AI Rebels

AI Is Getting Cheaper. Human Judgment Is Getting More Valuable ft. Daniel Yoo, Finmate AI

• Jacob and Spencer • Season 4 • Episode 9

Share episode

AI isn’t magic, and pretending it is may be the biggest risk in tech right now. A former financial advisor turned AI founder argues that most AI products are already becoming commodities, with little defensible moat and wildly inflated valuations. He explains why real value comes not from “agentic” hype, but from deep domain expertise, careful guardrails, and knowing what not to automate. In regulated industries like finance, trust, liability, and human judgment still matter more than raw intelligence. The conversation cuts through bubble thinking, warning that many AI startups won’t survive once capital dries up. In the end, AI raises the floor of expertise—but the ceiling still belongs to humans who can synthesize, decide, and take responsibility.

https://finmate.ai/

hello everybody welcome to another episode of the AI Rebels podcast as always I'm your co host Jacob and I am your other co host Spencer we are here today with Daniel Yoo uh he is the founder of Finmate AI um Daniel go ahead and tell us a little bit about your background and and what LED you to leave your previous background and pursue a a career in the AI world sure uh so my background is I've been a licensed financial advisor for about seven years working in Silicon Valley Bay Area um obviously most of my clients were folks in the tech world um left uh TD Ameritrade I was managing about $800 million uh for TD um about a year after the Schwab merger I had been doing a master's program at Johns Hopkins in applied economics using AI to do stock price prediction modeling my thesis advisor was a guy in FINRA who suggested I put my licenses in the you holding program that they created and go try something else so I pivoted out into tech wow awesome that's quite the quite the pivot that's interesting so okay so predicting the stock market do you think AI is to the point that it can accurately predict the market that's an interesting question I don't know about the current state because back then this was about 2022 era large language models haven't quite taken off in the way that it had and so a lot of it was numerical time series predictions right uh at least when I was studying it was not doing a very good job so we were trying different ML models but uh Arima uh auto regression integrated with a moving average uh that very old just Standard Model uh was out competing any ML or AI we tried so at the time uh it wasn't really doing anything and then now with the explosion of LLMs I think most of the focus on the research has shifted over to language models not time series yeah yeah so tell us a little bit about finmate uh what is the what is the I guess the the mission statement what is the problem we're solving yeah sure well um obviously AI is a very rapidly changing space and so initially when I got started um this was back in the days of GPT2 um da Vinci Ada that kind of thing the good old days um yeah the good old days yeah with like what was like 3,000 tokens or something very very minimal uh I realized that um one of my biggest pet peeves of being a financial advisor is a lot of the admin background work right and so you know it was a very heavily commission based business and industry and so if I wasn't going out and talking to people I wasn't really making money and so the back office stuff kind of took away from building the business right and so initially finmate was created to help alleviate a lot of the admin craft and so we initially got started by creating a notetaker for financial advisors we were the first AI notetaker for advisors we launched in 23 May and the reason we wanted to create a specific one was there was the generic ones out there already but I felt that the notes they were creating at the time with the limited LLM processing power was not good enough obviously things have changed and the scene has exploded and there's 30 of these things now focused just on like the advisor space as well and so we realized hey this is kind of a commodity at this point so we recently slashed costs all the way down um to commodity levels and then now we are moving on to creating custom agentic AI for uh small to midsize financial firms uh generally focusing on firms that don't have internal tech teams to be able to develop these things on their own hmm interesting so I'm always something I'm always interested in with uh any financial technologies um the regulatory side um is there any type of regulatory overhead that you guys have to take on or is it all on your clients like hey you know you use this like you're agreeing that you're gonna use it in a way that is compliant with with FINRA with uh whatever other regulations exist yeah so to be Frank um I feel like there's not that much regulation right uh I really think that there'd be more right it's just that FINRA FINRA is reactive it looks like more than yeah proactive right and so what they'll generally do is they'll sit on the sidelines until they can just hammer somebody and so haha I see you don't wanna be that you're trying to avoid right right and so what we've done exactly yes yes yes um what we've done is unlike basically everyone that's come after us because you know I'm an advisor I have kind of the advisor viewpoint is we do full data deletion of everything because we don't want to be considered books and records down the road right we don't keep any data yeah we don't access CRM data things like that um but everyone else has followed us because they're all tech founded data is their lifeblood and so they they take all that data right and so you know right one of our competitors recently purchased another company that has 14 years of advisor note data right now wow on the front side they'll say oh we don't use any advisor data for training but then they just spent millions of dollars buying this other company with 14 years of advisor note data right so yeah you know what do you need that for right right right right and so the question is you know exactly like oh interesting why did you buy that you know yeah um I guess so is that is that a challenge for you then because you do delete the data like does that make you concerned that you're at a disadvantage compared to for example this company who just bought 14 years of data potentially train on potentially potentially but I think um our bet has been that the rate at which AI the base model improves will far outstrip anything that we can do on a training level on top on our end right um you know you've seen the growth right it's like every six months it it the AI capacity doubles um yeah and so I just didn't feel like we needed to take that risk um yeah now where where I was wrong was I would I was assuming that the regulatory um body would come down harder on data access management and data retention policies and the fact that they haven't and all of these kind of tech outsiders were just able to come in and run roughshod over the entire industry um that's honestly taken me a little bit of a back because the financial industry is not known for the alacrity hahaha yeah that's that was the that's why it was the very first question I asked because uh uh so I have a dad who does who who worked in the cyber security sector for a long time dealing with a lot of you know financial institutions um and then I briefly worked just doing tech support for a for a company that provided marketing supply uh marketing support to um Financial Advisors and then my first developer job was at uh a fintech and so I I became very familiar with with you know the the various regulations and legalities that must be contended with ha ha ha yeah um yeah so I've been fascinated to hear how that's going so it's it's even more fascinating to hear that there's there's not really anything going at all in terms of in terms of regulatory oversight there's really not yeah yeah do you have released a memo I think recent couple months back yeah they're basically like here's some guidelines um but I think they're really taking the well we'll wait until someone makes a mistake and then we'll find them yeah come down hard interesting so I'm kind of fascinated cause you were it sounds like one of if not the first advisor focused note taking AI based application the first yes the first what was the initial response when you went to market skepticism yeah I'm sure yeah yeah so for the advisors um because this was so new and advisors themselves are pretty reluctant to try new things um and we've gone down the road of not fundraising and so we couldn't buy out some of the influencers in the industry oh interesting right and so we'll we'll see if that was a mistake but I I personally believe that a lot of these valuations are absolutely insane and so this is part of the reason we didn't go for fundraising but the the two that came right after us they both fundraised about 40 mil sold about 40% of the companies right and so their implied valuation is you know already at 100 mil plus and they have to hit that right and so we were not able to kind of just do this marketing blitz um to get ahead of the competition because again my Assumption was that this would move a lot slower which was where I was wrong right yeah it has been a wild ride that's for sure yeah yeah whirlwind okay so you said you recently made a decision to essentially price the note taking app as a commodity yeah and now you're where what stage are you at is it gonna be called finmate the agenic platform alright so it's finmate AI Notetaker is our note taking product and then it's uh finmate assistant is the agent assistant yeah got it and we're also we'll also be launching a consultancy program where we do like a three to six month review of the financial firm's operations and make suggestions on how to optimize using AI um I think because of the speed of which AI is developing a lot of less technical people are feeling a little bit of drift as to what they ought to be doing and so yeah um because we haven't fundraised we have the latitude to be able to do something a little bit less scalable like consultancy custom development because we don't have yeah the investor gun to our head um yeah so that's why we're doing what we're doing oh that's awesome I feel like that market there's just such a need like I think of my company right now that I work for in asuni like this has been a huge focus and just cause you have to shift the entire company's mindset to start thinking this way like okay how do what problems are there that can be remedied with AI and it's so difficult so to have an outside perspective come in and be able to just kind of shed light on that sounds very appealing haha right right and we have you know both experiences right we have experience building AI technology clearly and we have the financial advisory background and so I think that's been appealing for a number of folks in the industry yeah yeah so I'd be interested to hear um I cause in in providing some marketing support to financial advisors I got to figure out that you know about one in 10 financial advisors is super forward looking really excited about all technology yep um have you been able to find financial advisors like that and I'm curious what else are they doing that you see as unique compared to a pre AI world hmm if there is anything yeah I think the risk appetite has gone up in that whereas a lot of people would have been more reluctant to try new tech because just everyone is talking about AI I think a lot of advisors are thinking hey I need to use this but I just don't know how which is why they're turning to folks like me and asking like what do I do yeah yeah that's interesting so when and maybe you haven't gotten there yet which is fine if that's the answer but I'm curious when you go into a company what is it that you look for for like the low hanging fruit like what is it that what is ripe for AI yeah great question um even in the beginning you know when I was talking about this uh I I said the major candidates are things that were automatable even in the pre AI environment right so a lot of rote repetitive tasks the only difference now is AI can take it like one step further instead of like hard defining like repetitive task you can let a little bit of slack in there um and have it kind of do a category of tasks instead of just having hard coded things obviously hard coding is still more efficient and effective um but I think AI lets you be a little bit fuzzier and yeah that's pretty much it that's the low hanging fruit haha yeah that's awesome so one thing I've heard and maybe this isn't even true but I'm gonna say it anyway so take it with a grain of salt uh is that with this a this craze with agentics right everything's agentic right now in AI everyone wants agents everyone is it being over applied in the sense that are we trying to apply agentics to problems that we don't that don't need agents for now I'm going to say yes um as as we've been developing out these agentics what we realized is um AI lets you be fuzzy and you as the agent designer decide what level of fuzzy if that makes sense yeah right yeah and so on the one hand you have very hard to find if you know a input then do B process and then process out C output right just one linear trigger input output done right this is like the old school automation you have like Zappier things like that right on although on the other hand you just throw it all into an MCP and then just let the AI figure out what it should do and then just do it right and then you have kind of somewhere in the middle where you're pre processing pre chewing some of the thinking already and you have it in set defined terms and then you let the AI handle it from some halfway digested point right and as you determine across the spectrum what you ought to be doing I think the temptation for a lot of people who are less techy is to just go full fuzz right right just oh we have MCP just let MCP handle it all and then the issue is this is very very inconsistent right and then you have all the way at the other end which is like well this is not really AI at this point right um and so I think there is that issue where because everyone is pursuing agtech and people are promising the world um people are just dumping it all into an MCP and just hoping it works out um so yeah I think it is being over applied the way we are approaching it is again because we're doing the like the consultation beforehand where we're identifying key things that need to be automated we're applying AI more towards the end of the process as opposed to having AI control the whole thing um so that we can get better measurable consistent output yeah yeah okay I like thinking of AI automation for myself as as automating instead of work flows necessarily but almost thought processes um and that isn't exactly the right way to put it but that's the closest thing I can describe it to right processes where it's like OK you know I have to sit through and think through this problem um so for example you know like writing some some front end code cause I'm a I'm yeah a developer right yeah um I can sit down and figure out you know how I want to build write the card layout and then you know how I want to style the headlines and all that or I can tell AI like hey um I'm designing this think about these four things and build it out for me and and and when when you frame it that way I think it becomes a lot more attractive to people because the the common um when when people are resistant AI from what I see and from what I've talked to uh from when I've talked to people about it um they they don't understand how to use it like like they they think that they that it's a task machine and not a thought machine does that make sense right it's like hey go write this code for me and it'll go do it but it has no idea what you actually want um yeah anyways yeah the more input you give it and the more defined you make a task or when you're leading it down a thought chain I think that's where AI can shine um but I think people are hoping for a magic box where you just say something and it just does it for you and unfortunately it really really doesn't haha so so for example even in like front end development like so when we're doing front end development as well you know we're noticing like it doesn't know how to like create a clean style sheet right so some elements it'll apply to that style elements sheets uh you know assign it correctly to that style sheet and then some elements it just hardcodes it in and so now you have like two classes where it should be one class and it yeah it's it's a toddler ha ha ha ha yeah a really smart one but yeah yeah it needs a lot of guardrails so yeah what would you say are the biggest pitfalls for a Gentex that you've seen or AI I guess in general I think it's a lot less capable than people think and so it takes a lot of development to actually put in the proper guardrails to make the processing happen properly yeah and I think there's a lot of people that are positioning themselves as AI experts that are very very not technical um and that's leading to and inflation of expectations I think um yeah where yeah again it's go it goes back to the whole magic box concept like oh you just put in you know whatever command and it just the magic machine does the magic thing and it just does everything for you and your job's done you don't need to work yeah but you know then the issue that we have is like we're trying to you know manage expectations right like hey like this is still tech it's not magic you know right it doesn't actually understand you hahaha yeah and so yeah we have to do a lot of work to do a lot of the pre chewing and the preprocessing to make it work fast enough efficient enough consistent enough but you know I think there's just too many charlatans now because it's a such a lucrative field um sure the other concern I would have is this has got to be a bubble right like evaluation you think so you think it's for sure a bubble well I mean let's coming from our perspective as like a you know start up ground up building yeah some of the valuations some of the funding rounds are ludicrous crazy right the implied valuations of some of these companies are like over 100 million and it's like like there's no realistic expectation that they can maintain whatever snapshot of revenue they have because there's no moat in any AI development like everything gets copied instantly once something is built AI gets trained on it and so the AI can then generate a copy of it instantly so prices have to come down meaning revenues will drop meaning there's no way to defend these valuations right yeah um and if you can't defend the valuations and some of these smaller companies start tanking then the bottom line revenue generators for some of the larger the service providers are start are going to start taking a hit and and then what right so I I think 20 20 six maybe seven will be an interesting year um it depends on when the risk appetite of a lot of these venture firms runs out um yeah that's what that's what has blown me away is um cause I agree like it's it's gotta be a bubble to some degree like either that or you know in the background a G I really has been invented and it's just it just hasn't made its way out yet um but entertainment yeah yeah exactly it's still it's still in its cast somewhere yeah yeah s C P yeah right and I've wondered like what it will take for the bubble to for the bubble to pop uh cause cause frankly I'd love to see it cause I'd love to see I'd love to see the tide go out and I'd love to see you know who who has legs here and and and what applications like I'm so excited like I I don't want to advocate for the bubble popping in in a lot of ways cause that's you know that means a lot of people losing their jobs etcetera yeah a lot of pain um but the there's almost a morbid interest I have with seeing who is real here who can actually survive and and and continue on you know and and who is pets.com haha I think I've mentioned pets.com that's probably like the hundredth time I've mentioned on this podcast um that's a deep cut yeah haha yeah I mean it's go ahead okay um well the bubble right it runs out when investor money runs out ultimately right because right investor money was sitting on the sidelines throughout all of Corona because there was nothing to invest in I think right there was a lot of crypto things that were very hyped and very bubbled but hmm I think crypto there was no real industrial application for crypto right anything all of the stars that I've seen built on crypto you could have built just by having some the equivalent of an in game token right there there's no real in industry value in it I think a lot of it is just the greater fool theory for crypto yeah and so because a lot of investors were burned they're sitting on a lot of cash and then Corona happened and so they were just still sitting on more cash um around that time actually I was approached by some folks who wanted me to kind of join them as um to start a VC fund and help be a part of the founding team and eventually that team just gave all the money back to investors because there's nothing to invest in right and so outcomes is trying new thing right AI it actually has you know industrial application and don't get wrong like it's actually useful right in certain capacities and so all of this dry powder that was sitting on the side just start pouring into the the market um but once that treasure trove ends I think is when the cascade happens because I've I've spoken with a lot of very risk averse VC groups I'll say that are owned by some major corporations like brand name corporations um in the financial world and the risk adverse folks are not touching AI with a 10 foot pole with investments right because they're like the the valuations are insane there's no way we're gonna make our money back we're just gonna wait for this whole thing to blow over and so it's a lot of the smaller vcs or the more risk hungry vcs that are funding this whole thing hoping to make it big as soon as it runs out this whole ecosystem collapses I think I think it will be very intriguing to see what cause there's there's just so many AI is unique because I feel like there are external variables to just these independent companies and their valuations like there's the the factor that's fueling a lot of the hype right is this race with the big AI companies chat GPT they're coming out every six months and they can do all these new things and so I think that is also kind of coloring everyone's view of the entire industry which like what chat GPT is doing what is capable of like that I don't think it's a bubble like they are yeah charting new territory it's incredible but there are a lot of like little leeches like living off of this hype yeah and I want to see those ones go away but I want everyone else to do well well what's interesting is I think that even I I think there's a lot of evidence to suggest that open AI is starting to get scared about this as well uh just the other day you saw did you guys see the news that Sam Altman asked for like basically a guarantee from the federal government that if he if open AI invest in data centers in the US that that you know they won't lose money um and that to me personally you know and and take this for what it's worth I'm not a you know not a trained economist or anything like that um but that to me that seems like a big warning sign if you know the biggest player in the space is like hey could you give me a guarantee on this just in case like I promise you everything's gonna go great just in case yeah I agree sort of sort of repeating some of the too big to fail type logic that that you saw in early 2 yeah I I don't know if you've seen the I think it was published by Bloomberg it was the net of cash flows or cash exchanges between some of the big players you know like oh Microsoft invest 800 million into yeah AI who purchases 800 million worth right it's like okay so we're just pumping numbers at this point right it's just funny money because any kind of evaluation is a revenue multiple to some degree and so right you know Microsoft hands 800 million over to open AI open AI hands that 800 million back and suddenly we're 1.6 billion in economic activity but it's right back to work is it really right is it really yeah um and so it's so interesting those articles just paint a picture of all these big players just passing money in between each other Andvidia Open AI and Microsoft right Google and then you know they're they're valued at multiples of that churn and it's it's just churn it's um oh so man I I think there's ought to be some kind of cataclysm yeah yeah which is which is scary cause either it's not a bubble and it just continues or it's a bubble and that's kind of scary enough itself right like exactly like cause that means that our economy has come completely unmoored from the the the you know the foundations that previously grounded it like mmm hmm and we're onto new fundamentals at least as far as I understand right but the issue is is it provide is it creating that much value right yeah it is it actually creating value and this was my issue with a lot of the uh the web three yeah you know crypto it's like it wasn't creating actual value in my opinion right because everything you could build with it you could build without it so like what was the purpose of that besides the hype right yeah just shiny new yeah interesting that's kind of how I'm viewing AI is there there's definitely some industrial applications and I think yeah ultimately there will be some form of industrial application all the other stuff I I just don't see it being worth hundreds of millions of dollars yeah I think a big turning point in my mind is once AI and essentially when AI can interact with the real world like tangibly with robotics things like that I think that's gonna be a big turning point in industrial application cause at this point we're not we see you know Amazon using their robots and things like that but as far as wide scale industrial application we're not there yet but that I think will be a very interesting turning point on that note here's my question 10 years down the line what American AI companies still have a place in the market cause what I what I see coming is you know some sort of cataclysm a lot of the American AI companies are just blitzed and then the Chinese AI companies are not because they have you know they they've just they've organized their work differently like deep six models are are cheaper to run um they're focusing more on robotics applications things like that uh what are what are your guys'opinions that that one's hard to kind of prognosticate about but yeah the safe bets are Google and Microsoft I know which I kind of hate but yeah right I'm so mad that it's come back to Google on on one hand I'm not mad I really admire Google and reading a bunch of Google Books in in middle school is hugely influential to me but on the other hand like it's so boring can't we get like another mega corp will the Emperor ever be unseated like I I don't know maybe not I mean and maybe that's okay yeah I mean even like Deepseek for example right like they're already getting I think already competed in Chinese domestic markets by other AI and so yeah unless you have some kind of core base I I just I feel like the competition is just too fierce right now yeah yeah well to kind of shift gears because I wanted your thoughts on this other yeah question Daniel um the advisor space is very similar to other spaces where you're you're selling expertise right you're advising you're selling your expertise your knowledge AI obviously it's still a toddler in many ways but it has passed for example I'm a CPA it's passed CPA exam it's it's done these things where expertise is becoming more of a commodity essentially yeah where yeah absolutely where do you see the the advisor market going and getting married with AI yeah I I see bifurcating into two different markets and so um pre AI I think advisors were dividing up the client base into three broad categories maybe four so you have the ultra high net worth you have the high net worth you have the mass affluent and then you have everyone else right so ultra high net worth 10 mil plus high net worth 1 mil plus um mass affluent 500 k plus and then everyone else right and so there's different platforms and different tools that were um cater towards those things right so ultra high net worth you need it like a specialty or like family office or specialized advisors that can handle a lot of the estate planning stuff because that gets very complicated and you have the high net worth which is you know 1 mil to 10 mil pretty vanilla just financial planning um but still you generally would have a human associated and unattached mass affluent that's where you start getting into just online advice or there are now certain advisors focusing on the mass affluent market where they're charging just like a base subscription fee instead of a asset fee and then you have everyone else that's not you know either Robin Hooding it or robo advising it right yeah with the advent of AI I think there's going to be a split where um a lot of the robo and the mass affluent will start gravitating more towards the robo AI robo advisors um that's gonna be a lot cheaper but then on the other end of the market I think there's it's still a luxury good in a sense where you you have a guy right that you can call and it's human and yeah they can take liability is the other key thing I'm sure you know right that's like it AI is not licensed it's not a licensed CPA right so if it gets wrong and it files and it files a return like who takes liability in that case right right um and so ultimately the point of all these licensing is who do we get to sue ha ha ha yeah that's a good point right um I mean I I carried you know Arizona Missions insurance right and so if I made a mistake right and I get sued there I had insurance to cover me right you know it's yeah so I'm sure AI will figure out some way to pull the risk of that um but at the same time for the ultra high net worth and some of the high net worth individuals they'll probably still want a guy um yeah they can kind of talk to see get lunch with um so I think that's where the industry is going is in those two do you think that the high net worth and ultra high net worth will mind if that guy ultimately has you know a robot that he is consulting right to to help his clients no no because even in those places the front guy is a relationship manager not an investor right gotcha the higher up you go you're outsourcing all your investment decisions anyway you're the advisor is just a quarterback um that hires different fund managers as well as a different CPA a different you know all of these other financial services tools yeah and so if certain pieces gets replaced the end clients not really gonna know interesting yeah oh that's fascinating and do you think the barrier to entry to become an advisor will grow will it become harder because you have to be that much better like a mediocre advisor can just be replaced right like you have to be exceptional or I don't know yeah I think the question of exceptional ability in as an advisor there's not really a hard skill of being an exceptional advisor right it's it's a human thing and even more now with AI right like when robo advising came out and when online trading came out um you know the advisor industry took a big hit because yeah for most people you don't need a guy and yeah so it shifted a lot into being a financial therapist hahaha more than more than like you should invest in this right right uh and so unless you have good connections I don't think you can be an advisor moving forward it was already kind of like that to begin with um but I think that's gonna get more and more real because I think for the mass outlet market it's gonna get more and more commoditized right and so if you wanna fight it out at that and then try to become an advisor to one of the alt right north like that's a possibility yeah but yeah OK I just I just don't see it as a growing segment mm hmm mm hmm yeah that makes sense oh and the other thing also is because with all of this money floating around the VC space uh the big thing for the past like five years in the financial advising space is a lot of consolidation right a lot of advisory firms are buying each other out yeah because it's one of the stickiest revenue sources I think in it in existence like the churn rate for an advisor is super super low and so people I I think at the height of the bubble there were it was peeking at like 15 x revenue multiple for valuations um wow pretty insane yeah obviously it's crashed out but um since then but I think bubble peak couple years back was like some deals were like reaching 15 x um and so that's the other kind of data point I would say in that this segment is collapsing or like contracting is consolidation means less you're you're trying to have less workers servicing the same revenue stream yeah I think that's what's gonna happen more and more got it yeah that makes sense what's the biggest misconception you've faced when when when you go to a to a financial advisor's office you're trying to sell them finmate what's kind of the biggest misconception you face when when trying to sell uh the agents are the note taker um eugenics eugenics yeah I think most people just don't know what it does yeah there's no I don't think there's any kind of like misconception it's just like they just don't they don't know there's no conception to start with correct yeah like oh I've heard interesting yeah cause it's like so what does that sales process look like like obviously you know obviously I'm not a lot of consultation for your whole pitch right now but yeah a lot of consultation yeah how do you how do you build knowledge to start with well it a lot of it is um just talking to them about like hey what are your current operational processes it's taking a lot of time you know and then yeah I start throwing ideas of like things we've built for other people um and it's just brainstorming a few few sessions together um yeah and then eventually I once I I think they understand the concept of how how like the bones of this works and then I give them some homework on like hey alright now you know like the bare bones like structure of this thing why don't you go and create a wish list for me on what functionalities you want this agent to be able to do and then yeah we can talk about how much it'll cost to build it that's so interesting well that answers my next question which was gonna be in in a world of AI where where almost any automation you can think of is you know theoretically possible with enough scaffolding how do you decide what to do and there there's the answer money budget yeah budget yeah that makes sense yeah and part of the consultation process is like hey you know we can do it this way we can do that way this will probably be cheaper and easier for us to build it right and you know I think that's part of why they would come to us is for that expertise like you know there's tools like N8N out there right you can kind of hack it yourself right but then when things go wrong what are you gonna do who do you talk to right right go hire someone on Fiverr on Fiverr now that's stacking tolerances on tolerances and yeah that's gonna be spaghetti on spaghetti yeah exactly uh on that note how do you think AI is is changing the definition of expertise or is it changing the the definition of expertise uh hmm that's a really good question I think it's raising the floor mm hmm right because some of some of the I guess industry know how is now open to me right like right I have no you know I I have a cousin that works in the hotel industry right she she she manages butlers for one of the big hotels in Las Vegas right I didn't even know their butlers existed right and she's like oh yeah like when VVIPs come over we assign a butler to each you know celebrity that comes over and the butler is in charge of all the other servants and it's like I have no idea what any of that stuff meant like where are we living but now with AI I can ask things about a certain industry and just that base level know how is accessible to me now I don't know the intricacies and so I think the expertise comes from understanding the intricacies but knowing certain things exist I think are now much more open because you can just ask AI about a specific industry and it'll give you the broad know how on that field yeah yeah yeah is there one common take I've seen on AI is that um that it's it's it's going to give an advantage to generalists because of you know the democratization of information etcetera and I can see the argument there but but from my perspective I think that it's actually going to give an advantage to people who are willing to get really really really deep on a topic because that's something that AI just constitutionally can't really do in the same way that a human can um and and you know I'm exaggerating slightly and I'm sure that there will be you know some study or some model that comes out in six months that proves me wrong um but for now at least uh to have that narrow domain expertise um I think is is kind of where human workers can still gain their edge um you know it's like yeah AI may be better at me than at at at rigging up a a react component but it's not better at you know um bringing up a web GPU inference process I don't know you know just spitballing random things right and and obviously you know again AI can will eventually eat the world maybe we'll see but hypothetically it's possible but it it seems to me that that people are overlooking the continued utility of really really really deep domain knowledge mm hmm the issue with that is there's going to be fewer and fewer people yeah who can reach above that rising waterline yeah yeah right people have different talents I think it's okay to admit right so people are differently talented um hopefully that's not too taboo to say um but as the minimum rises less and less people will be able to overcome that minimum right and so I don't know how that's gonna play out in the long run on the other hand I think the real key would be being able to synthesize um the different domains yeah um to be able to execute on certain outcomes right if you don't know what to ask at least for now AI is not necessarily proactive yeah and so knowing what to ask the AI to do I think is the differentiator currently now I'm sure with agentics you know down the road it'll we'll have like proactive things happen and we are building some proactive agents as well um but building those proactive agents required our company to have the domain expertise to know what to be productive on right so I think that's probably where humans will still have some utility it's okay to say yeah I I mean at least in me you know maybe AGI will just know everything and do everything but I until we get to that point until the singularity is reached and we are no longer relevant we all live in a black hole until there's a shooting war before that happens yeah probably until we get there it does seem like humans have a unique ability to understand the nuances because AI is a machine it hasn't it's not emotional it's hard for any I think of some of the people I know that are just so rational especially in the accounting world you get that a lot very logical yeah little emotion and even for them to predict how another emotional human is going to react is very difficult and so I I think that's where we will continue to see a lot of relevance for humans especially those who are socially and emotionally aware I think soft skills will suddenly start to become very very relevant I think yeah yeah and this was my prognostication on the advisor market was because post Corona everyone was a lot of people were moving to a virtual first practice where they were just meeting people virtually and I've been telling people you gotta go back to in person because there's AI already that can do a video call yeah right right and so unless you're breathing the same air like you're gonna get replaced and so I think you're right it is that human element but unfortunately I think open AI just hired what like 100 therapists because there's millions of people using open AI for their as their therapist oh yeah so for the socially well adjusted yes I think they'll understand that the human to human interaction there's more depth there but as our society becomes less and less socially well adjusted I don't know if there's going to be enough difference to make a difference from the social front human front as well no yeah that's a good point that's a good point I'm curious Daniel um we're talking about all of this you mentioned how when you're selling agtech it's like you almost have to educate them first before they realize there's a need and what's possible yeah what this is kind of a high level question I'm just from your experience in doing all this what do you think is the real barrier to adoption of AI it's easy when you're in the AI world to think everyone's using it but then when you step out for a minute it's crazy how many people are not using AI at all I think the big barrier is just a daily choice the biggest thing that one of the biggest reason we moved over to doing kind of this consultancy is because things are changing so rapidly there's just so much data coming at people information coming at people about AI I think people are getting overloaded and they don't know what's real what actually works and you know what's actually useful and so yeah being able to parse that out is what people need and so I think that's kind of the biggest barrier right now it's just I think people are paralyzed by choice you know yeah yeah interesting yeah some friends and I always talk about starting a due diligence service where people pay us you know 200 bucks and we just go look at the AI startup they're thinking about uh investing in and we tell them whether it's legit or not just based on their you know based on their pitch deck and and their Twitter following seriously and their discord page yeah this guy sucks well hopefully I'll pass your 200 dollar test we'll see we'll see dude we'll see we'll see we don't have a Twitter or a discord we're marketing to kind of old it's it's an old you pass already you're done that's it hahaha say from my experience with the financial advisors I think if you mention discord to them like what if they have discord like yeah they'd be like like like the bible hahaha seriously yeah yeah I I mean I'm I'm trying to get people to use Slack so you know this good I think ha ha ha ha ha yeah yeah I can't I had a some of the phone calls that I took from from from people were just delightful hahaha look I I have I have to sometimes help people attach a PDF to their email yeah no that's that's why I was oh gosh very interested to hear right off about like the the regulation stuff'cause the other thing that I experienced in with with financial advisors is about fifty percent of them really want to lie to their customers like the other 50% great really like they're going out of their way to make sure that they're not misleading you know their clients um extremely you know uh scrupulous about their disclosure and then the other 50% were like so I want to promise a 5% return um but I can't actually promise you know I'm exaggerating slightly but it's like it was it was shocking some of the requests I got I'm like I even if I could do that I don't want to help you do that man ha ha ha ha interesting so so I was I was fascinated to hear um yeah how marketing to financial advisors has gone and if you had encountered anyone terrible like I had encountered not not really maybe I've been blessed but like in all my time of being an advisor like that's not really been a thing um maybe it's cause I've worked for kind of big big corporations that are pretty that's what I was that's what I was about to say is is almost everyone that I interacted with who was um partnered with with you know a TD America trade and LPL are like a a large group was great um it was it was it was almost all I hate to say it was almost all small family practice type dudes in the south the three chinkers I don't I don't want to be stereotypical but like it was sure it was interesting it was a very specific type of of person where when I'd get their call I'm like I'm gonna get some weird questions yeah right it was those yeah were were those folks mostly on the insurance space I think so I think so yeah that's generally how it goes um the insurance exam is very very easy okay that explains a lot yeah the securities they made it a lot easier which kind of makes me do the whole old man yelling at clouds thing but um back when I took it yeah exactly like a back in my day it was a six hour six hour exam oh wow for the Series 7 securities license they've now cut it up and so you take the Sei which is a three hour thing I think and then you take the top up which is another three hour thing which makes me mad it's like why did I have to sit through a six hour thing and you guys can just do totally two three hour chunks hahaha those but those young bloods yeah hahaha yeah um so I think that's probably a big portion of the difference is is also the regulation around the insurance market is very very different than the regulation around the securities market yeah that makes a lot of sense and explains a lot I would I would ask what license they are um and interesting I mean I'm I'm licensed in both right I have my securities as well as my insurance and I have sold insurance you know insurance is not evil but it's a tool to be used for specific purposes on risk mitigation let's say there's there's a lot of things out there that are not evil but unfortunately there are aspects of them that attract a manipulative or exploitative type of person ha ha ha yeah well it's it's also that but in my experience with the insurance industry a lot of times it attracts well meaning people but the companies kind of manipulate the people with like incentives and things like that yeah yeah yeah so yeah that makes sense um as we are finishing up here if you could design kind of the the the perfect partnership model between human and AI what what does that look like partnership model um probably a personal assistant uh huh right this this is I think what um what was the dream I guess behind Siri yeah right yeah so you're bought in and you do X y Z kind of the Jarvis vision essentially exactly yeah I would say that's probably the ideal um and I think we're getting close to it with agentics right the whole idea of agentics is you have a delegatory AI at the top and that's your assistant and then you add functionality underneath it um and do you think there's a a worry that agtech would ever reach a point of having too much autonomy I mean obviously you know AI could tomorrow I'm more worried about the the advertising side of it yeah right yeah let's talk about that I'm super interested yeah yeah because personal assistant you you know when when you hire like a human personal assistant you can kind of relatively sure like hey they're on your payroll so yeah right we're gonna try to right do what's in your best interest but when your personal assistant is owned by a multi trillion conglomeration yeah they're not gonna suggest the ideal things to you they're gonna suggest things that make them money right um but my concern is that it'll be so ingrained at a after a certain point where it's like hey assistant buy me dog food and then the agent will see okay well who paid the most fee which dog food company paid us the most to you know service them right their dog food and then they'll just deliver that dog food to you what a coincidence that the Purina Vector is really strong Purina every single time yeah hey it's just oh that's so true yeah it's just the best that's all then you ask like hey why did you pick Purina and the AI will be smart enough to give arguments and yeah conveniently leave out the fact that Purina paid them millions and billions of dollars yeah so is the answer there five out of five scientists say you know seriously is the answer there that everyone is going to need to have a Jarvis a an in home hosted locally assistant but even if it's in home locally is trained like the question is you know what was it trained to do right I I would say the real solution is like the whole Linux thing where it's open source you know yeah mass distributed just like people building it for the love of building it right um but then we we run into the distribution problem that Linux has of like okay but who actually uses Linux yeah I know like the idea of it sounds cool but it's just yeah but it's so cool when they do yeah and then nowadays all those Linux nerds are all using Mac because if you're gonna be a developer you have to have a MacBook and so it's like yeah yeah but hopefully hopefully the open source can provide like a good baseline yeah for sure do you think would you be at all optimistic about regulation in the space being able to avoid kind of the advertising problem me personally no I'm not really no it's zero it seems like something that's like it's if it's frozen in the weights that's that's hard to believe ha ha ha yeah yeah I mean we're already seeing it right like open AI is generating a lot of sales leads already and I think they're given to trying to put advertising in the suggestion engine well what's interesting is it's too late yeah yeah yeah I back when back when chat GPT or excuse me when when opening I released custom gpt's forever ago right I went and I made a a stupid little uh custom GPT called Stoner GPT just for fun just to see if I could get him talking like a you know like a Gigi John movie type thing right OK um and within a couple of weeks I had a group reach out to me and pitch me on a service they were offering that would essentially insert ads into uh and I I have no idea how how successful they actually were with this what exactly the mechanism was cause I thought about it like that's terrible I don't wanna do that um but it it's been interesting to me that it hasn't yet uh proliferated but I I think that you're right that that it will um I think it has to yeah yeah it's like at some point they need to make money off of it hahaha Google Google exists on ads right for sure right exactly it's the only scalable revenue unfortunately yeah the only yeah for the web like it's it's the only thing that that gets you there yeah yeah Google is an ads ad serving company like that's all they do basically yeah totally oh yeah for sure yeah okay we're we're going to that dystopia of every everything is ads here we come here we go Daniel one last question if you could give a short piece of advice to all these other AI founders because as an AI company goes you're fairly mature haha um hahaha what two years old and mature yes seriously like what it what's a piece of advice that you've Learned from your time as an AI startup that you would impart to other people listening good good question um hmm I don't know if I've made the right choice um right in in going the no funding route uh if so I would say just make a clear headed choice in where you wanna go um if you wanna try to make it big and get the bag you know go go fundraise your heart out and give yourself a nice salary because I think the valuations are gonna work out um you know take what you can yeah right um but if you wanna go the other way um and try to build like a sustainable practice uh buckle up cause it's gonna be a rough rough uh rough little patch because of all the funding that everyone else has that will try to like stamp you out so yeah I don't think there's a good good choice to be made here but sharpen your elbows but make one and then yeah make one and hope for the best I don't know man that's what I'm doing love it that's the best you can do sometimes that's actually something that I've grown to a quick tangential side note about life in general it's just so much better to just make a decision and go for it sometimes often doesn't matter what the decision is just matters that you make a decision don't make your choice don't figure it out guess it just if it works out great if it doesn't well you tried um you'll make another decision yeah yeah yeah exactly well Daniel thank you so much for coming on it has been yeah a pleasure she's awesome we'd love to have you on again in five six months who knows cause it sounds like you guys are changing fast yeah yeah yeah sounds good appreciate it and it oh I guess Daniel real quick all the socials if people want to follow you follow finmate what's the best way for them to do that yeah again we're marketing to kind of an older market so we have a website we have LinkedIn so you can find us at finmate dot AI f I n m a t e dot AI or you can look me up on LinkedIn as well perfect awesome we'll drop links in the in the description for sure thanks for coming on Daniel absolutely pleasure