Kitces & Carl Ep 154: The Limits Where Technology Can’t Scale Financial Planning Advice

***Editor's Note: Can't get enough of Kitces & Carl? Neither can we, which is why we've released it as a podcast as well! Check it out on all the usual podcast platforms, including Apple Podcasts (iTunes), Spotify, and Stitcher.

Show Notes

Carl: Michael Kitces, how are you?

Michael: I'm doing well, David Carl Richards the III.

Carl: That's right. That's right. That is right. I would like more people to address me this way. David Carl Richards the III.

Michael: Can we just say DCR3?

Carl: Yeah, that's okay too. That's okay too.

Michael: It's got kind of like R2D2 vibe to it. DC-R3. DC-R3.

Carl: Yeah, yeah. You had to take it that way, didn't you? You just had to take it...

Michael: I...what you're talking about, of course I had to take it that way.

How AI Can Be Used To Expedite An Advisor's Work [00:41]

Carl: Well, listen, I'm hesitant to make stretch intros, but that actually leads us to today's topic. The connection to robots. I've been using ChatGPT...

Michael: Oh, I've heard of that.

Carl: ...yeah, quite a bit lately.

Michael: That's that artificial intelligence thing.

Carl: I've been super surprised at how helpful it can be, especially with, as almost a writing partner and editor, etc.

Michael: I was going to ask how you're using it. So editor...

Carl: Yeah, coming up with, for instance...

Michael: ...like, here's a thing, tell me how to make it better?

Carl: Yeah. For instance, maybe I'll put it in a draft and say, "Hey, I'm not sure I nailed this word here, or "This sentence isn't..." or "I think this..." I'm almost always looking for what we internally call it the zinger. All of The New York Times columns, we were like, what's the zinger? In the normal world, you'd call that the lead in the paper world, in the newspaper world, they'd say, don't bury the lead. I was always after what's the one idea. That, to me, is the crunchy bit. And so, I could put that in Chat and say, "Hey, I think the zinger is this, but I'm worried about this word. What other words do you think I could use?" Things like that, very specific ideas.

Michael: It's like a brainstorming buddy for you...

Carl: It's pretty amazing.

Michael: ...that's better than a thesaurus.

Carl: Pretty shocking actually. I've been shocked at how good it is. So today I was like, "Oh, I wonder what it would say about what we should talk about." So I told it to...

Michael: For Kitces and Carl?

Carl: Yeah, I told it. I said, "Hey..." I think the exact prompt was, "I have this podcast called 'Kitces and Carl.' Go listen to it." I knew it was just going to go read the transcripts essentially.

Michael: You feed it episodes, you're just abstractly saying, "Go listen to it."

Carl: No, this time I didn't. What I've built in other cases is I've fed a bunch of my own work into it and said, "I want you to analyze this and call it Carl's writing style." And then, I've given it a bunch of prompts to learn Carl's writing style. In this case, I didn't even do that. I just said, "Go." It gave a bunch of really good ideas about topics, but I want to just focus on the first one because I thought the first one is...

Michael: We've got a future lineup as well. So what was the first one that ChatGPT recommended.

Carl: The first one is really funny. I said, "I need some topics for today's episode." It gave me a list of some really good topics, but the very first one was this. I'm just reading verbatim. It says, "For today's episode, how about exploring the topic of the future of financial advice, how AI and technology will shape financial planning." How about you talk about me?

Michael: Wait. So ChatGPT's suggestion of what we should talk about is that we should talk about ChatGPT.

Carl: That's exactly right.

Michael: It volunteered itself into the conversation.

Carl: I sort of feel like there was somebody in the background, "Hey, me, me," raising their hand, like me. But this is interesting. "Given your expertise, it would be fascinating to discuss how new technology can enhance accessibility, reduce costs and improve client engagement. This aligns with your 2 –" Kitces and Carl "– belief in technology's potential to make financial planning accessible to more people, especially those with lower asset minimums. You could also" – and I love this – "you could also delve into the ethical considerations and limits of AI in offering personalized advice." So nice of it to point out its own shortcomings after volunteering to be the center of the conversation.

Michael: Yeah, after making itself the center of attention. That was very balanced of it.

Carl: Yes, it was. What do you think of that? First of all, the fact that it volunteered itself, I'd like to talk about that a bit. And then, let's talk a little bit about tech and maybe dive into a little bit of AI in terms of the future of financial advice.

Michael: Well, look. First and foremost, I'd say, so there's been a lot of discussion out there just of AI, uses of AI, of ChatGPT and its ilk, these large-language model things, what do you do with this version of AI that we can interface with. And I will say, I've used as well and one of the primary places I use it is as a brainstorming buddy. Very, very similar to what you said, right, you were working on what's the zinger? What's another way to write this idea or make this point? Coming up with ideas, coming up with titles, coming up with things to talk about or write about or covering conversations with clients. I think it's really powerful from that end as well as taking in that information and figuring out what to do with it. More and more of us are using it as a meeting summarizer with Jump, Zocks, FinMate, Fathom, Zoom AI, right? There's industry versions and generic versions.

There was all this discussion of, "Oh my gosh, you can talk to ChatGPT like a person. So how long is it going to take for people to ask it financial advice and then get financial answers?" And I think what a lot of us discovered really quickly is the AI is sort of mostly right, but not really, totally, completely always right, which is not good when you're using it to get advice to make financial decisions. It's not a good advice partner, but it's pretty good at supporting some aspects of creativity around writing and ideas or synthesizing when a lot of words and things have been said already.

And so, I find it to continue to be powerful there. Super excited for all the ways that we continue to use it for meeting notes, meeting summaries, which then turns into post-meeting emails and compliance documentation and maybe even kicking off some workflows, and then, synthesize every meeting I've had with the client in the past and give me some reminders about what's going on their lives when I'm going into the next meeting. Right? Last meeting summary is next meeting's meeting prep. So I do think there's some really cool use cases there. And it's not trivial. Our research has shown for years that across the whole industry, the average advisor, we spend 1 to 1-and-a-half hours of prep and follow up for every 1 hour we're actually in a meeting.

Carl: That's amazing.

Michael: And so, if you even just make a dent in that, I don't know that we would...I'm certainly not one that would totally automate a post-meeting email from AI. But if you want to capture the meeting, synthesize it and make the draft, and I take 5 minutes to draft the email instead of 20 or 30 minutes to write it out, that is an immense amount of time savings. And I do think...

Carl: Let me just...

Michael: AI is much more expedite than automate.

Why Advisory Firms Haven't Become More Profitable Despite Expanding Technologies [07:56]

Carl: Let me just chime in on that note. That is one area where I am just, with a very small investment of time in sort of prompt engineering, just learning how to ask it the right things. Hey, I need to write an email that does this, this, this and this. Can you please draft it? Incredibly helpful.

Michael: Yep. So I love it from that end. Now the interesting part of this, and I guess ChatGPT's response to you about topic kind of alluded to it, right? Then this gets into all sorts of domains around reducing costs, boosting productivity, enhancing accessibility, right? There's a lot of downstream effects that start coming when you start saving a whole bunch of time on things that previously took a lot of time as an advisor. And I will say this is an area where I'm a little bit more mixed or at least I think how this plays out for advisory firms is a lot more nuanced than people realize, than certainly the industry and AI adherence give it credit for. So I look at this relative to the past couple of decades. Right? If you go back to the 1990s when the RIA model was first emerging, the assets under management, recurring revenue, served clients for ongoing fees was emerging. And Mark Tibergien and Moss Adams started doing the first industry benchmarking studies on it. And the average advisory firm was known as the 40/30/30 rule, 40% of revenue direct expenses, 30% to overhead, 30% profit margin was how advisory firms ran. Since he published those studies, we had the internet, the smartphone, robo-algorithm business process automation, and at least wave one of AI. And then, you look at industry benchmarking studies today and the average profit margin isn't even 30% now, it's like 25%. So I look at that I'm like, huh. We had a billion-X increase in computing power and created the internet, the smartphone and robo-automation, and we're less profitable now on average, slightly less profitable now than we were 30 years ago. Oh, and the median advisory fee was 1% then, it's still 1%. So, fees didn't move. Technology improved a bajillion percent and we did not have some giant productivity profit boom. And so, to me it's interesting to ask what happened.

Carl: Yeah, that's exactly what I was going to ask you.

Michael: So when I look at what happened, we took all the time that we save on our clients, and then we turned it around and did more for them. I go back then, plans were just you print the financial planning software output. There's no customization. There's no additional detail. We're not building 1-page plans or summaries. You're just printing out a 150-page thing which was kind of both comically long and greatly expedited because you just printed the software output and brought it in. We spent a huge amount of time doing investment analysis to manage client portfolios. But we could only meet with them once or twice a year because it was just time consuming literally to do the investment research around the practice. We weren't that deep in the financial planning really beyond the plan software output. We didn't do tax preparation. We didn't do estate documents that so many firms now are starting to add in. A lot of us had one-meeting closes. Now the average firm is a three-meeting financial planning process and some of them do four, five, six meetings. So we took all the free time that we freed up from the tech and we just reinvested it back into doing even more for clients and apparently doing even more, more for clients because now our profitability actually went down a little bit because we're doing so, so, so much more for clients that even when you take the tech efficiencies, we apparently, maybe as an industry, invest a little bit more into clients than we've gotten in tech savings. And so, the reason why that's such a big deal to me, when I then look with the lens of all we're talking about AI right now, let's say, if the past is any predictor of the future...haha, I'm not talking about investment returns.

Carl: Right. Sounded funny.

Michael: The past is any predictor of the future for human behavior, which it tends to be really good at, we don't get this giant productivity boom when we get AI. We free up a whole bunch of time so we can do more for the clients that we've already got, so we can hold on to them and keep serving them in a competitive world. Because frankly if we don't, we say, "Look at how much time I'm saving. I really have to work for...I can work 1 hour a year for my clients. Look at all these fees I make." Some other firm up the street says, "Well, I'm going to take all that free time and do awesome new things for my clients. I'm going to take all your clients." It's like it's the Jeff Bezos, "Your margins are my opportunity." We can't sit back...one advisor, one at a time, maybe find some cool tech efficiencies for ourselves. But as a profession, we can't rest on our technology-efficiency laurels. Because someone else comes along and does more for the clients than we do if we sit on that, if you don't reinvest the savings back into doing more for clients, that you've got the capacity to do by virtue of the technology automation, the speed and the efficiency and all the rest. And so, the bar of what it takes to serve clients well just keeps going up. You get service inflation, to use the term du jour. And we don't end out more productive. To me, that's the fascinating thing. I think clients are now way better served. I would argue our value prop now is so much more than what we did 20 or 30 years ago. So I think clients were a huge beneficiary in a lift of what we do for them. But we got remarkably little improvement in profitability and productivity as advisors from this amazing technology boom of the past 30 years.

Carl: Yeah, it's so interesting. I don't want to take us down a different path. So I just want to mention this and maybe we can talk about it in the future. I still always think about...I get what you're saying about the whole industry. But I still love all my favorite people out there that are running, and proud of it, proud of it, running lifestyle firms and especially the solo lifestyle firms. And I still remember having, and I've had recent conversations with people, "I don't get what everybody else is doing. What are they doing with all their time?" And they always want to do this whispery because they don't want anybody to know because they're scared people will see them as slackers. But like, "I work 15 hours a week. I make more money than I ever dreamed. My clients are getting better service than they'd get anywhere. What are the rest of you doing with all your time?" And let's not address that because that's a whole episode unto itself. But it's so interesting to me. So I always wonder, taking...because I think we've seen this across, not just our industry, but across...and I'm not in any position to comment on the entire economy but it just seems like what we're doing is taking these gains and running faster. Life hasn't gotten slower. We're taking these gains and running faster. Pieces of...

Michael: Societal malaise. Again, we've had all this amazing technology efficiency that allows to get worked on better and more efficiently than ever.

Carl: And you're busier than ever.

Michael: Yeah, we're busier than ever and having more trouble and never disconnecting.

Carl: More frantic and all those things.

The Upper Limit Of Automation In The Financial Planning Process [16:23]

Carl: So that's interesting stuff to think about. But here's one thing I want to make sure we cover here is this last line that Chat said we should talk about. We should also...You know what? Sorry, Chat did say you could. I was about to get a little judgy with Chat. You should? Who are you to tell me that we should. It says, "You could also delve into the ethical considerations," but I really want to talk about the second part, "and the limits of AI in offering personal advice." And I think we could just swap out AI here and say tech. What is the limit? I've had some interesting conversations lately about what's the upper limit of, let's assume you had all the resources. Let's say you're a giant firm that has all the resources that you could build tech. You're kind of starting from scratch. You don't have, let's pretend like you don't have legacy issues. And you care deeply about the clients. So we make all those assumptions. What's the upper limit and what is the limiting factor? What is it that we bump up against that we can't scale past 100, 150, 200. What is that thing?

Michael: I'd frame this in two ways. I map this to other professions. So in the simplistic view of financial advice, clients come with problems and we analyze and give them answers. Which means if you build an analytical tech tool that's good enough, you just input the data into the software and it tells you the same answers.

Carl: I have a sketch about this.

Michael: And so, any problem for which the only question is literally what is the answer after a path of analysis, becomes the domain of technology, for which the only limit in that context is problems for which there really is no right answer because there's true complexity in the client situation, like the technical definition of complexity.

So for instance, what's the most federally estate tax-efficient way to transfer money to my grandchildren. The answer is a generation-skipping trust and let's talk about what that is and how that's structured. What's the best way to divide my assets amongst the grandchildren, when two of them really want to take over the family business and the third is estranged due to a drug problem? There's not a calculator answer for this. There's no right answer to how you do this. At best, I can combine some tools and techniques, asset protection trusts and family trusts and generation-skipping trusts, to help navigate some of the rules around how do I not have the business...how do I get the business down without a huge estate tax liquidity issue, and how do I protect the assets from the child that's got some drug problems so that the dollars are not lost in unfortunate ways. But how much should I leave each of them in this structure, there's no right answer. This is something that the client's got to figure out for themselves. I can show them some tools and techniques to facilitate the process. I can perhaps be a helpful thinking partner to give them new views and new perspectives. But there's not a right answer. And so, on the one end to me, there's a complexity threshold to tech that not all financial advice problems have an answer that you can calculate.

Carl: Yeah. Yeah. I have a joke that I've been to mortgage payoff parties. I've never been to an efficient portfolio party. And we all know that almost every spreadsheet you could ever build, obviously depending on interest rates, but almost every spreadsheet you could ever build in the last 10 years tells you that paying off your mortgage wasn't necessarily the most efficient tool. But I...

Michael: Don't you know what your portfolio earns compared to the rate on your mortgage?

Carl: Totally. But I've been to those parties.

Michael: Your mortgage is tax-deductible. Dude. I would take that off.

Carl: Yeah. I've been to those parties. And some people say that I hang out with the wrong crew because they've been to lots of efficient portfolio parties. But I think that's wrong. I think they're making that up. But yeah, I think there's the complexity. I just think, the way I would phrase that is human, right? We bump up against...and I think there's also humanity... and then there's this other piece. My editor at "The New York Times," Ron Lieber, his favorite sketch of mine was not really a sketch. It was just the word, money with the equal symbol, feelings. So money equals feelings. If money actually equals feelings in lots of our minds than being heard, listen to, asking clarifying questions, empathy, those start to get on the outer edge of limits of capacity too.

Michael: Absolutely. To me, that's the second domain of it, right. Part of it is just some questions are not just complicated but actually complex. There is no calculable right answer. At best, you can help someone think through the issues and choices and they're going to make their own decision. But the second is, I'm not even sure what the right word is. It's the feelings, it's the empathy. I like to just call it, it's the relationship. It's the relationship with another human being at a level of trust that I can have a discussion about how you're thinking about the fact that one of your grandchildren is estranged, and you're thinking about disinheriting them and you're trying to figure out what to do with that. That's a weighty conversation. For a lot of families, it's not even a thing you talk about in the family about the child or grandchild who's estranged. So we're going to have a conversation about all your family drama and all your money, and we're going to put it together, just literally a very, very challenging conversation for a lot of people to have with anyone, anywhere, ever. And if I'm supposed to have that conversation with a client, then there has to be a certain level, yes, of empathy, of emotional IQ that you bring to the conversation. But I just think of it in the pure sense, we have to have a relationship that's deep enough and trusting enough that you're willing to have that conversation with me, or I can have that conversation with you and we can navigate that discussion together.

So when I think about how far does tech go, the limit on tech is the threshold of time that it takes for 2 human beings to still have a relationship of trust and rapport. And I don't know quite where that threshold is. It's not a small number of hours of investing into the relationship every year ongoing, to make that happen. It's a couple of conversations a year, an hour or 2 at a time, to really be able to unpack some of the messiest issues. And then, per the earlier discussion, it takes me an hour of prep and follow-up for every single hour meeting I have with clients. And so, you can very quickly get to a realm where I need four, six, eight, ten hours per year, per client of doing things that are communication, interaction, relationship, meeting, email, phone call, text, whatever it is, that invest into the relationship and support the trust and support the connection. And when you just start doing the math on how many working hours there are in a year, that you can even be client-facing, because we can't literally be...client meeting 8 hours a day would be freaking exhausting. To me, you very quickly come down to a point where if the ultimate threshold is the time it takes to have a relationship, you just reverse engineer the math and your absolute cap is somewhere between 100 and 200 clients.

Carl: Yeah. Super interesting.

Michael: And that's actually higher than what most advisory firms have as they scale up.

How (Delegating To) AI Can Make More Room For The Human [25:51]

Carl: Yeah. So the way I think about tech that gets me so excited, like ChatGPT, my experience even today using it, gets me so excited because it gets me super excited to try and continue to ring out from my life all the things that aren't uniquely human so that I'm left with more capacity, time, even the ability to rest, so that I can do the things that don't scale. I'm going to do the things that don't scale that only I uniquely can do. And for me, I've gotten pretty clear about what those things are. Anything that's not that, I'm trying to figure out how to either delete, delegate, or really delete or delegate, because I don't want to do it anymore. So I think all these tools are giving us ways to be like, "Wow, could I be faster with that so that I can leave the time to do only the things that only I can uniquely do."

Michael: One, look, I look from that frame, again, looking at the evolution of technology. So I remember, so the second firm I worked at 22 years ago now, was an independent broker-dealer firm. 3 advisors. About $1.3 million of revenue, which back then was a really good-size shop. And I believe there were 13 of us in that office. Basically 3 advisors, 10 support staff, 3 client service administrators, 2 other support, 1 person in the front office. 1 and a half people supporting marketing. All the way around. We had a woman whose name was Betty, bless her soul, adored Betty. The bulk of Betty's job was open all the envelopes that came in every day, file each client statements in their folder, because we literally had a room that was just the folders, file folders for all the clients, paper files, check all the envelopes for checks, because you can't sit on a check overnight, and then greet all the clients as they came in for their meetings. Because everything was in-person.

Betty's job wouldn't exist today, certainly not as a full-time job, supporting three advisors. In fact, if I look at a typical advisory firm today that had 3 advisors and $1.3 million of revenue, they would probably have, at most, 2 support staff, maybe 1. And this firm, it was ten of us. And so, to me that's the needle that tech has moved. It used to be three staff to support one advisor. And now three advisors can be supported by one staff member. It's like a nine-times improvement in the ratio of staff support that it takes to run an advisory firm. And I think that's all tech. All the things Betty did have been automated out of her job or out of existence. Maybe there's still a few client meetings where someone has to go up to greet the clients, but you don't need a full-time greeter.

And so, there to me, the tech has shown up. But the remarkable piece of that is, and the advisors back then spent the bulk of their time either in client meetings or doing prep and follow-up and a little bit of managing their team. And now advisors spend their time in client meetings or doing prep and follow-up or managing their team. The advisor role in time basically didn't change. A lot around us did. But to me, that fundamental constraint of, look, where do we tend to show up at the end of the day, in places of complexity that need a relationship to navigate. And that hasn't changed. And tech doesn't really move that part of the needle.

Carl: And I think that past is a good prologue for the future in that way. And so, to me, I've been asked a bunch, "How do you feel about tech? I'm so scared." To me, all I think is the solution to feeling like tech may be a threat is to just simply lean into being more human, right. Use the tech as a tool to allow yourself to get more clear about what the value was in the first place. And it turns out the value that you really provide, the technical stuff is incredibly important. Don't get me wrong. But the value you really provide is your own humanity, and the empathy and the ability to think through complex problems, to ask the right prompt of the Excel spreadsheet, right. All of that sort of stuff is still incredibly valuable. So I think, if maybe we could just wrap up on this idea of, yeah, use tech so you can be more human.

Michael: I like that framing. Use tech so you can be more human.

Carl: Cheers, Michael. Super fun.

Michael: Thank you, Carl.

Carl: Thanks to ChatGPT too, just for this...

Michael: Yes, thank you, ChatGPT...

Carl: ...thank you. We're glad you were here.

Michael: ...for the wonderful topic about yourself. Thank you.

Carl: Cheers. See ya.