The WealthTech Podcast Transcript

Host: Mark Wickersham

Guest: Kelly Wagman, Director of AI Strategy | Cresset

Title: AI Strategy, Shadow Tools & the Future of the Advisor Relationship | Kelly Wagman, Cresset

00:00:35.190 --> 00:00:50.639

Mark Wickersham: All right, Kelly, welcome to the Wealth Tech Podcast. I am excited to have you on the show. I think this is going to be a fascinating discussion. Kelly is Director of AI Business Enablement at Cresset.

00:00:50.640 --> 00:01:02.649

Mark Wickersham: Which I really think is a fascinating role, and I'm looking forward to digging into how firms can adopt AI in a practical and safe manner.

00:01:02.710 --> 00:01:07.809

Mark Wickersham: Kelly, would you mind, kind of, giving a brief introduction of yourself and your role at Cresset?

00:01:08.520 --> 00:01:22.230

Kelly Wagman: Thank you, Mark. Thank you for having me on the podcast. So like you said, I'm the Director of AI Business Enablement here at Cresset, and I really focus on our internal AI efforts, which I think of in kind of 3 categories.

00:01:22.230 --> 00:01:36.909

Kelly Wagman: One, looking at our foundational technical and policy strategy, another looking at adoption and training, and then the third being thinking more specifically about where we want to invest in particular AI solutions.

00:01:36.950 --> 00:01:47.270

Kelly Wagman: And prior to Cresset, I did a PhD in computer science at the University of Chicago, where I focused on AI, and particularly human-AI interaction.

 00:01:47.380 --> 00:01:58.420

Kelly Wagman: I also did research at MIT, and I worked at Microsoft building enterprise products, so I am new to the wealth tech space, but very excited to be here.

 00:01:59.810 --> 00:02:06.560

Mark Wickersham: Well, you may be new to the wealth tech space, but obviously you're not new to AI, and I think that we're able to bring that…

 00:02:06.960 --> 00:02:18.469

Mark Wickersham: perspective into the industry is great. Why did Cresset create this role? What is the vision and purpose behind the role at Cresset?

 00:02:19.160 --> 00:02:32.130

Kelly Wagman: So, Cresset decided that they wanted to bring AI expertise in-house, which I think makes a lot of sense. They want to be looking at how do we set AI vision, how do we develop our strategy.

 00:02:32.130 --> 00:02:44.210

Kelly Wagman: I think right now, AI is moving so quickly, and there's frankly, a lot of hype. I'm really looking forward to helping us navigate, you know, what matters and how do we drive really practical impact.

 00:02:46.380 --> 00:02:52.540

Mark Wickersham: That makes sense. There is a lot of hype around AI, and then, you know, how to get…

 00:02:52.920 --> 00:03:03.770

Mark Wickersham: results from AI that are meaningful, I think, is important. First off, let's talk about how to adopt AI safely and securely.

00:03:04.290 --> 00:03:12.969

Mark Wickersham: What should firms think about when it comes to adoption? What sort of guardrails and policies should they have in place?

00:03:13.250 --> 00:03:31.460

Mark Wickersham: With their employees, corporate-wide, to be able to ensure that their, you know, client information remains private, that information's not inadvertently leaking out, that there is kind of a… kind of a stated and known

00:03:31.460 --> 00:03:33.590

Mark Wickersham: Way in which people

00:03:33.610 --> 00:03:39.099

Mark Wickersham: can use AI, so there isn't, like, a shadow docket that's getting created with AI, but…

00:03:39.490 --> 00:03:56.299

Mark Wickersham: How can firms balance the ability for, kind of, employees to be able to experiment with AI, to be able to leverage the capabilities of it, yet do it in a way that adheres to the needs of compliance of the industry, as well as the needs of the firm and the clients?

00:03:57.020 --> 00:04:08.479

Kelly Wagman: So I think something you mentioned is actually the first most important thing, which is that it's important to give employees access to AI tools, because there is this risk that if you don't provide people with

00:04:08.480 --> 00:04:22.420

Kelly Wagman: best-in-class tools, they will likely try to use these shadow tools on the side, and that's actually a huge risk to having data accidentally leak out. So we provide everyone with things like ChatGBT,

00:04:22.620 --> 00:04:27.669

Kelly Wagman: And then, I think part two, to your point, is it's important to

00:04:27.670 --> 00:04:43.210

Kelly Wagman: do that along with a really robust policy and training program. So we have an AI policy that covers what you can do, what you can't do, and the thing that that really reinforces is that it's important to have a human in the loop at this stage.

00:04:43.260 --> 00:04:58.350

Kelly Wagman: Nothing that AI produces should be just sent out without having someone very closely look over it, and that AI should really be used as a support tool. So, it's not here to do your job, it's here to help provide you inputs for you to do your job.

00:04:58.350 --> 00:05:07.959

Kelly Wagman: That is kind of the policy side, and then on the training side, I think really helping people to understand, you know, how they can leverage it best.

00:05:07.970 --> 00:05:18.439

Kelly Wagman: What it is actually useful for, and doing all of this on tools that we have vetted, and that we feel good about the contracts, and that we are sure that the data is really safe.

32

00:05:19.430 --> 00:05:21.759

Mark Wickersham: That makes sense.

33

00:05:22.090 --> 00:05:35.869

Mark Wickersham: What would you say to the 50% of advisors that have not adopted AI, you know, in terms of your advice and where they should be at in their journey?

34

00:05:36.880 --> 00:05:40.210

Kelly Wagman: I think that it… it's pretty…

35

00:05:40.830 --> 00:05:51.969

Kelly Wagman: easy to just start experimenting, even if you're not using the outputs, and so I think that even if you're not sure this is something you want to build into your workflow.

36

00:05:51.990 --> 00:05:57.689

Kelly Wagman: you know, go out, get a, you know, try it with ChatGPT, try to have it help you write an email.

37

00:05:57.690 --> 00:06:17.800

Kelly Wagman: just start playing with it, because I think that it's kind of a new way of interacting with technology, it's a new way of working, and as Ethan Mollick, who's a professor, likes to say, you know, this is the worst AI you'll ever use, and so it is going to keep getting better, and I think it's best to try to get on the learning curve now.

38

00:06:18.820 --> 00:06:24.720

Mark Wickersham: Yeah, I think that makes sense. It does seem to be getting better and getting better rapidly,

39

00:06:25.020 --> 00:06:26.200

Mark Wickersham: I think…

40

00:06:26.560 --> 00:06:33.630

Mark Wickersham: In terms of the… and it's also from a, you know, a user standpoint, too, that the more…

41

00:06:33.730 --> 00:06:42.409

Mark Wickersham: you interact with AI, the better results that you can get, because you understand, like, how to craft that prompt better,

42

00:06:42.640 --> 00:06:53.470

Mark Wickersham: and even just basic stuff, like, hey, if you're gonna do a search, why not try, try ChatGPT or another,

43

00:06:53.670 --> 00:07:08.409

Mark Wickersham: AI engine to be able to see what the results are. I know when I go back to Google search results, I'm like, my god, this is such a poor experience. I don't know if, you know, Google search has gotten worse, or AI is just so much better, or a combination of the two, but

44

00:07:08.410 --> 00:07:18.359

Mark Wickersham: How would you recommend, you know, people personally getting an understanding of how to use AI better, whether in their personal life or professionally?

45

00:07:19.030 --> 00:07:34.619

Kelly Wagman: Well, one thing with your comment on Google, I think the AI has also gotten better at citing its sources, and that's something I tell people is, you know, ask it whatever you want, but then tell it to cite its sources, and then click on the sources, and make sure it's actually delivering the kind of valuable information that you're looking for.

46

00:07:34.640 --> 00:07:46.970

Kelly Wagman: So some of the things we do internally to try to get people comfortable with it is I release a prompt every week that's really targeted at something an advisor might find useful.

47

00:07:47.280 --> 00:08:10.899

Kelly Wagman: We also run training programs where we explain, you know, there's the interactive part of something like ChatGPT, but there's also ways to connect it to tools and other information, and I think a lot of it is not intuitive, and so there's actually… there's great YouTube videos, OpenAI releases, resources. It's worth kind of looking at some of these resources, so you are getting past just the initial

48

00:08:11.160 --> 00:08:13.370

Kelly Wagman: You know, most obvious way to use it.

49

00:08:13.400 --> 00:08:34.250

Kelly Wagman: And then we're really encouraging people to share ideas that they come up with. You know, this happens to me, as I'll hear a way somebody's used it, and I'll go, wow, you know, it seems obvious now, but why didn't I think of that? So we have, we have mechanisms for just anybody at Crescent who has a great idea or wants to share a prompt to do that, so that we can all learn from each other.

50

00:08:34.890 --> 00:08:37.830

Mark Wickersham: Love it. Prompt of the week, that's great.

51

00:08:38.289 --> 00:08:49.150

Mark Wickersham: What are some use cases where AI can really deliver value? And talk to me a little bit about how you gotta do that identification and prioritization of that.

52

00:08:49.880 --> 00:09:02.169

Kelly Wagman: Yeah, I think that that really starts with looking both at what are the high-level business goals and the things we're trying to achieve, and then also looking very closely at what are people's workflows like, and

53

00:09:02.170 --> 00:09:12.920

Kelly Wagman: Where might AI be able to unlock some sort of advantage that, you know, lets people refocus on the types of work they want to do and takes over some of the grunt work?

54

00:09:12.990 --> 00:09:24.929

Kelly Wagman: But really, I think… I think it's important to start with those pain points, and then try to figure out what is the best solution. So, you know, there's some generic things, like note-taking has been really valuable,

55

00:09:25.230 --> 00:09:28.689

Kelly Wagman: Document extraction is better than it used to be.

56

00:09:29.290 --> 00:09:46.910

Kelly Wagman: ways that people can ask questions about things like HR documents that it used to be you had to go search for the information in a file somewhere or bug a colleague, and now, you know, you can ask questions in pretty natural language and get a good response. So, these are kind of the first

57

00:09:46.960 --> 00:09:54.240

Kelly Wagman: Sets of things we're looking at, as we explore more closely, you know, what different workflows can really benefit.

58

00:09:57.140 --> 00:10:01.980

Mark Wickersham: Talk to me about note-taking. Has Crescent adopted…

59

00:10:02.130 --> 00:10:08.479

Mark Wickersham: A client note-taking tool, and if so, like, what have the results been with that?

60

00:10:09.210 --> 00:10:19.249

Kelly Wagman: We have been exploring note-taking, and we are piloting, a tool, and we found that it… it really…

61

00:10:19.250 --> 00:10:38.400

Kelly Wagman: Anything that we do, we want to make sure it ultimately is going to benefit the advisor and benefit the client. So, this is a case where I think, in some cases, it lets us understand the full scope of what happened in a meeting, and then the advisor can make sure that they're staying on top of that, addressing it, and

62

00:10:38.600 --> 00:10:41.609

Kelly Wagman: You know, are able to look at it for the next meeting.

63

00:10:41.750 --> 00:10:54.599

Kelly Wagman: And, of course, people were taking notes before, but, I think that this lets the person really engage in the human part of the interaction without having to worry about, you know, did they catch everything?

64

00:10:54.600 --> 00:11:07.610

Kelly Wagman: Of course, we also… this is something we are building a specific policy around, though, because there are times where you may not want to be recording, and you want to be very sensitive to client preferences, so…

65

00:11:09.490 --> 00:11:18.849

Kelly Wagman: I think, yes, it's definitely something we're exploring, but in the way that we're doing everything with AI, we're making sure we're pairing it with training and the appropriate policy support.

66

00:11:20.280 --> 00:11:21.570

Mark Wickersham: Yeah, I think…

67

00:11:22.070 --> 00:11:36.830

Mark Wickersham: to your point, the ability to have an AI note-taking tool allows the advisor to be more present, right? Or you don't need the 3 or 4 people that you might have needed in the meeting before, because you had to have somebody

68

00:11:37.040 --> 00:11:51.760

Mark Wickersham: taking notes, that the advisor can really just focus on the conversation, knowing that the follow-up items will be captured. Also, I think some of it with the integrations into the CRM and the ability that

69

00:11:52.070 --> 00:11:54.629

Mark Wickersham: Have that kind of follow-up items, kind of.

70

00:11:55.000 --> 00:11:59.460

Mark Wickersham: mind from that conversation could be incredibly powerful. I think we're seeing some…

71

00:11:59.750 --> 00:12:10.279

Mark Wickersham: studies out there that have shown it's saving 3 to 4 hours a week of advisors' time, which is a lot, right? You compound that by the number of advisors you have, you compound that by the number of…

72

00:12:10.280 --> 00:12:21.130

Mark Wickersham: That's a pretty significant ROI. Are you seeing any other kind of high-impact ROI use cases that… where AI can play a significant role?

73

00:12:22.590 --> 00:12:29.649

Kelly Wagman: I think that the… the document extraction one is also really valuable. You know, we…

74

00:12:29.700 --> 00:12:44.460

Kelly Wagman: a lot of data gets stored in documents, and being able to pull that data out and reorganize it, and make sure it's stored somewhere that is easily searchable, and, you know, that can be fed to other kinds of AI to be able to be searched,

75

00:12:44.820 --> 00:12:50.309

Kelly Wagman: is another one that, especially generative AI has been making a lot better.

76

00:12:51.770 --> 00:12:59.579

Mark Wickersham: Yeah, there is a lot of… in the wealth management industry, there is a lot of information that's tied up in documents, right?

77

00:12:59.890 --> 00:13:18.069

Mark Wickersham: Trust, state documents, legal documents, alternative investments, like, it's just chock full of… it's a, it's a, you know, it's the land of PDFs there. And AI does seem to be able to be really good at being able to extract that information consistently, accurately.

78

00:13:18.110 --> 00:13:23.979

Mark Wickersham: How does Crescent go about creating an AI roadmap? What is…

79

00:13:24.250 --> 00:13:34.219

Mark Wickersham: What is the mechanism that you guys have put in place to be able to identify the initiatives that you want to take on and be able to prioritize those?

80

00:13:35.360 --> 00:13:45.700

Kelly Wagman: So we're… I think back to the combination of a bottom-up approach and a top-down approach. We want to make sure… excuse me, we're building the roadmap based on

81

00:13:45.970 --> 00:14:00.159

Kelly Wagman: the long-term vision, but then we're also going out to people across the organization. We have, an AI Champions program. We've got representation across the organization to help us surface what are the really high-value use cases.

82

00:14:00.270 --> 00:14:01.510

Kelly Wagman: And…

83

00:14:02.940 --> 00:14:11.739

Kelly Wagman: I would say we're really focused on the advisory space right now, but I think there's also a lot of opportunities across all areas of the business, and I…

84

00:14:11.890 --> 00:14:17.629

Kelly Wagman: Because it changes so quickly, I think the important thing with the roadmap here is it has to be

85

00:14:17.990 --> 00:14:34.490

Kelly Wagman: iterative and frequently updated, so if there are things we could be delivering as we go, I think the best approach is release something that works today, make sure we can iterate on it tomorrow as better things come out, make sure that that's flexible.

86

00:14:34.490 --> 00:14:45.839

Kelly Wagman: And it's, you know, it's a learning process for everyone because, you know, there might be a tool next week, or OpenAI might release something next week that we didn't have last week, and that could unlock a whole new capability.

87

00:14:46.990 --> 00:14:52.530

Mark Wickersham: Yeah, I think that makes a lot of sense. I think an iterative approach is always a good approach. I also think…

88

00:14:52.700 --> 00:15:00.010

Mark Wickersham: you know, a roadmap needs to be a living, breathing document, it's not something that's set in stone, right? New information comes up.

89

00:15:00.210 --> 00:15:11.190

Mark Wickersham: I think there certainly needs to be room in roadmaps for… yeah, not only did we do this initiative, but there's gonna be follow-up on that initiative. We're gonna need to tweak it, we're gonna need to refine it, we're gonna need to really…

90

00:15:11.220 --> 00:15:27.119

Mark Wickersham: kind of optimize it, versus just kind of automatically moving on to the next thing. There needs to be some sort of reserve put in place there. I think your change management approach to the bottom-up and top-down approach makes a lot of sense, whether it's AI or…

91

00:15:27.280 --> 00:15:45.649

Mark Wickersham: Any other type of technology initiative? It seems like the struggle, not so much with the technology, is the change management process, the human process of that. Talk to me a little bit about Crescent's change management process, and how you're getting adoption and buy-in.

92

00:15:46.500 --> 00:15:48.390

Kelly Wagman: I've been really thrilled that

93

00:15:48.670 --> 00:15:55.200

Kelly Wagman: pretty much everyone I've talked to has been very excited about AI, so that's a nice starting point.

94

00:15:55.540 --> 00:16:13.349

Kelly Wagman: And I think people are just really excited to learn, so I think what we're trying to do is meet that demand for learning and curiosity that's coming from folks with things like training, support, outlets for sharing ideas,

95

00:16:13.900 --> 00:16:27.960

Kelly Wagman: It's really a strategy built around saying we've got, you know, great people, and they're already doing great things, and we want to empower those people to help us figure out what we can be doing better, because they're the experts in their workflows.

96

00:16:27.960 --> 00:16:40.530

Kelly Wagman: And so, yeah, things like the AI Champions Program, you know, ways for sharing ideas. When people do something great, we really try to celebrate it and make sure we're giving those people credit. So at our recent.

97

00:16:40.620 --> 00:16:50.079

Kelly Wagman: a recent meeting, we shared a couple of GPTs that folks had made that were really cool, and it showed other people that they could also go make these things, so,

98

00:16:50.760 --> 00:16:56.860

Kelly Wagman: I… it's been really fun, actually, being in an environment where everyone's excited to be building things.

99

00:16:57.440 --> 00:17:01.650

Mark Wickersham: I love it, I love the fact that you're celebrating your successes and highlighting,

100

00:17:01.850 --> 00:17:07.369

Mark Wickersham: some of those kind of team member contributions.

101

00:17:07.700 --> 00:17:17.970

Mark Wickersham: human-centric design. Obviously, when you were talking about your background, it's a key part of it. Maybe folks might not be as familiar with

102

00:17:18.270 --> 00:17:27.470

Mark Wickersham: what is human-centric design? But talk to me about human-centric design, and how does that guide your approach to development and implementation of AI?

103

00:17:28.319 --> 00:17:31.289

Kelly Wagman: So, human-centered design really says.

104

00:17:31.349 --> 00:17:44.289

Kelly Wagman: it sounds kind of obvious when you say it, but you start with the human, and then the technology follows. But I think it's easy to get into the trap of, wow, there's this really cool new technology, and then you've kind of got a hammer, and you go looking for nails.

105

00:17:44.289 --> 00:17:56.359

Kelly Wagman: This is saying it's important to figure out what the nails are first, and then select the appropriate hammer. And I think that that is really important to my philosophy for how I approach AI at Crescent, in terms of thinking about

106

00:17:56.359 --> 00:18:06.869

Kelly Wagman: What is it that people need? What are the workflows that people want help with? What is it that people… what parts of people's jobs do they wish that they could automate? And then say, okay, given that.

107

00:18:06.869 --> 00:18:17.799

Kelly Wagman: what is the best solution? You know, I'm not coming in saying, we must adopt AI because it's cool and new and shiny. It's really trying to change how people work for the better.

108

00:18:18.580 --> 00:18:24.209

Mark Wickersham: Not technology for technology's sake, but, you know, focusing on problems people have.

109

00:18:24.210 --> 00:18:41.400

Kelly Wagman: Exactly. And in research, there's, you know, there's kind of ways that we've developed to go about that, you know, whether that's going out and interviewing people, or doing surveys, or making sure we engage. So I spend a lot of time talking to people, and making sure I do really understand where folks are, and where they're coming from, and what they're looking for.

110

00:18:42.480 --> 00:18:49.710

Mark Wickersham: Love it. Talk to me about AI and data, the relationship between the two.

111

00:18:49.880 --> 00:18:58.730

Mark Wickersham: What do firms need to do from a data governance standpoint to make sure that they're getting good results from their AI initiatives?

112

00:18:59.610 --> 00:19:18.040

Kelly Wagman: So, I think AI and data are kind of like two sides of the same coin. Data is really critical to having AI that works well. In the machine learning era, you know, data was used as a predictive mechanism, and I think there are still cases where machine learning can be valuable. In a generative AI con…

113

00:19:18.190 --> 00:19:30.480

Kelly Wagman: paradigm, I would say. You think of the word context, so you have to provide a generative AI agent the right context to be able to help you answer whatever question you're looking for.

114

00:19:30.480 --> 00:19:48.839

Kelly Wagman: And so, as I think about how we want to organize our data, I think, you know, what are the types of questions people are asking, and how do you make sure you have all that context in one place so that something like an AI assistant can go out and retrieve that information? So, for example, if you're thinking, you know.

115

00:19:49.400 --> 00:20:07.570

Kelly Wagman: provide me, you know, the note-taking was an example of, like, tell me about the last meeting, you know, what did we talk about? You have to make sure that that context is somewhere that's accessible. And I think a strong approach to data governance is really important. We actually have a really fantastic data governance

116

00:20:07.580 --> 00:20:14.750

Kelly Wagman: head of data governance at Crescent, who's working on making sure we have all of that data organized and really clean.

117

00:20:15.640 --> 00:20:30.999

Mark Wickersham: I give credit. Not only do they have your role with AI enablement, but they also have a data governance role, not only a function, but somebody that's really kind of a steward of that, that's overseeing that, which is great.

118

00:20:31.160 --> 00:20:39.629

Mark Wickersham: How do you see AI impacting advisors, and in particular, the advisor-client relationship?

119

00:20:41.270 --> 00:20:52.719

Kelly Wagman: I think so. Something we've touched on is giving advisors time back to do the human part of their job, which is interacting with clients, which is, I think, super high value and something that AI

120

00:20:53.040 --> 00:20:59.879

Kelly Wagman: can't and shouldn't ever try to do. So if we can just drive efficiency, that ends up being great for clients.

121

00:20:59.950 --> 00:21:17.879

Kelly Wagman: I also think it can actually help advisors deliver more tailored and personalized content. So, for example, if you know that a client is really excited about a particular topic or a type of philanthropy or something, you may not have had in the past, you know, half a day to go write a report on that topic.

122

00:21:17.880 --> 00:21:35.470

Kelly Wagman: But AI can actually help go out and do that. You know, of course, you're going to vet it and go over and make sure that the content is good, but you can… I think we'll see advisors able to really better serve clients with that type of personalized information that they might not have time to go out and work on.

123

00:21:36.270 --> 00:21:53.519

Mark Wickersham: I mean, I can be really great at doing deep research, and be able to, like you said, might not have had a day or a day and a half to be able to go deep on a particular subject that maybe a client is interested in, but can certainly get up to speed a lot quicker on that.

124

00:21:53.770 --> 00:22:00.679

Mark Wickersham: How can AI personalize the communication between the advisor and the client?

125

00:22:03.100 --> 00:22:19.649

Kelly Wagman: I think it's about, yeah, providing that kind of tailored content, so whether it's a topic, or something recent that's happened in the news, or that might affect the client,

126

00:22:19.870 --> 00:22:38.060

Kelly Wagman: I think AI is very good right now at summarizing, you know, you could say, here's some papers that I have, or here's a news source that I like, or go out and look at these websites, and it can actually do the work of trying to understand, you know, given the client context, what about this would that person care about, and it can draft

127

00:22:38.060 --> 00:22:46.959

Kelly Wagman: content to share, and again, the advisor will have to go over that, but it will be much faster than having to have them write it from scratch, and I think that kind of…

128

00:22:47.010 --> 00:22:52.880

Kelly Wagman: Personalized updates and support, can be really valuable for the client.

129

00:22:54.060 --> 00:22:57.150

Mark Wickersham: Where do you see AI,

130

00:22:58.020 --> 00:23:04.010

Mark Wickersham: contributing or helping a firm like Crescent grow on the prospecting side?

131

00:23:06.940 --> 00:23:08.789

Kelly Wagman: Excuse me,

132

00:23:09.780 --> 00:23:23.300

Kelly Wagman: I think that… well, first, that kind of concept relates to prospecting in the sense that maybe you have a new client coming in, and you want to make sure that you've thought through all of their… the issues that they care about, and maybe AI can both help

133

00:23:23.320 --> 00:23:36.880

Kelly Wagman: understand externally what resources, if there's some sort of research that has to be done, but also internally. So, Crescent offers, a number of services to clients as part of our family office services arm.

134

00:23:36.880 --> 00:23:47.339

Kelly Wagman: And I think AI can help say, you know, this person has this need, you know, maybe you should reach out to this person, or here's the type of service that Crescent offers.

135

00:23:48.340 --> 00:23:57.370

Kelly Wagman: And it will better inform the way that the advisor can explain, you know, what types of things Crescent can do for that person.

136

00:24:00.320 --> 00:24:09.590

Mark Wickersham: What is the next phase, the next wave of AI innovation? I know we hear a lot about agenic AI.

137

00:24:10.290 --> 00:24:24.120

Mark Wickersham: Can you talk to me a little bit about agenic AI, how you see agenic AI impacting the advisor practice, and also, what do you see in terms of, you know, line of sight, in terms of where AI is going?

138

00:24:25.200 --> 00:24:42.029

Kelly Wagman: So I think of generative AI in general as you essentially have some combination of a large language model, a prompt, and access to some knowledge or data or context. And that, combined, allows you to have some sort of action that is repeated.

139

00:24:42.150 --> 00:24:59.849

Kelly Wagman: Agentic AI, I know it's used in a lot of different ways, but the way I think about it is you have that structure, but it can also now go access tools, it can make decisions, it might actually take some sort of action on behalf of the person who created that, the agent.

140

00:25:00.470 --> 00:25:11.340

Kelly Wagman: And… I think that, as we look at the, the, advisory space going forward.

141

00:25:12.600 --> 00:25:25.800

Kelly Wagman: essentially, these agents will be able to do some of the workflows that currently are… have to be very manual. I really see it as we want to figure out how they can do things that are,

142

00:25:27.060 --> 00:25:41.450

Kelly Wagman: things that people are not excited about doing. You know, it might be things related to, you know, opening accounts, or back to data extraction, or organizing forms or documents,

143

00:25:42.720 --> 00:25:48.150

Kelly Wagman: Just trying to understand how you… you… cre… create…

144

00:25:49.020 --> 00:25:50.229

Kelly Wagman: how do I want to say it?

145

00:25:51.210 --> 00:25:56.569

Kelly Wagman: It's a flexible workflow that can be repeated, that's how I think about it. I don't think…

146

00:25:56.700 --> 00:26:11.050

Kelly Wagman: I think there's a lot of ways to make it sound really hypey, but I think it's a… at its core, it's a flexible workflow that can be repeated, and that does need some nurturing over time to make sure it's performing well, but that will be able to support advisors.

147

00:26:12.160 --> 00:26:23.950

Mark Wickersham: Yeah, it's a workflow with a little bit of intelligence, right? I think it's gonna be definitely… I think it already is starting to impact, low-value, first-mile type of activities that…

148

00:26:24.160 --> 00:26:40.229

Mark Wickersham: Allows to be able to free up staff to be able to work on more higher value type of activities. I think you see a lot of it, a lot of room in that area around data aggregation, back office reconciliation, as you mentioned, new account setup.

149

00:26:40.360 --> 00:26:45.880

Mark Wickersham: There's a whole host of things that I think that can kind of really help

150

00:26:46.040 --> 00:27:03.460

Mark Wickersham: automate and free up staff to work on a higher value type of activities. I think… I did hear somebody was saying that the people that manage… the future of management, that the people are gonna not only manage people, but they're gonna have to be used to being, managing,

151

00:27:03.650 --> 00:27:06.540

Mark Wickersham: AI agents, as well, as part of their job.

152

00:27:06.540 --> 00:27:06.940

Kelly Wagman: Yeah.

153

00:27:06.940 --> 00:27:08.050

Mark Wickersham: description.

154

00:27:08.050 --> 00:27:16.650

Kelly Wagman: I think the way to think about that is just that these workflows, because they're using generative AI, are non-deterministic, and so…

155

00:27:16.650 --> 00:27:32.109

Kelly Wagman: the benefit that they're very flexible also means that they can flexibly, you know, start to go in the wrong direction. So, I think by managing them, you know, it means we'll need things like evaluations, we'll need audits, there's a whole host of ways that we'll need to put

156

00:27:32.210 --> 00:27:35.209

Kelly Wagman: Come out with to understand how to manage them.

157

00:27:36.320 --> 00:27:53.339

Mark Wickersham: No, you really want the same result every time, sometimes, for certain functions, right? And some… that's… we're not always there, and I think it's important, to your point, to have that human in the loop to ensure that you have that auditability, that kind of quality check.

158

00:27:53.580 --> 00:28:03.770

Mark Wickersham: Where do you see… how do you see AI impacting wealth management on a larger arc?

159

00:28:04.250 --> 00:28:10.330

Mark Wickersham: 2, 3, I mean, it's moving so quickly, it's tough to even say what's gonna happen two years from now, but…

160

00:28:10.580 --> 00:28:17.130

Mark Wickersham: You know, taking a look at the longer arc of the wealth management industry, how do you see AI impacting it?

161

00:28:17.540 --> 00:28:34.189

Kelly Wagman: I think given that there's new releases every week, you know, I don't want to pretend that I know what's going to happen in 2 years. I don't think anybody knows what's going to happen in 2 years, let alone 6 months. So, I think the best thing we can do is make sure that we are…

162

00:28:34.190 --> 00:28:45.009

Kelly Wagman: Kind of prepared for change, and that our organization is able to quickly, you know, come up to speed and adapt as things change.

163

00:28:45.030 --> 00:28:54.279

Kelly Wagman: And start building, kind of, those core competencies around how to work in this type of new interaction paradigm.

164

00:28:55.450 --> 00:29:08.909

Kelly Wagman: So, I don't… I don't wanna… I don't wanna be in the business of forecasting something so rapidly changing in two years, but I do… I do think Ethan Mollack is right on this is the worst AI you will ever use, so that's… that's something to keep in mind.

165

00:29:09.870 --> 00:29:22.979

Mark Wickersham: Yeah, it's interesting to see where it's gonna go. I mean, we, you know, those big announcements this week with Sora. They have the AI actress that, that, that came out that has…

166

00:29:23.220 --> 00:29:26.960

Mark Wickersham: Kind of people up in arms a little bit.

167

00:29:27.470 --> 00:29:34.989

Mark Wickersham: It will be interesting to see how it impacts the industry. I think, ultimately, at the end of the day, I think that human…

168

00:29:35.190 --> 00:29:40.640

Mark Wickersham: Connection becomes even more important, that advisor-to-client relationship.

169

00:29:40.970 --> 00:29:49.159

Mark Wickersham: should be enhanced, will be enhanced, but it is, you know, that's the focus, you know, that's the thing that AI can't replace, right, is that human…

170

00:29:49.160 --> 00:29:49.560

Kelly Wagman: Oh, yeah.

171

00:29:49.560 --> 00:29:51.499

Mark Wickersham: That element, that human touch, right?

172

00:29:51.500 --> 00:30:04.239

Kelly Wagman: And there's… I just see no world where we're interested in replacing advisors. I mean, they're fantastic at their jobs, and we need… I think the core of the advisory relationship is that human interaction, so…

173

00:30:04.550 --> 00:30:13.420

Kelly Wagman: Can we be more personalized, more efficient? Of course, and we will work towards that, but it's all in service of helping them better serve clients.

174

00:30:14.690 --> 00:30:23.490

Mark Wickersham: Right, yeah, ultimately it's a people business, that empathy aspect is extremely important. I think you even see the wealth management industry kind of

175

00:30:23.710 --> 00:30:35.309

Mark Wickersham: Shifting, where it's becoming much more of a… like a hu… like a lifestyle coach, or, you know, there's more of a coaching aspect for clients versus just focused more on the…

176

00:30:35.310 --> 00:30:52.800

Mark Wickersham: in the past, it was more purely technical, like, hey, here's what we're doing with investment performance, and your risk, and shop ratios, and it allows that, I think the advisor really focused on, what are your goals, what is the meaning of the good life, you know, those type of broader questions that AI is not going to be able to handle.

177

00:30:53.010 --> 00:31:00.590

Mark Wickersham: Ai in your daily life, in your personal life, how do you use AI?

178

00:31:01.900 --> 00:31:14.819

Kelly Wagman: I am always really impressed with things that let me go out and interact in the real world instead of being behind a computer, so I love being able to take photos of things and ask questions about things in the world, whether that's

179

00:31:14.920 --> 00:31:36.470

Kelly Wagman: tell me about the flavor profile of these oysters on this menu, you know, while I'm sitting in the restaurant, or… I asked the other day, you know, it looks like this tree was hit by lightning. Was it hit by lightning, or is there some sort of disease in this part of the world? And it gave me, actually, a really good answer, or I was in Japan earlier in the year, and I could have it translate entire menus for me.

180

00:31:36.470 --> 00:31:37.280

Mark Wickersham: Nice.

181

00:31:37.690 --> 00:31:40.339

Kelly Wagman: So I… I'm… I'm…

182

00:31:41.480 --> 00:32:03.600

Kelly Wagman: I'm always optimistic when I feel like technology can be kind of woven into daily life. I also like using the voice feature, so, you know, you can actually talk to things like ChatGPT, and it will talk back to you, and that's another interaction that I will expect to get better in the coming years, because it's much more natural for a lot of things to just speak instead of having to type everything.

183

00:32:04.380 --> 00:32:06.640

Mark Wickersham: I found it particularly useful with,

184

00:32:07.130 --> 00:32:13.040

Mark Wickersham: Getting pronunciations of people's names correct sometimes, especially if you're dealing with,

185

00:32:13.510 --> 00:32:30.609

Mark Wickersham: you know, people from different countries, names that you're not normally familiar with, you know, you want to make sure that you get somebody's name right, and it can be really good at that. If you're traveling, you know, hey, can I have a build, you know, a travel itinerary for me? Obviously, like.

186

00:32:30.610 --> 00:32:43.260

Mark Wickersham: the quality of your prompt is… will directly impact your itinerary. You need to tell it what you like and what you want, and I'll still probably recommend a restaurant or two that's been closed for two years, but .

187

00:32:43.260 --> 00:32:46.260

Kelly Wagman: Yes, I've had that happen. That's pretty good.

188

00:32:46.470 --> 00:32:51.210

Mark Wickersham: It can be good for that. Like, you know, for the podcast, like, it helps me…

189

00:32:51.490 --> 00:33:08.240

Mark Wickersham: you know, if I feed it a transcript, to be, you know, to come up with, like, a 60-second kind of summary, you know, give me two paragraphs on… on what's gonna be a 40-minute conversation, like, it's hard to do, right? To be able to do that first draft, it takes a long time, and you might miss something.

190

00:33:08.240 --> 00:33:09.160

Kelly Wagman: Yes.

191

00:33:09.430 --> 00:33:25.649

Mark Wickersham: So I think it could be great for that as well. Images, although, you know, I've added… I've added some of my profiles, I feel like it adds 10 pounds onto me when I ask it to give me a different headshot, or turn myself into a painting, or something like that.

192

00:33:25.650 --> 00:33:34.249

Kelly Wagman: Yeah, I tried to get it to give me a different haircut to see what it would look like, and that did not go well, but again, you know, in a week or two, maybe it'll be better.

193

00:33:34.500 --> 00:33:35.280

Mark Wickersham: There you go.

194

00:33:35.800 --> 00:33:51.330

Mark Wickersham: All right, I like… Kelly, I like to end these podcasts on a personal note with 3 questions that have nothing to do with, wealth tech, or in this case, AI. You're in Chicago. What do you love about Chicago?

195

00:33:52.510 --> 00:34:02.179

Kelly Wagman: One is that I really love the lake. I feel like we live… it looks like an ocean from the vantage point of standing on the shore, which is really wonderful.

196

00:34:02.180 --> 00:34:13.829

Kelly Wagman: And the amount of just art, culture, food that's in the city is incredible. I went to a very experimental play last week, which was just a cool thing that I could do without

197

00:34:13.830 --> 00:34:16.710

Kelly Wagman: Having to leave the city, and

198

00:34:16.830 --> 00:34:19.559

Kelly Wagman: There's just a lot to explore, which is fun.

199

00:34:20.389 --> 00:34:36.389

Mark Wickersham: It is a great city. It is kind of unique how you have the, you know, these huge skyscrapers, and then it's, you know, you've got to have your back to what looks like the ocean, right, because it's so big. There are some great, great spots there, and second city, that's always up.

200

00:34:36.389 --> 00:34:37.589

Kelly Wagman: Yeah, so fun.

201

00:34:38.570 --> 00:34:44.400

Mark Wickersham: What's one spot in Chicago that people might not know about from an outsider?

202

00:34:45.300 --> 00:34:53.489

Kelly Wagman: So, there's a place that I really like, I guess it's technically Chicagoland, not in the city of Chicago, called the Morton Arboretum.

203

00:34:53.520 --> 00:35:05.059

Kelly Wagman: Which is a huge park that has tons of different kinds of trees, and they have a rotating, sculpture exhibit, and they get these enormous sculptures. Right now, there's one that's…

204

00:35:05.060 --> 00:35:17.589

Kelly Wagman: maybe two stories tall of a giant snail and rainbow colors, which is just incredible. So, if I want to get out and do a big walk or something, I go… it's about a, I don't know, 30-minute drive outside the city.

205

00:35:18.290 --> 00:35:20.989

Mark Wickersham: Okay, well, I'm gonna have to add that onto my list.

206

00:35:21.110 --> 00:35:26.080

Mark Wickersham: Is there a book that you've been reading lately that you'd recommend to the audience?

207

00:35:27.930 --> 00:35:38.430

Kelly Wagman: I think given that I've mentioned Ethan Mollick, the book that I typically recommend to people interested in understanding generative AI is his book called Co-Intelligence.

208

00:35:38.520 --> 00:35:40.549

Kelly Wagman: Because I think it provides…

209

00:35:40.620 --> 00:35:52.369

Kelly Wagman: ways of thinking about the type of interaction and the types of things you can do with generative AI. I think often when we approach a new technology.

210

00:35:52.370 --> 00:36:05.360

Kelly Wagman: you either come with some mental model of how it works, which informs how you interact with it, or you don't have a mental model and you just have no idea what to do with it. And I think he has a lot of really concrete, helpful,

211

00:36:05.590 --> 00:36:10.710

Kelly Wagman: Ways of thinking about the mental model of the interaction, that is a really good place to start.

212

00:36:11.780 --> 00:36:13.950

Mark Wickersham: Alright, that's a good one, I'm gonna look that one up.

213

00:36:14.160 --> 00:36:18.869

Mark Wickersham: This has been great, Kelly. I really appreciate your time, and thank you so much for sharing.

214

00:36:19.030 --> 00:36:20.880

Kelly Wagman: Thank you, thanks for having me.