Industry consultant, Suzanne Siracuse sits down with Michael Kitces, Chief Financial Planning Nerd at Kitces.com, to discuss how AI will impact the future of Financial Advice and Wealth Management. He will share his views on how advisors and firms should be thinking about AI in its current state, the tools that advisors can plug into right away and the where he thinks AI has the most opportunity and will be the most disruptive to an advisor's practice. Michael is never shy, so you won't want to miss this.
Transcription:
Suzanne Siracuse (00:00:12):
Okay, great. Well, thank you Brian. I really want to give kudos to Financial Planning Magazine and Arizent at their parent company for seeing an opportunity, a niche of covering a conference, creating a conference that is completely focused on AI because there's, as you could see, there's just so much engagement here. There's so much for everyone to learn, and I really think that it was a smart decision and thank you for the opportunity to be up here and to talk to one of the leading voices in our industry. I know Michael is very opinionated. You've got lots of thoughts around this, so I'm going to get right into the questions. So AI has really swept the tech industry and quite frankly, our world, but except for the people in this room, it's not been as embraced in the wealth management industry. We're a little slow to adapt as well as being a little cautious about AI and its applications. Why do you think that is?
Michael Kitces (00:01:22):
So I think this kind of by analogy, so imagine for a moment you downtown in major metropolitan area, New York City, Los Angeles, the like, you got to get across town for a meeting. So the cool thing about the modern technology era is I can pull out my smartphone, hit a button in a few minutes, like a car literally just pulls up in front of me to take me wherever I want to go. So you pull out your phone, you hit the button, the car pulls up, you walk over to the car, you pull the back door open. As we do, you kind of look into the front seat because you got to make sure that the driver up there matches the driver on your phone that you're getting in the right vehicle. There's no driver in the front seat. It's a driverless car.
(00:02:04):
Do you still get in the car? Now most of this people in the room are going to be like, yeah, driverless cars, it's the future. I can't wait to not have to drive around anymore. This is awesome. General consumer sentiment. The majority of Americans would not get in the car and the number of people who don't trust it is actually on the rise as just we see more nuances and details about how driverless cars show up. Just there's been some complications from the ideal that we thought it was going to be a few years ago and consumer sentiment and willingness to get the driver. This car is actually declining right now and I think it's an interesting map to a broader range of the challenges around ai and I'll say how AI shows up in high stakes situations. If you think about that for a moment, I think like most of us in the room grew up in the stranger danger era. So to be clear, I'll get in the car with a stranger, just not a computer is what most of us will say. I'll get in the car with the emotional, irrational, neurotic, crazy fellow human being that may or may not be insane and pray that they did enough vetting on this driver, but I won't get in the car with the computer.
(00:03:24):
I think it comes from what frankly is this unfair bias that we put against technology. We all understand humans are not perfect, but we expect the technology to be right. Humans can have accidents because humans have accidents, but it's completely unacceptable if a driverless car ever has an accident ever anywhere and then we all just play the game of when it has to do the trolley problem, who's it going to kill? And I call this the AI trust gap that we actually have this strange standard, I don't know where it comes from in our brains. Someone better about psychology research maybe can help me figure it out that the standard we set for technology is actually higher than the standard that we set for a fellow human in anything that has relatively high stakes. If it's a simple problem, just be like, fine, I'll let the technology do it. If it doesn't, I'll correct it. But when the stakes get high, it suddenly matters more and at least when there's a human driving, if something's going wrong, I know that the odds are decent, I'm going to survive because they want to survive too and we're in this car together. Same reason why I really like that There's a pilot in that airplane and they're at the front, so at least if we're going to hit anything, they're going to see it first and try to avoid it.
(00:04:42):
When this shows up in financial services, I think we start to get a very similar kind of problem in how this shows up. Most of us as financial advisors are in the long-term relationship business. We take on clients, we have 95, 97% plus retention rates once we get established in our firms. So our average client tenure is 20 or 30 years, which means most of us spend the first 10 years growing our client base in the second 20 to 30 years simply serving them. Awesome. And maybe if you decide to add a few more and add some other advisors to handle the capacity, but you cap out pretty quickly, we can only do so many relationships and in a relationship based business like that, I can't be wrong. The whole relationship is trust based and I can't lose the foundation of trust. So in a financial advisor context, if someone comes to me with AI and says, I made an amazing AI piece of technology that's 99% accurate, I'm like, cool.
(00:05:36):
So I'm going to be sued once every year. My goal is to not be sued once in 30 years because the moment I do that goes on my regulatory record in every disclosure document I have to present in front of every future client for the rest of my life. And even though I'm still a fallible human, you are an AI technology so you're not allowed to be fallible. And that kind of gap I think creates a real headwind around how advisors adopt technology now as we'll Talk more in a bit, I think there are other ways that technology can show up around what we do as financial advisors in the whole spectrum of what happens in advisory firms. But this fundamental premise that I know a lot of companies started out with, let us come in, we'll build your portfolios and do your financial planning for you.
(00:06:19):
We'll supplement the advice. All you have to do is deliver it to your clients. For most of us, you can't be accurate enough to get us to trust it and especially not when right or wrong, we require the tech to have a higher level of accuracy than we do in the first place because of that strange AI trust gap phenomenon. So I think a lot of the headwinds around AI coming into the advisor world is a lot of this started with we'll automate your advice. I'm like, guess what? Wasn't looking for that. In fact, that's what I do that's valuable for my clients. I am not looking to automate that part. There's maybe some other things in my business life that I would like to get easier that I would be happy to have technology use, but I think we started off on the wrong foot and now we're iterating in a better direction around it.
Suzanne Siracuse (00:07:08):
Yeah, I think that you made so many great points there and it is ironic that we are really challenging technology to be better than humans and that just your analogy around Uber drivers and driverless cars I think is something that we can all identify with. So given some of the challenges that you just brought out, let's actually focus on some of the positive. I was here all day yesterday. There were some amazing demos, amazing use cases, just the discussions, the engagement. Fantastic. So what are your thoughts around what's currently really successful regarding AI in wealth management? And let's kind of divide this into two areas like more the kind of operational efficiencies or the back mid office as well as the solutions that are out there that are going to either help an advisor drive more revenues to their firm and or provide more offerings to their clients. So can you start out with the operational side?
Michael Kitces (00:08:07):
Yeah, look, so the operational side, I think hands down our clearest strongest use case that's unequivocally got traction right now is meeting notes and just like the related layers of client meeting support, there's a pre-meeting aspect, an in meeting aspect and a post-meeting aspect. And different companies have come at the space from different directions of that more and more eventually covering the full range of it. I mean, we were talking about this when ChatGPT first came forth last year on the Kitti platform and me, this is a great use case when we've done research on on our platform for a long time, the average financial advisor spends one to one and a half hours of meeting prep and follow up for every one hour that they're actually meeting with the client.
(00:08:57):
I got to get back up to speed on who they are and what's going on. Maybe there's some quick analysis before I go into the meeting. I got to remember all the stuff that was going on, so I'm mentally present in the meeting. Again, we're in a trust relationship based business, so like, oh, remind me the name of your kids again, even though you pay me $10,000 a year doesn't fly very well. So there's a lot that we have to do to get prepared for the meeting. There's a lot that happens in the meeting. And then afterwards, there's a lot of post meeting follow-up that we have to do. We've got documentation requirements for compliance purposes. We've got follow-up action items to the team to make sure that we follow through on what we said we were going to do. Best practices at least is that there's some kind of post-meeting message out to the clients to let them know, recap the meeting and what happened.
(00:09:40):
And so there's a lot of time that goes into that. Advisory firms have historically said, we will solve this by bringing a second person into the room, so I'm just going to focus on the client and my associate advisor can take all the notes and do the CRM tasks and the follow-up. But this is very conducive to automation in general and me is especially conducive to the sweet spot of AI in an LLM oriented world. It's a big old ream of talking stuff and I mean if there's one thing that LLMs are really good at, it's taking a whole bunch of words of talking and transform that into something else that's useful, like meeting notes, summaries, takeaways, action items, et cetera. Now I think meeting notes are also an interesting use case because a lot of the meeting notes providers are now seeing where some of the challenges crop up when you get from the immediate stage of capture and summarize notes and into what a lot of us talk about ideally as automating the tasks that come after.
(00:10:35):
Because the reality for a lot of advisory firms is we're not that systematized in the first place. That's kind of the deep, dark, dirty secret of a lot of advisory firms. And so as we may touch on a little bit more later, I think ironically there's a lot of challenges around true automation in advisory firms because the whole nature and flavor of what we do is every client is unique and special and different and we have a deep relationship with them. I don't need you to create personalization with scale. I've been to their weddings and funerals. Personalization is not what I need, but prep and help and support is absolutely what I need and what I can use. And to me, one of the interesting phenomenon that's starting to crop up around this is what I see in practice is that a lot of tech platforms are trying to automate things that we actually don't need fully automated.
(00:11:26):
We just need expedited. You can't automate meeting prep. My brain still has to remember the people and the stuff. If you get me down to zero minutes of meeting prep, I'm going to make an idiot of myself in the meeting, but you could take my meeting prep down like 80% and by getting me all the information faster, easier, more packaged and maybe even prepping it in a useful format so that I can get up to speed faster and easier. And so if there's one takeaway I would give for a lot of the vendors in the room around many, many of the advisor use cases like move away from automate and think expedite, I'll actually pay you more to do 80% of the solution than a hundred percent of the solution.
Suzanne Siracuse (00:12:06):
That is
Michael Kitces (00:12:06):
Word yes, because if you do a hundred percent, I got to be afraid you did it wrong. If you do 80%, you just saved me time and I was going to have to spend some time on this anyways, so now we're all winning.
Suzanne Siracuse (00:12:19):
You have such a good way of instilling common sense into some of the things that we are all thinking about and talk seriously that is just such a small nuance to change that way of thinking and to kind of change your marketing to reflect some of the things that Michael just said. And we had a few demos yesterday that were specifically around meeting notes. One of the things I was sitting next to an advisor after the demos and after the demos I asked him, which one of these demos really resonated with you? What did you think? And he was pretty fascinated with the fact or he was talking about the compliance implications and the regulation of all of these things. And so I guess when you have a transcription that all is discoverable and then there was one solution that said, well, we don't record it so you don't have to. That's not discoverable. What are your thoughts around, you were specifically calling out meeting notes around the discoverable transcripts and not having a recording?
Michael Kitces (00:13:28):
Yeah. Look, I think this is going to be a really interesting regulatory battleground candidly in the next couple of years and to say it hearkens back to me to what happened when email first showed up in advisory firms 25 years ago because historically, essentially we have three ways of communicating with clients. I can sit in front of them in a meeting which is live and contemporaneous and there's no capture. I can call 'em on the phone, which generally is not recorded, or I could write them a letter or type out a letter and that was correspondence and that had to be archived and reviewed and compliance to its thing. Then email showed up and email was like, well, it's just faster for a lot of clients to drop 'em an email than a telephone call. And now all of a sudden advisors are like, whoa, compliance is now all up in my client communication in a way that they never really were before because I went from telephone to email and I'm basically saying all the same things that I did before, but if I call my client and say, I'm confirming the meeting for Tuesday, that's fine, but if I send 'em an email to say, I'm confirming the meeting on Tuesday, I have to use a compliant email provider, run it through the compliance system for review and do all these other things, and now we're even seeing the same thing like, cool, you can email your client the confirmation, but don't you dare text message them that confirmation unless you've gone through all the lists of the compliance things.
(00:14:44):
And to me, this is just the tension of look, as advisors, we generally just want to do our business and interact and relate with our clients, and regulator is going to regulate and the general default of a regulator is any opportunity, but we can see more stuff that's going on. It's more opportunity to supervise and capture wrongdoing, which as a general case is good and in practice gets a little messy when you start getting again into approved channels and how to review an archive. I think this is going to be a really interesting challenge when it comes to what happens when whole meetings start getting recorded right from the pure use case of an advisor. Yay, no more meeting notes. I don't have to spend 30 minutes typing up the follow-up email to the clients. I just pull it up on my screen pre-populated from a platform and I just spend five minutes editing and sending it out.
(00:15:28):
I often take two minutes to review the CM notes instead of 20 minutes to write them up. This saves me all sorts of time except my meeting is recorded and discoverable and I think this is going to show up in a bunch of different ways. For some advisory firms, they're going to get just nervous about the amount of discoverable stuff that's out there and the way that lawyer, plaintiff's attorneys may come in and go after it. Some large financial services firms that know deep, deep down that maybe not every single rep in their firm is quite doing the best thing now really increase their own internal regulatory and compliance burden. It's like, cool, are you ready to record all your meetings and actually fire a segment of your advisors when you find out what they're really saying to some clients? The pro consumerist in me says, yes, please.
(00:16:16):
That's literally the point of compliance. But a lot of firms are not necessarily ready to actually deal with the consequences of what happens when this all gets recorded and then we have to figure out how consumers actually feel about this, right? Some consumers by and large just don't like anything recorded much less when you get into, so I'm literally sharing my personal deepest, darkest secrets and family strife about money. These are really personal conversations to clients. And so I can, I'm not quite sure how this plays out between consumer protections, regulators want to regulate. Firms have, I'll just say varying views about how much they really, really want to actually do compliance, and advisors simply want to get some good efficiencies out of this, and so do we record meetings, transcribe them, and then delete the recordings as a consumer protectionist slash Then we don't have to see what we don't want to see thing. Are we ultimately going to say, Nope, recordable format has to be recorded because that's just the compliance reality. I'm really not sure how it's going to play out, but I think this is going to become, we talk about it a little bit in the industry. I think this is going to become a really active regulatory debate over the next 24 months when regulators really start seeing this and asking questions and compliance attorneys really start getting into it. They have to decide how much do we actually want recorded on the record or not.
Suzanne Siracuse (00:17:35):
I agree, and I think that that's one of the biggest challenges for this industry in terms of wealth management and financial services is the heavy, heavy regulation. And so a lot of times we get kind of knocked, oh, you guys are slow to adapt. Well, it's not really, we're slow to adapt. It's, we have so many regulation challenges.
Michael Kitces (00:17:56):
We literally hold people's life savings. We can destroy 30 years of someone's wealth accumulation, like 30 minutes of stupidity. So I'm frankly of the view that I think we should be a really, really highly regulated industry. I don't love every regulation the way it happens to be written. I've got my grief with some of it as well, but I think we should be a highly regulated industry. You move fast break things does not work well when people's life savings are at stake.
Suzanne Siracuse (00:18:24):
Yeah, I think next year we need to have someone from the SEC and FINRA here to actually listen to some of the use cases, listen to some of the challenges, to really figure out how they can be helpful in solving for some of this. That would be, I think a really important move for them to be here, for them to get involved in the actual solution versus creating more headaches, I think for a lot of the advisors here.
Michael Kitces (00:18:55):
Now, the one thing I would note is a dovetail on that. There are also many AI use cases that don't even necessarily involve getting terribly deep or at all into personally identifiable information into client data, into meeting notes. I mean, it's cropping this particular issue because the first use case we all hit out of the gate was meeting notes and like, oh, there's the transcription now we got to deal with it right now. So I maybe just ironically by circumstance of how this played out, we happen to do the really, really recordy thing first with all sorts of client privacy and regulatory aspects to it, but there's a lot of other use cases. This is not going to be the issue. It literally doesn't have to do with recordings and transcriptions and what's being said to clients, and it might not even touch on PII at all. Right? There's all sorts of marketing use cases around how we build marketing materials for clients that I've got a compliance obligation to review advertising. I am not implicating all the rest of client data and communications in the same way.
Suzanne Siracuse (00:19:57):
Exactly. We had a great session here yesterday, AI and marketing where we had some individuals and advisor and two marketing experts talking about the various ways in which you can use AI to really help create content, which I thought was fascinating, and you won't have that same level of compliance or areas issues. Those were really good use cases. So I want to go back to are there any AI solutions out there or offerings that you feel are set up to help an advisor drive revenue to their practice?
Michael Kitces (00:20:37):
That was more of a side than I meant as a statement about it. So it was like, yes, but I think there were a couple of interesting places where AI starts showing up in marketing and business development, everything from just help me compose and create content more easily. I can basically go to chat BT and do that directly, but some tools are building standalones. Frankly. I think a lot of that's probably going to show up within the existing marketing platforms. I'm probably not going to buy an AI tool that does social media headlines. I'm going to go to whatever software I use for social media and I'm expecting it to have a generate 10 alternatives to this headline that gives me ideas kind of thing. So I'll call, I'm expecting more embedded AI than standalone AI apps, but lots of marketing communication I think has very good use cases here at its core.
(00:21:30):
Anytime you have to deal with a blank page, AI's really good at making it blank, whether that's, I need a social media headline, I need to write an article, I need to write an email to a client, right? Post meeting notes follow up is a version of this as well. It's so much easier to edit than it is to create that when AI just gets you past the blank page in something that you need to edit, you tend to be greatly expedited. Again, not automated. I'm not letting the AI literally write my marketing communications, but you can expedite me. Don't make me spend an hour trying to figure out how to draft an article when I'm not very writing inclined. Give me a sample of an article around the themes that I told it in the prompt and then let me spend 10 or 20 minutes editing it and getting it to where it needs to be, and you saved me 50 to 75% of my time in getting there with a greatly expedited process. So phenomenons where there are blank page phenomenons, I think there's great opportunity, but I'm a little bit skeptical of how much is standalone AI tools versus embedded AI within existing software and vendors that already do some of that.
(00:22:34):
The other realm that I think there's at least interesting opportunities around AI that where we're starting to see a couple of players crop up is broadly what I'll call the prospecting realm, like figuring out who to call on in the first place. Now, notably, not a lot of advisors actually do aggressive outbound marketing. At least for most of us, we do that because we have to when we start and our singular goal is get to the point where we have enough clients to refer you, then you never need to do that again, and only a small percentage of advisors. Our estimate from some of our internal ads, maybe 10 to 20% of advisors I'll call it, they get the thrill of the hunt. They love going out there and doing business development and winning another new client and that's how they're wired and they'll keep using those tools.
(00:23:18):
Everyone else, as soon as they don't have to do prospecting, they want to move away from it. But there are interesting opportunities around prospecting tools and use cases. Candidly, I'm not sure how much of this is really AI, and it's certainly not necessarily an LLM aspect of AI, but anything, if I'm going outbound and prospecting, anything that gets me closer to what the industry would call a warm prospect than a cold prospect, the better. And generally, there's two ways that I can expedite this. The first is find me someone that's warmer rather than cold because I have some connection or relation to them. So here's my prospecting list, cross-reference it against marketing databases, LinkedIn, everybody else, and find me some way to connect in commonalities like, oh, you're into golf. I'm in a golf. I didn't realize we have the same alma mater. Anything that lets me create some kind of connection.
(00:24:06):
Oh, you actually know them through a LinkedIn connection. Like, cool, I'm going to ask Bob to introduce me so I'm not going out to 'em cold. So anything that makes a cold connection warmer and there's lots of, I'll just call big data ways that we can try to find those connections becomes valuable that anybody does prospecting. The second version of this, I think that crops up is how to better qualify a prospect. So most advisory firms have some kind of minimums. Either they have a minimum fee, they have an asset minimum just to run an advisory business. I need a certain amount of revenue per client to run the business feasibly. And so one of the first and most important things that anybody learns when they come in the business and they start getting clients is don't spend your time with people who aren't qualified to do business with you.
(00:24:48):
If you can't afford my minimum fee, I really shouldn't be spending time talking to you, chasing you, prospecting with you. Like no offense, I wish you the best in your life. I need to spend my marketing business development time on people who can actually afford what I do and might have a need for it. And so there was another set of opportunities. Again, I don't know if I would strictly call it AI because there's different ways that you can tackle this problem, but tools that help us better pre-qualify the prospects that we're talking to and against. Lots of different ways you can do that with marketing databases and the other stuff that's out there, how to predict when people have money in motion opportunities that create a chance there. So there's tools that help me get to warmer prospects than cold tools that help me put better pre-qualify my prospects.
(00:25:32):
There's some ways to wed that together, give me warm leads that are pre-qualified, my list of leads, make them warmer. Here's my list of leads pre-qualify them. And I do think we'll see some opportunities there because at the end of the day, even advisory firms with low minimums at the end of the day typically have to generate at least about three to $4,000 of revenue per client to really scale the business. Most advisory firms work in a five to $10,000 revenue per client range and higher end firms go up from there. And when you live in a world where an advisor might have a $10,000 client for a million dollar minimum, and we have 90 to 90%, 95 to 97% retention rates, we live in a world where one client is like a 200,000 plus dollars revenue opportunity over their lifetime. And at the end of the day, when one client is a $200,000 lifetime revenue opportunity with a 30% profit margin, we can spend a lot of money on prospecting software if we actually want two prospect and business develop that way.
(00:26:32):
If you've got $5,000 software and I get one client, I get a bajillion percent ROI. So it creates a lot of room in that space. But I think the challenge for a lot of vendors coming into that space is there's only so many advisors that actually like to do that level of continuous outbound marketing growth. So we see it in these pockets, new advisors that need to get going, the small population of advisors who just love business development and continuous growth and just getting to going out there and getting more clients and some of the more transactional models in the industry like the annuity FMO channels and the like that have a more transactional business model and always need a continuous flow of new revenue. If you're on a commission-based model, your income goes to zero every January 1st. You are always out there on the hunt.
Suzanne Siracuse (00:27:18):
And in fact, great, great use case. There's a session later today that is focused exactly on that AI and marketing where how do you qualify prospects easier, et cetera. But to your point, in fact, we were on a panel last year at the CFP board where you likened financial advisors to doctors, right? You're like doctors, like financial advisors are treating a problem or a symptom like you are treating and helping a client, but you don't expect doctors to go out and market to get patients. I thought this was a really interesting analogy. And so the same thing happens with advisors, right? You came into this business not necessarily to go knocking on doors and figuring out the marketing 101. So I think a lot of what I see is some trends that'll happen are firms hiring business development associates to focus on this kind of stuff, right?
Michael Kitces (00:28:16):
Yes. And there's an interesting phenomenon out there now, particularly amongst some of the larger advisory firms where, look, if you look at almost any industry besides financial services, sales is separate from service by any large product enterprise solution, whatever it is. You do sales with the sales team and then after you say essence on the contract, you're handed off to customer service team, customer success team that actually runs with you and helps you do it and implement it and get the thing you're going to do from that point forward. Almost every industry separates sales from service except ours and ours doesn't because historically we were all salespeople working on commission. There really was no service because we just had to go and sell more new people. We migrated from these commission-based models into ongoing revenue models with assets under management now subscription fees, and we went from a sales-based business to a service-based business that still needs business development.
(00:29:11):
And because we all grew up in a business development world, if you've been doing this more than 15 to 20 years, we tend to still attach the business developers to the service. And what's starting to happen now amongst some larger firms is that's problematic for many reasons. Some people really are just great at the service, not the sales. Even the people who are great at the sales, often they get to a point where I make enough money, I don't really need to grow your firm anymore. I'm going to work with my client base and make good money and chill out. I know you're not going to fire me anyways because then the clients are going to come with me and you're going to lose money, so I'm just going to not grow and you're going to leave me alone and works great at the advisor level.
(00:29:44):
Very problematic at the enterprise level. So what's starting to happen in some of the larger firms in particular, because the more advisors you have, the more this problem compounds on you is they're beginning to separate sales from service and making the business development sales, marketing function centralized to the firm and putting advisors in more pure service roles. I think the only caveat I would attach to that relative to the AI conversation is when advisory firms start building centralized marketing and sales mechanisms, they generally don't do the same outbound prospecting approach that we were all trained on because it's very labor intensive and not terribly scalable for all the reasons that we're actually moving away from in the first place. It tends to be more centralized marketing. Often there's inbound marketing approaches. It's a heavier focus on branding. It's something that a lot of firms struggle with because they don't have core capabilities around it. But to bring it back to the AI context of the discussion, I suspect many of those firms will do less of the AI prospecting tools and more of other centralized marketing functions just because of how that particular use case shows up in a large enterprise.
Suzanne Siracuse (00:30:50):
Yeah, I think that's a great point. And again, just part of an overall trend of what you're seeing in the industry with smaller firms versus larger firms. So we just talked about a couple of, I would say successful or fairly successful use cases or where you feel that there's an opportunity. What about, and this is not going to be popular probably, what about some of the use cases or some of the solutions that you think are really not there yet?
Michael Kitces (00:31:23):
The biggest, I'll say the biggest problem case to me that I've seen crop up in the industry when we got the first wave of ChatGPT and the like, and it was kind of cool, you can literally talk to it. You can have it a conversation, you can ask it questions and it answers. It's not a far jump from there to say, well, you could not just ask it a knowledge question, get an answer. You could ask it to do something and have it do something for you. And we got this whole wave and trend of the future is going to be ai, digital assistance where you just tell your phone what you want done and it does all this cool stuff for you, and look maybe someday, someday distant out. Okay, I'm not going to foreclose on that in some 10 to 15 year time horizon because we all tend to collectively underestimate how much technology moves in the next 10 to 15 years.
(00:32:13):
But when it started to show up in the advisor realm, I started seeing the vendors coming forward with these use cases of like, Hey, this is so cool. If you want to know what meetings are coming up for your clients next week, you just typed the bot, tell me what my meetings are next week. And then it gives you a list of meetings. Then you could type, great, send the agenda out to client A and give me the prep for client B, and then it would do that because it could translate your text into commands and do all the cool things that modern chatbots can do. Okay. But you realize most advisors are two finger typists.
(00:32:53):
Like this is not a very effective use case to give me a list of my client meetings next week. If I want to list my client meetings next week, I just want my CRM to have a screen that has a list of all my client meetings and a button next to it that says, kick off the workflow for the meeting agenda. I don't need a natural language processing chatbot to do this. In fact, it's demonstratively slower for most advisors to do it that way. And so the chatbot realm to me in particular became very, very challenging, both because, oh, just a lot of advisors at the end of the day are not fast typists. Well, some companies to me were basically like they were using chatbots as a substitute for good ui, then just make the CRM UI better. Just give me a screen that shows me my upcoming meetings and a button to click off my workflows and I don't need a chat bot to type back and forth. I can just one click a button on my screen to get to the actual thing that I was trying to accomplish Anyways. So I think there were real challenges in how those types of chatbots showed up in our space.
(00:34:03):
Now, I want to differentiate that a little bit from, I know there are some tools that have built chatbot style interfaces for things like investment research. I would view that as a little bit different. I'm not trying to use it as my assistance as a substitute for CRM when I could just have a better CRM interface. I might be asking more novel questions and novel prompts. But even in use cases like that, what I ultimately see a lot of these vendors quickly evolving into is they basically have to teach advisors how to be effective prompt engineers in order to do that. If I need to be an effective prompt engineer, I'd rather you just make a better UI that makes me not need to type a prompt is what crops up for most advisors. I didn't sign up for this job to be a prompt engineer.
(00:34:46):
I signed up to give advice to my clients. There are maybe a subset of, I'll call 'em hardcore investment nerds that will just swim in that data all day long and ask 47,000 different prompts and play around in that space. And I'm sure there's a narrow use case of a couple of people that super love nerding out deep into the data that way. But to me, it's still a fundamental challenge that I see a lot of chatbots coming up as what to me is basically a substitute for making a better UI that wouldn't necessitate the chatbot in the first place and would actually take the pressure off of me of figuring out how to be a prompt engineer that I didn't want to be in the first place. That's not to totally eliminate the relevance of chat style interfaces, but frankly I think it would drastically cut down on them.
Suzanne Siracuse (00:35:34):
Yeah, that's an interesting point. I think that you're going to have lots of schools of thought and different schools of thought based on how you like to work.
Michael Kitces (00:35:42):
Someone's going to come at me when I go out the door at the end of the session.
Suzanne Siracuse (00:35:44):
Okay, well, that's part of why you're such a good person to have a conversation with. It does make you think about things and people are different. The whole thing, if you've met one advisor, you've met one advisor, everybody's got a different opinion so,
Michael Kitces (00:35:59):
But just I will harken back to me. I'm a use case guy at the end of the day, I'm just, what does it do? What does it do for me? Does this actually make my life easier and more useful? And so again, that's why I don't say pick on, but that's why I have a challenge with a lot of the chatbot interfaces. Again, why do I need a chat bot if I can just have a good screen that shows me my upcoming client meetings? Why do I need a chatbot if I can just have a good query engine or a good interface to drill down on my investment research? And again, that doesn't eliminate all chatbot use cases, but I feel like a lot of vendors initially came up with this technology is so cool, you can make a chatbot. And I'm like, yes. But just because we can doesn't mean it's actually going to save us time to do it.
Suzanne Siracuse (00:36:45):
Yeah. I think that your philosophy, and correct me if I'm wrong here, is about start with the question. What are you trying to solve for? And don't focus in on the actual solution that is out there, but what is the problem you are trying to solve for in your own firm?
(00:37:03):
What keeps you up at night?
Michael Kitces (00:37:04):
Which again can even go so far as, and I actually wasn't looking to automate that, right per earlier. Sometimes we want to expedite but not automate. I don't want to make that thing go away. I want to just do it better for my clients. So sometimes it's expedite, sometimes it's better. Sometimes I just need a different interface in the first place. There's lots of different ways that get us to our incremental improvements and efficiencies and AI may interlace with those, but to me it always comes back to what's the actual advisor use case and pain point that we're trying to solve for. And in a technology world, it feels like the answer virtually always is automate. And so that's part of why I'm picking out automate a little bit here because for a lot of advisor use cases it's not.
Suzanne Siracuse (00:37:46):
So I want to switch gears a little bit to talk about the firm of the future. And for all of you that are here, kudos to you for having the foresight to understand that AI is not going away. It's more about how will it be integrated it into your day to day. And so taking the time out of your practice to be here, to be learning about some of these things, to challenge your thinking is I think a really important characteristic to have to be successful in the future. So there was a panel yesterday that had Samuel Dean who's a solo practitioner, and then next to him was Amanda Lott, who runs JP Morgan's innovation department. So two very, very different types of firms. And I think that's what's so cool about this event so far is that you have smaller firms, larger firms, the characteristic is learning about AI. So how do you think the advancements in AI and some of the solutions that are here are going to affect smaller firms versus larger firms that may have a lot more resources to test out all of these use cases?
Michael Kitces (00:39:01):
So I'd frame this in two different ways. I think there's two notable aspects of how this plays out. The first is, and I won't peg this specifically on AI, but the evolution technology in general, it has never been easier to be a high profit solo advisor than it is today. And I think that's a big deal because huge swaths of the industry maintain and insist that the only way to survive and thrive in the future is that you have to consolidate and merge and gain giant economies of scale. And then you go and look at their P&Ls and their profitability isn't better than anybody else's, and their overhead expense ratio is exactly the same as everybody else's until you get down to solo advisors that run drastically higher margin businesses than anybody else.
(00:39:47):
When I look at this, even relative to my career, so 20 years, a little over 20 years ago, I worked in an advisory firm that was three advisors and about 1.3 million of GDC, which back 20 years ago was a pretty sizable independent broker dealer shop, and we had a total staff of 13 to do all the things that it took to run that business. We had a woman, bless her soul, her name was Betty. Betty's entire job was greet clients when they came in for in-person meetings and gather all the reams of mail that comes in every day and open all the envelopes, find the checks, because you can't hold those by accidentally take all the statements file in every single client's folder. There was literally a room that was just the file cabinets for all of the clients. Betty's job does not exist anymore.
(00:40:34):
That whole thing, her whole job is gone thanks to technology and just like the digitization of paper statements and the fact that we can meet with clients virtually. In fact, when I look at most advisory firms today, if you were a three advisor firm with 1.3 million of revenue, you probably have a total headcount of five. It's probably you, the three of you and maybe two administrative support staff that you share between the two of you. Now, the interesting thing about that is it still might be three advisors, although you could probably do that revenue with two, but we went from a staff of 13 to a staff of five. That's all technology. That's all technology that has made smaller firms better, more efficient, more effective than they have ever been at any point in history. And so I think the whole industry narrative of you have to consolidate in order to survive, frankly, if you look at where the message is coming from, it almost entirely comes from two places.
(00:41:26):
The people who want to buy you, who insist that the only way you can survive is to sell yourself to them and large advisor platforms, broker dealers and RA custodians, who at the end of the day say, you know what? Instead of serving like 10,000 of you, it'd be so much easier if you all just consolidated into one. And then we would just have one giant client because 10,000 individual firms would pay in the backside to serve. And so the big players are mostly talking their book around the necessity of consolidation. And when you just drill down to the actual underlying practices of solo advisors, they've never been more profitable, or at least the advisory firms, they're struggling with profitability. It's not because you can't run efficiently as a solo. It's because they take too many clients who can't afford their services. They provide too much service relative to the fees that they're charging.
(00:42:12):
Their pricing is out of whack, but not because you can't run an incredibly successful solo practice. Now, large firms benefit from the technology as well. It shows up in a little bit of a different way. If I'm a small firm environment, I love technology and the internet because I can run my practice from my office or the beach and no one even knows the difference. I got a virtual background behind me and I'm having the same meaningful conversation with my client. If I'm in a large firm environment, I'm getting some of the complexities that come with larger firm and scale. I'm trying to find efficiencies. Now I really live at the level where I have a process that's repeated across seven people. And if we could automate that and have one person who oversees it, that's a really meaningful cost savings for me so I can reallocate that cost to somewhere else in my organization.
(00:42:54):
So from the development end and the resources comment you made, that does mean it also shows up differently because the smaller practitioners, we're going to buy the stuff off the shelf that many vendors here are demonstrating, we're never going to build our own tech. We need to buy the tools that are available. And the fact that SaaS models exists is an amazing thing. I buy this and I buy that, and I like to buy that, and I buy exactly things that I need from my practice. And I think larger firms are going to engage with a lot of these providers in deeper relationships. I think we'll see a material segmentation between retail advisor vendors and enterprise vendors. I think we'll see some companies that do all of their work in custom strategic project work around AI with large enterprises. And that's not really different to how new technology leaps emerge as we've gone through the cycles when the internet showed up, some people sold retail solutions and then some sold private clouds to enterprises and built that way.
(00:43:50):
So I think we'll see the same thing playing out in the AI space, but to me, how it shows up is a little bit different. Large firms really look for efficiencies of repeatable scale problems. Smaller advisors, it tends to be less about scale. I don't really have no matter how efficient you make me, my brain can literally only keep track of so many clients before I start forgetting who they are. It's much more about how do you not automate but expedite so I can just do this more efficiently? It's about how do you enrich the relationships that I can just get better at what I do with my clients. And so the flavors of the technology will show up a little bit differently between the two.
Suzanne Siracuse (00:44:29):
Yeah, great point. So you mentioned on your last point, how do I serve my clients better? And that goes to my next question, what AI solutions, how is AI going to really help advisors deliver more personalized or customized solutions to their individual client needs as client demands are changing? Right? Yeah.
Michael Kitces (00:44:55):
So this again is something that I look at very differently depending on whether you're talking about the individual advisor environment, which is kind of my world, my peeps, and how I find large, highly scaled enterprises think about this problem. So for most of us who are individual advisors, my primary asset value proposition is the advice I give to my clients. That's how I differentiate myself. And this whole idea of personalization at scales, an individual advisor is sort of comical to me. Do you not understand that I already sit across from all of my individual clients as individual human beings, and that the literal definition of advice is that I'm giving a recommendation for you specific to your needs and circumstances. It's already personalized. I know all of them. I have relationships with all of them. I give them advice that's literally specific to their circumstances, and that's the primary thing that they sought me out for.
(00:45:56):
I need zero assistance with personalization and zero assistance with customization. In fact, the primary challenge for most advisors is that they over personalize, over customize, and therefore they can't automate anything because you go into their firms as vendors and you try to automate and you're like, tell me about your process. Well, here's the a hundred different processes I have for my a hundred different clients. I'm like, well, we can't automate that. It's like, yeah, that's why I'm not automated because I already do personalization and customization for every client, right? It's like the leading practice management advice is basically stop doing that and at least be a little bit more systematized about how you serve your clients when you get to the, well, sorry, and dovetailing on that quickly. So what that means from the advisor perspective is if you then want to help me get better, because now it's really about better, now we get into things like, how do you help me take my relationships deeper?
(00:46:53):
So I've got a client meeting coming up, I can use some of the meeting prep tools to help go through what was going on in the meetings. But ideally, you're looking at the client's social media feed so that if they recently announced a baby or a wedding or something, I know that coming in to ask them about it, I saw your pictures from the trip to Africa. Tell me more about how that went. If I'm working with executive, give me all of the information about their publicly traded company and any notable filings that have been coming up so I can talk, talk about the company, give me other ways to enrich the relationship if I'm getting into my planning analysis, are there other strategies? Are there other ideas that I might've missed that I don't necessarily need? Fancy ai? You can just get down to checklists.
(00:47:38):
Have you considered all these things that you might be talking to your client about, but it's much more about deeper enrichment when we get to the other end of the spectrum. So the largest of the large firms, to me, I find there's a fundamental mentality shift. Large scaled organizations don't treat advice as their asset. They treat advice as their primary liability. I don't give advice because it's my value proposition advice is a liability risk to be managed because if one of my zillion advisors does this wrong, we're going to get sued. And when we say we're going to get sued, I mean our corporation is going to get sued because you don't, as a plaintiff's attorney, you don't go after the individual advisor or 10,000 advisor organization. You go after the national holding company that's got really, really, really deep pockets. So large scaled organizations tend to treat this very, very differently.
(00:48:26):
I don't want my advisors to blossom into brilliant individual advice giving machines. I wanted my advisors to give the same advice to every single client in every single situation ideally generated by technology so that they basically can't say anything that I didn't want them to say or do it wrong. I just want them to be the advice delivery machines because apparently clients take it better when a human says it than when a piece of technology delivers it to them. And I'm being a little bit hyperbolic in the description, but not that hyperbolic from where I know some firms kind of view what their ideal is because they're terrified at the level of advice that their advisors might give and the liability exposure that they have. And so the challenge that comes then if you're a large scaled organization, is I really want all my advisors to do the exact same thing in the exact way, same way, in an incredibly consistent basis.
(00:49:15):
They don't go out the lines. There's a dramatic reduction in my risk of getting sued and having liability, but now I lose all personalization. I lose customization. I just turned my offering into a national scale cookie cutter. And so for a lot of large national firms, they're now trying to figure out, okay, we've swung the pendulum very far in this direction. We put a lot of boxes and guardrails around our advice. We've done it in a way that's allowed us to control our liability, but we're now losing some of the personalization and customization, and they're trying to figure out how do we bring back personalization at scale, customization at scale, in the way that they can still control and manage from a liability exposure perspective. So it's two really different frames and dimensions around what this looks like. If I'm an individual advisor, I want deeper, I want better, I want expedited, but necessarily automated.
(00:50:06):
I don't need one iota personalization, customization. That's what I breathe all day every day with my clients in the first place. In fact, I need to be less personalized and customized. So you can maybe automate some of my workflows. But if I'm a large scale financial services organization, I've probably spent a lot of time trying to put guardrails around what my advisors do because I have to do that to manage my own viability, but it's limiting my ability to do customization and personalization scale. And so I really need it on the large enterprise realm. And the larger the enterprise gets, the more that crops up. And then you even get a group in the middle that are kind of what I'll call the mega RIAs. They're bigger than small firms, so they're trying to systematize their advice. They're not as large as mega firms where they've potentially over systematized their advice and they're in a messy middle of trying to find that balancing point between the two. We want all clients of the firm to have a common consistent experience in being inclined to the firm because as Peter Malik recently demonstrated, you get a real premium in the marketplace if you can show that to investors, except if you do that too much, your advisors turn into automatons and clients actually don't feel the personalization of the advice and you risk falling off in the other direction.
Suzanne Siracuse (00:51:12):
Yeah, that's an amazing perspective and I think a real challenge for companies that are trying to do business with all of the various types of models that the advisors fit into. I think that it's one of the biggest challenges of selling to you as financial advisors is just this segmentation of your own types of firms, and a lot of times the lack of process and one advisor does it this way, another advisor does it this way. So I have not a lot of time left. And I have three questions I really want to get to because this flew by. Okay. I want to ask about can AI contribute to new investment strategy? So we haven't talked a lot about the investment strategy side of things. In fact, it was actually really hard to get speakers to talk about the investment side of things. I think it is still a little new. So can AI contribute to new investment strategies by analyzing data patterns that human minds may miss?
(00:52:17):
This is one question.
Michael Kitces (00:52:19):
Oh, now I get to get in more trouble. If you have a software solution that sells this, I don't believe you, and let me explain why. If you have built AI software that can actually find novel and original investment patterns that can be invested and exploited in the marketplace, and you sell SaaS software as the way to monetize that, you are not a good business owner. No offense, you should be raising a hedge fund and seeing if you can beat Ray Dalio because that's what he's done for 30 years. He took every type of investment rule in pattern recognition thing he could figure out how to do, and he put it into, I don't even know what they have now, a hundred billion dollars plus hedge funds earned 20% carry on it and became one of the a hundred versus people on the planet. If you can make technology that actually figures out novel investment opportunities that can be exploited in the marketplace, that's how you monetize it.
(00:53:18):
You don't do it by selling software because it's so ludicrously, brutally, insanely difficult to do in a sustainable matter because markets become remarkably efficient and anomalies and patterns get exploited and reduce out of the marketplace very quickly. If you want to get to the meta level of the investment marketplace, if you can actually create software that does that, I sort of know why you might've had trouble getting anyone onto the podium, because if you can do that, the only people you tell are people who will invest a minimum of a billion dollars with you and you don't tell anybody else because you don't want anybody to find out what you're doing and take advantage of the same opportunity and arbitrage it out of the market. You want to invest it as effectively and aggressively as you can and make a bajillion dollars. To me, just the whole premise doesn't really work as a software solution.
(00:54:09):
And on top of that, if you're doing that, good luck beating a lot of hedge funds that literally put billions and billions of dollars into this and have been doing it for 15 years, since the very earliest stages of anything that we started to call machine learning and AI. They have been trying to exploit this for a long time because they can blow a billion dollars a year at it and still make a zillion percent ROI when they find a one exploitable market opportunity that they tell no one. Now, I do want to ask her that by saying that's not to say that there isn't value in investment research and analytics tools that have some aspects of AI and either how I interface in it or how I scan the data. Cool. If I've got some ideas about a certain company and whether there's something in its SEC filings that could be a trends that I've already got as an idea, but I don't want to read 10,000 pages of SEC filings and I can tell the bot, go find me all dimensions of this and where it shows up like, great, what a wonderful use case where AI can scan a bajillion pages of natural language and surface some insights that I was looking for and how to hypothesis on, but I'm not expecting the tech to find the pattern and originate the pattern and generate the strategy.
(00:55:19):
If I was, I would monetize that very differently. But I do think there are very valid use cases around AI as a investment research assistant with just the caveat that I mentioned earlier, which is chatbot and prompts is just one of many ways to interface with an investment research and analytics engine, and that's not always the way that I would ideally like to interact with it.
Suzanne Siracuse (00:55:39):
Yeah, yeah. Well, thanks for your perspective and there was a great demo yesterday about a solution that I think is on the positive side of that that can really help. Another question that I have that was a big popular conversation yesterday and it's becoming much more of a discussion point than quite frankly I ever thought that was going to, and it's around data. So let's talk about data. So can AI help financial firms better utilize the vast amounts of data that they have to generate more insights? How is AI playing into the data side of things?
Michael Kitces (00:56:23):
So here again, I have to do a little bit of the small firm, big firm split, or I'll even say like the small firm, big firm, mega firm split. The average advisor does not have a big data problem. We don't have enough data to have big data problem. We don't have enough data to have a data problem, like I got 82 clients and their personal information. I don't need an AI bot to figure out, maybe you should scan all the clients and notice when they have significant the milestones like RMDs coming up and then queue something up for them. I can just literally say, make a safe search of my CRM that shows me every year that my clients are turning 73 and FRMDs are beginning and then I'll just call them and do my thing.
(00:57:02):
One layer of intuition and knowledge about the industry in a relatively simple search query kind of covers what I need. We don't have big data problems. We have small data problems, which are very different, larger advisory firms, so I'll kind of put this in the call like the mega RIA category firms that billions, maybe tens of billions of dollars. The data gets bigger and messier when you start rolling up that many advisors, but we don't really have what I would characterize as big data problems. Frankly, our data sets usually still aren't really large enough to do what big data really does when you turn some machine learning tools into the data and tell it to start finding things. We definitely have data challenges. We have tons of data challenges, but the biggest data challenge for most larger firms is simply the centralization of warehousing function in the first place.
(00:57:48):
My problem isn't how to get insights. My problem is I can't even get them into one darn place to get insights. And even if I get into one place, the data is really messy because our industry is not normalized. Its data standards and I don't really want to spend the time normalizing it because I'm an advisory firm, not a data warehousing firm. So a lot of advisory firms have challenges in their data as they get larger and they would like insights, but to me these are not really AI driven big data analysis problems. I know the insights I want. I want to know my revenue per advisor and I want to know their productivity and I want to know their growth trends. I could come up with a couple dozen queries that I would like to make on the database to get insights about my business, but I'm not looking for new novel insights and patterns where I turn AI onto it and say, find me cool things.
(00:58:35):
I just literally have questions about my business and can't figure out how to centralize and normalize the data. And so maybe there's some edge cases of if you can make AI that could figure out non-normal data so I don't have to normalize my data because you can figure out when Schwab says it this way and Fidelity says it this way. That's actually the same thing in my data downloads. Cool. So there might be some edge cases around it, but at least as I think a lot of people classically talk about turning AI loose on big data. Even big advisory firms really don't have big data problems. They have a high volume of small data problems and a struggle and how to warehouse and normalize it. When you get to mega firms, the Schwabs and Merrill Lynches, when the dollars are measured in the trillions and the client households are measured in the millions now you have legit big data sets where you can turn something on and just have it swim through the data and see what it finds and wishy luck. May you find something cool that we can all learn from. But there's actually relatively few of those enterprises out there. Some of them will build things, either they'll do it internally or they'll do strategic partnerships with some AI sharp companies that want to do this. But most advisors I find we really don't have big data AI problems and opportunities. We either have small data problems or we have warehousing and normalization problems against which we just want to run relatively basic queries that are really hard to do right now.
Suzanne Siracuse (00:59:59):
Yeah, I think your point about the warehousing is right on with the majority of the firms that are here. So one last question. I'm going to take one more minute. One minute we have.
Michael Kitces (01:00:09):
Alright.
Suzanne Siracuse (01:00:10):
There are a lot of, I think firms here that are developing solutions for financial advisors utilizing ai. That's one of the reasons you're here. What's your last word? What's your one or two bits of advice for those companies that are trying to market to and sell solutions to advisors like you?
Michael Kitces (01:00:33):
So aside from the comment earlier, ironically, be careful with automate, expedite, better, deeper, richer are actually better verbs than automates because a lot of things we're really not actually looking to have it automated or it's not automateable because we literally do a different thing for every client. The biggest thing that I would say, so we did a poll on this out to our advisor community earlier this year. We just asked advisors, when a company says they're an AI solution, is that a good thing or not? Do you like it when companies say we're ai? And what we found was almost a perfectly even split. A third of advisors said, cool, love AI. It's the future. Sign me up for some AI stuff. I want to check this out. A third of advisors said, if you say ai, I'm gone. It's the same person like open the door, look in the front seat, there's no driver there.
(01:01:36):
Close the door, walk away from the Uber, just not getting in a driverless car. The moment you say AI, I see all the risks and you can't possibly benefit me enough. If you literally get me sued once, you'll destroy my career and ain't worth it. I'm, I'm not trying to save five minutes a week at the risk of my career. The last third said, I really couldn't care less that you're AI. Just tell me what you actually do. I don't care whether you do it with AI, 10,000 monkeys and magic or outsource to some place that just doesn't really cheap. I don't care. But care is what you do. Just what do you actually do? What does your software do that helps me be more successful in my business? Now the thing I would note about that, if you kind of roll those together, that means when companies lead with AI, a third of advisors like it and two thirds don't.
(01:02:23):
You're literally turning off the majority of your audience. Now, you will get your early adopters and frankly, if you're an early phase company and you're still trying to iterate and find product market fit, great thing like lead with ai, find your early adopters, get them in, do your rapid products cycle iterations and find your thing. But if and when you're ready to go mainstream and broader, AI is still very much a liability in company advertising and marketing. And at best is not helpful because the advisors really don't care that your ai, they just want to know what you do. What is your my language, not advisor language? What is your use case that you solve? What is at the end of the day that you actually do to make my firm and me more successful? And I really don't care whether you do it by AI or some other means. I just want to know what it does for me. So less about the AI. I know all of your investors love it. It's in everybody's investor deck because if it's AI, that must mean it's the future and you get a bajillion dollar opportunity and you get funding. But what goes in your investor deck does not go in your marketing deck to advisors. Focus on the actual problems that you're solving and it's way more impactful than leading with AI.
Suzanne Siracuse (01:03:33):
Well, that was great advice, great last line. I want to thank you for always giving us something to think about. I think part of these types of conversations are meant for you to play devil's advocate for you to consider some of the things that Michael was discussing today. Thank you all for being here and for your attention and for your interest in really utilizing AI in the future of your business. So thank you. I think we have a break next, but appreciate your attention. Thanks, Michael.
AI in Financial Advice: A Fireside Chat with Michael Kitces
November 12, 2024 3:18 PM
1:04:14