AI-Native Finance: The Next Financial Evolution

Transcript

Hi everyone, good morning. Um, thanks for uh still make it to uh the last day's uh morning session. You guys are really passionate about learning uh AI, crypto and finance. Um I'm very happy to be here. Um and I just realized my outfit fits my uh the color scheme of my presentation. So uh I'm very uh coordinated today. Um so I want to talk about uh what does AI first fintech mean? Um you know fintech has been one of the biggest innovation um in the past 15 years. Do you guys know since I mean you're familiar with the term right?

Since last 15 years how much progress have we made in terms of the percentage of fintech as a total percentage of finance. Do you guys think it's 5%. Less than 5%. What do you think the percentage is? Maybe two. That's a very very good uh very good estimation. Can you believe after all the work we've done the past 15 20 years or less than 1% of the total financial markets? I don't think this number includes um some of the crypto revenues. I think if we can include the crypto exchanges revenue and some of the onchain uh revenue or maybe bitcoin I think we're actually a bit more we

probably um make it a 1.5% crypto plus fintech so we have a huge way to go long long way to go in fact you're not late at all we're all really really early in trying to find uh bring real technology to finance Yeah. So right now um the moment we really are in is a financial system is incredibly complex and it's beyond most of our comprehension or capability to even understand how a financial market works. But who is the best at handling complexity that they created? I think it's AI themselves.

So look at the the 2010s uh first wave fintech companies. They have done a lot to improve both reducing cost, improving performance and also improving user experience in how we interact with finance. But AI native finance will look very different from workflow to intentbased. I really think finance is AI's perfect target. There's a lot of structured data of course in finance that machine learning is approaching is is already working on most of financial execution are done by algos as you probably

already know but there's a lot of unstructured data that is really complex to for human to understand and there mountains of the data not only um social um economic data but also a lot of regulatory no uh new laws and new regulations is really making startups even harder to enter the financial markets. So I think actually regulation is a mode for AI instead of um what we think today crypto people think crypto is on the other side of regulation but I actually think crypto is better built for regulation

crypto and uh plus AI of course. So the AI first operating model um I think we all know now like we don't really trust AI bullying. um probably now I would think 5050 right also do some of the work for us and we do the approval process right we still need to monitor what AI uh does for us but in the future I think AI core will be 90% of the work and human will only do help to manage and monitor um and probably setting the guard rails instead of um being in the center of it I think biggest re revolution for me is

really the interface because most of the interface we do today whether it's crypto whether it's Robin Hood is still very much um we learn how to we have to still learn how the interface works and we need to adapt to uh the interface I think the next stage will be AI learn how we interact how they need to build something to customize to us so maybe for me I still prefer using a mobile phone, using a browser, but maybe the next generation only wants voice. So I think native AI native interface is actually built customized for each one of us. Um I have uh actually my kids in the audience and

I do believe when they start to manage their money it will look very different than uh what what I what we do today. I really do want to talk about regulation. I think regulation is um is actually a competitive advantage for AI first companies. Like I mentioned earlier, regulation is extremely expensive. We I came from um traditional finance. I used to work at Goldman. When we first started, compliance is really a small part of the bank and in the end it's really runs the bank. Everything has to go through compliance. Many deals are blocked or delayed by compliance because it's a ever more complex society and

it's difficult to understand if we're compliant or not. But who else is better at cracking compliance? It's really the code, the automated, consistent, and extremely scalable. So I think crypto's competitive advantage is actually smart contract and be compliant because reinforcement is built in the code. Yes, this is a just a quick chart of where I think um AI natives will be. They will be very fast but smaller amount.

Um I really like to talk about Robin Hood. is one of the companies I really admire um who has been at the cutting edge of innovation. Well, I think I think everybody here has heard what Robin Hood has announced. they have not only adopted crypto but adopted blockchain and building their own L2 and owning the full stack of not only from users, distributions, block space and um and of course uh they'll be using AI in driving their um products and performances. So just this year they also announced a fully 100% AI managed asset management product.

So if Robin Hood is already experimenting with the strategies, I think most of fintech companies in the next few years will start to adopt AI first and OA only investment uh investment strategy. Has anybody use Robin Hood? Have you guys tried the AI native uh fund yet? Yeah, I think I'm going to start to put in a little bit of money in and see how it goes. I think it's u before we know it, I think they're going to start to manage a lot more of our money.

So, um I picked three companies um I want to tell you about. Um I'll come back to a bit more uh features that I think are important. Um and I will cover one of the companies more extensively because we are building um super fund inhouse. We're trying to use AI and crypto to disrupt our own business. Um so super fund is an air power prediction market and um a a fund matching platform. We want to automate the 500 billion venture capital industry. Payman is one of our portfolio companies. They are they believe human will be working for AI. So they're focusing on building

a payment reel for AI to pay human. Skyfi is not our portfolio but I there are many companies uh we have seen in the past year are building AI to AAI payment you'll probably have seen or maybe building one of these another thing that um I want to come back to AI is that the number one question is like do we really trust our wallet I think this only talks about uh fraud because uh fraud is probably the number one problem that tech companies have been using in the past probably decade to detect to protect our money.

Uh I think the next hardest thing is do we trust AI with managing our money on the strategy side. But I do think at least for now this is very well used uh AI is very well penetrated in um the security or the protection or the fraud industry. Um so I think the next thing we see is so we definitely know that AI can protect your money. Next thing is whether they can make money for you. Um, I think we're still probably just here act with oversight. Uh, I don't think we're at full autonomy. Um, when just quick uh question, when do you think we'll be uh we'll achieve full autonomy?

Do you guys What do you guys think? 10 years. Sorry, 20 years. What do you guys think? I think it'll be much sooner than uh 20 years. I think maybe five to 10 years we will probably get to a full autonomy. I think these are just very small like incrementals that we probably want to work on in the next few years. Um I actually have some other ideas I will show you guys at in the at the end of the talk. Um that goes beyond what I believe is AI assisted economy to AI native um economy.

Um so if you are a big company or if you are a small company trying to build a lot of people I ask me this question like what is really a moat for the fintech companies now because we believe like data is the moat because the more data we have the existing users we have which means the existing companies win which means like the it's harder for the um smaller companies to come and build in the space. I do think proprietary data network is very important in terms of driving insights right because if your customer your bank has your data naturally you would think they have a better understanding of what you prefer what you like um I also do think I

actually have one of the investment uh we have one portfolio companies building AI native um kind of trading strategies and uh they actually told me they don't even think by prior data has an emote because they use synthetic data to create a lot more scenarios. So I think it's both true and false that data is a mode. Um because I think for especially like certain use cases like trading um synthetic data will become even a bigger um force to help improve the algo. But I do think compliance is a code is a is a really as a mode. um network effects of course

specialized AI models. I think this is um this going to be um for me one of the biggest mode and also why um we're also building the model for ourselves right now even though we know AI is not perfect um because we think the the more so we're training our own uh or improving our own models for um our VC uh for super fund because I think now is the right time to start uh building your model so for so we are actually Spartan even though we're a fund uh but we're also a founder ourselves. Uh we're also incumbent you know we've been quite successful in this past seven years investing um and advising projects. Uh but we know now like oh we have three

co-founders we all are very um concerned about our own business model being disrupt by AI. So we have already started building. So let me uh show you uh very quickly. So the first thing we thought as an investor what can we do to scale our business. So this was why we built a super fund. We believe with the power of AI as uh the best coding companion we can build a lot more products and a lot more people can become founders. So this product is a vibe coded by my head of design over two weeks and um he used it to build a prototype and since he built

it he's able to sign 10 partners. So now we have invested uh additional capital to hire people to build this out. But this is actually the first version after two weeks of VIP coding. He's never coded before. Does it? Uh, I can't seem to click the Maybe I'll see if I can show you guys a screenshot. Let me see. One sec. Sorry about that.

You guys are looking at my working working dot. Oh. Oh, but it's then it's not the Well, maybe I go through this one then. It's done. It's not the video, but it's uh I want to show you the H. I don't have the product. That's actually much better to see the product itself, but it's um it's actually a marketplace for um I don't know why it doesn't work. I'll just talk about it. It's um it's like a LinkedIn for um companies to apply for hackathons and grants. Um so you can upload everything that you have

about your company, your deck, your website, your LinkedIn just connected. And on the other side, we have hundreds and thousands of grants that have special application process, forms, requirements. And um with one click you can apply to like hundreds and thousands of grants. It's just like what LinkedIn does for people applying for jobs and companies selecting candidates. We believe like the number of entrepreneurs and founders will at least 10x thanks to the you know the claude and the Gemini's uh coding co-pilot products. So we want to build a product ready for uh helping these uh vip coders or the next generations of founders to become better entrepreneurs. Um and the next thing

we're going to build is a prediction market because I think one of the hardest thing to uh predict is longtail assets and especially valuation and uh and progress of early stage companies because they are not uh there are no metrics no clear metrics that are consistent and measurable. Uh but we are deploying swarms of agents to do this um to do this analysis because for a human to do this analysis the cost extremely high but for agents to do this analysis and bet against each other to find the optimize the answer sometimes takes $2 $3. So we are rethinking about early stage VC from a first AI first principle. Um and we are using token and

uh um and AI agent to help us become better early stage investors. Um unfortunately I cannot show you the video but I have to say this is a vip coded by my head of design who's never been an engineer and I think if he can do it um and many of us uh can as well and he's really an inspiration for us to build this product out and serve um all the founders out there because I think every one of us will be um will be building products in the future. Um thank you very much and uh we um we do have a website uh superfund.xyz XYZ. So please uh do follow us and if you are looking to apply for grants and even VC capital um please uh check it out. We might be able to help you. Thank you so

much. Um do we have time for question? Yes, maybe one. Anyone? Yeah, sorry. Um because right now for example you can get some know virtual card in Robin good or revolute but if you speak about I know normal bank ordinary bank we still don't work in weekend for example you can get you can get any feedback from their support they don't use some AI bots etc and the problem will be not the technology mostly about the people what do you

think about this um people sorry I didn't I couldn't yeah let me explain I mean like right now we have a lot of technologies from the fintech which Robin good revolute has but in the same time if you speak about some big banks in Europe, Asia I think Asia maybe it's we can exclude it know Europe and US we still use some old technology they don't work in some weekends despite of the that we have ability to use the technology right now yeah yeah I I do it's actually a good question I think the biggest um change will be um will be um um Gen Z taking over I think it's actually um the older people will pass away and the money will be handed over to their children who are very used to modern technology and very feel very

comfortable with using tech. So I think uh I think that just come naturally as uh as old um old people die and new people start managing money and fortunately I don't have a better answer for this. I don't think it's easy to change you know our parents or grandparents um you know I don't think it's I think it's too hard so I think just wait until money get passed down. Okay. Thank you so much. Oh sorry. Uh yeah, I have questions. Uh I think like there are many AI uh coding tools that allows you to build from scratch.

But if for a company you already already have like thousands uh sever thousands lines of code, it's actually not that easy. like either you have to uh have everything on the tool in this case you really need to uh make confidence uh to the tool. So my question is uh I think it's there are tools allowing you to build from scratch but basically only from scratch if you have already based uh for the moment I don't find much tools so but it's not real world it's not the real world that we always build from scratch right um yeah I'm not a developer so I don't know how to like actually answer your

question um better um I have seen I mean I If you if you if you see news, I know Salesforce said 30% of the code is built by AI already and Amazon said uh probably 30 to 40% of the code is already built by AI. So I you need to as a founder you need to find where the angle is right for me. Uh when we built when we wipe coded it yes the we have to we we actually started building this in January. So the tools are already like 2x better. uh we did have to refactor the code with real like human engineers but they manage and use AI to help them build much faster. So I think it depends it really depends on you you probably

should know you know who your like who your CTO is how he use AI like I think good CTOs know how to use AI so and unfortunately I'm not but I know uh they all managed to find a way any okay uh when doing the uh specified AI model training so how do you solve the uh false positive and false negative data problem. Um, first I mean we collect a lot of data ourselves. I think that's the moat that we are building ourselves right. We have a product that is working and we can get to uh our our customers faster um and we will get the data quicker. We are already um so actually yesterday we

are the official partner for the um hackathon or the shoulder build case at Ethereum um here. So I was a judge and uh before we started um the session uh we have produced a very extensive research for on each of the presenters to give it to the judge for to assist their uh assessment right of the projects. Um and then as a user myself sitting there and using my own product to help me judge all the presenters I can see like oh this is might be too not very useful. the questions are very good um and um um but the format is not good right so I'm already giving my um team feedback how I can exactly improve this product in this use case so that's very

much that it's like get your product into the hands of your clients as soon as possible and AI built so fast and you know focus on small problem and start collecting collecting data and improve your uh workflow and improve your data there's no other Hey. Yeah. Get him to use it.