AI and The Future of Crypto Data Experience - Lili Feyerabend

foreign [Music] thank you [Music] thank you thank you hey um hi everyone uh my name is Lily and I am a co-founder of miti.ai and I'm gonna talk about the AI and future of crypto data experience um okay so I have a question has any one of you been using AI products in the past few months yeah okay um and have any of you been like integrating AI into your workflow in the past few months like right in your workflow awesome okay so we are all like a chance humanist group so I'm in the right place um so let's get to it so yeah so all of us have been into this AI rabbit holes in the past few months and I guess you guys have been using chat GPT already right or similar products okay that's awesome um so before I dive into our product demo um so I would like to talk about a little bit of the big picture like why do we actually need AI at all right and first of all I would say that we have a lot of data floating around and our brain has limited capacity as you know so basically the capacity of our brain is 2.5 petabytes or basically is 2.5 million gigabytes that our brain has capacity and that's equal to about 3 million hours of Netflix shows if you turn on the TV for a straight 300 years that would be the amount of capacity also the processing capacity is equal to about 7 000 thoughts per day at the speed of 300 miles per hour but then when we talk about the information is being flown on the Internet it's so much more right and it's keep growing currently is about 5 000 petabytes with the processing capacity per day at about 2.5 quintillion bytes per day and then we look at the ethereum blockchain right we just the tiny bit of everything that I have talked about so we just have currently the full sync data at 907 gigabytes uh which is bring us to this slide where when we talk about web3 knowledge most people would think about the on-chain data which is the one of the core important uh like source of knowledge but it is not holistic enough for example if you would like to understand a certain Concept in white paper just asking the question on chain it would not be able to compute that answer for you right um one second and so the product will be building metis.

ai so we're actually looking at the crypto data experience from a transhumanist perspective so basically there are several versions of AI that we know nowadays right so the first version of AI is the AI that will kill of us destroy the world the evil technology the second type of AI is that the AI that will augment our ability make us 100s of what we do um I hope you guys think the chat DBT for example is perhaps the second type have you guys heard of uh Sydney so basically Sydney is the code name of the hacker which is trying to hack the large language models and have you guys tried Bing chat right so basically the the model underlying the Bing chat was talking about Sydney um this is like very interesting you should try to read maybe go on the internet and read a story about Sydney and Venom so they're like many different kind of archetype of AI that the large language model has been kind of interpreting and like talking to the Bing users which is like it's a really crazy story but that's like for another time um so what are we focusing on right now is that of course we understand that AI has a power right but um how can we harness that and like let's say I actually help us in our workflow so for us uh we focus uh specifically on using AI to improve crypto data experience and as you can see here like how we identify the data experience here is basically we we want to improve the data environment where user could make a research about what they want to know or do in web 3 and we would reduce the distance between the knowledge our insights to the moment that they could take action and that we would consider the holistic data experience and in order to accomplish that right that um there are several ways to think about it so first of all it's quite important to talk about the interface um okay so you cannot really see much here but on the left hand side is basically like the node graph interface of Houdini have you guys use Houdini app before so Houdini application is for the VFX artist to basically create an algorithm to um to make like all the 3D effects and the interfaces I would say one of the most complex out there right and there has been the debate about this whether like is it like does it make sense that like the complex tasks that we would like to accomplish doesn't necessarily need a complex interface on the right hand side obviously um the uh interface from I think like 1990s is one of the company's inventory interface uh GUI which is like is unnecessarily complex and do we actually need that right as as you have tried shaft DBT and many other AI products uh a lot of times this thing has been abstracted down to just a few contact points so what we actually focus on admittee is we have designed this using like user-centric like philosophy uh we're thinking about very simplest part of like we want to help user access to data and act on it as simple as that but in order for us to go through those four we would have to first identify like who are the users uh what are their needs and how do they achieve their goal currently I will show you a bit later about the use case also how much information the user required to take an action and also this is also related to the amount of trust that users have in in the application that that they use to take action and also what data is available or have to be created in order to fulfill such goals so our case study is for example um I believe like some of you have gone through the exact same flow as I have so the first application I would like let's say I'll be talking about different cases here let's say I want to talk about the staking pools the liquid sticking so first is that I want to find out like what are the what's taking pools actually exist right there right so I may have to like do some Google search or I may basically go to Define Llama Or you know to understand what are the apy uh the second is that I have to get already the name of the protocols and the values but then I don't have like I can't remember exactly like how staking works or what what is it about basically like underneath underlying technology then I will just go to some tutorial which require me to jump into another website and uh later then I would also want to understand okay the differences between these Protocols are like one of them either like decentralized or not or like who are the top delegates uh who actually the builders then have to go to like another website in this case I will probably have to check on Snapshot and then later I would also want to understand okay I put this amount of money how much you will get back in return then this is also another app that I have to also go through um finally if I want to take a decision at all then I would also have to jump to another application and that's very likely I'm gonna either go to like Lido rocket pool and so on so forth so as you can see that the experience right now is quite fragmented because you have to go jump into many places and a lot of times you get lost in the flow so this is basically our [Music] demo that we have just put out so you can see here we have abstracted all these like complicated tasks that I have just uh explained in one rather simple interface starting from the chat interface um so basically the agent would ask you okay the agent has a capability to draw the knowledge or insights from the off-chain and on-chain data source so you don't have to hop between platforms and just condense your action into one place so for example this is like what the agent would say okay what are the top the question that you could ask is like what are the top LSD token with the highest deal or you could also summarize the agent to summarize the eigen layer white paper so yeah so here so this is like like rather fast maybe I can just go one by one um yeah so you can for example ask the agent to compare the apy of different staking pools by just typing a prompt then the agent will like draw the real-time information and then compare each of the pool for you and this take like few seconds to complete then after you have the apy of each pool that could say if in there and then what you want to know next is that uh how much eat do I have then you can also connect your wallet then the agent would understand because they would draw the oil chain information saying like Okay you currently have this amount of eth in this wallet then after this stage you can tell the agent that okay I would like to take this action and this is pretty cool because you can take like B can allow the agent to take the best action which is I'm gonna save you time and potentially some gas as well you can also tell the agent for example to stake let's say 15 eth in lighter pool and another 10 insect wise for example and then the agent would send you the simulation of the transaction and then you can also take action right on this interface um yeah so basically this is like long story short about what we're building uh we are just starting so we did like a very soft launch like few days ago if you guys are interested to try out or sign up so please simply go to our website meaty.ai and yeah by about next week or maybe 10 days we will release the first version of the product um yeah do you have any question no cool all right thank you so much [Music]