AI vs Human Traders in 2026 | Innovantage Podcast #49 | Powered by BMI Executive Institute

Why, hello there and welcome to another episode of the Innovantage Podcast where business meets tech to bring you your competitive edge. As always, I'm your host Max, and it is my ongoing job and duty to explore that edge between business and between tech and to find out something interesting for you to listen to. Today is one of a very interesting series of episodes that we are doing in conjunction with the BMI, Business Management Institute, here in Vilnius and in Belgium. And we are going to be talking about interesting people that are connected to BMI in one way or the other. Today is an alum or a soon-to-be alum. Honestly, I don't remember the timelines. Vas will correct me very, very soon. He is the CEO and co-founder of DHF Capital, and we are going to be talking about AI in trading, in finance. We're going to be talking about his history with telecom and transitions from that, and where does tech and business coalesce between telecom and AI and high-volume trading. So, Bas. Thank you for having me. Welcome to the show. Yes. Tell us a little bit about yourself. Well, entrepreneur for many, many years. I started very young and I always had my interest in tech, in IT, and I had my IT company at a young age. Moving forward to telecom because I saw the shift from IT to telecom, that telecom became really an IT activity. And moving forward from that later into finance. So trading for 11 years at this moment, still on the desk. Sometimes. But yeah, in general, love the technology move now that IT comes into finance. Yeah, like, I feel that those two are now so connected that they cannot be disconnected. But from a layman's point of view, you don't really think about it too much. So can you give us a little bit more insight as to how now technology and finance work in reality? Well, I think everything is tech at the background. So it doesn't matter what kind of company you have or where you work. At the end, the tech side has to work. Your computer has to work, the printer has to work, and that are the obvious ones. But of course, there's always a lot of servers included and there is a lot of data included. And I think that made the biggest difference with AI now, where big data and AI are the biggest partners in crime on the tech side. So you said that you're like yourself behind the table, right? Still trading. How does that look like from your table? Where does the tech come in? Where do you use it and where maybe does it limit you? Well, I think on the tech side, as a trader, you want to have a good setup. You want to make sure that you're connected with the right software to the right brokers and you're able to do your analysis in mostly still a manual way. That switched maybe already like 5, 10 years ago, it already switched very quickly to algorithmic trading for me, which means that you program an algorithm one time and it keeps doing the same thing. You just check as a manual trader, you're checking the entries, you start it, you stop it, you pause it, you watch the news, you make sure that your algorithm doesn't do things that you don't want it to do because it cannot really think, it just executes. And nowadays we are kind of making the switch that it can improve itself, it can think, it can analyze. And that's the AI side of trading that we are rolling into. Hmm. So it's very hard to grasp, right? Like, how do you give an AI so much freedom and does it have enough bandwidth to do that, to actually make those conscious trades, to analyze all of that data? So how do you control that? How do you control the risks of AI, you know, thinking for you and making some bad decisions? Well, that's the question, because what I like to always share is that people also make bad decisions. And when you have a lot of colleagues or you have a lot of people on a trading desk or in general people in a company, that they also make mistakes. So I think it's more the question if you really like AI to be as adaptive as a person or you want it to be perfect. And we are currently at the level that AI can start doing more and more things in different departments. It can write a blog post about this podcast, and it might not be perfect, but it does give you maybe a two-pager where you just have to make small changes yourself. Writing that from scratch, like let's say the old days, would just take you way more time. So it's that combination of the human and the AI where AI can help any person's job, including that from a trader. Yeah, that makes sense. That makes sense indeed. Like a lot of what even we here do, we do a lot of things with the help of AI. And I joked many times on the show that I think AI has enough training data of me, so I can just use AI to do those episodes without actually being here. I think that can be arranged. Though I am still here. This is me talking. I do exist, but You know, maybe not for long. But in any case, I do want to take a little bit of a step back before we delve into the whole AI thing. Over your career, you have made several big shifts, and you have been long enough around the block to see many trends come and go. And AI is the latest of those, but not naturally the only one of those. So throughout your own personal experience, how do you notice those kinds of trends? Like looking back, Did most trends, you know, become what they promised in the beginning, or did most trends just move away into something more? That's a good question. How to spot a trend? I think it comes from natural interest in a certain topic. When you saw telecom like 10, 20 years ago, you basically still saw two wires coming out of a plug that you connected an analog phone to with a fixed number. Then you see the switch to mobile phone. Then you see the switch of office phones becoming more of like a central system. And then guess what? The central system is an IT system. It's not a telecom system. So I saw many IT people making the move to say, okay, I want to learn this kind of systems. When you call at this moment, you call and you say, press 1 for sales, press 2 for another department. That's an IT system. That's not the telecom system. IT people came into the telecom space. And I think that when you see those things happening, you can ask the question, will it ever go back? And if it does not go back, then you're not only on a trend, you're just already on the next level of where that technology is going to go. And finance the same. You can still trade with paper and email it to your broker and execute the trade that way. But, and now you have a buy and a sell button, and the buy and the sell button is now automatically clicked. So it does go just step by step as an evolution of the market, basically. That's a good point. That's a good point indeed. When, if there is no chargebacks, if there is no walking back on the implementations, that is a very good indicator that this is basically no longer a trend, but the new normal, and either you get used to it or you will be replaced by it, I guess. Well, that's where jobs change. I think jobs never really vanish, but they are two different people. They would change the job into something that is useful in the new way of working. But as I said before, you would call your broker and you say, I want 10 shares of Tesla. But now, why would you pick up the phone and do an international phone call that costs you €2, €3 for just that call to buy 10 shares of Tesla. Well, we have a mobile phone and an app, and you can do exactly the same on Revolut or any other platform. So that's the change. And there were people picking up those calls. Now they probably work in risk management, or they monitor client accounts, or they changed into advice. So the jobs are not gone, they just change. True, true. And also to that point, trading existed before the phone existed, right? Like paper, paper trading was a thing where you had to go and mail a specific order for a specific stock, and then the specific stock certificate would need to be specifically mailed from one way to the other. So making one purchase of one stock would take you, I don't know, weeks, weeks, right? And then it transitioned to phones and phones disrupted this whole industry. And now you could do it with just one phone call. And now we look at a phone call and like you said, like, why would you pick up a phone and call somebody to make a trade? That doesn't make any sense. So I am resisting my urge to start talking about the AI's role in all of this. I do want to talk a little bit about your personal journey. So you also mentioned that you are like an entrepreneur basically, or a serial entrepreneur, if I can call you that. From an entrepreneurial mindset, how do you make that choice, you know, where there is so much disruption, so much risk, so much uncertainty. How do you decide that, okay, I'm going to be in even more uncertain, I'm going to start my own business, I'm going to work on my own business, and I'm going to survive in all of that? I think my choices to create a business always came from almost already a proven idea. So just people who are asking to say like, okay, if 10 people ask you, fix my computer, my antivirus, and your printer, and now you have an IT company. And if you do that with a lot of people, you will need some colleagues and you grow like that. For me, the same actually in the telecom space. I just saw the move and I chose to educate myself. I saw the telecom was moving to data centers, which is IT again. And you can always, I think, find the business needs based on needs that you hear in the market. So my telecom company at the end was in 35 different countries. In exactly the countries where the big one, the big telcos did not want to go. So I started with just a list of the countries from Vodafone, British Telecom, where are they? So I don't have to go into those countries. I will go into all the other ones. So it's very hard for a very big corporation like Vodafone to fly to Turkey and make a local server, local data center, and a local connection and a local agreement with Turk Telekom, for example, but as a small flexible company you can, and then you sell it to companies like Vodafone, Verizon, because they are not in Turkey. So I think if you, you need to find that gap of like, where do you, where can you go where others didn't? Oh yeah, that's, that's a good point. You would think that somebody like Vodafone would have all the resources in the world to do anything that they want, but I do agree as a fellow entrepreneur as well, like Finding a niche that is untaken for one reason or the other, be it a geographical niche or a service niche or any other type of niche, they always exist. That's like just how the market works. There's always something that is unaddressed. There's always some problem to be solved. As long as you understand that problem and you can solve it reliably, then yeah, it's a good mindset to start a business. Yeah. And another motto I always hold on to is that I want to be the best Second, so a lot of people are very innovative. They want to be the first into a certain niche, or they want to be the first one who creates an electric car. And that's great, but it's going to cost you a crazy amount of money, R&D development, and a lot of mistakes on the way. It's maybe very honorable to be the first, but it's way faster to be the second because you see everything that they did and you basically continue in the same market, maybe in a different area or different specialism, but if you are the second best, you're still in a very good place. Normian, that's nicely said. Nicer said indeed. That is true. And there are many examples. I can't think of any at the top of my head, unfortunately, but historically there have been many examples where somebody invents something first, but the company that succeeds is the one that standardizes it, that scales it, that controls it predictably, and so on and so forth. So yeah, that's definitely something that is more valuable, I guess, in the long run. So what happened to your company in the end? Did you, what did you do with it? Well, the telecom company, I was fortunate enough to be able to sell to one of the larger telcos, Vodafone, because to a daughter-related company of Vodafone, every, all those companies have a wholesale department. So it's basically, it's normal for a Lithuanian person to call to a Lithuanian number, but when you call outside of Lithuania, there is a big chance that they are using a third party or their own network if they have it. So if you build a network with all the countries they don't have, plus the software, because I found that the biggest issue was the currency part, because it's not hard for a company, a big company to go to Turkey, hang up a server in a rack and plug in a cable. The issue is that you also have to deal with a foreign currency, foreign policy, a lot of relationship building. And I think that it's not worth it for them. It was too small to go to that specific country. But in the total volumes of everyone combined, there was still a lot of volume going to different countries. So I did select on volume how much is being called to those different countries. And then you go there and you create, you create your setup. And what was very interesting that after about 3 years, I got my first offer from another company. And I think when an entrepreneur gets an offer like that, it's very important to say no, because for me it was just a confirmation that you're on the right way. I do think it takes a little bit more than 3 years to build a successful company. It's very nice if you're profitable even in 2 or 3 years. Nowadays that can go faster, but at those times it would take a significant investment to build up a company. And then, yeah, it's good to keep going. Maybe say no to the first offer you get and see what comes 1, 2, 3, in my case, 5 years later. So after 8 years in telecom, I did make that decision to, yeah, to hand over the company to a bigger firm. And it's a success story. That's what most entrepreneurs would want. Because again, like culturally speaking, you hear a lot about those visionary entrepreneurs that disrupt markets and change trends and become the top of the top of the top, you know, your Facebooks, your Googles and whatever. But those are just singular examples. There are many entrepreneurs, thousands upon thousands of people that have successful businesses that are profitable, that are growing, that are, you know, that are solving real problems on the market. And to exit from that company is a sound strategy a lot of the time. And the question then becomes when to do it. And I very much like what you said about saying no to the first offer because the first offer is definitely not going to be your best one unless of course you're, you know, have that already. Yeah, or even if it would be the best one in numbers, it would maybe not be the best moment for you. I think it's a nice crown jewel even to get an offer. There's a lot of companies who would never be sold. They just vanish if they are not able to have generational continuance with family members. But I think it's very nice that when you see that someone is that level interested in your project, in your company, that you're just on the the right way. Yeah, that's true, that's true. So my question for you would be, how do you find out which offer is the best one? Is it a feeling? Is it based on data? Do you hire consultants who can tell you the answer to that question? Like, what do you do? Well, I would definitely advise people now, knowing my experience, that you should have a consultant because the first thing what happens is that they will let you sign an NDA. And you probably have to give the people that you are going to talk to about this deal in that NDA. And at that moment, as I was the only owner of the firm, I did not sign anyone else, which means you cannot talk to anyone about your offer. So it took me about 6 months of negotiating, I think, and I did not list any advisor, I did not list list any family member. I didn't list anyone, which means that it's a very lonely process if the other side has 17 lawyers. So it's definitely another advice that I learned the hard way, that it's very important to get support on your own side. List your own CFO, which like is still not sure because they all have double motivation as well. So an external firm is very good advice to get on board. Yeah, I've never thought about that, honestly, like, but it is a good point that you make indeed. Like, once you're in, you're in. So how did you survive? That sounds like very hard to do. It was a very interesting period, I must say. What is very nice is that the person who negotiated the deal at the end gave me a book and they said we basically just used every trick of this book one by one. We used all the different valuations, chose the lowest one, tried to convince you. I still remember a very interesting fact when I had actually a meeting with two owners of the firm and one of them would not show up. The other one was like, well, you can sit in my car, we will wait with the other guy. He's not coming. It's like a whole playbook, basically one by one for, okay, we will just call him on the car kit. Very negative conversation, then okay, we will go and have something to eat together. So you have one friend, you have one enemy. There is a lot of plays that are happening when you negotiate price and terms and conditions at those moments. So what was the book? Do you remember? I do know it was a Dutch book, so it is not an international one that I can recommend, but it was like basically a summary of the 7 or 8 valuation types for small-medium businesses plus ways to negotiate, let's say. Okay, okay. Well, kudos to you. You survived, you did it, you actually succeeded, and then you went into finance. So why finance? Why was that the next venture for you? Well, for me, it was like I basically developed a software telecom company, and the software's main purpose was to handle 7 different currencies. So what we did is the moment you picked up the call, we would already analyze how long are you normally calling to this number if we have that data. Otherwise, would you call to this country more often? If you never did, maybe your peers or other people of the same profile call for average 32 minutes to Turkey, for example. How much Turkish lira would we need if this call is going to be 32 minutes and we would lock in the currency trade at the start of your call. And then when you hang up, we would exactly know how much Turkish lira was spent for our euro-paying customer. So we would lock in the trade, reserve the currency, had 7 different currency accounts at the bank, which means that when the invoice came, I already knew how much was going to be on the invoice before the invoice of my suppliers would would arrive, which would mean we could pay the same day. And any other company would at that moment start like, uh-oh, we need this foreign currency. Let's go and convert it, easily lose a week or two. And in telecom at that time, 60 or 90 days payment terms were normal. But if you are the only one who pays the same day as the invoice comes, or you can actually negotiate about the invoice, because if your data gives a different number, we would spot the differences and mostly they were spot on, everything automated, and then you can get a lot better conditions with your suppliers. So my goal was basically to do that and then compare to other suppliers who had 60 days or 90-day payment terms or prepaid even. Yeah, they were not knowing what they were spending in all those foreign currencies. So that's what this system was actually solving. And that way wholesale telecom is about 7 to 9% gross margin if you're lucky. And we were on 11 to 13. Oh, wow. And it was mostly payment conditions and doing the currency at the right time and not a month later after the call actually happened. That's a very great example of that competitive advantage that comes from unifying business and tech, basically, so you understand the business problem so well, because, you know, I would never have thought of that. I don't know that the disparity between when you start a call and then you end a call specifically for like currencies that are more volatile, like the Turkish lira or any other volatile currency, like that there's a real difference in the— with smaller margins on every single transaction, those differences actually make makes sense in their— If you have 7% gross margin and you pay 5% currency conversion, then there you go. Yeah, yeah, yeah. And it's a market that is unaddressed or under-addressed. That's very smart. That's very smart. So how does the company look like today? What countries do you work with? So Turkey, you mentioned, like, are you distributed or are you centralized? No, so that company I left in 2015. So that's 11 years ago now. And basically, I think that that if you pick up the phone and you have Telia or Beta at this moment and you call to certain countries that might still run over the servers that I started at that point because that is just an international system that got integrated. Okay, okay. And what about DHF Capital? How does that look like? Yeah, so I, because I was actually way more in finance at the end than in tech in the telecom company. We were so much more working on the currencies and on the currency conversion and on the markets, I started trading as a hobby in the FX currencies, in oil, and in gold. Yeah, so those became my specialties as a hobby. And at some point, when you leave your company and you keep doing that, people start to say, well, can you not— or teach me how to trade, I do like that as well, or just trade for me. So we started with a family, friends and family fund, which in Netherlands is allowed up to 50 people to just like manage funds for friends and family. And that got full very quickly. And then we were like, well, we have to do something in a more professional way. And that started DHF Capital. So who do you work with in DHF Capital? What are your target markets? Kind of verticals? As a couple of the founders are from the Netherlands and also some C-level staff is from the Netherlands, we have a lot of Dutch clients, but they are all from Benelux, they are from Europe, they are from the GCC, Middle Eastern regions, they are from America. We have clients in 19 different countries at this moment. Wow. Wow. Yeah. So does like trading change over those countries? So like you mentioned a very specific difference with Turkey, like different currencies, but the US is going to be the same. And even inside of the EU, things are very different. As far as I understand, like even inside of like Belgium, you have so many different things and different regions in terms of financial regulations and stuff. Definitely. Every country has its own local regulator. Now, the nice thing is that we are a professional investor-only fund. So the retail regulations are in general that are per-country-based do not fully apply to us. Yeah, we do make ourselves known to the regulator, but we are a Luxembourg-based company and Luxembourg-based funds as well. And any professional investor is welcome in Luxembourg. So that's also why we did choose Luxembourg as basically the number one active funds country. There are some other countries like Cayman Islands and other locations that have have in quantity more funds, but a lot of them for single transaction purpose or things like that. So Luxembourg is kind of the place for funds to host themselves. And a lot of people, when they think they invest in Switzerland or they invest even at BlackRock America, if you actually look to the ISIN code of the financial products, a lot of them start with LU, which means that it doesn't mean that that you actually invest in Switzerland. You do. The manager can be in Switzerland, but the fund registration and the domiciliation of that fund might just be in Luxembourg. That's fun. I never thought about that. Luxembourg is such a small country that doesn't really come up in a lot of conversations, but that's a very interesting point. Anyway, so then what's your goal for the company right now? Is it still more of a hobby where you're just having fun and that's it, or like an entrepreneurial mindset where you have to reach a specific goal? No, we definitely have been growing a lot. We basically doubled every year over the last 3 years in assets under management. So AUM is what for financial companies mostly counts as importance, as most of the fees are based on how much money you manage. And we definitely plan to grow times 2 for the next 2-3 years. A lot of our larger competitors, they are also all stock listed. So I cannot say that that is not part of a dream as well. So the company is growing and we are doing really well. I'm happy that we have a Lithuanian team. Dutch team, Switzerland team, team in the UAE as well. So we are having a serious amount of funds under management and manage that with a great team of different departments together. Well, happy to hear that. Happy to hear that. And I hope that you will succeed in getting a ticker out there for DHF. So then let's finally talk about AI and the current role of innovation. So where do you see, like, first and foremost, right? What's your opinion on AI? Is it disruptive or is it augmentative? Is it like a very noticeable change in how trading and finance is currently done under your management, or is it still not there just yet? No, I would say it's there. And if you asked me that 2 years ago, it was all new and people people discovered ChatGPT and they were like, oh my God, what is AI going to do? And is AI going to change our lives? And I'm all going— everyone is going to lose their job. Were very interesting questions at that time because it was so unknown. I think that is changing too. It's more— it's not a trend. It's just something that's here, maybe was already here, but not so known to the public. And you start seeing some companies that just use it or employees who use it to do their work better. They might do it faster. Example of the blog post, it's now done in 10 minutes with some edits. So I think that AI is definitely here to stay. It is improving by the day. I am subscribed to some news bulletins that show the last 24 hours of AI and every day it's a new surprise. So from that perspective, it's growing tremendously. But also only when you look. So that's also a very interesting fact I saw that because if I open any social media, for me, my feed is filled with the latest AI tech things. Same. But for people in general who are not looking at this topic, they are completely missing all that kind of technical innovation. So And I think there were some studies about how many people are actually using AI in, for example, America, and it's like 2%. And we in AI maybe think that everybody should use it or everybody is using it, but that's absolutely not the case. And I think it was less than 1% uses AI on a daily basis. So it has not hit the markets. It did not hit or influence everyone. And some people changed Google for ChatGPT, and that's their new way of searching for information, dealing with bigger data, getting multiple sources. So some people have figured out, but what's actually interesting is how many people have not. Yeah, that's an interesting statistic. I would, you would think that a lot more people use AI and a lot more companies, you know, mandate that people use AI in their day-to-day work. Well, we saw some studies at the BMI from various professors as well there that companies are forcing people to use AI and they have whole adaptive programs and training programs and everything. And at the end, for example, a company says, okay, everybody's going to use Copilot for their work, part of the Microsoft package, free. So why not? And at the end it was still shockingly low numbers. Not more than 5% would actually use it on a daily basis. Is it the same that you view internally in your company? No, I think at our company 99% of people are using AI on a daily basis. So what's the difference then? How do you make that change if most people don't and then people in your company do? Well, even though we are with a seriously sized team for SME companies, about 40, 45 people. I think that we are acting as if we would have 200 colleagues if we would not have AI. So I think that the pressure is high. The amount of things we want to get done on a daily basis is high. I really like small teams that are having a goal and they just want to achieve it faster than any other larger institution can do. And that's how you catch up to second place. So when you have a small team, you can mostly run way faster. The decision methods are way shorter and people are having real autonomy as well. No 5 different checks on your work. You do your work, you're specialized in that area and you execute. Makes sense. Makes sense. You know, one thing that I noticed professionally when I work with different companies when implementing innovation in AI as well is that a lot of companies struggle to count the impact of implementing AI solutions or AI workflows or just AI in general into their companies. So like, yeah, the mandate can be there and people may be using it or maybe they are not using it, but it's very hard to quantify like to what degree do they use it and whether or not it brings actual business value. Is that something that you struggle with as well, or have you solved that? No, it might be hard to quantify if you look at the larger KPIs. And I would say that it's maybe better to check on the how much things can you get done without hiring more people. So if you have a certain team and you work together, but you want to grow, the normal way is always, okay, we'll hire someone. So what we did at DHF, we kind of put two rules in place. One is a higher stop in general. So if we have anything that we feel a person is overworked on manual work, we first try automation. We first try to see if processes can be improved before directly saying, okay, we need more people. Yeah. Second, very interesting one is when someone would leave for whatever reason, We ask an IT guy first to say like, okay, we will describe what this person did. There might be SOPs, there might be manuals, there might be all kinds of paperwork that supported that person in their role. And if you would have 3, 4, up to 6 months of salary of that person as an IT budget, would you actually be able to automate this role? And the interesting thing is that people don't realize how cheap it is to actually automate a person's role. So it's never been 6 months. Well, normally it would be at least 1 or 2 months of hiring process. Then the new colleague starts to work, takes 1 or 2 months to get fully familiar with the job. So you lose those 4 to 6 months of salaries anyway. But if you tell a person, let's say around, it's a €3,000 gross salary employee. If you give €12,000 to an IT guy, he goes through the roof. He's never seen such a budget for one simple task. And that task is do everything that that person did, but in an automated way. And we've not heard no yet of a job that cannot be automated. Okay, that's both very fun and very scary. Yes, I agree. So what kind of jobs do you think are resistant to that, at least in your own understanding and scope? Well, on the drive here, reading your questions, I actually thought about that because my first answer from the AI perspective is like basically any job. Of course, you can have physical jobs, firefighters, people drive the ambulance, even that may be robots later. But if we talk about AI, I think that most jobs behind the desk, those are the jobs that are currently the first one to be more and more automated. Very simple example is open a new bank account. The ways 10 years ago, you would probably have to go to the bank. Then a few years ago, it was an online application with the person reviewing it. At this moment, no one is reviewing your bank application anymore. Your KYC, you make a picture of your passport, you do a selfie. After the selfie matches the passport, you have your IBAN in 5 minutes. There is no person. If you do that at 3 o'clock at night, there is no person that accepts that client into the bank. And of course, at some point there are all kinds of manual checks and activities, but compared to a couple of years ago, a lot of that became, or just IT automated, or even with the help of AI matching if this new client should be a good new client for the bank. Yeah, that's true. That's true. And I think a lot of criticism of AI comes down to the fact that it still hallucinates a lot. Like that's just physical limitation of any LLM, generally speaking. And AI is much larger than just LLMs, of course. But the reliability of those processes are always in question. From your experience, again, when automating those kinds of roles, were they as reliable or just reliable enough Well, I think that this kind of systems all work with, with certain flags of what would be approved. Like perfect case gets approved, uh, 99% score maybe still gets approved, but when it goes to 95%, 90%, or AI is not sure, that's when you build the right system. So if the system asks for human approval, um, it does mean that if you have 200 new applications for a financial product or anything, maybe at least 180 of them can just go completely without human interaction. And then you only need a human for the last 20. So I think that that's where we are going, that more and more does get automated or does get its green flag based on the support of AI. But there's always a human factor for the exceptions. That's true. That's true. There's always going to be a human factor regardless. Like, you can have the best tool out there, but if the wrong person is using it in the wrong way, then it's not going to work. It's interesting because I do feel that I can add an extra colleague to your Teams or your Slack or whatever communication system you use, and he can introduce himself as a new colleague. He can have the exact skill set for the role that is being planned is basically AI agentic level. And you wouldn't know that it's not an actual person that just happens to live somewhere else. So that can start to happen that actually like colleagues and people who have more of the agentic AI, the agents, that they actually can really work as if it's a person. And then from management perspective, you should accept that that that person makes mistakes. True, true. Yes, yes. And I guess then you have to change the mindset of how you manage those kinds of people. Correct. Yeah, because it's not much different than managing real people. So what's the main difference then? Like, what do you look out for when managing AI agents versus actual people? So for example, if you have an AI agent, everybody is so happy that they work 24/7. But if you are not able to manage, check, or control this AI agent 24/7 because you want to sleep, you can also make your AI agents sleep for a couple of hours or say, do this during the night, give me a message in the morning. In the morning, you have a summary. I've seen multiple AI agents have a Scrum meeting together. They literally walk to a table and you see this person's talking, person, this AI is talking, this AI is talking, this AI is talking. You are observing, and they literally walk back to their desk in a visual way and start working on their own tasks. But you can say, okay, I want to have this Scrum meeting every morning at 9 o'clock. I want to read the summary and then till I approve the tasks of today. So it's all about approval. Give an AI agent $200, but still ask it every time it wants to spend something to press approve or not. It's a way of still controlling. And I think that it just gets better and better that you at some point have an approve all button and you do not approve single things anymore. Or you build limits, everything less than $5 approve, but more than $5 ask for permission. Yeah, makes sense. Makes sense. And it also kind of builds up trust, I'm guessing, right? If you just see everything that's going on, click approve every single time. And then you understand that, hey, I haven't clicked disapprove in like weeks at this point. So yeah, why not automate it? Yeah. Yeah, it makes sense. And maybe a good thing to mention is also that AI is not always cheaper. So I've also seen a programmer that went quite active with his Claude AI agentic bot. But after 2 days he woke up to a bill of $450 of AI spend. Of course, after that he, like, let's say, improved it and made sure that that would not happen again. Sure. And the very interesting question for me to this guy was like, what, are you happy that it spent that much? And he was like, okay, I'm not happy with the number, but it does mean it did a certain amount of tasks. That I would, I would still have to do. Yeah, they are now done. So it's like I'm not happy with the bill, but I'm still happy with all the things that, that got realized in those two days. Yeah, makes sense. Makes sense. I want to talk a little bit about like the perspective of the company and the management of the company when introducing and implementing AI processes and tools. So in that scenario that you've described, right, you have an engineer and you have access to AI tooling as like a manager or a business owner, how do you make sure that people actually, you know, create value with AI tools and not just spend $420 or whatever just on doing tasks that can be done cheaply or faster in some other way? Yeah, it's a good question. I think that when entrepreneurs start with AI, they want to start with the low-hanging fruit, like the processes that are done every day the same. There's a lot of quick wins you can do. What we did is we emailed basically everyone and say like, if you have any repetitive task, let us know now. And then you already get like an insane list of things that people just happen to do every day. You might not even know. And those are the low-hanging fruit for automatization because because you still have the same people. Now those tasks are done and they have more time. They have more time of brainpower and things that AI cannot do yet. Logical thinking, strategic thinking, and they have more time for that. And then they get again repetitive tasks and you just keep improving. That's good advice. That's good advice, honestly. I'm looking back at how we implemented AI internally, at least into processes, and it was very, very similar to how you described it. We went for the low-hanging fruit, naturally speaking, and we saw what things are repeatable and easily repeatable and how we can replace those things with automation and processes and whether or not the output is going to change. And even if the output got worse, but how worse did it get and how much free time did somebody else gain in return so they can do something else, something that is more valuable because, you know, like issuing, I don't know, invoices, for example, fairly repetitive, fairly predictable. You only need to check if something really changes. That can be automated a lot of the time, even without the use of AI, but it's still a very critical business process where, you know, it's the invoices, it's the actual money coming into the company and there's a balance between what you can automate and what you shouldn't automate. But for something like, I don't know, announcements across the company of what changed in every single department, an AI bot can do that much more faster and much more efficiently than anybody else that would have assigned that task to. Yeah, and everything with a certain level of human oversight and approval, because it can make all your invoices and before checking, one button approves to send them out. And very interesting way, the other side is the amount of invoices a company receives. They all come to the same mailbox. AI can scrape them, they can read the invoices, they can make an Excel sheet, they can probably upload it to your bank and you only press approve from the bank's perspective to actually pay those invoices. But all the steps before that, there's really no human need. True, true. Do you have like maybe any specific examples of processes that you've seen that were at first glance very human dependent that you couldn't automate and then you ended up automating somehow? Well, I think that every question when someone says this is a repetitive task for me is a little challenge by itself to see like, is this actually possible? Where does the data need to come from? How are we sure that it's going to be consistent, that it doesn't make a report of the S&P 500 pricing that you think is the same, but it's still on this website, it's a different price than the other website, and that it doesn't take different sources and that you get inconsistent data. So I think every process still takes time to one by one execute those automations and create the flows for that. But when they are done, they will never have to be done again. That's true. That's true. Aren't you a little bit afraid maybe or cautious about the potential for your clients basically understanding that? You know, like creating AI agents that do trading for them and automating the whole kind of process of analyzing the market, understanding the market trends, building exit strategies, maintaining portfolios, and so on and so on and so forth. No, we've been very clear to our clients. So on the 5th of January this year, we opened a separate fund, a separate hedge fund called the DHF Nova Fund. And it's 100% AI. There is zero human touch. It built itself at this moment 187 agents that are all trading their own one strategy. It has multiple agents as colleagues, one for risk analysis, one for watching the news, one for taking money from one trader that's not doing so well and give it to the trader that does better. It's a no human touch fund. While our other products are still just the old way as we've been doing for 5, 6 years. Because there we made a promise that we would do things a certain way and we keep doing them the same, which was already algorithmic trading, not fully manual, a bit of a mix, but still the use of technology, but not the use of AI where it can go either way. Surprisingly, in November and December, we had quite a lot of investors interested. No track record, no idea where this would go. It could double the money or lose it all. We did build in risk factors as you do with humans as well. So the fund has a 30% risk, which is about 3% a day maximum risk per agent and 9% a week. So there's automatic levels where if an AI agent does completely the wrong thing, that it just gets blocked on the daily or weekly levels and it will probably never reach that 30 because we manually would see that. We get alerts and reports and everything. So we have manual oversight, but there is, yeah, the rule is not to touch it. So when will the experiment, if I can call it an experiment, end? Like, what are you, what's the next checkpoint? I would say that when we started in January with the initial few million in funding, we basically said, okay, let the AI start itself. And as every other trader, you need some track record. So it's, it has a very slow start because you cannot really optimize what you don't know yet. So the most important one is the optimizer agent that is looking at a certain trader after every 100 trades and says like, okay, you should do something different. So change the algorithmic settings. So for now, end of March, only a few agents have reached that level of 100 orders getting optimized for the next month. So we expected about 3 months of initiation phase, which is almost over. So the exciting part is going to come now because we see small, small negatives and small positives, but they kind of leveled each other out. Now the goal is that the positive ones are staying positive and the negative ones are getting optimized. And then we do have great expectations for, for the fund as a whole. Well, it would be interesting to hear what ended up happening in the end. I'm very curious to hear whether or not the AI trading can, like, maintain more or less the same, the same level as human-first trading, if I can call it that. But regardless, even if it doesn't happen right now, I do believe that this is one of those areas where AIs can be much more competent than humans. We tested with our own data and this AI system in 2025. So it was already running without client funds for, for almost a year. And what we basically saw was 32% more efficient. So I don't say better, but as a trader, we have big news moments and we mostly know them before they happen, like on Farmers Payroll every first Friday of the month at 2:30. What we do is in the morning, you already don't trade anymore. You at 12 o'clock, you go for lunch, you probably close out all your euro and dollar orders. You wait for the news. The news happens after the news. It's Friday afternoon. You maybe don't want to open that much news. New orders anymore. So basically that whole Friday is gone. Mm-hmm. AI stops 30 seconds before the news, waits for the news, checks what the news did, acts directly upon the news maybe even, and starts trading 30 seconds later. So instead of losing a day, you lose 2 minutes. So that is a very big efficiency level where AI can just outperform a human that does have lunch, that does go to the bathroom, that does sleep. There's simply an efficiency there. That's a very good example as well. Interesting. Yeah, just by volume of actionable things that an AI can process, they're always going to outperform a human. Yeah. Yeah. Well, like I said, both interesting and scary because then at the end of the day, like, what's left, right? Like, what do you do when AI automates everything? It does the predictive thinking, it does the trading, it does like most of things in your life, like what's left? Yeah, that's an utopia that some people are trying to predict. What are we going to do? I think that what for us was mostly very interesting as a company to go also this this direction, as again, we didn't want to be the first, but it looks like we are kind of one of the first who are going on this scale, at least for companies of our size. Yeah. But what we saw, I think it was 2 or 3 years ago, that there was research done by eToro clients, which is a copy trading platform where people can invest and copy their money into another trader because they do not want to trade themselves. And they interviewed a lot of their customers under 25 years old and asked at that moment, would you accept if your whole portfolio would be traded by AI? Would you trust that AI does a better job than anyone on here? And I think it was 70, 80% said yes. So that generation is going to come. So all the companies in finance or in trading that do not use AI in any form They are just going to be behind on the ones that do. Yeah, yeah. Time will tell, I guess. We will see. I am cautious about one more thing that AI gets criticized for is the resource drain, because right now a lot of technology providers in one way or the other, like OpenAI and Google and NVIDIA as well, like They operate in a principle of cornering the markets with low prices and then raising the prices once the market is being dominated. The classical Silicon Valley playbook thing. If AI prices rise to a point where it is cheaper to just hire, you know, 2, 3, maybe even humans instead of one AI trader, when, like, where does that level of efficiency need to be for both AI and humans to balance that out? Yeah, that will be a question. I think there will always be different models for different segments. And I think when companies are using AI, they need to more think about security, about data protection, and not give AI the full keys to the kingdom, let's say. We also build a data center here together with Delsky in Lithuania. We have our own hardware and you can run your own LLMs locally. You don't have to upload everything to the larger data centers. Since November last year, we had a collaboration with xAI from Elon Musk, and we basically outsourced the heavy tasks because he built a data center for $6 billion, and we did not have that. Yet. So not yet. But these big companies, they are basically outsourcing their computing power. And that's what you're paying for. What's still very interesting is that as of today, maybe about one request that you use in a high-level request that we would use would be like 5 million tokens, which is like a euro. So instead of even buying your own servers and paying the electricity and paying everything that comes with it, it is still very far away from that. That's actually a cheaper option than to outsource it to the big ones. So it's a combination. So we send only anonymous data away and all the high-value data we do locally. So that's been our choice on the security side and safety. That's a good point. Yeah, we haven't really talked about security and safety so far, and it is a big concern, especially when it comes to trading to finance to personal data that is involved in trading and finance. Yeah. Yeah. Well, if we give you 100 trades that already happened, there's not much you can do with that kind of data. If you're talking about onboarding clients, if you're talking about KYC, know your customer, if you talk about passports securely saved, I think none of that should be in any cloud, even though a lot of big companies do have that outsourced as well. Yeah, but we always, we have the IT background. We, at this moment, I think we start to have more IT people than other departments shortly. So we are able to manage this all in-house. And I think any company can find a way to do that in the most safe and secure way as at least a private cloud or something safe to connect and then use AI to deal with that data. Yeah, true, true. And as like somebody who works in the space of tech and security and cybersecurity and all of that stuff, there are many different workflows and specific solutions and approaches that you can utilize to be more safe. So using AI by default does not mean that you are vulnerable. There are ways of addressing that. And I'm glad that you guys are. But yeah, like I do want to talk a little bit about the future and your feeling towards how it's going to look like in terms of like where technology and people will coalesce and what will change. So how do you see it moving forward? At some point, will you just be a one-person company that controls a lot of AI agents and that's it? Or a balance somewhere in between the two? Or it will pass when, you know, Elon Musk decides that now xAI will cost like 10 times what it used to cost? Yeah, well, it needs to go a lot more than that before the price of IT would be too high, let's say. I think that we get to a nice economy there. Running a company alone with only AI agents, that's no fun either. I do see setups where companies from scratch are built with only AI. So if you have a company that leans for that, that. I do believe that's possible, but I've always believed in the combination of man and machine and not versus. So it's maybe not too many humans, but adding more AI, adding more technology that can scale companies that don't have to grow in people but can scale as a company in revenue or AUM or any other metric that they have. In combination with using AI. So that will be interesting. So what then is your advice to people? We have talked a lot about how AI and other types of tech, like, disrupt the industries that people just work in. You know, traders trade. Now AIs will trade probably much more than people, and that's just one example. So how do people then adapt? What kind of skills do they need to invest in, or what do they need to do? Well, 2 years ago, I would say go and learn Python programming and just turn yourself into an IT guy or girl. But that will, that will not be the advice of today. I would say that it's very good that if you can use any form of AI in your current job by just doing it faster, that helps you standing out from people who don't. And on a whole company level, the companies who do adapt versus the ones who don't. And that probably makes you more scalable, able to do more things. And I think that that is mostly the future for individuals and companies to look at. Yeah, makes sense. Makes sense. Or become an entrepreneur and build those kinds of one-person companies with a bunch of AI agents. Well, then you need to start to learn Python again. Well, thank you, Bas. Thank you very much for your insights. Where do people find you? Where can people read or hear you, more of your thoughts? Well, I am online, of course, on the different platforms, LinkedIn, Instagram, and at DHF Capital. People know where to find us, maybe, to see how those new developments are going. We have on the DHF Capital website a separate blog purely on the NOVA fund and the whole AI project. And we have a human-readable version and a tech version. And I think it's an interesting project to follow. And at some point, as I said, we only work with professional investors for now, but the goal always has been to make our products available to as many people as possible. So on the long run, that might change as well. Yeah, yeah, we will see. We will see. Thank you very much, Buzz. Thank you. Thank you for having me. And thank you all as well for tuning in and listening to our episode here today. What do you think about it? What do you think is the future of trading? Is it going to be AI only? Is it going to be human augmented? Is it going to be old school style when we have Skynet over us where we will have to mail in ballots? From across the world, and only then will trade happen? Tell us in the comments. Tell us in our private conversations on LinkedIn. Tell us anywhere that you can reach us and subscribe for more of our episodes of InnoVantage. Thank you very much and bye-bye.