AMA | How AI Is Reshaping Trading in Crypto and Stocks (Featuring Edgen)

In this AMA, Joules Barragan, Marketing Lead at Sahara AI is joined by Sean Tao, Co-founder of Edgen, a company building an AI market analyst that bridges crypto and stocks to surface real cross-asset intelligence. They discuss how the democratization of data affects the competitive landscape, where the new "edge" comes from, and how platforms like Edgen are building a market operating system to be a personalized, 24/7 assistant for all traders. They also explore the surprising correlations Edgen is finding between stocks and crypto, and how they use a multi-agent system to cut through the immense noise of both markets.

Link: https://x.com/SaharaLabsAI/status/1965218425367929299 

Transcript

Joules: Okay, we are officially live. Thank you everybody for showing up. We will get started in just a few moments. Let’s just wait for people to start showing up.

Joules: Where are you at right now? Anyway, Sean looks like a pretty fun, pretty fun spot.

Sean: Yeah, I'm in Korea right now, so there's a big conference today.

Joules: So nice. Are you gonna be there for KBW as well in a few weeks? 

Sean: So yeah, are you coming too? Yeah.

Joules: I'll be there, definitely. Hopefully we run into each other. We'll have to make time.

Sean: So actually, we’ve invited, we have one, like, super hot K-Pop, like we will have a club and invite one of the hottest K-Pop groups to the club.

Joules: Yeah, we have, I hope so we're doing an opening party. So if anybody that's listening right now is it's going to, KBW, yeah, we're doing that opening party. And we'll also have some cool K-Pop celebrity singers. So we're doing, like a mini concert. We'll have like three singers. We just announced two of them. The third one we’ll be announcing in a few days. So that's pretty exciting.

Sean: Yeah, that's super cool.

Joules: Yeah, I'm pretty excited. I went last year and KBW was a blast. I love being in Korea. I can't wait to go back.

Sean: Yeah? Oh, we also have one event in Singapore as well. Oh, nice, yeah.

Joules: So we're partnering up with Kraken and at Marquee, Singapore, Marquee, so last year, when we host the event with Bear Chain and OKX, like, there was a, like, a 10K people queue outside our key, and I saw that people sharing that on Twitter a lot.

Joules: Yeah, that's really cool. Okay, we should probably get started. Looks like people are starting to trickle in. Hello everybody. Welcome to today's AMA. I'm Joules with Sahara Labs, Sahara AI. So for decades, access to advanced trading tools and algorithms have been pretty limited to, you know, high frequency traders, hedge funds, institutions, anybody with these like massive, massive resources, and the average trader honestly didn't have much to rely on other than gut instinct, maybe a handful of charts and whatever news you could pretty much parse from the open media that everyone else has access to. Right now, we're starting to see with, you know, the advent of AI, that that imbalance is sort of shifting. And you know, today's AI systems, they're being used by everyday people to analyze blockchain transactions, stock order books, fundraise flows, even global sentiment. And a lot of this has been done in real time as well. And these are basically all capabilities that used to be locked behind these, you know, expensive, closed doors that only trading desks and high frequency traders could actually use and it's bringing all of that into the accessibility of everyday people. And this democratization of this market intelligence is starting to also raise some new questions, right? If everybody has access to these AI powered insights, how exactly does the competitive landscape evolve? Are we headed into this new race to the top? Where does the edge actually come from now that everybody has access to AI? These are some of the questions we hope to answer today with Sean, our amazing speaker. So Sean, I am joined today by you. You are the co-founder of Edgen AI. Edgen is, you know, building an AI market analyst that bridges crypto and stocks, and it's designed to cut through the noise and surface real cross-asset intelligence. Sean, thank you so much for being here today. I'd love for you to, you know, share Edgen's mission, give us a better, maybe, understanding of what you guys are doing better than I can do in one sentence.

Sean: So Hi everyone. I'm Sean. I'm co-founder of Edgen. So Edgen's mission is we want to build a market operating system. Because in this cycle, especially in this cycle, we see that there are no barriers between like traditional traders and blockchain traders, right? We see all these D.A.T. companies emerging, ETFs, like Wall Street money flooding into crypto. So we think that in this cycle, the barrier between stock and crypto is kind of diminishing. So what we want to build is a market operating system for everyone to use. For stock traders and for crypto traders, Edgen will know you, and will know what you're trading at, and it will be your personalized assistant. So you don't need to open Goldman Sachs or like Morgan Stanley to get their super detailed insight. But Edgen will be your 24/7 personal assistant. It will give you advice based on your portfolio, and we'll give you market insight immediately. And in the future, we want to trade everything on-chain, on Edgen as well. So that's Edgen's mission.

Joules: Yeah, you guys are doing great work over there. I'd love to set the baseline right for those who maybe aren't as aware with where trading used to be and where it's at now. What did trading used to look like for let's say your average investor before AI, you know, when institutional players had all of the best algorithms, they dominated the field. What were things like before, and how are they like now?

Sean: Yeah, so things before AI is something that like AI is definitely a game changer for this, right? So before AI, for these institutions, they need so many researchers based on different segments and different categories, and even for crypto as well, right? There is just so much information you need to be aware of or you need to learn to understand the sector, because we know that the edge for information is how you can earn money, right? So that's why we call it Edgen. But, but in AI like this, AI helps people so, so like finally retailers have a chance to get institutional level of information, thanks to AI and Edgen is shaping that. And we are giving like retail, so in our October version, when we have this personalized version, you can integrate your watch list, you can integrate your MetaMask, Robinhood, Coinbase account to Edgen, and we'll know you, and we'll give you real time information and insight and recommendation analysis based on your portfolio. So this is something we want to do. We want to just make sure that people will have access to institutional level information.

Joules: Yeah, that's amazing. I didn't realize that's where you got your name from. That's really cool. I like that connection right there. That's a great story. I wonder. 

Sean: So, you know, you know, I think Edgen is also Edgen also makes sense. There's like, since I'm in Korea right now, right? Yeah, Korea, there's a meme called Edgen-nam. It means that an Edgen girl, Edgen boy, stuff like that, right? So that's a really popular meme in Korea to my understanding.

Joules: Oh, wow, that's cool. How auspicious. I am curious, though, with bringing these tools to more people, right now, every day, people have access to this, like you said, institutional-grade AI insights. You're basically you have your own quant, your own analysis in your pocket that you can rely on. If everyone has access to these AI powered insights, now does this actually level the playing field for everybody, or are we creating some new kind of race, you know, to the top, where the real edge not doesn't really become like, who has and doesn't have AI, but it's more everyone has access to AI, so it's all about how you use it and who's using it the best?

Sean: Yeah, definitely. I think at the end of the day it is who uses AI the best. So that's something we want to provide to users, right? Like in the past, if you want to learn something about stocks or investment or tradings or ETFs, you need to go through like you need to check out Bloomberg, or you need to check out like Seeking Alpha, or like schools.com. It's just so much information you need to, you need to absorb, right? But through Edgen AI, we help you to do all the jobs, and you don't need to spend so much time and you will have more time to learn more sectors. Or like, even like, enhance your trading. Trading levels as a quant, or you can write your own, like quant code in the future. So this is something we wanted to do. So I think, I think, like AI, makes everyone's learning curves much more possible, right? Like people can, can have a better learning curve than before AI. So definitely, like in the future, I think that everyone will have so everyone's trading skill will increase, right? And so for like traders, the trading level will increase too. It's just how we can, we can catch these institution guys.

Joules: I love that. I'm wondering, are you starting to see traders start to build out these, like AI strategy portfolios, so like collections of prompts, workflows, model settings, anything like that, that now that's kind of what they're trying to guard as, like their secrets. This is what separates me from you as like an AI power trader.

Sean: So, so right now, like Edgen currently, doesn't support writing codes or strategies on Edgen, but we are testing that internally. We also, we also look all the, so, so there are a lot of strategies on GitHub and all these open source platforms right, like Hugging Face and stuff. So we actually tried all of that. And so most of them are, are okay in a period of time, like back the back test works in a period of time, but it doesn't work like for lifetime, right? So yeah, so I think that like quant trading and also like CO trading, is still pretty hard. The barrier is still high, and there's no one quant trading, or CO trading, that can guarantee you to have 100% win rate, or to guarantee you to win money. So I think the quant firm will definitely still exist, because this is super hard. It relies a lot on math, mathematical work and they need to adjust, they need to adjust the model, like, from time to time to make sure it works for lifetime. So this is something we're testing it, but we're not sure if we want to provide this to users, because, like, this can make you earn in a period of time, but like, probably like in the next period you cannot do. So we're still pretty cautious on that right now.

Joules: Yeah, now that makes sense. I'm curious you mentioned all of the data that you know you have to sift through, and it definitely I've been in crypto for a while now, since 2015 as, like, on a personal level, and you're right. Like, it just as the years go by, there's just so much more every day. I feel like there's something crazy new going on, and I can't even keep up with, like, how many tokens? I'm like, oh, I should have bought that, but I haven't even heard of it. And it's like, already gone. You guys are not only just doing crypto, but you're doing stocks as well. That's so much information. One, I guess I have two questions. Sorry, I have ADHD. So the questions, they come out, when they come out, I have two questions that that makes me think of right? One is, with all of that data, how do you prevent people from being one, overwhelmed by all of the noise. But then two, crypto alone is like a lot of information for people to, like, go through. What Edgen is doing is not just, you know, giving you insights into crypto, but stocks as well. Is there like a correlation that you guys are seeing a high correlation between, like, stock and crypto prices? Is there a reason why you're putting both of those on one platform? Is it just because you want to cover all of trading like, what? What is the connection between that?

Sean: Yeah, something to why we are doing this. Because we found a very interesting correlation between stocks and crypto. So if you are trading crypto, you're very likely to trade AI-related stocks, nuclear or E.Z. or like quantum computing, or like rocket-related stocks, right? Or like crypto, crypto blockchain stocks too, right? And D.A.T. stocks. So we found a super high correlation between these because, like, Edgen is from EVG and EVG itself, we have another, like an institution platform called Aspen, right? So we found a lot of correlation through this platform. And so that's why. We think that, oh, so in this cycle, if you bought all these sectors, you're very likely to buy crypto too, yeah, and so that's why we decide to put everything together. And I think, and the other thing is that we think that in this cycle, we care about mass adoption. And in this cycle, I think most of the retailers who invest so so most of the retails, when they come to crypto, they usually buy like they invest in D.A.T. stocks like Korea itself. I know that they have 600 million inflow, USD inflow to Bitmine, the Tom Lee Bitmine, right? So that crazy, right? Because Bitmine's like market cap is like 6 billion, and Korean like, it's 10% of it. So we see a super strong demand for Web2 retails to Web3, but through these D.A.T. from stocks, from these new tools that we’ve never seen in the last cycle. So that's why we put all these together. Yeah, it's fascinating.

Joules: You say that I was just thinking about how D.A.T. is so massive in Korea, specifically, it's kind of crazy, really crazy.

Sean:  To my knowledge, like Bitmain is one of, I think it's in the top three inflow and among all so in Korea, right? So Koreans, they're really fascinated about these D.A.T. stocks, and they generate such a high inflow to D.A.T. stocks.

Joules: Is it crypto specifically, or is it like AI, a combination of both? Or you're just seeing a little bit of everything.

Sean: So, wow, Bitmine is top three of among everything, yeah, so that's crazy. That's really crazy, yeah. So, so the demand, so, actually, the mass adoption for crypto drives from Web2, right? So that's why, and, and also on Edgen, so actually, like, so we have payments right now, right and 90% of our users are actually using Stripe to pay our membership fee. So that's something that pretty surprised me too, because we support both crypto and Stripe, and 90% of users are using Stripe. That means that like Web2 users, they're looking for a solution to check out like stocks, to do an AI analysis for stocks and crypto at the same time, so that's something I think is really interesting. Yeah.

Joules: That is really fascinating. I've worked in Web3 and for a while too, and it's always crazy to see how many people still prefer to make payments, and you know, through Stripe or other like payment processing platforms rather than crypto, either because you're bringing in a lot more Web2 people that aren't really comfortable holding their own tokens yet, or you get the opposite end where it's Web3 people who just don't want to, they don't want to spend their tokens for payments. They're just like, No, I want to hold as much as possible. Yeah, the market dynamics are kind of wild in this space. Yeah, exactly. I'm wondering with, you know, stocks and crypto, it sounds like there's a lot of noise that can be happening when you're trying to do any kind of analysis for whatever kind of trading you're looking to do? How does Edgen deal with cutting out all of that extra noise?

Sean: We tried so this is a really complex, complex question, to be honest, but we try to make it much more simplified. So if the token, so stocks is much easier, right, because we have more data to support, there are a lot of Wall Street analysts giving forecasts for the stocks, and also combined with technical analysis, fundamental analysis, it will have, like we have this forecast section, right? So it will give us an easier way to forecast the price for the stocks in next year or in the next five years. So stocks are easier, but crypto is really hard because there's no, like, there's no fundamental analysis for crypto, only technical analysis, and the prediction is pretty wild for crypto as well. Like some predict BTC will go to like 40 million, and some predict BTC will go to like 150 million. So it's crazy, right? And then some predict BTC will go to 150K. So the range is really wild for crypto. So I will say crypto is harder, and we need to focus more on the on-chain side, the smart money as well, for all these small tokens, like meme coins and something like so. So we basically break down into a lot of sectors using our multi-agent system. So like BTC, Ethereum, we have multi-agents, like ETF, inflow, outflow, and we have agents, and we have a technical analysis agent, we have D.A.T. agents, so we'll have a better prediction for these to help us to have a bigger picture of how to predict these, like large tokens price prediction. But for small ones, we will have an on-chain Smart Money agent and probably like tracking their liquidity pool and tracking like the top holders, like holding changes, right? So we'll have agents like that, and to make sure that we will probably have, I cannot say, price prediction, but a better understanding of how the price will go for these small tokens. So it's based on different multi-agent, and we like based on different tokens. We will give different kinds of multi-agent to make sure that we have a better price prediction.

Joules: I'm curious, especially since you're doing all of this dealing with multiple agents at one time, all interacting with each other. How do you approach that, like reliability within the different agents and models that you're using, what you're capable of, speaking of, obviously you don't give me all the inside details for how your models work. But how does that model validation work with you guys, what are you doing to ensure trust and reliability in the models you're using, the agents that you've built?

Sean: Yeah, so for the model validation, right? So I think the first thing is that we need to make sure the data is accurate. So the data is the most important thing. The second is how we can categorize all this data, and because like data is a mess, it's chaos, right? So we do have researchers from Wall Street in our team to make sure that all these data are very categorized and compiled so and the third is that putting all these data into multi agents and the floor will be validating if these data makes sense or like, because we have our own private model as well, right? So the fourth will be like validating all this data and to see if this data is actually accurate, because sometimes, like, we know that models, if there's no accurate data, it will have illusions, right? So making sure that the data is accurate is probably the number one thing, and compiling is the second.

Joules: I love that this is definitely something that we work with a lot of people on, so I cannot stress the need for accurate, high quality data enough. That's kind of why I'm excited to, you know, speak to the next thing I want to talk about, which is our data services platform and the task that we have going on with you guys at Edgen, where essentially, we have people going in and essentially using your tool to do a few things and then sharing their prompts. And essentially the way this works is through, essentially what we kind of just discussed right is model, it's a type of model validation. It might look like, Oh, hey, I'm just like, using, you know, Edgen's tools, and then getting paid for it through our data services platform. But in reality, on the back end, what you're doing is validating that, hey, this model is working the way it's supposed to. I'm interacting with it the way I want to. And, you know, I agree with the things that it's like sending out to me, and I love that we're able to do that.

Sean: Yeah, and I think this is probably the first cooperation between the Sahara and Edgen, because I can see that we have so many things we need to validate in the future. So right now, like, because this is the first time, so all of the questions are really easy, right? Like, as Edgen develops and we have new features probably next time, like, the task you see on the DSP platform will be, oh, like, it will be theme based, or we'll put something like quant related inside. We want to, we want to do like, pretty in depth validations in the future, and to make sure that all the data and all the information by our project model is provided accurately to users, because I think that's something we can discuss the whole AMA, and that's the most important thing.

Joules: Yeah, no, I love it. I'm wondering how new some of this is, have you learned anything surprising from how users are already interacting with Edgen that's maybe different than what you expected.

Sean: Um, I think, I think something makes me feel very excited, is that, like we have 10 questions on DSP platforms right now, and usually so the average user will ask, like, three to four follow-up questions after that question. So that means that like, because we are also testing like, a user will ask more in-depth follow-up questions for the first question, right? So a user will ask three to four follow-up questions. And it pretty surprised us, it means that there's a huge demand for a user to understand more information, so that so, so I think that helps a lot in building our AI results in the future as well. So maybe we can give better recommendations based on how a user behaves. So this is really interesting, and helps us on the product side as well.

Joules: Yeah, I love that. You actually, you answered a, I think, a common misconception that I've seen, which is, you know, how can, how can just testing one prompt actually do model valid work for model validation, especially when the prompts are already predetermined, right? And you kind of just answer the question, which is, like, it's not just about the single prompt and how, you know the AI is answering that question. It's, what's the user behavior after, like, what follow up questions are they asking that we didn't ask them to ask her, ask them to question. And it's, it's, I find the whole like, model validation process, like, really, really interesting. And I think it's something that doesn't get talked about enough for, like a general audience. So yeah, thank you for some of those insights.

Sean: Yeah.

Joules: So I do want to make sure, yeah, I know we're closing in on time, so I want to make sure, if you guys have any questions, please leave them in the comments, and we will get to as many as we can. I know we're running a little short on time. I do have a few questions, if you're ready for them.

Sean: Sure 

Joules: Okay, let me see, going through some of these. What's one of the most interesting insights you've seen Edgen create?

Sean: I think the most interesting thing Edgen created is that we help users, so if you go on Edgen right, you can find that every category is compiled by themes. And the reason why we want to compose, so, so the reason why we compile by themes is because the Wall Street researcher, they said that's something working in Web2, in traditional finance. So we compile all these for crypto ads and also for stocks. So user, you can come, so for you, you can come to Edgen and see that, oh, I want to invest in like D.A.T. or E.T.F.s, you know, crypto, and we will give you answers directly, because for most of the time, like we if you give a chat box to user and tell them what to ask, most of the time, user will don't know like what to ask about, right? So giving usera compiled preset themes, right? I think it is really important. We're not guiding you to answers to ask something like set up by us, but, but should drive by accurate themes that you can just look inside of said. Oh, D.A.T. There is so much money flooding in D.A.T. to buy bitcoin today. There are so many inflows, money plugging into Ethereum to buy Ethereum today. There's 95% that Solana, Litecoin, XRP, will pass the ETF before October 3, right. So this is some accurate information we want to provide to users. I think that's really important. So that's something that's the value that Edgen generated right now. But some value we haven't generated is we haven't generated personalized value so and just like what I what I told Joules early in this AMA, like in the future, you can connect your account, your MetaMask, or your Robinhood, your Coinbase, and we will give you personalized AI analysis and servers, for example, if we found out that, oh, you're only buying meme coins, and we will probably recommend you to buy other large, large tokens, like BTC, ETH, Solana, to increase your Sharpe ratio. So this is something that Edgen wants to provide in the long term, to give you information advantage and to give you analysis advantage.

Joules: Cool, I have another question here. How do you personally see traders best integrating AI in the future through full automation or as a Copilot?

Sean: I think it depends on the trader himself. So if the trader can adjust their quant model on time then AI can definitely help with that. But if the trader is not so good at math, then AI can help more on the information side and to help create the first on-trading code. 

Joules: I think that’s very well said. We have time for one more question. We have somebody on X asking if they can use Edgen for retail trading today.

Sean: Not right now. But we will onboard all the on-chain execution in Q4. Because Edgen is still in beta. So right now you will see that Edgen is developing, it’s upgrading every 2 weeks. So every 2 weeks we will have something new, something huge launching so give us some time yeah. Like eventually what Edgen wants to do is a mass adoption AI product. And this AI product can even be super useful in Web2. And that’s how we can make sure that this platform can survive even their market and we have mass adoption users, clear monetization model. So I think this is really important for a crypto project. We want to be one that breaks the Web2 and Web3 barriers.

Joules: Yeah I love it. Thank you. And that is all the time we have. So I want to say to Sean, thank you for sharing all the amazing insights. It’s been fascinating looking at how AI has been reshaping both crypto and traditional markets. For everyone that’s just tuning in or has tuned in for the entire time thank you for being part of this journey. Make sure you go and test out Edgen AI through the Sahara Data Services Platform. Remember by generating outputs and sharing your feedback, you will help provide Edgen’s models and earn some of Edgen’s tokens as well as Sahara tokens for your contributions. Sean, is there anything else you would like to shout out?

Sean: No thank you, Thank you for the AMA and thank you to every user for participating in this AMA too.

Joules: Thank you everybody, thank you for joining us. We will see you next time. 

Sean: Okay cheers bye!

Joules: Bye!