Podcast: Play in new window | Download (38.5MB)
Subscribe: Apple Podcasts | Spotify | Amazon Music | iHeartRadio | Email | RSS | More
Alexander Harmsen – Portfolio Pilot
On Playing the Game: “How do we set a level playing field so that we negate the effects of all these high frequency traders?”
Investing strategies can be challenging. Typically it is you either trying to beat or keep up with the market, or you paying someone that vaguely promises to beat or keep up with the market. You almost need a computer to help you crunch all of the numbers.
Alexander Harmsen, cofounder of Portfolio Pilot, has had some successful exits and wanted to find places to invest. But the market can be daunting, and all of the so-called pros didn’t seem to be all that talented, at least as far as offering well calculated advice. When you meet a financial planner that isn’t dominating in the financial field, what are they actually offering you?
So Alex helped build Portfolio Pilot.
PortfolioPilot.com is a platform that uses AI to look at all of the numbers you have, what you want, where you’re going and calculates, in the fancy way AI is known for, finally pumping out some investment action step suggestions for you. Ultimately you decide, but it gives you well calculated advice.
If you’ve ever wondered if you should buy or sell a certain investment product, using AI to help figure out all, or most, of the variables can help you choose what to do and when.
Alexander shares insights on how PortfolioPilot empowers DIY investors by providing a comprehensive view of their financial landscape and offering personalized, automated advice without taking control. With over 30,000 users and $30 billion in assets on their platform, PortfolioPilot is changing the game by offering tailored financial recommendations based on a user’s age, net worth, and risk preference.
Listen as Alex illustrates the shortfalls of other options, and the cost of ignoring the opportunity to help get your money to work for you.
Enjoy!
Visit Alex at: PortfolioPilot.com
Podcast Overview:
00:00 Exploring Financial Ventures and Investments
04:12 PortfolioPilot: Custom Investment Platform
08:32 Monthly Tax-Loss Harvesting Tips
10:54 AI Evolution and Financial Modeling
15:36 AI Entrepreneurship: Navigating Practical Realities
18:11 Simulating with GPUs for FAA Approval
20:39 Compute Needs and Costs Rising
25:16 User Feedback Drives Product Success
28:41 “Obsessed with AB Testing”
31:54 Resisting Financial Advice on Tesla
34:10 Investment Diversification Dilemma
38:54 Macro-Driven Volatility Analysis
41:21 Market History and Transformations
44:12 High-Frequency Trading Equalization Strategies
48:56 Distrust in Financial Institutions
49:44 AI: Future of Personal Advice
Podcast Transcription:
Alexander Harmsen [00:00:00]:
And it goes through and it talks about, like, retirement readiness, and In scores your portfolio, and it, you know, calls out what’s what’s really good and, like, where are some opportunities for improvement. And, you know, importantly, it bends it benchmarks you. It compares you to other investors on your age, your net worth, your risk preference. The difference that we made, this absolutely massive change
James Kademan [00:00:24]:
You have found Authentic Business Adventures, the business program that brings you the struggle stories and triumph and successes of business owners across the land. Downloadable audio episodes of the podcast can be found at drawincustomers.com. We are locally underwritten by the Bank of Sun Prairie, calls on call extraordinary answering service, as well as the Bold Business Book. And today, we’re welcoming slash preparing to learn from Alexander Harmsen, cofounder of portfoliopilot.com, and we’re gonna learn about investing in AI. So, Alex, how are you doing today?
Alexander Harmsen [00:00:56]:
Doing really well. James, thanks for having me on the pod.
James Kademan [00:01:00]:
Thank you know, I’m excited because I like to think of myself as an investor. I’m not gonna say a smart or good investor, though up until maybe a month ago, I thought In was the best in the world. Right? And then Yeah. Then things changed. So let’s just start with what is PortfolioPilot?
Alexander Harmsen [00:01:20]:
Yeah. It’s, so much of it is about, our own psychology, right, of perceptions, and, you know, investing feels like it’s a lot about, you know, controlling fear and probably controlling greed as Call, you know, depending on, whether things are green or red at any given moment. So, you know, we can get into this later, but I I sold my previous company, an AI startup that I ran, you know, for six and a half years. We sold In 2021. And then, you know, basically, we bought a house in the Bay Alex, started a family. We have two toddlers now. My wife runs a biotech startup as well, like, really cool post series a. They’ve raised a lot of money for that.
Alexander Harmsen [00:02:05]:
And, in 2021, I sort of found myself trying to figure out what’s next and, you know, took out my coding gloves for the first time in a little while and, you know, built a bunch of companies on the back of a napkin and prototype things. And sort of at the same time, we started talking to financial Adventures and wealth managers, and I don’t know. It sort of feels like I got put on some list at some point because I was building calls every other couple days. You know, I don’t know. Felt like the thing to do. Right? First time in life, we had any sort of real liquidity, and, like, let’s get a professional involved and, yeah, let’s get an expert involved in managing our money and investing. And I did a little bit of crypto investing, and I had a Robinhood account and, you know, 401. And, you know, we were thinking about setting up a five two nine for the kids, and, you know, we took the house and all all these things, like, it felt like there must be things we can do to optimize this, improve In.
Alexander Harmsen [00:03:08]:
But it just felt like every single financial adviser I was talking to was more of a salesperson than they were, you know, an expert that was ready to just, you know, dig into my finances and investing and, like, put me on the right path. And I assumed at first I was just talking to the wrong people. And after talking about 20 to 30 of these wealth managers, financial advisors, I realized that this is the game. You know, they have some central funds, and they’re mostly just salespeople trying to get you access to that fund, and they’re charging 1%, sometimes 2% fees plus mutual fund fees and kinda feels like there’s conflicts of interest and human bias. And crazy thing, it was it wasn’t even tied to performance. It was just, you know, 1% every single year regardless of In happens. Yeah. And then they’re sort of on to the next client to try to figure out, like, how do we pull in more assets under management? And, you know, what I wanted was a coach.
Alexander Harmsen [00:04:12]:
I wanted to be somewhat involved and make some of my own decisions, and robo advisors didn’t really feel like a reasonable alternative. It was only a small part of my money, sort of lowest common denominator advice, you know, putting some money into to ETFs and, you know, equities and bonds. And then there’s, you know, a thousand and one trading apps and Bloomberg terminal for retail investors and Wall Street Bets and Motley Fool and CNBC and, you know, all kinds of trackers and screeners and but I didn’t wanna make this a hobby or a full time thing. I wasn’t excited about day trading and technical analysis and, you know, looking at all the numbers. What I wanted was something between. And so, you know, basically, I built something for myself in a couple months. And over the last three and a half years, that’s turned into PortfolioPilot, which now has, you know, over 30,000 users and $30,000,000,000 of assets on the platform. We’re actually a registered investment adviser with the SEC, so we can give real fiduciary financial advice.
Alexander Harmsen [00:05:18]:
And the main idea is PortfolioPilot is a personal AI financial adviser that doesn’t have control over your finances. You stay in control. You know, it’s best for DIY investors, self directed investors. You plug in your net worth. You know, you answer a bunch of questions. You can connect accounts. You can take a screenshot of your account, whatever it is, retirement accounts, crypto, cash, credit cards, real estate. It all gets connected.
Alexander Harmsen [00:05:46]:
You can see it all in one place. It scores everything for you. It gives you a full assessment. And then we can actually give you personalized financial advice and say, you know, James, based on your goals, based on what’s happening in markets, you should sell, you know, $4,000 of this stock you hold in your portfolio, or you should buy, you know, 8 and a half thousand dollars of this commodities ETF because, like, this is what it’s gonna do for your portfolio. And here’s how it’s diversifying, and here’s, you know, how it gonna, like, increase your downside protection and expect the returns. And there’s forecasts and a view to the markets. And, you know, because it’s completely automated advice, we can get very personal to you and your situation itself because we’re registered investment adviser. There’s, you know, an alignment there In interest, and there’s a whole bunch of compliance stuff we need to do to make sure, you know, we’re giving compliant advice.
Alexander Harmsen [00:06:49]:
And, you know, that seems to fill this, like, incredible niche of people that want to be somewhat involved, but, you know, like the idea that you could, Gavin, AI, you know, do 95% of the work for you and then sort of bring you a menu of different options that you can act on. And it’s, you know, it’s investment advice. It’s tax optimization. It’s thinking about cash management, long term, short term liquidity needs, estate man you know, estate planning. It’s thinking about scenario modeling and retirement. It’s running crisis simulation through, like, historical crises, you know, fee optimization, all, like, in one sort of proactive platform. And so it’s not just, you know, chat GPT, you know, open box. Like, you know, we’ve we’ve tried something like that two years ago, but we found that what people really want is they want our system.
Alexander Harmsen [00:07:51]:
They want an adviser to sort of bring different opportunities to them because they don’t actually know the questions asked. I mean Mhmm. If you know the right question to ask, you can get the answers. Right? There’s a thousand other tools In Google search and ChatGPT itself that can answer the questions for you. The big question is, what should I ask? You know, what should I be paying attention to right now? Right? If the markets have crashed, which, you know, last couple weeks, like, probably people should be looking at tax loss harvesting opportunities. And that seems crazy to say say because that feels like a year end, you know, December thing.
James Kademan [00:08:28]:
Yeah. We’re, yeah, we’re in March right now. Yeah. But, like, there are
Alexander Harmsen [00:08:32]:
tax loss harvesting opportunities that open up throughout the year that you could take advantage of in moments like this, and you could book those losses, you know, trade into something similar, and, you know, those potential opportunities might go away three weeks from now. And if the market is up compared to where it is today, those opportunities might be gone if you just look at it once at the end of the year. And so there’s things like that that, like, there’s lots and lots of those little tricks and optimizations that you could be applying to your own net worth financial situation, you know, on a monthly basis. Right? We send an email out. You know, you can log in to the platform, but we send an email out once a month, start of every month just saying, like, James, here’s the top three things, you know, we think you should do. It takes ten minutes to act on this. You know? And even just that helps people feel, like, feel a little bit more confident, feel like they’re a little bit more on top of their finances. And, you know, in some ways, it feels very niche.
Alexander Harmsen [00:09:37]:
But, like, by our estimate, you know, there’s probably somewhere between twenty and fifty million Americans that are actually in, like, this situation. But they don’t wanna be completely passive handed off to someone. I don’t wanna look at my money until retirement. And, like, on the other extreme, they’re sort of like the day traders and hobbyists. I think most people actually fall somewhere in between where they wanna be somewhat In, and, you know, they just need a guide. The more personalized it gets, the better. You know, kinda like every other consumer product we use in our life. Right? Netflix actually gets better every single time I use Netflix.
Alexander Harmsen [00:10:18]:
Right? Every time I say, like, no. Ignore. No. Oh, that’s an interesting show. That’s a movie that fits. And so this In, like, this idea of recommender engines throughout everything that we do that just keep learning about us and keep getting better and better and better.
James Kademan [00:10:32]:
So I wanna ask you a quizy about the AI stuff a little bit. Let’s downshift a little bit and talk about your first business or, I guess, I don’t know if it was necessarily your first, but that business you sold that was involved in AI in ’21. AI I mean, chat g p t came, what, November, something like that. Is that right? Right. So you were way ahead of the curve there, it sounds like.
Alexander Harmsen [00:10:54]:
I was, I was doing AI, you know, when I was still called deep learning. Even before deep learning, you know, neural nets. Even before neural nets, he was working on expert systems and machine learning algorithms. I mean, even when I say AI for portfolio pilots, like, there’s some stuff that we’re using LLMs for. Some web scraping, some summarization, you know, thinking about, like, just translating stuff In English for, you know, regular people and investors. But, like, when I say AI, I also mean, like, there’s these dynamic factor models that we use to think about modeling interest rates and thinking about GDP and inflation and like, this is stuff like the central bank uses. You know, the Fed uses dynamic factor models for some of its simulations and, like, mapping out macro conditions. We use multivariate machine learning forecasting models, you know, for some of the underlying risk assumptions and expected returns across the system.
Alexander Harmsen [00:12:04]:
We are using regression models to think about, like, exposures of your portfolio to changing inflation conditions or, you know, GDP or raw materials risk. You know, what happens if the price of oil spikes? You know, that’s going to affect the downside protection in your portfolio. And so AI is like I mean, it’s a whole spectrum of things, and then recently, it feels co opted for, you know, LLM everything. But even then, like, everything we do on the coding side, right, we’re using large language models for and we’re a 10 person team right now. I think we’re operating at the same output as our 25 person team with Iris Automation in 2019.
James Kademan [00:12:47]:
Wow. Okay.
Alexander Harmsen [00:12:48]:
And, like, I like to think I’m a better manager and, you know
James Kademan [00:12:52]:
all do. It’s all good.
Alexander Harmsen [00:12:53]:
Yeah. Of course. Right? I mean, if I think back to some of the mistakes, but, like, that’s not where this, you know, 50% increase in efficiency is coming from. It’s not like we’ve hired smarter people. You know? I’d like to think we hired for some pretty smart people at Iris as well as, you know, with PortfolioPilot. It’s the tools we’re using. It’s, like, it’s everything coding, marketing, product design. And, I mean, that’s Kademan, like, an absolutely massive difference.
Alexander Harmsen [00:13:25]:
But even then, like, even in 2015, ’20 probably 2014, I was working at NASA as part of, where, basically, we were using neural nets and computer vision models, and this is, like, super, super early. Like, if you trace LLMs all the way back, you can trace them back to these, like, very early computer vision models around ImageNet and AlexNet. And, like, this is the kind of stuff we were doing, like, using to do object recognition. We were working on something called Mars helicopter. Basically, a little drone that landed on the Martian surface a couple years ago. I don’t know if you remember that in the news. No. Basically, like, with the rover, there was a little helicopter drone that, like, you know, was exploring the Martian surface, and we were working on that in 2014.
Alexander Harmsen [00:14:14]:
And you might think that the biggest challenge there is, like, the atmosphere is so thin. How do you, like, keep it in the air? The bigger problem is, like, there’s no GPS system. Positioning on Mars is really difficult.
James Kademan [00:14:27]:
Oh, no.
Alexander Harmsen [00:14:27]:
You know, how do you keep track? How do you know how close you are to the surface? And, so we were we were building computer vision, like, navigation algorithms to be able to, like, figure out just how close we were, you know, and just imagine rocks in the desert, you know, outside of Pasadena, you know, on LA, and, you know, trying to simulate, like, what it might look like on the Martian surface using these advanced machine learning computer vision algorithms. I mean, absolutely, like, super, super early AI stuff, robotics. And then my my previous company was, like, all computer vision, navigation, massive simulation environments, autonomous Calls. And, it I mean, it feels like we were really pushing the edge of, like, what we could do with AI at that time. And, like, so much of it translates. Right? It may feel completely different. Fintech for consumers versus, you know, autonomous Calls, but, like, we have to get regulatory approvals. We needed to work with the FAA and other civil civil aviation authorities.
Alexander Harmsen [00:15:36]:
Running a business can’t just be some r and d research lab. Right? We don’t have the resources of, you know, Waymo and Google, and and we needed to make it work and, you know, deal with sort of practical realities of, like, AI is not gonna be perfect every time. So how do we think about complex bounding functions, and how do we make sure that the AI, you know, has some sort of physical like, there’s some reality of, like, sort of the physical world around it. Right? What’s up? What’s down? Where’s the Earth? How do we actually navigate around some situation? And, you know, we got regulatory approvals in nine countries. We’re potentially the first autonomous vehicle company to, like, truly expand commercially to hundreds of clients, get regulatory approvals, like, around the world, and, sort of opportunistically sold In 2021. Like, even just building out the kind of evaluation and testing system needed to be able to general like, make sure we had some generalized system Feels very similar to sort of all the eval stuff that people are talking about with LLMs and, you know, as a core part of what we’ve built with PortfolioPilot is just how do we assess it properly? How do we have some sort of auditing system to make sure that the advice that we give, you know, we can actually stand behind even though it’s personalized and, you know, there’s hundreds of thousands of different variations, you know, that might pop out, you know, the other end of this AI financial adviser we’ve built.
James Kademan [00:17:04]:
Yeah. Tell me I wanna look or talk to you about the processing power because you’re talking about way back when when processing power I mean, I remember Pentium two. What are we talking? Like, 300, four hundred megahertz, and you were like,
Alexander Harmsen [00:17:17]:
ah, I
James Kademan [00:17:18]:
can do whatever. It’s my supercomputer. And now the processing powder, when you’re looking at these In chips, it’s like, oh my gosh. And the size of them and the power they use. So forget about the energy usage. But way back when, the actual processing power I mean, it seems to me like you would take rooms full of computers to make these to make these AI models actually work.
Alexander Harmsen [00:17:40]:
I mean, even in 2017, ’20 ’18, we had, like I mean, we had an office in San Francisco, and I’d say probably 20% of our office was, like with just, like, a server room with stacks, like, NVIDIA GPUs, and, like, we couldn’t buy them fast enough.
James Kademan [00:18:09]:
Back then?
Alexander Harmsen [00:18:11]:
Like, I mean, we, like, we went I remember, like, calling up other companies and saying, like, can we buy your GPUs from you to, like, rack and stack in the server room? And it was like part of it was machine learning and, like, deep learning training and, like, these models. A lot of it was simulations as well. Like, we needed to prove to the FAA that we could do, you know, millions of collision scenarios and map this out properly in different lighting conditions and different, you know, vehicle configurations and colors and weather patterns and like, that was a huge part of it. And then the actual, like, onboard like, there’s sort of there’s a pretraining, right, to build the models themselves, and then they’re sort of, like, actually running it, the inference time. And so we were doing all this inference on this little embedded GPU called, the Jetson chip. And, like, Call, we moved to these, like, sort of embedded Xavier models, and we’d have sort of streaming video in this hardware in the loop setup. And we would have, like, a hundred of our little devices, you know, on these, like, you know, these little embedded processors and little box we built, a hundred of them just on a Call. And they were just like there was a simulated video stream being fed into these in real time.
Alexander Harmsen [00:19:35]:
And then once in a while, it would trigger actions and, like, a change in the navigation. I mean, absolutely crazy. I mean, with PortfolioPilot, we are now, like, a fully, fully remote team. Everything happens in the cloud, you know, but, like, we just so we actually just went through a massive infrastructure upgrade, where, like, we now have all these dynamic queues and being able to Call calculate and recalculate recommendations and run through tens of thousands of different potential changes to someone’s portfolio and look at market conditions of forecasts. And, you know, we have all these, like there’s one main thing, you know, forecasting model that sort of reprocesses every week, and I think it takes sixteen hours to process, you know, across a whole bunch of different services. It’s like this parallel computing framework. Absolutely crazy. Probably looks something similar to that server room, you know, if it was physically, you know, here in my house or in an office.
Alexander Harmsen [00:20:39]:
Compute requirements have, like, grown tremendously, and, this is kind of like a wild thing, I think. A couple weeks ago, a lot of people were talking about, like, oh, now we have this sort of cheaper model. Yeah. You don’t need to train as much. And, like, a video stock dropped, which was wild to me because, like, every tile every single time, like, we figure out how to do something cheaper, we’re, you know, 10 x In the amount of that we’re doing it. Right? If we Call do error checking, you know, in our AI assistant or in some of the recommendations that we do, you know, some of it is just too computationally expensive. It slows it down. You know? People, you know, individuals want to see recommendations generated and, you know, max twenty seconds.
Alexander Harmsen [00:21:25]:
Right? If you’re waiting more than twenty seconds for something to spin, you know, I was like, ugh. This doesn’t Under the next,
James Kademan [00:21:31]:
man. There’s different squirrels.
Alexander Harmsen [00:21:32]:
Yeah. Even if it’s the most advanced thing, you know, we gotta load in twenty seconds. And so every time we figure out how to make it a little bit faster, we’re just packing more into that. And I think it will just keep accelerating like that. You know? There’s never gonna be, you know, enough compute.
James Kademan [00:21:50]:
That’s fair. I can tell when I look at the processing power of just a regular desktop now compared to twenty years ago. Twenty years ago when I had to click and you wait for the little In circle to do stuff, you’re just like, oh, that sucks. But now you look at the processing power now. Anytime I see a little hourglass come up, I think why? We have 50 times the processing power. There’s no reason for us to see that hourglass anymore. Yet here we Alex.
Alexander Harmsen [00:22:19]:
Like, waiting for a song to download on LimeWire. Like, it’s like, hey. It happened in two or three minutes because we’re, like, we just got some, like, upgrade to, like, DSL. Like, we just went from I remember going from Draw up to DSL in our house. I was like, wow. It’s taking two minutes. This is insane. You know how many songs I’m gonna be able to download this week? Like, dozens, maybe hundreds.
Alexander Harmsen [00:22:46]:
Like, this is a game changer. Mhmm. Absolutely wild. Right? I just listened to this, there was, like, a little dramatization of, the Spotify story.
James Kademan [00:22:56]:
Oh.
Alexander Harmsen [00:22:57]:
And I I love that. I I forgot what it’s called, but, it was, like, sort of the Swedish series dubbed over, absolutely well done. And, it was, like, six six different episodes.
James Kademan [00:23:09]:
I watched that. I watched that because they were talking about the streaming and the delay.
Alexander Harmsen [00:23:15]:
Yeah. Like, I mean, the founder just, like it’s told, like, there’s six different episodes each, like, half an hour long, and they’re all told from the perspective of someone else on the team.
James Kademan [00:23:26]:
Yeah. I mean, that was an incredible
Alexander Harmsen [00:23:28]:
episode In, like, you know, the founder saying, like, this is how it happened. Here’s the story. And at the end, you’re like sorry. It cuts to the CTO. He’s like, nope. That’s not how it happened. You know? This is how it happened. And he tells, like, the big vision, you know, the big vision is nice, but at the end of the day, you need someone to execute and get Spotify to work, you know, within two hundred milliseconds.
Alexander Harmsen [00:23:52]:
Because, like, as soon as it drops under two hundred milliseconds, then it feels instant. And then, like, the end of the episode, you know, it’s the lawyer that says, that’s not how it happened. Like, the only reason why Spotify was successful is because, you know, I figured out all the legal stuff and, you know, I honestly, I love it. Right? There’s there’s so many I mean, others there’s this expression. Right? There’s, like, you know, there’s his truth and her truth and then what actually happened. Right? And everyone has their own perception. But, like, this idea that it’s, like, two hundred milliseconds, that’s, like, as fast as human reaction time. And, like, you just see the founder constantly saying, it’s not fast enough.
Alexander Harmsen [00:24:36]:
You know, everybody else is like, holy like, this is, like, this is the Pirate Bay era or the buy CDs Draw, and Spotify came in and said, like, we’ll just give you instant access to any song on the planet. And, like, nope. There’s still I could still sense the delay. It’s gotta be quicker than that. And every time, they’re just shaving off a little bit, a little bit, a little bit. It matters a tremendous amount in human psychology and consumer products. And I think, like, it’s a big you know, there’s so many of these, like, psychological pieces that I’ve learned just over the last couple years. I’ve I’ve talked to thousands of investors at this point.
Alexander Harmsen [00:25:16]:
Right? And I was talking to someone right before this call. I probably do, I don’t know, two or three customer calls every single day, and I I love it. Right? I’m talking to people across America, completely different financial backgrounds, level of expertise using this product, and some people absolutely love In. And how could I have survived before this? And some people say, like, I can’t believe it’s so difficult to use and, like, why don’t you put this button on this page and, like, this view isn’t the way I want it set up. And, and so it’s like it’s, like, it’s tremendous it’s amazing because people have such different reactions, and we made a huge difference recently. It’s like night and day difference in the product. As far as we can tell, like, the test isn’t completely finished. As far as we can tell, it is a five x increase in conversion rates to get people to actually start a free trial, like, sign up for the product.
Alexander Harmsen [00:26:14]:
Wow. And it is the difference between it is the difference you come into like, basically, you sign up for the product. You answer a bunch of onboarding questions. You link an account, and then you come into the product and sort of, like, see the dashboard. And there’s a big banner on the top of the page that says, like, check out your portfolio assessment. And it goes through and it talks about, like, retirement readiness and its Call In your portfolio, and it, you know, calls out what’s what’s really good and, like, where are some opportunities for improvement. And, you know, importantly, it bends it benchmarks you. It compares you to other investors on your age, your net worth, your risk preference.
Alexander Harmsen [00:26:54]:
The difference that we made, this absolutely massive change was preemptively opening this assessment for people or leaving it as just a banner for people to click on their own. Turning opening it, like, as far as we can tell, like, it’ll it’s gonna be a massive improvement. You know, we don’t know if it’s probably somewhere gonna be between, you know, three to five x conversion rate.
James Kademan [00:27:19]:
Wow. That is impressive.
Alexander Harmsen [00:27:21]:
But it’s like this is
James Kademan [00:27:23]:
The little tiny thing.
Alexander Harmsen [00:27:24]:
Technology. Right? I think it’s opening that automatically for folks and because I think I don’t know. People miss it or it feels like extra work. And, like, once you see the assessment for many people, this In, like, this is the light this is the, you know, what do you call In? The moments. Like, okay. There’s there’s something here. Like, there’s something very interesting. This AI model isn’t just, you know, chat GPT.
Alexander Harmsen [00:27:50]:
If it’s able to synthesize everything here across my life from taxes to fees to retirement readiness to, like, calling out opportunities, then, you know, maybe it’s worth, you know, actually starting a free trial, actually exploring the rest of this product versus just saying like, oh, it’s just like, you know, every other AI investing product out there.
James Kademan [00:28:12]:
You know, it’s so interesting that you go through, let’s say, 99% of all that stuff, the technology, the software, the coding, the chips, figuring it all out. And someone’s gonna say yay or nay based on the color or the font that you chose or the size of the font or where the button is. It’s crazy how much psychology can make or break a business regardless of how good that business is. It’s It is. Insane. Insane. I get it.
Alexander Harmsen [00:28:41]:
Absolutely insane. Right? Like, we’re we were we’re running so many different AB tests right now. Like, on a in the last six months, I’ve, like, become a little bit obsessed with just AB testing things. I think we’ve hit a point where, like, it works, and now it’s a lot of optimization, and we have enough volume where it actually makes sense to AB test stuff. You know, we’re it’s, I think when people think about AB testing, I think we’re, like, oftentimes thinking about, like, you know, is it a red button on is it a light red button, you know, this sort of AB testing. But, like, we’re think like, we’re testing, like, which image to lead with on our landing page. You know? Above the fold, h one on our landing page. On our sign up page, right now we have a test that’s running that says there’s three different headings on the sign up page.
Alexander Harmsen [00:29:31]:
One of them is, like, improve your financial strategy, And another one is, like, the one that’s winning right now, and it’s Call looks like it’s gonna improve our conversion rate by about 22%. Instead of improve your financial strategy, optimize your investments. Oh. Difference between those two. Like, they’re basically the same thing, but, like, there must be something that makes, like, financial strategy feels like just a little bit more technical, feels like I don’t you know, if I I’m thinking as a retail investor, I don’t really have a financial strategy. But the idea that AI could come in and optimize my investments, that makes sense. Like, there’s there’s inevitably something I can optimize about the investments that I’m doing right now.
James Kademan [00:30:20]:
Yeah. That is interesting. Little tweaks like that. To me, I wanna I wanna shift into the investments themselves. Is this exclusively for stocks and mutual funds and stuff like that, or does it go beyond into real estate or crypto or other I mean All of the above. Million things.
Alexander Harmsen [00:30:38]:
Income, crypto, real estate, private equity, debts, student loans. Right.
James Kademan [00:30:44]:
Put In some overtime. I don’t know.
Alexander Harmsen [00:30:46]:
Something like that. Cards, Call, retirement accounts, tax free, Roth, conversions, everything. You know? I think, in some ways, it’s very contrarian to, like, the standard start up advice. Like, I advise a few, tech companies. And, honestly, sometimes it feels like all of my advice just boils down to just focus, stop trying to boil the ocean, and that’s exactly what we’re doing. I mean Alright. I love that. What we’re not doing.
Alexander Harmsen [00:31:22]:
Right? Like, we’re boiling the ocean. It almost feels like we’re doing everything. Right? It’s comprehensive network tracking across everything. It’s like advice on all these different things. It’s, you know, taxes and estate planning and fees and, you know, investing. And the thing I’ve come to realize, like, we started very narrow, just ETFs and stocks, investing recommendations. And then, like, I still remember two years ago, I was talking to this guy. I was showing him, like, sort of one of our early, like, sort of beta models.
Alexander Harmsen [00:31:54]:
So In everything was working technically, but we’re, like, trying to figure out how do we actually give the financial advice to people, how do we show people recommendations. We weren’t charging for it yet. And, this guy had $900,000 net worth. 40% of it was in Tesla, and the system told them, you should sell 80% of your Tesla position. Feels like no brainer financial advice, like, for so many reasons. You know, it’s not efficient, and now we’re close to the efficient frontier. There’s lack of downside protection, and you have, like, all this, you know, you have all this, idiosyncratic risk, you know, in an individual investing position. And he said, no.
Alexander Harmsen [00:32:41]:
Ignore. That’s a that’s a bad recommendation. And I’m, like, on a call with him, and I’m just sort of like, hey. Let me watch you use the system, react to the investing advice. And I asked, like, why why like, why did you press that button? And he says, I’m not gonna do that. That’s crazy. Yeah. I’ve I’ve $350,000.
Alexander Harmsen [00:33:01]:
I’m not gonna sell $300,000 of Tesla. And, so okay. Like, do you think you’re on a like, is your portfolio set up in exactly the ideal way? In that, oh, no. No. No. It’s, like, super problematic that I have all this Tesla. Like, okay. Why don’t you take the advice then? He’s like, well, I’m gonna pay a whole bunch of taxes on this, first of all.
Alexander Harmsen [00:33:29]:
And if I sell it, I’m not just gonna sit on cash. And so I need to figure out, like, what to trade into. You know, what am I gonna invest instead of Tesla? I’m not gonna buy bonds. I’m not gonna, like I like, I’m 28 years old. Like, I’m perfectly fine taking a whole bunch of risk. I’m probably gonna end up putting it in Meta or Google or like, it’s basically gonna be, like, 90% similar correlation to Tesla. It’s gonna be some, you know, mag seven type, you know, growth stock. It’s gonna have like, if the market crashes, Tesla’s gonna go down and Google’s gonna go down at the same time.
Alexander Harmsen [00:34:10]:
Like, am I really more diversified? All I’m doing is just wasting a whole bunch of time doing research and, you know, paying taxes for, like, a similar exposure and, like, In into it a little more. His uncle worked at Tesla, so he kinda felt like he had a little bit of In. Alright. And then, like, I mean, there’s all these things that you need to consider before you like, someone will actually take investment advice. And so after probably six iterations with him over the course of the next six months, you know, him and a couple other investors that like, the advice that he actually ended up taking was something along the lines of, like, sell 5% of your Tesla position. Here In the taxes you’re gonna pay. With that money, you should buy this, like, commodities index fund. You know, it’s everything from, like, oil to, like, a wheat, and it’s like it’s an ETF.
Alexander Harmsen [00:35:07]:
It’s a low expense ratio. It’s gonna do wonders for your downside protection. Here’s, like, a metric, you know, showing you how your downside protection is gonna increase. It’s gonna make your whole portfolio more resilient. Here’s the expected return. You know, basically, the exact same, but the efficiency of your pro portfolio goes way up. And, like and he said, okay. That actually makes sense.
Alexander Harmsen [00:35:29]:
Sort of it checks all the boxes. It like, I could trust that it makes my portfolio a little bit better. The taxes, like, I can swallow that. That’s sort of like a onetime hit. And, I mean, you sort of need to do everything else to get one piece of advice correct for someone. You know, you can’t just Calls like, I think this is the biggest problem with, like, any company like Motley Fool that’s just saying, well, you know, In is a buy today or NVIDIA In a sell. You know, it’s undervalued. Whether you, James, should buy NVIDIA or sell NVIDIA is very personal to your preferences and what’s in the rest of your portfolio and what you care about and, you know, whether you’re gonna buy or sell your house next year or what are you saving for, how your age, and how close you are to retirement.
Alexander Harmsen [00:36:19]:
And and so it’s like you kinda need to do all the things. I think about a lot like this iceberg. Right? There’s so much under the surface that you need to think through before you can give one very simple piece of financial advice to someone that they will actually trust and, you know, and act on.
James Kademan [00:36:38]:
Fair. Tell me the learning model. When I look at the market over the past, let’s say, fifty years, it’s doing whatever. But then it seems to me pandemic ish is when it starts going crazy roller coaster y. Like, before, there were roller coasters, but our idea of a roller coaster even ten years ago compared to today I mean, we’re talking plus or minus five to 10% over the course of a month where twenty years ago, people would be screaming. And now we’re just like, hey. It’s gonna come back up. It’s not a thing.
James Kademan [00:37:14]:
So how do you teach a model that this is this is what’s happening because this is the way the world is going now versus twenty, thirty years ago. That’s the way the world was with the market. I don’t know if it was I don’t know why. If it was more retail investors got in or what happened pandemic wise that it started to become so, Rocky. I have theories, but I In don’t know. I’m certain
Alexander Harmsen [00:37:37]:
you know more. There’s there’s two parts of this answer. I think it’s a it’s a good question. Right? And, I think I think the short answer is you know, as far as I can tell based on, like, a lot of the work that we do with, like, these historical trends, things aren’t any more volatile these days than they were twenty years ago.
James Kademan [00:38:00]:
They are not.
Alexander Harmsen [00:38:02]:
No. And so look at
James Kademan [00:38:03]:
the graphs, and there’s definitely it’s a five year old with a crayon right now.
Alexander Harmsen [00:38:08]:
I honestly think I also feel that, and I I honestly think it’s just in our head. Like, we are just like, we remember the past differently than we remember now.
James Kademan [00:38:19]:
I mean, like, you look at a graph of the S and P 500, and it’s it’s Yeah.
Alexander Harmsen [00:38:25]:
Well, okay. This is like if you if you wait it like, if if you take a log graph, I don’t think it looks very different. Like, it’s just it’s exponentially larger and sort of swings look larger. Okay. But, like, looking back over time, like, the markets aren’t really more or less volatile than they they were before. And, like, I think it’s Okay. Like, feels like just the hindsight bias. Maybe it’s just like when you look at the graph, it’s like exponentially Calls like it’s large like, the magnitude swings are larger, but the the relative swings are the same.
Alexander Harmsen [00:38:54]:
And then, you know, practically, the system is looking at historical volatility, recent volatility. We have this exponentially weighted model for, like, how we assess volatility. So it gives a little bit more weight to recent events, but then we also tie it to macro trends. And so, like, every single forecast and prediction and volatility model, and if we’re analyzing, you know, correlation and betas and everything like this, it’s all tied to macro trends as well. And so we’re tracking something like 20,000 different macro series from US GDP to, you know, Canadian CPI to building permits in South Korea. You know? Like, these are Call sort of, like, publicly reported, you know, macro indicators that, like, all connect together in a way and as much as possible, you wanna try to find leading indicators for, like, how things are gonna happen. Right? And, like, if you make the model big enough, you can capture more and more of these sort of, like, these edge Calls, and you start sensing anomalies. And all of that gets fed through to, like, the underlying risks and expect the returns for the portfolios.
Alexander Harmsen [00:40:05]:
You know, we as much as possible in, like, what we, you know, what we design into the core system, we try to make it as accurate as we possibly can even if we have to sacrifice some precision. And so what you’ll see is sort of, like, the more uncertainty there is in the world and markets, what happens oftentimes is that the uncertainty and the error bars in our own models and forecast grows as well. And that means that the, like, the expected return to volatility volatility trade offs that it’s making, like this risk adjusted return component ends up getting worse and worse and worse, like, as you go into recessions and more volatile periods, which is a really good thing in general because that feeds into the recognition engine and, you know, ends up being threaded through to, like, your individual financial advice.
James Kademan [00:40:58]:
So how far back is the model looking? Is it going nineteen hundreds, early depression?
Alexander Harmsen [00:41:04]:
It really depends on, like, which series, but, I mean, we have datasets that go back really, really far. I mean, definitely, like, well into the, like, the nineteen hundreds, you know, deep into the nineteen hundreds.
James Kademan [00:41:20]:
Okay. Okay. I don’t
Alexander Harmsen [00:41:21]:
know if we have any data that, like, goes back to the eighteen hundreds. Maybe there’s some gold stuff that is, like, that far back. But, like Okay. I think you definitely have, like, there’s certain moments over the last two hundred years that have, like, really defined market conditions and, like, there’s, like, the creation of the SEC and, like, thinking about all the stuff that happened after, like, the 1929, like, you know, the Black Friday and the stock market crash and a lot of the regulations that came In. Things changed tremendously after World War two In markets and, like, population boom. And then things really changed again in 1970 after the, like, you know, moving off the gold standard. And, Oil embargo and A lot of it, I think. Like, it’s it’s really difficult to, like, trust trends, like, going back before 1970 in particular.
Alexander Harmsen [00:42:16]:
Like, I think most people, most economists, most people, like, in the investing world are really just thinking about, like, the last fifty five years as sort of, like, as far back sort of in the current regime regime.
James Kademan [00:42:32]:
Fair. Fair. Yeah. I think it was a different game when computers weren’t really involved. You look at the little videos of our, oh my gosh, the depression stuff when they had the tapes, and the price couldn’t keep up with the speed of the tape.
Alexander Harmsen [00:42:49]:
Right. Right. Right.
James Kademan [00:42:49]:
So people would make a trade, but it would I mean, it’s so chaotic. And I was thinking, man, they did that without computers. Where today
Alexander Harmsen [00:42:59]:
What an absolutely different world. You know? Yeah. One of these books behind me here is, it’s a book called Flash Boys, and, like, it’s basically all about all about, like, getting closer to the exchange.
James Kademan [00:43:14]:
And Yeah.
Alexander Harmsen [00:43:15]:
Yeah. Yeah.
James Kademan [00:43:16]:
Know, like
Alexander Harmsen [00:43:16]:
Yeah. Microseconds, like, start to matter and people building cables and, like like, people realized that, like, they were running a trading Business, and, they realized that they were getting different prices from different exchanges. And depending on which exchange you sent your orders to, you could sort of undercut other exchanges. And there was, like, this arbitrage opportunity happening in milliseconds, and they ended up creating this, like, completely separate exchange that, you know, executed everything at the same time to make it sort of, like, a little bit more neutral for everybody, like, all market participants. And, like, that was a big deal because, like, in their trading business, they realized that there was all these high frequency traders that sort of, like, undercutting all their trades. And so they were trying to figure
James Kademan [00:44:10]:
out, like, how do we Yeah.
Alexander Harmsen [00:44:12]:
How do we set a level playing field so that, you know, we sort of negate the effect of all these high frequency traders and, absolutely fascinating. Sort of like, as you dig down I mean, you keep digging down. Right? There’s like there are arbitrage opportunities on, like, the systematic macro fields that, like, you know, company hedge funds like Bridgewater are playing In, and then there’s sort of, like, day to day dislocations. And I’ve talked to data providers that will, like I mean, they they promise to, you you know, give certain anonymized credit card data for retail transactions so you can bet against, you know, Target announcing earnings, you know, a day ahead of time, you know, based on credit card data. And then, like, you get closer and closer, like, lower lower, down and, like, it matters which you know, if you’re running an FPGA or, like, a CPU and, like, how many feet away are you from the central server to the New York Stock Exchange? And, you know, it turns out that the the Stock Exchange is obviously in in Manhattan, but all the servers are across the river in New Jersey. And so, like, all the buildings like, some of the most valuable real estate in the world is actually, like, the buildings right around that exchange. And then even the New York Stock Exchange, like, a number of years ago, realized that they could actually start clearing away offices in the building itself and just let people collocate their servers, like, a floor above the central servers.
James Kademan [00:45:49]:
Wow.
Alexander Harmsen [00:45:49]:
And that was incredibly valuable real estate. And then they started clearing away space, like, around like, in the room itself around the servers. Absolutely wild.
James Kademan [00:45:59]:
Yeah. I remember watching a documentary about them running a cable. I wanna say Chicago, New York, something like that. But it was it was interesting because how much time, money, and effort they went in putting that in, and it was all a game of fractions of a second. But it’s just like you’re talking about that arbitrage where they’re essentially somebody makes a trade, and they’re shaving off that little thing every single time you make a trade.
Alexander Harmsen [00:46:25]:
Yeah. And this is obviously, like, high frequency traders of tens of millions of dollars or more, you know, on Wall Street and, I mean, probably some of the smartest people in the world doing that. Like, knowing all that, I actually think it’s very difficult for retail investors to believe that they really have some opportunity to beat the market. Agreed. And Totally agreed. It’s, you know, I mean, this I think this is, like, a big part of the motivation for us as well. Right? Part of it is just education. Part of it is helping people think about asset allocation.
Alexander Harmsen [00:47:00]:
How do we rebalance stuff? How do we get these sort of these other optimization opportunities around taxes and scenario planning? And, you you know, if you could just shave off fees, that might be the best thing that you could do, you know, over the next thirty years. You know? Yeah. Even if you miss every single, you know, major trade and, you know, you miss NVIDIA, like, could be the difference between having, you know, $1,600,000, you know, in retirement and $800,000 in retirement.
James Kademan [00:47:30]:
Yeah. You know, it’s interesting. I had a financial planner way back when, and I learned about Vanguard funds through the John Oliver segment, which you probably know way back when.
Alexander Harmsen [00:47:39]:
Yeah. Yeah. Exactly.
James Kademan [00:47:40]:
And I looked at what I was invested in with my financial planner guy, and he just had me in index funds.
Alexander Harmsen [00:47:46]:
Right.
James Kademan [00:47:47]:
And so I looked at Vanguard index funds, and I saw this, like, what the hell? I’m paying this guy a percent and a half, whatever it was. So I moved all my stuff. And he calls me up and he’s like, James, what’s going on? I saw you moved all your stuff. What’s going on? And I’m like, hey, man. You got me In index funds. I just moved it to index funds because I can just do that on my own.
Alexander Harmsen [00:48:06]:
Right.
James Kademan [00:48:06]:
And he’s like, no, man. But I’m your guy. And I’m like, okay. You’re my guy. That’s cool. Great. If you can tell me the value that I’m getting for that one and a half percent, I’ll switch it back right now. And he couldn’t.
Alexander Harmsen [00:48:20]:
Yeah.
James Kademan [00:48:20]:
And he’s like, no. You can call me anytime. Like, for what? For what? There’s no reason I need to call you. I don’t wanna call you.
Alexander Harmsen [00:48:28]:
I I read a report recently that said that, in 2010, the amount of 40 year olds that were using a human financial adviser was something like fifty percent. Now it’s down to thirty percent. Wow. Like, for 40 year olds, you know, today. Yeah. And, I don’t know. I talked to millennials and gen z In particular. Like, I don’t know.
Alexander Harmsen [00:48:56]:
There’s, like, there’s just a distrust in institutions. There’s more than enough, you know, brokerages that are offering solutions like this. I think a lot of shows like this, right, that, like, tell people, like, hey. You can do it yourself. Like, you gotta justify the 1% fees. Like, I have no idea what’s gonna happen to this, you know, $240,000,000,000 human financial advisory market. And I think I honestly think it’s, like, it’s doing well mostly because people are living longer, and there’s, like, there’s sort of just, like, the baby boomers and older generations that are, like, accumulating wealth. And I think as soon as it gets passed down, this industry just like, something needs to replace In, and, like, that’s our big bets.
Alexander Harmsen [00:49:44]:
Right? They’re like, people still need and want guidance and financial advice. And if I’m doing it myself, wouldn’t it be nice to get a second opinion assessment? And, like, I mean, that’s that’s, like, our big advantage. Right? Like, that’s sort of the the big macro play here. Like, why I can’t imagine that, you know, there won’t be lots of competitors. And if we fast forward ten years from now, everybody’s gonna have some personal AI doctor that they consult and a personal AI lawyer that works with them and a personal AI financial adviser that is just, like, constantly looking out for threats and opportunities. And, like, they’re mostly just in the background. I don’t need to chat with them like I’m, you know, pulling up chat g v t, but an email, an alert, a notification every once in a while. Maybe some simple automations as well.
Alexander Harmsen [00:50:41]:
I don’t know. I feel like it’s it’s so hard to imagine a future that isn’t like that.
James Kademan [00:50:46]:
Fair. Totally agree. Totally agree. Yeah. AI changed the game. He’s changing the game.
Alexander Harmsen [00:50:51]:
Yeah.
James Kademan [00:50:52]:
Yeah. Very true. Alex, where can people find you?
Alexander Harmsen [00:50:58]:
On, I, you know, obviously, I would love for people to sign up for portfoliopilot.com. And, you know, feel free to to shoot me a message on on Twitter. You know, always I feel like that’s the go to place. Reach out to support. We’re, like, constantly trying to get feedback. You know, truly, it’s like we’re trying to make it simple but powerful. Right? That’s the that’s the mantra In 2025. And, you know, it’s all these little optimizations In the interface and how do we deliver the advice.
Alexander Harmsen [00:51:33]:
And, Yeah. My, I mean, Portfolio Pilot is the Twitter handle or, you know, my personal Twitter handle is, or x handle, Alex Harmsen, h a r m.
James Kademan [00:51:46]:
Simple? What did you call it? Simple but powerful? Simple but powerful. Yeah. I love that. That’s awesome.
Alexander Harmsen [00:51:52]:
James, thanks so much for having me on.
James Kademan [00:51:54]:
Yeah. Alex, thank you. This has been Authentic Business Adventures, the business program that brings you the struggle stories and triumphant successes of business owners across the land. Downloadable audio episodes can be found in the podcast link found at drawincustomers.com. We are brought to you by Calls on Call, offering call answering and reception services for service businesses across the country on the web at https://callsoncall.com. And, of course, The Bold Business Book, a book for the entrepreneur in all of us, available wherever fine books are sold. We’d like to thank you, our wonderful listeners, as well as our guest, Alexander Harmsen of portfoliopilot.com. Alex, Call you tell us that handle one more time?
Alexander Harmsen [00:52:33]:
Yeah. Just portfolio pilot.
James Kademan [00:52:37]:
Easy enough, man. All one word, no dashes, no lines, whatever.
Alexander Harmsen [00:52:40]:
All one word.
James Kademan [00:52:41]:
It doesn’t get easier than that. I love it. Past episodes can be found morning, noon, and night at the podcast link found at drawincustomers.com. Thank you for joining us. We will see you next week. I want you to stay awesome. And if you do nothing else, enjoy your business.