# Alex Ungureanu - Smart trading: Using AI to boost your trading experience - Channel: [ETHCluj Meetup](https://streameth.org/ethcluj-meetup) - Date: 2025-10-07 - Duration: 14:14 - Watch: https://streameth.org/watch/yt-N8_CT1Elz2M - YouTube: https://www.youtube.com/watch?v=N8_CT1Elz2M ## Description How Wolfswap managed to save user funds by leveraging AI and Smart trading principles, by warning and stopping people trading bad tokens. The importance of a good user experience flow in the trading experience. ## Transcript Today I want to talk about a simple concept, but one concept that allowed us to save over $100,000 in user funds. And it's not revolutionary. It's not crazy, but it's the way the tech needs to be presented to the end user. First of all, first of all, let me tell you a bit about who we are. We are a team of four core developers. We are from Romania from Tennylvania like 100 kilometers away from here and we have been building in blockchain from 2020 and uh in the past two years we have been building world swap. What is world swap? Actually it's a place where we gathered a lot of web three people a lot of members of the community and we try to grow build with the community for the community. That means that we are self-funded. I'm the leader of of this amazing team of over 21 people and the leader of the community which we have over 21,000 member in our community. What we are at the core we are an aggregator deck. What does that mean? We take and use all of the liquidity on the chain to be able to give the user the best rates from all of the pools. You can see here V3, V2. And the reason why I'm telling all of this is to show how how we made that the end user does not have to care about anything. I think that's the that's the next step that we all have to take in order to grow web three evolution. And um the way we did it is that we do everything for the end user. We take the we take the pools, we take the best rates from V3, from V2, we combine them and we deliver them in one simple and powerful trade. What we do uh next is we said okay this year it's the year where the people needs to be incentivized to do something. So our unique value proposition on the market since two years was the gamification. We not we we don't have only one, we have three and four concepts that are coming this year. Mystery boxes, we have trading contests, we have seasons with airdrops and all of the stuffs are combined around our whole concept of wolves and the moon. That's the that's the magic header where we use AI and we are getting to the subject of the of the of the pitch. We use AI first contract analysis. That's the that's the that's the key part in in my in my presentation and we will discuss later in this. We have another product which is a compounder aggregator. We use AI in order to find the best time to compound the user the user and uh this way we manage to not only optimize the process but also to give an better API for the user while reducing the cost in terms of gas feeding. We have a social f product. I don't know how many of you heard about Yaps and Kaiito, but we also have a product like this for our community built and we use AI in content recognition. But let's go here. I want to ask a question. How many of you are trending on chain as of this moment? I don't no matter the chain. Is there anyone that's trading? Okay, we have one. It's good. Uh I'm not sure if you saw a chart that in the early phases was only green only green some red candles and after that it's only green in order to attract more people in order to attract more funds. The magic key here is that the developer had very have because it's still happening have very bad intentions and in the code of the token there is a part where it says only these wallets are able to sell the token. What does that mean is that he allows everyone to buy. Obviously he generate the hype from multiple wallets and then suddenly after he thinks that okay I'm getting everyone now I don't have any more liquidity to get he's selling and what's how can we stop that and how can we save those $85,000 the magic is in the AI and we use that and again we saved over $100,000 in user based on a report that we did last last month. We see a chart that it's pumping. Our algorithm is going on chain and scans the contract for the end user. Again, the end user should not know how to read a contract. I think that's that's the principle that we have to that we have to work with us as developers. End user should not have to care about reading an avm contract or Solana. They should be served data and informations in the natural language. So that's exactly what we are doing here. We are taking the contract, we are assigning a score, token safety score 10%. It's it's the best that we can get. The liquidity burned zero. That means that the the founder the the creator of the token can get it and sell it anytime he wants. But the magic source is in this part. What does the the token contract actually have? And we can see here the airdrop enabled means that the founder the the creator of the token can limit how many wallets can sell the token and how many wallets can actually make a profit on the token because everyone is trading for profit. And if we discover something like this, we instantly block the token from trading. I know onchain decentralization, everyone loves it, but no one loves it to lose money. And we are here for the end user to solve that and we are doing it successfully. So we can actually prevent this happening here. When the hype is going on, we actually put a very big warning and block you telling, hey, you can buy, you can spend $1,000. We don't mind, but you will not be able to sell and make any profit. Next next step that we took and uh this is still in beta because it's a lot of work in it is an AI companion. There is a lot of talks about onchain versus offchain, axis versus checks. And I think this year is a year where we will see more taxes coming to the to the onchain part and the users from the from the checks that are used with a lot of a lot of uh informations given to them. They will be hit by the amount of informations that will be find. So what we are building, we are actually building an AI companion that you can come and type a message. Hey, can you help me? Of course. What's What's the token? What's the token contract? Is the token contract safe? Can you buy $10 for me? Can you I don't know. Can you give me informations about the token uh team? How how active they are? How safe is it? Is a rockpool and all sorts of informations that we want to we want to provide the user the help that they need in a very natural language for them. And of course, it's going to ask, am I handsome? And I don't know the answer for that. We as a team, we are a team with uh with uh I think the the average year is is like 30 years. So we took all of the tech that we that we need from the from the ones that are younger than us. And we are trying to we are trying to leverage AI in every single possible way. We use it in marketing as you can see. We use a lot of VO. We use it in the lore buildup. We use it in uh in scanning token sentiment in order to provide our users what's trending, what's not trending and all sorts of stuff. And the best code and I think this is very important. AI won't steal your job as a developer neither mine but someone using AI will. And I think this also serve as a code for uh for us developers to think about our end user. how can we make better products using AI in order to serve the users that are coming to web three ether and u it's a year where we think that the web three space evolved but at the end of the day there is not many people on chain there are a lot of people that heard about crypto on checkis but I think this year is we we're going to see the movement on chain on self custody and we as we have the responsibility as developers to start using AI to make their life easier so we can accommodate and on board much more than we did so far. This was me. Thank you. I'm Alex from Wool Swap and uh for anything this is our Twitter and our website where you can trade on B Kronos Abstract and Polygon. Thank you very much. >> Awesome. Thank you so so much Alex. Um, well, I I certainly hope no one in the audience has been rugpulled as much as I have, but um, it's good to know that Alex and his company are out there to protect us. Um, so I think this was the last question, by the way. Is there could we go back to the previous questions? I think there were four that just came up. Ah, okay. Well, in the meantime, does anyone have an organic question they'd like to ask by putting their hand up so we could um inquire here with Alex? because I've got one while we wait. Um, do you think that some of the traditional trading platforms out there and the deck screeners of the world should have these kind of contract screeners built in? Is that a vision of the world that you want to get to? >> It's it's not only the vision, it's it's something that is happening right now, but unfortunately we can see on the screener. Dexter is the is the largest and the biggest uh price tracker in the world in the in the onchain. They are using stuffs that are not integrating the AI. So, what the the end results is that they are they are delaying in showing the best results over and I think that with AI we can actually leverage about the speed the calculation about everything that AI is bringing to us to be able to serve a better goal and we can clearly see that for example Binance is using as well. So if the major players in the in the in the onchain world is starting to get adoption rate I think yeah I think we are going into that direction. >> Okay good to know. Um, do you think that the use of of uh do you think the dangers of trading are going to essentially get worse and worse as time goes on, as more people get into the space, or do you feel like because this cycle was so involved with meme coins and these kind of high volatile out of nowhere tokens that everybody was jumping on, it was especially bad now and it's going to get better. What's your kind of And I know no one in crypto knows what's going to happen next cycle, but you know, would love to hear your thoughts. Well, this cycle was I think in my opinion it was a crazy cycle. We had the meme coins popping here and everywhere was popping meme coins. But unfortunately with that the bad actors in the space also saw an opportunity to come and uh steal money, steal funds and I think again we have to build better products in terms of user experience in order to don't to not allow anymore these bad actors because where there is money there are bad there are bad intended people to try to come and take advantage of that and that's I think it's going to happen more often than ever because the crypto is getting more adoption. Now we can see in the US what's happening that they are getting real traction there and u more and more bad actors will will join will will jump on web three but we have to we have to build better products in order to to counterattack that >> I think that's probably absolutely right. Okay, we've got a question here from Titus K who's who's gone through the app. How do you manage AI companion ident? Oh, have we got another one? How do you manage AI companion identity and data segregation for each user, Alex? >> It's based on we have three steps. It's first it's based on the user uh wallet address. So uh it's the simple onchain wallet address. Second, we do ask for anyone who wants the to connect the Twitter and the discord. So we can provide a full package data for for the user customized on his preferences, tweets and etc. So we are we we tried to combine everything that that we that we can and the user is free to choice what what what he wants to connect what data he wants to serve to us. >> Makes a lot of sense. Thank you. Um another question here from Gasguzzler 87. If there was a prize for for good usernames, I'd be giving it to that guy. What kind of AI models or algorithms did you find most effective in detecting bad tokens or suspicious trades? Uh that's that's a good question for my technical founder but from what I know is that we we tried we we developed a custom LLM based on CH GPT version that was uh last three months or so and what we discovered is that GPT was not good enough with AVM contracts. So we tried to seed him with over 10,000 contracts that we found on chain with bad code, malicious code and basically we have our own version of AI of LLM that is that is giving us the data in real time. >> Okay, fantastic. Thank you very much. Has anybody else got any other questions for Alex over here before we close up shop? >> Doesn't look like it. Okay, well thank you very much Alex. Let's put our hands together one more time for that fantastic talk. Thank you very much.