using it to achieve this holy grail of personalization at scale by two industry experts in and Andrew time is of the essence. So we're gonna dive in, so they understand the lens that you're coming at this through. Give the folks a little background on yourselves, role background, your firms, and then we'll dive into it.

Coley: (03:28)
Hi, I'm Coley. All right. I got a thumbs up, so I'm the head of, business transformation Alliance Bernstein, we are a global asset management firm. I sit in distribution business. So my role is thinking about how we best engage, both our third party advisors. So, retail channel as well as institutions, how we do that using digital to drive scale data for relevance and how we think about using that knowledge of the customer to expand into adjacencies. So personalization requires data. Data is really critical to how we think about things, both data and digital tools. I've been at AB for about four years, before that I was outside the industry, but prior to that had spent time in the industry. So bringing this combination of, consumer lens into a B2B practice is how I think about things.

Gabby: (04:25)
Great, Andrew.

Andrew Smith: (04:26)
Thank you. Good afternoon. My name is Andrew Smith Lewis. I'm a special advisor to case. Case is a platform in the alternative investment space. We sit between independent financial advisors and asset managers, facilitating education, access due diligence and transaction, we empower about 5,300 teams and firms that control about $2.5 trillion in network assets. My role at the firm has been in the field of education and innovation, and my focus has been on helping independent financial advisors, gain mastery and confidence in the space of alternative investments. So if you know anything about the alts market that the warehouse allocation versus the independent allocation is like this. And so part of that issue has been access to an alts platform and great products and due diligence, but a large part of it we found was education, independent financial advisors didn't have the knowledge required for them to advise their clients. And so we set about three years ago to create a platform to empower them, to learn. And my background, I don't come from the tribe of wealth management. Like most of you all, my professional background was spent in the field of kind of at the intersection of narrow artificial intelligence in cognitive behavioral science building systems to improve human performance. And that's what we've applied at case. And I think we'll talk about a little bit of that today.

Gabby: (05:53)
Fantastic. So I have an endless supply of questions that in our short time, we're never gonna get through, but more importantly, this is for you. So think your questions. I'm gonna ask if you get the ball rolling, but we'd love to get you involved in the conversation.

Coley: (06:07)
Let's hope we have an endless supply of answers.

Gabby: (06:09)
Yes, you do. Or you can leverage the AI just to generate exactly the answers, so let's, start with definitions. AI means different things to different people for some it's true artificial intelligence, Andrew sort of use the term narrow, AI others. AI is more about augmented intelligence, especially in terms of how we're helping advisors advise clients in the last panel. I heard that the term automation is it just automation mask rating as AI? So let's, let's hear from the experts here. What does AI mean to you?

Coley: (06:49)
So I like the augmented intelligence, right? I think we're never gonna, replace people and that's not our goal. Our goal really is to make all our people superheroes, how do we use the technology to enable our people, to perform at their best all the time, and to perform where they can't necessarily touch the customer too. So that, that idea of man plus machine I heard on the last panel, right, that's really important to me, data and AI absent context ends up with sometimes exercises in uselessness. For those of you who are watching, the hockey playoffs, the Rangers play, there was a statistic that the Rangers won more often when the opponent had more shots on goal that did not mean that the goal should be to have the opponent take more shots on goal, right? Like you didn't wanna increase that metric. It just was a fact that they were winning more because they had a really good goalie. So you can draw the wrong conclusion if you don't have that supplement of the person. And, the superhero analogy works really well for me, Batman, Ironman, they were wealthy people, right? That used technology to perform their outcomes. They actually didn't have natural superpowers, other superheroes did, but I think of, AI as giving us all superpowers, if we can use it the right way.

Gabby: (08:02)
Nice, Andrew.

Andrew Smith: (08:04)
I'm gonna disagree and say that I think we're about six months away from a complete Terminator style, event on this planet. And that we're all toast, especially in wealth management, just to sort of make this a little bit interesting. Hey, let's go no, I think, I think your point about augmented intelligence about this combination of, human and machine is the right way to look at it. I think there's too much emphasis on machines replacing the cognitive workload of people and that's unrealistic, as we've seen, there's just limitations to where the AI is gonna be anytime soon. So I agree with you. I specify narrow AI because I think there's a concept of big AI, which is generalized intelligence, and that's a very different, ambition than narrow AI, which is sort of scoped to machine learning, natural language processing. A lot of the, engines that drive the algorithms that many of us deal with.

Gabby: (08:55)
Great. I like that. So the holy grail in wealth management is personalization at scale. I've been at a bunch of conferences recently, and that phrase is being thrown around like crazy. So let's, break it down a little bit, talk about what it actually means and whether or not we're actually accomplishing it. We know clients want advice and engagement that is personalized to them. They don't want to just be a number and get generic advice, but we know that is hard to scale. So talk about maybe Coley, I start with you on this one as well, personalization at scale. Is it real? Is it happening? What are we doing? Well, what can be better?

Coley: (09:35)
I'll go first, but next time Andrew's going first.

Gabby: (09:37)
I Can't be, I've got a good one for him.

Coley: (09:40)
This scenario, but anyway I'll let myself be blown up no, but I think personalization at scale is real. And I think there are ways to do that, right? by learning the customer more by delivering off a chassis. And some of that is delivering to segments of one. As we know the customer more and more, as we can find them and engage them, we can start to personalize just like we do in human touch. Right. We in human touch, you're one to one. There's no reason to believe that we can't get there. We're not there yet. Right, and so we are, but we are getting closer to that where it feels one to one where you can personalize the touch and you can use the name, the attributes that you're the person is engaging with, that they're interested in, that they have purchased from you, right?

Coley: (10:25)
And you can deliver against that. You don't have to be in a mass email seven or eight years ago when I was at T I A A, an employee on my team or a younger employee came into an operating committee meeting and said to the operating committee who didn't believe they needed to personalize things at the time, again, companies can't find out. Now this is not an indictment of the firm. but at the time this person on my team said, I don't read anything if it's not addressed to me personally. And they were just stunned, but it was a fact. And now many people would say that, right? And there's an exchange of value. If I'm giving you my data, I would expect you to personalize your outreach to me, if I've shared data with you, if you give me a homogeneous, generic outreach, I'm much less likely to engage. So I think using an exchanging value for data is important in

Gabby: (11:14)
That relationship. And there's a lot of great research. I resisted, my consultants urge have lots of slides and data, but I do have a lot of slides of data on the fact that clients will gladly share quite a bit of personal data. If they get something of value to them, not you selling me product, but you advising me and I'll, I'll do that trade all day long, but you have to do that. Andrew your thoughts?

Andrew Smith: (11:36)
Yeah. On my side, the personalization is really a function of, my work with advisors in education. So again, my whole thing is that educating advisors needs to be done in a way that's distinct from what we've seen up until now. There's just way too much information hitting all of us in particular advisors for them to be able to really give their clients what they need. And so my focus has been using personalization, truly understanding on an individual by individual basis. What do they know? What do they not know? And what do they need to get the job done and delivering that just in time to them. And I think we've seen great examples of that come out of education, a lot of work in the military, and we're starting to see it apply to advisor learning. So we've taken 20,000 or so people through a very personalized experience that improves the outcome.

Coley: (12:28)
I just add a point to that? So it actually really ties closely to what you both just said. So we have a tool that we build called the digital coach and the digital coach is exactly where, we are providing value. It's a diagnostic tool to help you understand your highest priority practice management needs. As an advisor, you are helping us understand you and we're helping you understand those needs better. And so that is Gavin to your point, that's that exchange of value. You are engaging with us and sharing information. We are using that to create a personalized learning program for you, and then continue to add value to your practice along the way.

Gabby: (13:05)
Love it. All right. I'm gonna start with Andrew this time. So get prepared to disagree and I wanna go deeper into what you were just talking about around learning. So we're putting a huge burden on advisors. When do you think about it? There's so much just in this conference alone, you think about crypto and ESG and all in tax planning, estate planning. Some of the things that that Andrew talked about in the last session how can some of these tools that you're talking about around AI? How can that help move advisors through that learning curve and help them deal with that complexity?

Andrew Smith: (13:45)
Yeah, I think this is a challenge that's not unique to advisors. It's, we're all in this together. We are all getting inundated with information and education has kind of become information dissemination rather than knowledge retention and acquisition. And I think that probably has anybody in this room taken a college course or a high school course when you were kids, they ever have a course called how to learn. No. Right I mean, education is about the what, right? We're very, very focused on what in education, but how we learn is left up to us. Right? You gotta get stuff into your head, keep it there and be prepared for exams. Did any of you, when you were growing up, used an age old practice called craming maybe a couple, right? So craming is where you sort of party the whole semester, and then you stay up the night before and as much caffeine as your body can tolerate to retain information regurgitated on a test only to quickly forget it thereafter.

Andrew Smith: (14:42)
Right. And so you think about that, that's probably not a really good way to educate people, right? You don't want a doctor that's cramed for a medical procedure the night, right? You don't wanna fly home on a, with a pilot who just cramed to qualify for a triple seven and you probably don't wanna financial advisor, a wealth manager who cram before they tried to explain crypto to you and what you may or may not be getting into. So the whole thing about how we learn is really essential. And I think that this is a the core driver in one field of personalization.

Gabby: (15:14)
Love it.

Coley: (15:16)
I don't know if I can go after that, honestly he has much more expertise in education. I will say that for us, this personalization of education, right? You can't, maybe it is the craming analogy, but you can't absorb everything. Right? So for us where the digital coach is adding value, is that friction of what is your highest priority need, right. If I ask you, do you wanna grow your practice? Do you wanna learn how to have a tough conversation? Do you want to get more referrals? Yes, yes, yes. Right. If I say, which one's most important to you and work through that with you, then that's created ahha moment for you. And now I can help you with that because we've crystallized, what's important to you. Not what's generically important. Yes. I need to learn lots of subject matter and lots of products and lots of different ways of managing my practice. Now I've crystallized where I wanna focus and delivered against that. And that's where I think this man plus machine data plus human comes together is in creating relevant solutions that are important at that moment in time, not just important overall.

Andrew Smith: (16:21)
So You're kind of the front end, the funnel, right? So helping people figure out what they need. And then the question is how do you help them get that information into their head? So you need both.

Coley: (16:30)
You do need both. Yeah.

Gabby: (16:31)
Absolutely. All right, morning, I'm gonna ask one more question, then open it up and then I'll keep going. If you don't, but I want your questions. Let click more into actual advisor adoption, and maybe you can share some examples, Andrew of what you've done at case around alts. What have you learned both of you in terms of adoption, best practices things that you adjusted along the way?

Andrew Smith: (16:58)
Yeah, for us it was a real business need. And again, it's not just a matter of check, check, check and somebody gets a certain score in a quiz, and now they can start talking about crypto. We, we noticed that advisors really would not, move product for things they didn't understand, and they're not going a good, advisor's not gonna put their client into something as complex as a hedge fund without really understanding it. And so what we've seen is that an educated advisor will transact faster and more often. And that's good for everybody. It's good for us being in that, middleware layer. It's good for the asset manager. It's good for the advisor, more funds under management, greater wallet share. And it's good for the end client because they're getting a more nuanced experience recommendation.

Coley: (17:44)
So I guess I'll focus a little bit more on the sales team, although it gets to the advisor, right? So I have a sales team that works with the advisors and I first need to get the sales people to adopt something right before I can even. So I have, framework I use, which is adoption engagement outcome, right? If I can't activate someone, if they haven't used the tool where you can't create engagement, we can't get to a business outcome. And obviously at the end of the day, adoption's just activity. I really need the business outcome, but putting a score on the scoreboard requires me to measure all the way along the way, and then to think about all the audiences, right? So the digital coach, the first thing, and I've talked a lot about the coach. It's not our only solution, but the coach, the first thing I need to do was get the wholesalers, the sales team, all trained on it, right then I had to get them to use it with a customer.

Coley: (18:33)
Then I had to get the customer to finish a diagnostic experience. Then we could get to an outcome of whether or not we furthered and advanced the business relationship, same with social selling, right? So we have enabled our sales teams to use social as a way to engage advisors, first they had to update their LinkedIn profiles next. They had to share content. Then that content had to create engagement with an advisor base, right? And then they had to continue to build that audience and engage that audience with relevant content. So I think about it that way. And then I think about, look these are all really competitive people. Most people in this room are really competitive people, right? Sharing data from each other on how each other are performing. And then having the ones who are finding success, tell those stories is so much better than fully telling them what they need to do. Right. So, engaging that audience and creating the change is a really important part.

Gabby: (19:20)
I think both of those use cases are fantastic. Cuz Andrew was thinking about your point of helping them get comfortable. We know advisors, there's nothing. If they feel like they're risking looking dumb or embarrassing from the client, they are never gonna come near something, even if it's the right thing for the client. Yep. So it's a huge value. All right. Let me scan the room, brave souls to break the ice and get us, get us going. Come on back there.

Speaker 5: (19:50)
Question. So what are some examples of, to deliver on deliver social possible? We have tools in that business to scale or other examples.

Coley: (20:12)
Yeah, that's a great question. So I actually always think about the things that aren't data yet. Right? So how can I turn everything into data that I can then mind? So unstructured sources and structured sources. So call recordings queries, we run on Morningstar, right? We put things in lots of places that we don't collating and coalescing and using his data. So the first thing I'm trying to do is turn more things into data and then structure and hang that data off of something. So we're building a client data master, right? And then we can relate it all back social. Data's one of the things there's some things I can get and there's some things I can't and there are other things I can do in terms of social listening. It depends on the channel. LinkedIn's a very hard channel to get data from, but there are some things you can glean, right? Not yet for us In the i.S LinkedIn is the most valuable business channel so but we are trying to mine all those things and bring them together in this place. We're calling Oculus, which is our tool for sales enablement for the future.

Andrew Smith: (21:15)
Let me ask you a question, where does the world store its most valuable data

Gabby: (21:23)
Putting Zoho on the spot?

Andrew Smith: (21:25)
Where do you think, where do we start our most valuable data yep. In your head. Right. And what if we had a better window into that data, what would the implication of that be? And I think that's what, personalization and the right use of artificial intelligence can get us. If we truly understood what was in our heads, we would have much more actionable data to work with. And we're all running around with that. And the problem is that we don't have a very good lens into that. And as technologies improve our ability to access that, that's when I think it gets really interesting.

Gabby: (22:00)
And I'm seeing some, we were talking about this before. I'm seeing some nice developments with some of the firms in the room of more intelligent search. Yep. Leveraging AI where it's not like, if you don't know exactly how to ask a question, you're not gonna get the right answer. It's a little more nuance than that.

Andrew Smith: (22:17)
Yeah. Has anybody in the room seen G P T three or Dolly two? You know what it is? Have you seen Dolly two? Yeah. It's Dolly two is a system where you basically, you type in anything you want. I wanna see a Marilyn Monroe, and Maung playing poker and the system gives you an image like that. And it's absolutely it's uncanny. And it's the most like into the future thing I've ever seen anywhere. It can do photorealistic images. You can give it scenarios. It creates images for you. Soon. It will create video for you. And that type of capability is just remarkable. And it's something when that sort of hits the rest of us mainstream, it's gonna be, it's gonna be unbelievable what we're gonna be able to search for and what we're gonna be able to conjure or create. Because imagine just being able to storyboard by just talking to your computer and having it, create that, those images for you and text

Gabby: (23:14)
And someone's priority doing this, but just the notion of using deep, big technology to personalize videos, right. Where you can record one.

Andrew Smith: (23:23)
That's a really good idea.

Gabby: (23:25)
And apply it to clients.

Andrew Smith: (23:26)
Absolutely.

Gabby: (23:29)
Anyone else out there? Thank you right here. Yeah.

Speaker 5: (23:31)
Just one question. Things are turning out to be really complex. So how understand is this evolution is it's easy to for clients comprehend and from market perspective, what is your big implications from stock market, all markets, everything so political and that

Gabby: (23:59)
There's a lot in that one. Any anybody want to take that

Speaker 5: (24:04)
Pretty broad?

Coley: (24:05)
I think the challenge is, and this is where people still become important is the past is not, as we all say the past is not a good predictor of the future. Right. And so actually when you look at a lot of the models now with the market volatility, we're seeing, do they know how to absorb all that? Right. And so it is about training today. Maybe not in some of the worlds that we're gonna move into, but today's technology is trained by a human, by the way, it has all the failings that the human had trained it. Right. If you read about ethics and AI and other things, right. We bring our own biases into how we build the technology in many ways. So I think it'll be interesting as we can bring, as we get to this big AI, as opposed to smaller AI, how we bring in more data sources and how we inform things with more events that's where I think the human is still incredibly important.

Gabby: (24:55)
Yeah. There's a bunch of providers in the room in some way, shape or form. So what advice can you give them, as you think about the solutions that could be better for you or just build gaps in the space? Any advice for them?

Coley: (25:13)
For me it's really hard, right? And so all this work is hard and sometimes it's harder than it needs to be. And where I get value in partnering with people is when they can make it easy. Right? So a lot of times partnering with fintechs, they can work much more quickly, and Sharon knows this, having worked a little bit with us, right. They can work much more quickly than we can, they can help us experiment with things than they can. And showing me that is really, really helpful because we get, can get in our own way, right. We're trying to structure the data and hang it off this master. I mentioned about, that's like incredibly hard work for us. It really shouldn't be as hard as it is. And I'm sure someone else could solve that. Right. But once we get that right, we can introduce these other things. In the meantime, people can help us iterate and experiment around these other areas that we can bring it that's in my view is the biggest value you can bring me.

Gabby: (26:04)
Great. Andrew, any thoughts on that?

Coley: (26:06)
I think that was a Brilliant answer.

Gabby: (26:07)
It's just there. He's making up for before. Okay, are we out of time? Do we, oh, we got quiet. So, all right. Question please.

Speaker 6: (26:21)
I and the data bias themselves. Third component that, that engineer bias don't. So that to me is very concerning. How have you seen that manifest in a real world? If you have specific examples and what comes about?

Gabby: (27:09)
So the question is for everybody, the programmers are, have natural biases, data has biases. The recipient has their own biases. How do we responsibly and effectively deal with those?

Andrew Smith: (27:27)
I think your question is biased.

Andrew Smith: (27:33)
I mean, how do you deal with it? There, are last count. I think there were about 35 or 36 human biases and it's amazing. And we all, we fall prey to these things all the time. And some of 'em are very, very interesting. Right, and I think that, I mean we all do this you go you have the confirmation bias. How often do you Google for something where you're not looking for the answer you're looking for somebody else, who's gonna confirm that Elvis Pressley is actually alive right now in Memphis. And if you Google that, you're gonna find an article. He is, and you're gonna find it. Right. So the confirmation bias we do, I do this 12 times a day. So yes we are organic biased creatures. And it's really interesting when you think about things like how many people drive a Tesla, couple people, no one in the room drives a Tesla. Okay, sure. You're from California. I'm from California. I, there are no other cars than Tesla.

Gabby: (28:25)
My Uber in San Diego was a tesla.

Andrew Smith: (28:26)
Was there a human in there?

Spaeker 3: (28:28)
I think so.

Andrew Smith: (28:29)
Driving. Yeah. I mean, you think about the fact that there are, to your point, there are engineers that are making software that are making critical decisions, because at some point that car has to say, if there's an accident to be had, who's gonna eat it. Right. Is it the person who paid $120,000 for the car or some, or the kid crossing the road? And the kid cross illegally was the light red. Like there's a human being, a biased, human being, making these decisions. But I think that it's a function of data. I think it's a function of a lot of use cases and things sort of getting ironed out in the data. I think if you zoom in on one example, it becomes very zero sum and very bias. But if you can generalize out, I think that things begin to normalize. What do you think?

Coley: (29:15)
I think also you can test those assumptions, right? So how do you start to surface some of those? How do you, interrogate the data, the conclusions, the biases, right? How do you try and say, okay, I know when I look for this, I could be doing this, right? So I'll give two examples of that. One is, at a previous company, I won't name them again cuz whatever. But when I started, I had a team that was about competitive insight and market strategy. Right. And how are forming strategies based on what we saw in the marketplace and the person said, you're gonna be very careful. Are you finding the data that tells the story you wanna tell or are you telling the story of the data tells you, right. And continuing to question that, are you telling the story, the data is really surfacing or are you finding the data to tell the story you wanna tell? And by the way, with salespeople, you'll see lots of that. Right? People want the data to tell a story, they're very good story tellers, but how do you make sure. Right. And how do you continue to interrogate? And I think just opening our own eyes to the biases that can be there, help us. Right. Just like in the person before was talking about diversity and inclusion, it's the same in diverse situations, how do you look, check yourself or unconscious bias, right? How do we check the data for bias.

Gabby: (30:34)
With that? We need to wrap. Thank you for the questions Coley, Andrew. Thank you very much. There's a quick break. Am I right? There's a break.

Coley: (30:45)
Yes.

Gabby: (30:46)
I'm gonna, she's gonna tell you all about this, but come back for the rest of the discussion.

Sharon French: (30:50)
Right. I just wanna point out one thing that Coley said that I think speaks to a lot of the discussion today, specifically around data, this issue around capturing events and activities before they become data and converting them into data. That is just I cannot emphasize how important that is, in our third session to Gavin's point, it's sort of the third leg of our AI stool and Jack swift will come up here and talk to you about the, how there's a lot of people in the audience talking about, okay. So like how do we do that? Right. So, but I think this concept of converting events and activities kind of also gets to the bias question into data, helps get, I don't know all the way there, but a lot of the way there. So this was an awesome panel. Thank you to all of you. Thank you, if some of you just wanna stay in the room and network, because we only have 15 minutes feel free, but, our next session starts in 50 minutes. Thank you. Thank

Gabby: (31:53)
You.