In The Harbor
The Podcast where sports, leisure, and finance meet
In The Harbor
#20 Josh Pantony
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
In this episode of In the Harbor, we sit down with Josh Pantony, Founder and CEO of Boosted.ai, to explore how AI is reshaping not just investment research—but the business of advice itself.
Josh Pantony is the Founder and CEO of Boosted.ai, an AI-powered investment research platform built specifically for professional investors and financial advisors. With a background spanning capital markets, data science, and applied machine learning, Josh has spent his career focused on one core problem: helping advisors and investors make better decisions faster—without sacrificing rigor, transparency, or trust.
Under Josh’s leadership, Boosted.ai has developed Alfa, a conversational and agentic AI platform that integrates directly into real-world investment workflows. Beyond portfolio construction and macro analysis, Josh is especially focused on how AI can help advisors scale their impact—from generating differentiated insights to improving how they identify, engage, and communicate with prospective clients.
Josh is a frequent speaker on the future of AI in wealth management, with a practical, advisor-first perspective on using technology as a force multiplier—not a replacement—for human judgment and relationship-driven advice.
We start with Josh’s career arc and the founding of Boosted.ai, then unpack what agentic AI really means versus traditional large language models. From there, the conversation expands into how advisors are using AI to move faster and smarter across their workflows—from macro research and portfolio construction to prospecting and client acquisition.
Josh explains how AI can help advisors surface timely, relevant insights that spark better conversations with prospects, personalize outreach at scale, and spend less time searching for information—and more time building trust. We also dive into Boosted’s work around voice and conversational interfaces, why spoken interaction is a natural evolution for research and prospecting, and how advisors can stay compliant while embracing these tools.
We close with a forward-looking discussion on what the advisor of the future looks like—and why AI may ultimately make the human side of advice more valuable, not less.
Key themes include:
- Agentic AI vs. traditional LLMs
- AI as a force multiplier for advisor prospecting
- Turning research insights into client conversations
- Voice-driven workflows and efficiency gains
- Risk, compliance, and trust in AI-enabled advice
- The evolving role of the modern advisor
⚓️ Find us In The Harbor
Where sports, leisure, and finance come together — from the rink to the markets and everywhere in between.
🎙 Subscribe, follow, and join the conversation:
📸 Instagram & TikTok → @intheharborpod
💼 LinkedIn → Rock Harbor LLC
🐦 X & 📺 YouTube → @InTheHarborPod
#InTheHarbor #Podcast #Finance #Sports #Leisure #RockHarbor #FinancialAdvisors #Investing
Welcome back to In the Harbor, the podcast where sports, leisure, and finance meet. I'm your host, Jason Johns. Today I am joined by Sean Richland. Sean, welcome back to the co-host seat.
SPEAKER_03Thanks for having me, Jay. I appreciate you allowing me to come back on the show.
SPEAKER_00It's a heavy lift, but every now and then we we think we'll put you on. Thanks, man. I'm really excited about today's guest. Today's guest is Josh Pantany from Boosted AI.
SPEAKER_03Yeah, something that I think a lot of the world is curious about and what direction this AI world is going. It's so fast, evolving so quickly. And I'm interested to hear his spin on the financial world and how it can relay not only into analytics, but also helping financial advisors transform the way they look at their business.
SPEAKER_00Absolutely. I think it'll be a great conversation. I'm curious to hear what he has to say. He's focused on financial services at Boosted AI and enhancing everyone's practice. Let's uh let's get into it.
SPEAKER_02You bet.
SPEAKER_00Josh, welcome to the harbor. Thank you for being with us today. It's fantastic to be here. You know, I I think for our listeners, why don't we kick it off with a bit of your career arc and kind of where you where you guys started?
SPEAKER_01We could go way, way back. Um I I think I usually like to start actually as early as almost 2008, 2009 when I first got um exposed to machine learning. Um I was working at this company called I have International Financial Data Services, something like that. And they had this task related to a mail room. We had a bunch of people coming in that were like open up pieces of mail and then sort of sorting it into different piles. And as an intern, um I helped build this system that could basically put an 80 or 90% accuracy figure out which piece of mail belonged in which pile and could almost you know reproduce the labor of something like 20 humans or something like that. So I was really lucky where I had this opportunity to see an example of machine learning have a very big business impact in the wild. Uh that led to me starting my first company, a company called Maluba, back in 2009. Um, we were basically like a Siri competitor. We powered LG, Samsung, Blackberry, Windows Mobile devices. Uh that company eventually got uh bought up by Microsoft. I did a stint at Bloomberg where I helped Bloomberg start some of their earliest machine learning groups and then ultimately started this company.
SPEAKER_00Oh, that's that's fascinating. And then going back to college, you went to University of Waterloo. Correct. Which, you know, you have some distinguished alums from University of Waterloo and Chamath and then Kevin O'Leary. But what's interesting about that is it's a co-op educational school, which we don't have a lot of in the U.S.
SPEAKER_03Just explain that. What do you mean by by co-op and how does that work?
SPEAKER_01Yeah, so we do four months of school, four months of work, four months of school, four months of work. And so by the time you graduate, you've got like two and a half years of work experience. You don't get a summer. Uh, you know, it's like constant back-to-back. Um, and then the flip side is you've also worked at like six different companies. And in addition to working with six different companies, you've been exposed to six different types of engineering cultures. Um, and I think the other thing that was really, really kind of cool and exciting about building back then um is just an hour away to the north, you had the University of Toronto, uh, where you had Jeffrey Hinton. Um, Jeffrey Hinton actually invented neural nets, and a lot of his lab to today ended up going off and being very important. So, like Ilya Suskovitz over at um OpenAI, I think he was a Jeffrey Hinton student. Um, ultimately, neural nets led to attention networks, which led to large language models. Uh, and so we had this, you know, awesome ability to recruit tech talent, uh, very young. I think the average age of my first company was like 23, 24, combined with the ability to get access to like really, really early AI talent in one of the areas in which you know modern AI was really born.
SPEAKER_00Oh, absolutely. And then was Bloomberg your first foray into the financial markets?
SPEAKER_01Yeah. Um, and really something I didn't expect. Um, you know, after Maluba, I had an opportunity at Google and Amazon and a few other like big tech companies. Um financial markets weren't even on my mind. Um, and my uh actually my current co-founder, Jonathan Durando, kind of convinced me to join Bloomberg because at the time they had basically no AI, very little AI. They had nascent AI, but they had very little integration and huge amounts of data. Uh, and so I started thinking a little bit about what's the kind of workflows we could build there, and just a chance to build something from scratch at a giant company was just a lot of fun and something I didn't see elsewhere.
SPEAKER_00And I know in its infancy from you know my previous career, when AI was first coming into the forefront of financial services, one of the interesting applications was being able to, you know, similar to your mailroom application, being able to read transcripts across thousands of companies on earnings reports, and you could score them based on what words are typically positive towards earnings, what's negative, and give each company an individual score. So instead of having an analyst read through thousands of papers every earnings report, it could be automated.
SPEAKER_01Yeah, a lot of how I've thought about the arc of AI has really been around like retrieval, for lack of a better word, right? Um and you know, before I joined Bloomberg, a lot of how you would interact with it, and you still do to this day, would be typing in specific codes, going to different mnemonics, different functions, different things like that. Um, a lot of the stack we ended up building out was around the idea of just building like a search engine that could basically search around the entire terminal and get you access to the kind of information you want. Um, and that was you know a pretty huge shift from where you were before that, where you had to actually manually know exactly where on the terminal to get the kind of information you cared about. And if you think about modern AI, it's basically taken that kind of retrieval capability and then just supercharged it where you can inject a lot more intelligence to how you do that retrieval. Um and so I sort of think a lot of the arc of AI has actually been fairly consistent in the core mission and how it operates. It's just the kinds of ways that can solve that mission have been getting exponentially better as time goes on.
SPEAKER_03At what point did you realize that this was going to be something of an idea into a concept that was gonna go mainstream? Because now it's used every day in our business, meaning AI. Um, you know, you said back in 2008 and 2009, you kind of had this idea in the mailroom. But so when did it shift to, okay, this could be mainline streamed?
SPEAKER_01In 2008, 2009, a lot of how I was thinking about it was like there was gonna be very specific industries that were gonna get uh revolutionized, for lack of a better word. And then I think around 2010, 2011, and this is you know the Siri era, um, we started to get this thesis that you could start to solve bigger, bigger problems. Um, I at the time thought of it more as like a better way to do Google search or a better way to do Bing or something like that. Um, and you know, giving you some capacity to kind of uh use other types of knowledge beyond just the web in order to answer a question. Um and so that was the first time I thought something would be mainstream in a way. I sort of had this view that we would have something like a search engine that could, you know, index data and get you something smart and intelligent. And Maluba built that mission for for quite a while. Um and then there was this interesting thing where I thought the tech got really good and adoption wasn't actually as good as I expected. Um so you know, sort of jump forward to the like GPT-3 era, and most people that were actually in the AI space, uh, we were really surprised it didn't just like take over the world. Um and it actually, like a lot of people in the AI space were quite scared of the GPT-3. And the reason was because it could do all these really advanced uh capabilities, it could answer questions, it could trick you into thinking it was a human. Um, but it also would lie and you know, say like pretty scary things. Um and I think what was really interesting in that era is a lot of us were really just indexing to like, you need to build the most intelligent AI. And here was this thing that felt so intelligent, it could pass the Turing test, it could trick you into thinking it was human. Um, and yet nobody adopted it. Uh, and so that led to a little bit, again, not just me, I think a lot of researchers, um, into a little bit of an era of like, okay, maybe AI is actually not as important as we thought. Um, and then the really key unlock, the thing that OpenAI figured out that I think shocked all of us in the space, is if you can take that intelligence, but you can filter it through, I almost like to call it a conscience. You know, you can call it a jail, you can uh call it realignment to human values. Uh, you know, the technical term would be reinforcement learning from human feedback. Um, but whatever you want to call it, they figured out like if we take that intelligence and then we narrow it down so that it doesn't lie, so that um it doesn't say that it's human, so that it um, you know, doesn't seem as scary, that's actually the big unlock. That's when it becomes a valuable product. Um, and so that completely caught me off guard. I it did not occur to most of us that you didn't just need intelligence. Um, you needed, for lack of a better word, a conscience, or you, you know, you needed something through an HR filter. And that one was when you were gonna get the big unlock. And so it was this kind of funny thing where we actually got really, really optimistic, you know, 2013 era, let's say, up to about 2017, 2018. Um then GPT came up. There's this huge acceleration of expectation around what I could do. That didn't lead to the kind of commercial opportunities a lot of us thought. Um, you know, I started thinking about other types of problems, and then ChatGPT comes out, shocks all of us, uh, and sort of takes over the world. Um, so that would be the kind of arc that I think a lot of researchers probably went through.
SPEAKER_00Oh, definitely. And I would say now the interesting thing is a lot of people think about ChatGPT, but they use it as it's as if it's just a Google search query. And there's so many other applications. Totally agree.
unknownYeah.
SPEAKER_03Isn't that because the sim it takes the simple mundane uh daily activity that you do, makes it better and quicker, which uh gives it the best use case for anybody who wants to use ChatGPT or anything like that. So that's kind of the hurdle is to get to the simple use case where it can be repetitive and easy to make your day easier and fulfill a normal task that would take, you know, maybe 15 minutes, you can do it in one minute or 30 seconds. I think that's why you've seen such great adoption of it.
SPEAKER_01Yeah, I I think for what it's worth, and maybe you know, shifting a little bit more into the current business, one of our theses is like there's a lot of workflows that involve you know quicker answers. But when you think about finance, one of the fantastic things about finance is there's a lot of repeatability in workflows. Like it may be a little different, but every time an earnings call comes out, I'm more or less following the same process, right? Um, I'm looking at guidance ahead of time. I'm looking at the press release. Um, the earnings call comes out. I want to see, you know, how the new guidance aligns what they said before. I maybe want to see how the sell side's talking about it a little bit differently. Um, you know, if there's the big CPI report that comes out, maybe I want to send off a newsletter to my users. Uh, if the Fed does a cut, um, then you know, maybe I want to make some shifts or I want to communicate that in a particular way. So in finance, there actually tends to be a lot of these workflows that kind of repeat um based on some event or something happening in the world. And if you can find those repeating workflows, do them automatically on behalf of the user, and then just deliver that work product to the user before the user even thinks about it. We think that starts to become a very compelling experience and kind of the next generation of where AI is going.
SPEAKER_00Right. And it's a it's kind of a living, breathing idea behind it where it's constantly doing this in the background.
SPEAKER_03Yeah, and so again, making businesses more efficient. So with Boosted AI, your your company that you've um kind of built towards financial services, what what's the catalyst you see from you know all of this data? I mean, in our business, we do an amazing job of having data. That's one thing we're really good at, is an overwhelming amount of data. So taking that data, simplifying it, putting out information for asset managers, I can see as a humongous benefit as it gives them the ability to take that data that would take forever to analyze and go through in a simplistic form every time it comes out. Um when you're building these models for folks now and you're marketing the strategy, what is the the pitch to an asset manager, let's say, on how it can help them with their daily workflow? Is that it? Is it just taking the concentration of all the data out there and putting it in one place?
SPEAKER_01So I like to describe our product as like an AI-powered terminal. But unlike the terminal where you have to come in and figure out how to get value from it, uh, this is a terminal that acts more like a junior analyst, um, a really excited junior analyst, where every day you come in and it pitches ideas to you on how to make your life better. So depending on what you do as an asset manager, you know, it might be suggesting here's different ways to think about your allocation, or here's different ways that uh, you know, some of your managers are operating in the world, um, or here's macro events that are occurring, um, or you know, maybe more micro. Um, here's an update you might want to make to, you know, such or such financial model. Um and then as it suggests ideas on how to make your life better and you interact with some of them and discount other ones, again, just like a junior analyst, it's gonna get better and better and more and more excited at figuring out the kinds of stuff that you like and that you consume. Um, and then once you have that sort of base experience, you know, all the power of something like a terminal, but with a system that brings the value to you, suddenly you can start to unlock huge amounts of different persona types. Um, you know, an asset manager is gonna have a bunch of workflows they do. Um if I'm a wealth manager, I maybe care a little bit more about client acquisition, I maybe care more about communication, I maybe care about risk mitigation. Um, if I am an analyst, I maybe care about um, you know, specific catalyst events happening with a company. I care about my financial model. Um, I care about tracking different types of uh you know conferences or things that are happening, you know, orthogonal, doing um peer comp extraction and things like that. All these different workflows can now be solved with this very simple experience where I just show up and the system does its best to find its way to make my life better. And as I interact with those things, uh it gets smarter figuring out what I care about.
SPEAKER_00So just really you know, consolidating all the data and giving you the exact stats that you want to get out from the longhand work of doing all the research individually. Exactly.
SPEAKER_03And then is there uh uh one of the cases we hear quite a bit is hallucinations inside the overall model? How secure and do we know the knowledge inside of what's being uh exported out of the data once it's found quickly? Scanning through thousands of documents and millions of words. Um what's that process look like on your end to make sure the stuff that's coming out is actually factual?
SPEAKER_01Yeah, so hallucinations are a really interesting topic. I think number one, you do need to build a people make very big decisions on the back of this. So you need to build a system where every fact that the system states um can be backed up with a specific logic chain and a specific document. So just right off the bat, the system can't state a fact without giving you some exact sourcing uh to exactly where that fact came from or what math it followed in order to give that to you. Um but believe it or not, that's actually not enough. And the reason it's not enough is because citations actually vary in their quality. Uh, let's say I asked a question like, what is the GDP of Germany or something like that? Um and the system quotes uh a Yahoo article or or a Reddit post. Um, well, it's not a hallucination in the classic machine learning definition because it's citing it back to a source, uh, but you as a manager are probably not going to be super excited uh about a Reddit post uh telling you what the GDP of Germany is. You probably want something a little more high quality. You want this source back to a fact set data set or something like that. Um what if on the flip side you said, how many clients uh really like um uh this new hat that such and such company released, right? How many people really like this new hat that such and company released? Well, all of a sudden, if it's like quoting it back to you know 10,000 Reddit posts that are all talking about how much they love the hat, suddenly Reddit becomes a fantastic source of information. Um so the sort of first principle is every number, every fact, uh the machine's not allowed to invent facts. It's not allowed to have an opinion. It has to source it back to some we call ground truth, some specific sourcing in the world. But then the second principle is the source matters depending on the context, depending on what the user cares about, what the user would call intent is, um, and you know what uh what what gets the count as high quality for that particular problem um is actually quite variable depending on the type of problem you're trying to solve.
SPEAKER_00That's that's really interesting. Right, I like that. I especially the analogy with Reddit.
SPEAKER_03Yeah.
SPEAKER_00You know, on one item it's not a great source, but on another it's a perfectly reliable source.
SPEAKER_03Yeah, because you see the interaction from people that are users and you'll have a pretty good sense of what's causing optimism for a product or what would be causing negative comments.
SPEAKER_00Right. The hallucinations have been uh a topic of late in the legal profession, yeah, which has been interesting, citing just made-up cases. Right?
SPEAKER_03How does that even happen? Someone's going in there and putting a fake case and it's pulling that data.
SPEAKER_00Well, like anything, uh you think you'd fact check. Can you trick the machines?
SPEAKER_03Is there a way are people out there trying to trick the machines? Have you guys looked at that at all? Is someone out there putting stuff in there and you know it's not factual?
SPEAKER_01You know, the hardest problem to solve is actually the machine getting a source that's not factual. And then, you know, again, you have to kind of teach the machine to figure out what good fact checking is. I actually forgot it's worth. I feel like it should be easier in the legal department. I feel like those guys shouldn't have any solutions, hallucinations, but um so they need boosted AI. Um, but you know, like in our space, right, if you have a politician uh quote a wrong number, um, and then the AI picks that up, um people will blame the the AI for quoting a wrong number in a way they maybe wouldn't blame a Google engine uh that just gives a blue link to a politician with the wrong information. Um and so I actually think the onus and requirement on fact-checking is is harder uh in this space. And and so, you know, when when we think about people tricking it, it's not tricking it in terms of like direct adversarial uh attacks. Those can happen, but more at a quant level. It's more like there's a lot of sources of bad information in the world. And I actually don't think that problem's getting easier as time goes on. It's getting harder. Um, and and so like building a system that can figure out good quality uh is actually really, really challenging to do.
SPEAKER_03Oh, I would imagine. So I'd like to twist into like a real case here. So we deal with financial advisors on a daily basis who have clients out there, and you know, one of the biggest objections they have on a on a daily basis is client acquisition. So if I'm using the boosted AI system, what what's a use case for client acquisition that could be helpful for for an advisor inside of your system?
SPEAKER_01Love that. Um, I'm gonna talk about one that's like really specific to me. Um, I'll just say every advisor I've ever met, you know, they they almost act out operate as small businesses. So people are gonna have different types of profiles of people they look for, but I'll talk about one I think is really, really cool. Um before I sold my first business, uh I didn't know any wealth advisors, um almost none. Um I got an intro from a friend. Right when I heard that my company was about to get sold and met one particular advisor. Then the company got sold and I got flooded. I think almost 200 different wealth advisors reached out to me.
SPEAKER_03And they got the squeeze. They got the juice. They want to squeeze.
SPEAKER_01Yep. Yep. Yeah. And I'm in the news, whatever, right? And you can imagine that the wealth advisor I'd already been talking to, they got the business. And I I've thought about that a lot because there's actually a lot of people out there where they own a business and it could be they've been owning a family business for a long time. It could be people that are entrepreneurs like me. It could be people that come from a bunch of different backgrounds where they actually have a bunch of illiquid assets. They see a liquidity event and all of a sudden they go from not needing a wealth advisor to needing a wealth advisor. You can use AI to find people like that. And I think it's actually a very rich vein where a lot of advisors are there's a few, I've met a few that have gotten really savvy about this. Um, but most people aren't really looking for that. And so just right off the bat, can I find business owners who haven't seen a liquidity event who might see someone in the future in my area? Um, can I think about what they care about? And can you help me come up with a pitch for that person? That would be a really good example of a workflow that could probably revolutionize uh certain advisors out there.
SPEAKER_00Oh, that's incredible. Yeah, there we go. I could see a lot of value in that too if you can be dialed in and focused.
SPEAKER_03You know, especially if you focus in on one industry, for example, you could really, you know, depending on where you live, industries are going to vary, obviously. But uh here where we're located in Metro Detroit, you know, the autos are huge and there's a lot of suppliers that are in that space. So I imagine that that that would be a really good search locally and then nationally. You think about all the companies that are are out there right now and in potential, you know, maybe they're Series A or wherever they might be, and all of a sudden they could have a liquidity event coming up. So that that makes a lot of sense to me from a client acquisition standpoint.
SPEAKER_00No, I think that's a great use case for advisors specifically.
SPEAKER_03And then inside of Boosted AI, that's something that can be built. And then I guess as a follow-up to that question I had is what about with your uh if you're a financial advisor and you have a current practice and you're trying to manage, you know, these 500 families or how many families you currently have, um, is there any use case tools that potentially could be um brought to the forefront with Boosted AI?
SPEAKER_01Yeah, 100%. So a lot of those tend to be more around client communication. Um, the most simple of which is if you have a newsletter, something like that, what stuff that's happening with the portfolio, what stuff that's happening uh with the overall space. Um, you know, I've seen a bajillion different variations of that. Uh there's one of our users where um Canadian, obviously, they they have a hockey podcast kind of thing where it'll be.
SPEAKER_03Now we're talking, Josh. We're getting into the good stuff here. Let's go. The beliefs, by the way. What's going on with the beliefs?
SPEAKER_01Um, luckily I was born and raised in in uh in Calgary, and and uh uh, you know, some I'm a Flames fan by uh by right of uh legal requirement. Um yeah, it's it's been tough to be a Canadian hockey fan for a while. Anyway, um, you know, so the the system will kind of produce uh almost like a little podcast talking about here's what's going on with hockey this week, uh, and then here's what's happening with the markets, and it'll kind of like interleave it with these little like cute um you know hockey analogies, right? We're moving to where the we're skating to where the puck is going and stuff like that. Um you could use it to keep track of different events that are happening in people's lives. Um, you know, what's the next best action? Who should I be reaching out to right now? What kind of communication should I be using? Um, I think the other thing is depending on how you run your practice, uh, there's a number of advisors where they maybe have like one or two different portfolio types, and then they're trying to just manage that allocation kind of across everyone. Um, but especially as you start to get to folks that focus more on the high net worth and ultra high net worth, there starts to be more customization per user. Um, and for what it's worth, I actually think it's a harder problem than a hedge fund. Like at a hedge fund, I've got one portfolio I'm running, right? One strategy. Maybe I'm a multiple strat, but you know, if I'm uh if I have a few hundred million in client assets um and you know, like a hundred clients powering that, um, all these guys want something a little bit special, a little unique, they want to feel so like they're do something you know just for them. That's really hard to manage. But if you've got a bunch of agents that are kind of tracking and understanding the world, those agents can then actually feed back to you as an advisor and suggest different actions you may want to take and also communication on a very you know customized per user level. So that would be two ways to sort of think about it is client communication, um, client engagement, um, and then actual management of the assets with the clients.
SPEAKER_03That seems incredibly efficient. I mean, one of the things we talk about a lot are in our world is the value that's provided is staying in front of people, talking to them, getting their time, and having a vehicle behind you that can help organize efficiently, uh almost in a predictability way, and what you need to do and how you go about it would be uh, I think hugely beneficial for anyone in the practice of of people, you know, dealing with people.
SPEAKER_00Right, because it's a it's a complex problem, especially when you have you know a hundred different types of clients. So you have multiple models you're running. In the hedge fund, you have one strategy with one aligned goal, where here you have many different clients with different types of goals. So, how do you manage for each of that individual client bucket?
SPEAKER_03So, how would your team work with someone in that scenario? So let's say I'm an RIA, I have 500 clients, I call up Boosted AI and say, hey, I have this issue of which uh I would like to create a program where I can, you know, proactively build out thought leadership as well as uh a touch, a relationship, you know, I guess management tool. If that's fair.
SPEAKER_01Yeah, so uh the way it works is when you first log on, um it actually does an interview with you. Uh it starts to talk to you. Um I like to use the analogy of the movie Her, but more and more people have never seen the movie Her. Um but if you ever do see it, there's an opening scene where it comes in and he talks to the OS and it starts asking some questions. Um our questions are maybe a little more uh tailored to finance than that than the movie, but it's the same kind of thing where it comes in, it does an interview, it it talks to you with voice. And then at the end of the interview, it tries to get to the point where it sort of understands what you do, what your workflows are, what you care about, what your goals are. Um, you know, it can start to integrate with a whole bunch of you know data that's unique to you. Um, and then it just starts suggesting ideas. Um, and if you like those ideas, you can tell it. If you don't like those ideas, you can tell it. Um, and then just automatically, as you actually click and use stuff, um, it's sort of signal to the system, oh, you know, he really liked this. He really so that you know our system comes out and it says positive reinforcement.
SPEAKER_03Yeah.
SPEAKER_01Yeah. Yeah. Yeah. It it actually the experience is a little bit like like YouTube in a way, right? So, you know, I go into YouTube and it shows me a bunch of videos and I click into videos, and the more I click into videos, the more it figures out the kind of videos I care about. It's the same concept but applied to work where it suggests work to you, and the more I click and engage with it, the smarter it gets at figuring out the kind of work that I like.
SPEAKER_00It's really cool. Absolutely.
SPEAKER_03And then how do you figure? And uh so I I get it in the sense with financial advisors how they can execute uh a more sophisticated client management tool that will does it work within their CRM or how does it ultimately get into their practice where it's its use case on a daily basis is effective?
SPEAKER_01Uh a lot of how we integrate right now tends to be more through email and custom docs and things like that. Um but I think actually integrating with different CRM systems is going to be very important for fully conquering the workflows. Um a lot of how we think about it for different things is what we call last mile. So there are certain workflows that we're you know fully like when it comes to like client acquisition, we've got access to everything we need in order to do that. All we really need to know is like who are you, what do you care about, what kind of clients do you care about? And then we can help to build out workflows that can do it for you automatically. Um, we have a very good system for you know tracking and understanding portfolios and things like that. Um and then for client communication, we can hook into email, get a lot of context from that. Um, but I think the next step is integrating with CRM and getting those kind of technologies in. And we're not there yet, but that's something we're building out this year.
SPEAKER_00Well, and and one thing you brought up earlier that I kind of want to touch on is speaking to the AI. Right. So voice and conversation interaction. What have you seen as far as adoption? Has that changed people's workflows being able to talk to it versus type?
SPEAKER_01Yeah. So when we think about voice, people usually index the conversations like we're having right now. Um, but there's another element I like to think about, which is radio. Radio is is is you know, listen, it's it's here only, right? And um classically, I'd wake up in the morning and I'd turn on Bloomberg Radio. Um, and that'd be how I'd start my day, you know, what's going on in the world. Um, I want to we wanted to get to the point where we had a system where it's just like that, but it's a radio that's just for you. So again, same kind of thing. It knows who you are, it knows the kind of stuff that you care about. You wake up, you press a button, and now you're walking around or you're driving to the office, and it's just updating you on like everything that's happening. Here's the macro events that are happening, or you know, here's interesting news that might matter for you, or here's things, here's communications you've gotten from different clients, or whatever it is. So you just wake up and you basically hit play, and these agents will basically just give you a summary of the important set of stuff. And they'll try to do in a somewhat fun, playful way. So it's not fully monotone, but that's that's the most common, easiest way to actually get started with voice is just like a consumption-based system. And the analogy there is just radio. Um, and then from there, when it actually comes to conversation, one of the really interesting things we noticed with user behavior is when you type, we tend to be very terse. You know, get me a good stock. Uh, help me optimize, right? Whatever it is. Um, and it's challenging for an AI system to perfectly do what you want when all you say is like, get me good stock, right? Um, you know, maybe you're a value investor. So good stock for you means uh a value play. Maybe you're more technical. In that case, you actually want something that's got you know good momentum or something like that. Uh, maybe you're someone who cares about defensibility of stock. So you want low volatility or whatever it is, right? In order for the system to get a set of ideas that make sense for you, it needs much richer context than what you care about. Um, and this is one of the really interesting things we noticed about voice is as soon as you make it conversational, people just naturally, even when they're talking to a machine, give way more context. And then because it's a real-time interaction, it's easier for the system to ask you questions very quickly to sort of elucidate what it is you're looking for. Um and so that ability to actually converse starts to feel a little bit more natural. It gives you the ability to input more systems. And then, you know, this is thinking more broadly, yeah, we're not here today, but we also think voice um can be a way to help sort of communicate urgency and create an emotional attachment to a product. Um and so, you know, when we think about stuff happening in the world, today the main way the user the system can reach out to you is it'll email you or it'll show you notification or something like that. Um, but you know, imagine that uh a politician comes in and I don't know, bans all semiconductor products coming to the company, coming to the country, and 80% of your portfolio is semiconductors, and your portfolio is about to see like a 90% drawdown. Um, if it was a human who had that information, they'd probably give me a call. Um, well, voice lets you mimic that kind of interaction too. If if you can get to the point where the AI understands you deeply enough that it knows when it can uh grab your attention, all of a sudden you could actually have the AI itself reaching out to you, calling you, and now you're having a conversation and getting information that way. Um that's the other way we think about it is actually like expanding the modality some ways that you actually engage with it and making it, for lack of a better word, almost more human. Um, and so circling back, radio consumption, conversations get to workflows, and then long-term actually having it reach out to you.
SPEAKER_00I think that's fascinating because I would say if somebody's explaining something to you sitting across the table, they're going to be much more verbose. Where when they're typing, I think we've been preconditioned to keep it as short as possible.
SPEAKER_03I I'm I'm a believer in I like communicating via type sometimes though, because then there's no gray area. It's like it is what it is. Um, but I understand that you can create catch some sort of experience when you have the conversation face to face or when you have it radio.
SPEAKER_00Well, you could almost be having the stream of consciousness. Right. Right. When you try and put that into words, it's gonna be much different. Where when you're actually having the conversation, you can just have the ideas flowing and then it's asking follow-up questions. Frankly, Sean, we're we're gonna be out of a uh podcast hosting job soon.
SPEAKER_03Oh no. So the world can't exist without in the harbor. Come on. Uh so help me understand what's next. So we've talked, I mean, I I can't believe how much in one year, or maybe a year and a half, we've implemented these types of things into our practice, which were relatively mundane and not as advanced as what you're talking about. But what's next for uh boosted AI in the direction of this uh company you've built?
SPEAKER_01Um, okay, that that's a broad question. Maybe what I'll actually do is I'll actually gonna do something crazy here. I'll look three, four years out. Uh, where do we see the future just of the world? Yeah. Um and then I'll go from where do we see the future of the world down to where do we see our place in that world? And then from there I'll go to here's what's happening in the next three to six months. Um, you know, and and this actually builds really well on on the question you just guys just had. Are we gonna have podcasts in five years? Um, I actually think the answer to that is uh a 100% definite yes. Um the way that I kind of think about things is when you look at a lot of say white-collar work, um, you can break it into parts. There's a lot of work where it's just analysis on text or analysis on numerical data. Um, that's not 100% going away, but I'd say that's like 80, 90% going away. Like the vast majority of that is going to get switched over to AI just sort of processing and doing things for you automatically in the world. Um, there's a lot of processes right now that relate to creating text and and creating numerical uh analysis. That's also going to go away. A lot of that's going to go into the hands of AI. Um, but when we start thinking about, okay, well, in a world where you had perfect AI, whatever you want to call that, what are the human rule? Um, I actually think the new bottleneck starts to become human-to-human interactions, right? If we sort of zoom the lens out here a little bit, uh, we've had automobiles for a fairly long time now, but we all like to watch, or many of us still like to watch, the 100-meter dash. Uh, we've had chess engines that perform better than um human chess players since the 90s. A lot of us still love to watch professional chess players. Uh, we've had opening and closing doors for, you know, decades, but if you go to a high-end hotel or high-end restaurant, there still tends to be a dorm in there, right? And it's because humans like other humans. We have a competitive advantage in dealing with ourselves. Um, and the corollary of that is as long as there exists sets of information uh that um are only in other humans' brains, then humans will probably be the best way to get that information. So podcasts get to survive because I'm probably going to tell you guys a lot more than I would tell a robot, most likely. Um, but also, you know, I think wealth advisors become in some ways more important in this new world because they provide an emotional support. Uh, there's that, you know, human-to-human connection that is just harder to reproduce uh artificially. Um and when you think about analysts, like professional analysts, a lot of the parts of the job that relate to like doing management calls, going to conferences, meeting with people, those start to become very important jobs. The flip side is just about everything else, I think, starts to get built and consumed and done by AI. Um, and then that's kind of the long term, you know, three, four, or five years out. Um, I actually think for what it's worth, it's going to humanize most of our jobs. Uh, most of us will be spending most of our time interacting with other humans. Um and then when we start thinking about things, you know, kind of a little bit more uh in the short term. Um classically, you had products like the Bloomberg Terminal or uh like SP Cap IQ that were these really big professional products uh that you had to pay a lot of money to get. Um and I'd say they kind of had three classic modes. They had a data mode, they had a technology mode, and they had a distribution mode. Um, well, on the data side, you know, we can extract data as well, or I mean somebody's arguably better, um, than most of the professional human labelers at this point. I think there's something like a 99.9% alignment. Um, and so, you know, I think SP has something like 7,000 humans whose full-time job is to like go out and extract numbers from text. Like that that's going away. And and the the flip side to that is that means there's going to be massive downward margins on the cost of producing data. Once you can produce data cheaper, buying it also becomes cheaper, presumably. Um, you also have these systems where I think Bloomberg had like 15,000 functions, something like that. Uh, they've got whole engineering teams that are building and maintaining the terminal. Um, our system is increasingly gaining the ability to basically generate, like, you know, you've got this idea of an agent, but now these agents can actually generate UI. And they can generate UI on the fly, exactly how you think about things. Um, and so all of a sudden, those 15,000 agents that, you know, cost like a billion dollars plus a year in human engineering to build and maintain, that's going away too. Um, and then, you know, distribution, that's the toughest mode to crack. But I think if you've got something that's better and cheaper, it's just a matter of time before the distribution mode gets cracked too. Uh, and so in the absolute short term, you know, I use the idea of an AI-powered terminal, that's not tongue-in-cheek. I see these systems being able to very quickly uh unlock new patterns and new value that older systems can't, but very quickly as their coding capabilities get better, um, completely displace a lot of systems that already exist today. Um I sort of think the terminals of the workstations of the future are gonna be fundamentally different than the types of products that exist today. And even better, I think that the gap between you know your top-tier professional hedge funds, your wealth advisors, and retail, I think they're all going to close a bit. Um, and all of a sudden you're gonna get swaths of the population that had no access to this kind of stuff, getting like professional grade capabilities.
SPEAKER_00You know, one thing I'd be curious about, what do you feel like for the evolution of the workforce, just in the sense that a lot of these data-focused careers? So if you look at finance as a base or the law profession as a base, you'd get a lot of your training through going through the menial tasks of analyst or you know, the junior associate at the law firm looking up case law. How do you feel the training of the workforce evolves with a lot of those tasks being replaced? Made efficient.
SPEAKER_01Yeah. You know, what's really interesting, um a lot of the old tasks are gonna get replaced, but there's certain things where, even though they're numerical, uh, I don't think an AI can can actually truly replace. And it's things related to um ambiguity or where you're making a risk assessment. I guess to be like very, very specific here, uh, one of the workflows we look at a lot is investment banking. And investment banking is kind of interesting because you would think, oh, well, building a financial model is easy, Josh. You know, there's a ground truth, right? But actually, there's an element of storytelling. If I'm trying to help a company do a$300 billion bond issue. Um, I need to make sure that the documents I'm putting together uh don't make it look like the company's gonna go bankrupt in a month, right? Um, I probably wouldn't take on the deal if I thought I was gonna go bankrupt in a month, but you know, generally speaking, and this is assuming gap and you know, assuming I'm following like perfectly reasonable accounting standards, uh, there's different interpretations of the underlying numbers. There's different interpretations of what's happening. There's a lot of like fuzzy things, right? Like, you know, maybe the company has like eight or nine contracts, uh, some of them are in contractual view, but they haven't closed yet. How much do I want to consider that? What do I want to consider part of my like active deal pipeline? What do I not? Um, you know, maybe the company had uh some bad debt in the past that's been removed. Um, how much do I want to think about that in terms of their credit rating? Like there's a lot of these very nuanced decisions where an AI actually can't make the decision because there's a risk, right? Um, you could take one extreme where it's like, oh yeah, like everything's perfect. You know, the company has no problems, it has no bad history. Well, that's probably getting to the point where it's almost fraudulent, right? Um, or you could have a very pedantic AI where it's like, oh, the company's filled with problems, you should not give it any money. Well, then at that point, like, why am I hiring an investment banker? Uh, they just blew up my deal, right? Um, and so there's some middle ground there between perfectly rosy story and perfectly nevy story that is morally and legally acceptable. Um, and you can't let the AI make that moral decision for you. You need to understand what level of uh ambiguity you as the investment banker are comfortable with and the CEO as a user is comfortable with. Um, and for what it's worth, I think this will be true across most professions. When I'm doing um, you know, corporate contracts uh with very large clients, um I've got my lawyer, they've got their lawyer, but there's a lot of back and forth at the business level. And you know, the lawyers will go out and they'll duke it out, and then they'll come back with some clause and I'll say, no, that's a ridiculous clause. I don't give a crap about that. Um, or no, this is a clause that we're gonna die a hill, we're gonna, you know, spend time on the hill dying over, right? Um, and so there's that element of like ambiguity, let's say, that has to be worked out. And then the other thing that you like a good lawyer, at least a good corporate lawyer at least, will give me is advice. Hey, this firm, they really, really, really care about, you know, X, Y, Z clause. But every time we've pushed them on, you know, this other clause, they've caped 100% of the time. So why don't you come in and you know suggest a swap here, right? That sort of strategy of a deal, again, an AI can give context, it can be helpful, but you don't want the AI making the uh business trade-off decisions. That probably needs to be in the hands of a human. Um, and so I think the workforce is going to change. You know, I'm I'm not someone who thinks that there will be no unemployment that comes from this. I think that a lot of the workforce today is not equipped for the new world that we're headed into. Um, but at least for now, uh, I think there are gonna be some big, very important bottlenecks that only humans can fill in, even if we get to perfect AI. Um, and there's a way to sort of get yourself there. Um, you know, the comment I'd make is the more your job involves human-to-human interaction and um legally or morally ambiguous decision, the more likely uh an AI is not gonna be able to replace you.
SPEAKER_00Right. That makes perfect sense. You know, one thing as far as uh pushbacks on full-scale adoption within a company, for instance, I think would be around the security, specifically when you have non-material or non-material public material information. How do you protect against that? Are firms householding their own servers?
SPEAKER_01Yeah. Um interestingly enough, like we're we're SOC2 GDPR, et cetera compliant. Every one of our users um tends to be uh isolated from every other user. And so for like even some of the largest hedge funds, they've been more or less willing to um, you know, let us house and and uh uh store and and sort of protect that data. Um usually when we're doing uh a private cloud or something like that, that tends to be with a bank um or or some other like largely conservative institution. So we are seeing a little bit of that. Um I think when Chatter C first came out, there's a lot of people that were worried about LMs like even seeing uh their underlying data. That worry has lessened in the market. There's a lot more willingness to just get the best LM in the market um at the cost of some of you know some historic privacy concerns. Um, but uh there are still some that are really thinking about this in terms of like how's local solutions.
SPEAKER_03And so as you build out uh booze today, and one one question I have is is more on the job front. Um, and I know we discussed it briefly, but what skills and qualities are you looking for in your organization for someone upcoming uh that that is is a need right now?
SPEAKER_01Oh, I love that. Um I'm I'm very engineering focused and in my background, so I'll just talk about how I think about engineering. Um we changed our interview style, actually. So we we classically would have done a very similar interview process to like a Google or a Meta, um, where you come in, you do like six tech interviews, you're doing a bunch of tech uh uh tech queries, you know, we bring you in, we do architecture diagrams, et cetera. Um we haven't completely gotten rid of that, but I I really quickly figured out there was a negative selection bias because we want people who are AI forward. We want people who are really excited about AI. And if you are interviewing people and saying you're not allowed to use AI as part of your interview process, you're not going to get those people. Um, so we really took a different tract where we do almost like an open book style interview now, um, where please bring the absolute best LM to bear as you're doing this interview with us. Um, and then as you're doing it, you know, show us how you interact with the LM, show us how you generate, show us how you vibe code, if you will, or tell us how you, you know, actually develop stuff. But then what we're gonna do is okay, explain to me what it did here. Explain to me what it did there. Explain to me where the machine might go wrong or where it may go differently. Um so the way I kind of like to describe it is like we're very actively looking for people that can incorporate AI into their workflow and are excited about exploring uh new types of AI to start to enhance things. Um and this can kind of extend across different areas of the company. Um from a CRM stand, uh there's this uh software we use that uh every every time that we do a call with a client, um we'll try to record it, if they'll let us, um, and then we'll summarize what happened with the meeting, and then we have the ability to like take that learning and extract it out. So in the old world, when we were trying to build product, we'd do user interviews and you had you know humans coming in, doing what we call customer interview snapshots, aggregating it, storing it. Now we can do a lot of that like fully automatically. Um it gives us the ability from the sales cycle to sort of understand what's happening with pitches that are going well, what's happening with pitches that are going poorly. Um, and so you know, for us, we really think every role should be touched and impacted by this. And we're trying to find people that are actively exploring that already, excited about that. Um, but also, you know, understand its limitations and can work around the limitations, if that makes sense.
SPEAKER_03So potentially someone with an engineering background that you you can utilize AI to the benefit of your organization, but also have the skill set to incorporate that into their own knowledge and ultimately be beneficial for you guys. I guess I'm thinking more of this of uh of of like you know, my own kids when they're studying in school and what they're doing. Um, you know, the ability to get information through AI is so quick. And I'm I'm kind of of the standpoint like use it, get good at it, understand it because it's coming.
SPEAKER_01Yeah, yeah. I have uh a five-year-old about turn six and a nine-year-old. Um, and like any, you know, five and nine-year-old, they ask me questions all the time. Um, and uh I'll do my best to answer it. Um, but increasingly I'll like open up uh either you know chat GPT or if I can, sometimes alpha if it's relevant. Um Alpha Star product. Um and I'll just have my kids talk to it. Um and and it's just wild, you know, just like the like for us, we get exhausted, right? Maybe two hours of my kid asking me why questions, try to keep up with it. Um, but I you know, I don't have infinite patience in the way that an AI can literally just two hours straight question.
SPEAKER_02It's a babysitting tool, too. Maybe that's what we should create. I love it.
SPEAKER_00That's not a bad plan.
SPEAKER_02It's amazing. So good.
SPEAKER_00I I had my niece over. We put on uh Miss Rachel last night, so I was watching that. But I mean, if they could have full conversations with AI, yeah.
SPEAKER_01Uh another hint if you've got young kids, um my kids love me coming up with bedtime stories. Um, and same kind of thing. I've got this like chapter book at this point of prompts where uh I've I've got a few characters, you know. I made the characters came up with them, but I have the AI generated chapter, which I then read to my daughter uh to bed every night.
SPEAKER_03Same thing. I I I gotta I I do the same thing you do with uh characters creation. I like this, I gotta build this out now. My kids are getting a little too old for that, though.
SPEAKER_00So I I mean the the interesting thing, AI is gonna permeate every industry, every career. Yep. So what would you and this is asking for your personal opinion, Josh, but uh, you know, young students in college right now that are looking to optimize, regardless of which industry they choose, what do you think they should focus on as far as the AI capabilities and tools that are available to them right now?
SPEAKER_01Um, I mean, I'm gonna have a very finance and engineering uh biased perspective. I'm gonna be my little myopic, right? But I I'd say within my space, there's a few high-level comments I could make. Um first of all, uh new engineering grads, you new STEM grads have a once-in-a-lifetime opportunity to be the most productive cohort by far, right? Like if you become very, very, very good at learning how to integrate AI systems into your workflows, just given the pace of AI, uh, you have the ability to be more productive than engineers who have been doing this for 10, 20 years. Um, and and a sort of ironic thing here is a lot of big companies have indexed the other way. They're going, well, I want more senior people, et cetera. But it's actually the like highly talented new grads who are AI literate, who know how to incorporate this in, who have the ability to be like exponentially more productive than the peers above them. So um just right off the bat, be good at that and take advantage of it. Um, and then the other question I often get from like young engineers in particular is um, okay, how uh there's a lot of like open source projects and things like that that you can go out and and uh and interact with. Um, if you're in the financial markets um and you want to start a career there, I either you know trying to build out a wealth advisor or trying to be an investment advisor at a professional uh hedge fund, there's a lot of really hard problems in starting out that you can solve with AI. Um, when you're first starting off as an IRA, you know, I think something like 80% uh of new seeds, generally speaking, fail because it's really hard to do client acquisition. Yeah. Well, AI, as of right now, most people are not using AI for client acquisition. So if you're one of the few people that are, you have a massive competitive advantage, both over the older cohorts and other people of your age. Um if you're using it to do research, then you can produce institutional quality research in a way where people that are using older tools still can't, right? Um, and then similarly, if if you want to go the banker route or if you want to go the investment advisor route, sorry, the um the analyst route, pardon me, same kind of thing. What are the kinds of workflows that people are still doing in an advocated way? And can I leapfrog those people and find ways to be very, very productive? So I'm actually really excited for young people this day. That's a very controversial opinion. Um, but this is a generation that's growing up with this tech where it's been perfused. Um, and uh if you if you do a good enough job of it, it gives you a chance to actually massively accelerate um older some of your older cohorts. And I I think that the challenge that's happening is just there's a lot of that generation that's actually not taking advantage of that fact. Um, but the ones that are are having very good times.
SPEAKER_00Well, and you look at the rate of change throughout their lives for the younger generation, I would imagine they're more open-minded at the possibilities and more curious as a result.
SPEAKER_03Well, more willing to use it, right?
SPEAKER_00That too.
SPEAKER_03Yeah. And they're it's in their daily workflow. If they don't know an answer to something, they're certainly gonna ask the professional, whichever system they're using. Um, I experienced that on a daily basis in my household. They're definitely better at it than we are, and not even close. This is fascinating. I could talk about this all day with you. I feel like I'm learning so much. And um, I think we've taken a bit of your time today, so appreciate the information. And and uh do you want to finalize this with rapid fire or what are you thinking, Jay?
SPEAKER_00Yes, so Josh, a lot of our episodes we do a rapid fire QA. Uh you can you can answer as briefly or as uh verbose as as you'd like to, but I think we'll uh we'll kick it off. I know you said you're a Flames fan, but favorite Toronto sports team?
SPEAKER_01Well, it's gotta be Raptors, right? Um, I mean, come on. Uh you gotta go with the winner. Uh I guess I maybe I shouldn't be picking on Blue Jays too bad. Uh more Raptors than than Blue Jays, though, I think.
SPEAKER_03They're certainly doing great, that's for sure. Um, what what do you think is an underrated AI use case that uh you know advisors in our world that we deal with are not using yet or talking about?
SPEAKER_01I mean, definitely acquisition. Like, and and it and it's it's wild because you can see such a huge bifurcation between people that are doing it and that are not.
SPEAKER_00All right. I like this one. So you're focused all day, you know, very engineering mindset. How do you relax and reset at the end of a long work week?
SPEAKER_01Oh man, what does that even look like? Um, I mean, you know, I I I got two kids, and uh I I think for me, I I try to make sure uh there's an element of like living vicariously through your children, and and so for me it's just separating that line uh just trying to make sure that there is space to spend time with my kids. Uh and then the kind of stuff I do with my kids, I mean, that's gonna be super nerdy stuff. Um, you know, we've do some like Warhammer painting. Um, you know, I've introduced both of them to a bunch of board games. My son's getting quite good at chess, actually starting to beat me, which is scary. Um, so for me, it's it's time with my family.
SPEAKER_00Oh, that's great.
SPEAKER_03That is a good one. Uh uh with the you being a Flames fan, I gotta ask you, who's your all-time favorite Flames player? Is it Landy McDonald? Or you're probably a little young for Landy McDonald.
SPEAKER_01Uh, it would probably be Aginla. And and the reason is because that's when the red mile happened. That's when I was in high school. That's when things really started to take off. So maybe Kiffersoffer Againla, like those were the two players that really brought the flames to the forefront in my childhood.
SPEAKER_03Yeah, Jerome's son is turning it up for he put in the world junior team for Canada, TJ, this year. He had a great, great run. And and then I gotta stay on the hockey front here for the Olympics. It's coming up. Team Canada, probably the favorite going in. Um, where do you expect them to uh medal in Milan?
SPEAKER_01Gotta be gold, right? I mean, it's it's gonna it's gonna go all the way. I I uh I don't know. Gold or bust?
SPEAKER_03They're so good. I mean they're they're top line. McDavid uh McKinnon, two pretty good players. And then they can throw anybody else on the wing they want, but it's gonna be a it's gonna be a great Olympics, that's for sure.
SPEAKER_00All right, and lastly, where can our audience find you? Is there is it boosted.ai is the the company website. Any other way to reach out to you if there's any curiosity about how they could start implementing into their own practices?
SPEAKER_01Yeah, we've got a LinkedIn page as well, but boosted.ai it's got to contact us. You can sign up on the website too. Um, that's the easiest way to get in.
SPEAKER_00Excellent. And I'll put that in the show notes so it'll be listed there as well.
SPEAKER_03What a fascinating conversation. Thank you so much for taking the time today. It was great to uh chat with you about the future of financial planning inside of uh using AI as well as all the other use cases that you've uh built outside the business. And thanks for the education.
Podcasts we love
Check out these other fine podcasts recommended by us, not an algorithm.
