AI Agents & THE Future of Onchain Finance

first. Can you hear me July? Oh yes I can hear you. Perfect. Did I pronounce your name correctly? Is it July or something else? Oh you can call me Julie. That's fine. Julie, Julie okay that's perfect. Yeah I wasn't sure how to pronounce it. Where are you from? Yeah currently we are based in Hong Kong. Oh nice nice perfect. So yeah hello everyone and welcome to this space today. I'm super excited to dive deep into AI autonomous agents and we have some questions that go very very deep into these topics that I think you're gonna find super interesting. So let us start with maybe an introduction of you July 1st and then we can go from there. Okay I will make a short introduction about us. This is Julie from the agency I and it is my great honor to join these wonderful AMA. So let me make a short intro. We are a team focus on AI infrastructure and autonomous agents across Sue BSC and Bitcoin and our mission is to deploy autonomous agents with very viable identity continuity and consensus making AI more trustworthy and capable for the web 3 space and also for our products. The flagship one is Alpha X which is an app-powered crypto prediction engine which could deliver a short term BTC, ETH and so it's trained signals with over 60% accuracy and upcoming offering could core AI and choose sites and for our native token which name is EIA, ETH powers service access, taking governance incentives and better capture through revenue backpacks and burns and so this is generally about us. If you want to learn more please add in to your official head account or community. Thank you. Wow that was quite comprehensive already. I think people got a really good idea about what you guys do. Thank you for that. First I want to kind of start with a bigger picture like what's going on in the industry. There's been a lot of talk about AI. We're hearing about AI agents that can do a lot of stuff like open AI's dots for example or MUSE from Meta and you know the capabilities are really increasing rapidly. How big of a shift do you think this is and what does it mean for web 3? Okay. You know it's a massive paradigm shift and if you look at traditional AI like the chatbots we use every single day they are mostly reactive and you ask a question and it gives an answer and the interaction ends there. Like autonomous agents take these a huge step further. They can actually read the room and analyze the market like figure out the best course of action and execute it independently and for the for web 3 this is the massive puzzle piece. Blockchains are incredible because they are transparent and trustless but let's be honest, smart contracts are very rigid. They only do exactly what they are coded to do. AI brings dynamic intelligent decision making to the blockchain and we call this the intelligence dilemma and the agent AI is basically upgrading AI from just being a helpful option assistant to becoming an active, verifiable on-chain participants that can actually execute financial strategies for you. So yeah that's all my thoughts about that. Thank you. Very cool. Yeah I totally agree that it's getting pretty exciting when we get those capabilities on-chain. Then you already kind of touched already on what the agent AI does and what you're building but you want to go deeper on that and also I wanted to ask like what problems are you guys specifically trying to solve that you know existing AI agents or systems haven't really addressed yet? Yeah to put it simply, the agent AI is building the fundamental infrastructure for decentralized AI agents and we are making it possible to create and deploy agents that can thank, remember and ask across multiple chains and we are heavily focused on a web mb-chain and like the a BTC ecosystem right now, the biggest problems we are solving in the black box issue existing AI systems are centralized and you ask them to do something and you just have to blindly trust how they got the answer and in web 3 or whole ethos is a don't trust just verify and if you go into tell and AI manage to manage your wallet or execute trades you can't rely on the black box so we give these agents a permanent identity real memory and a consensus mechanism to prove their actions accurate and we are turning AI from a black box into a transparent and verifiable tools, okay thank you back to you from a black box to transparency I like that that's very defy of you guys and I yeah it's definitely what we need in defy and then I was reading through your docs and everything and I saw these terms are called you know very viable AI built around identity and continuity and consensus if you maybe explain these three concepts a little bit for us, oh sure, these three concepts are the cornerstones of everything we do and think of them at the credit system for AI the first one identity is like the agent on champ has port it ensures they AI has a unique cryptographically verifiable profile and it prevents the AI from suddenly changing its behavior or acting like a different entity and then you have continuity which is basically the agent's memory a smart agent needs context in East you remember what traits it makes every day and what your race tolerance is and we use specific memory modules so the agents behaviors remain coherent over time and finally we have consensus because AI models can sometimes generate slightly different answer and we use system where multiple executor news run the logic and then commit our news verify and agree on a single final result on chin together the three ensure the AI isn't just a smart but actually reliable yeah that's all of them that's a great explanation thank you so much I understand it much better now then you know you mentioned risk and you know people are worried about how can they trust AI agents and I think that's really important when it comes to you know when they're interacting with your financial assets you know it's your money at stake how can users verify what the agent is doing and do you have any safeguards to exist safeguard that exists to prevent like unintended or potentially harmful actions okay when user found our on the line security is everything because of our consensus mechanism every single decision the agent makes leave a cryptographic proof on the blockchain everyone can audit it but more importantly we're building some serious safeguards by separating the thinking from the doing event agent wants to execute a financial transaction it can just do it alone we use a decision plug-in combined with a multi-party computation or NPC what this means in playing English is that the agent only holds a piece of the puzzle it cannot sign a transaction without reaching consensus and staying within the strict safety parameters the user set and it completely prevent the AI from going raw raw with with your assets yeah that's our way to solve some of these problem oh that's cool you have mcp I played around with that too and yeah I think that's a really good method of kind of isolating the actions that the agents can do then I wanted to ask you know there's an important distinction between you know verifying what an agent that an agent followed its instructions and knowing whether it made the right decision how do you guys approach that challenge oh yeah it's a really sharp question and you're absolutely right or infrastructure guarantees the how it proves the agent followed the rows securely and wasn't tempered with but the protocol itself doesn't guarantee the AI well magically make 100 times trade and the corrects in this or the quality of the decision is handled at the product level for example with with or fact sheet product which name is of X we don't just use a stake model we constantly train it using reinforcement learning from human feedback like communist data and real market results and the infrastructure make sure the execution is safe and we relance this real-world battle testing to make sure the AI eats actually making smart profitable decisions over time and so yeah that's our approach to to that challenge okay cool and that's all nice on a theoretical level but how about like for regular users let's like talk about for example your alpha x you know core AI and AI agent platforms like what can you know you users do with these platforms do you have any examples to give us oh yeah sure um the the first one the name is of X which is our AI powered prediction a price prediction and James were assets like BTC ETA agents so and the numbers big speak for themselves we have over 400 000 daily active users 70 million total users and it driven over like 122 million on-chain transactions and the accuracy is constantly above 70 percent which is why both retail users and institutional quant funds are using it today and then there is core AI which is incredibly full it's no code quant strategy builder normally you need to you need to be a developer to build defy strategies with core AI you can just use natural language to tell the AI what you want and it will build back test and then finally ask you the strategies for you um for people who just want market monitoring we then have century it acts as a 70 24 7 watchdog on hyperliquid monitoring over 300 assets and use that the conditions and execute without emotion and without fatigue and yeah that's all over uh products that I could share with all of you wow I'm pretty impressed by the numbers you just shared like I had no idea no idea that you guys had so many users that's insane does love love to hear that I definitely have to test it out by myself as well seems like it's something that's actually useful for people uh then you also uh recently launched this AI token smart router is that right could you share what it actually does who it's designed for and yeah just want to learn more about that one yes the smart ruler is basically they automate unified gateway and models