Solving Cross-chain Interoperability Using AI

Transcript

hello folks I'm rahan I am currently an ecosystem manager at stockar uh if you don't know what stockar is stockar is a developer behind stocket which is a layer two I think one of the first ZK layer 2s in the ecosystem um that's U I'm not going to Deep dive into what stock is and what shockware does um but basically I think I wanted to take today to you know take this time to talk about at least what I feel um AI does uh enables in the in the crypto ecosystem and the blockchain ecosystem um and this is going to be more of a practical hands down guide of what actually uh gets unlocked through AI in crypto rather than hand baby gesture of saying you know what you can solve this problem you

can solve that problem in terms of um actually what we are doing as well as share um so let's Dive Right In um no um this shouldn't be a surprise but basically I just wanted to take it all the way back to like what the what the eum CIS was in the in the beginning um the thesis was basically to become a world computer um one Ledger um basically publicly verifiable anyone is able to participate in verification like this is the crusp of what ethereum was and is however along the way we found out that this is not the most scalable or the the the most viable option for us to get to a world computer so we changed a little bit it it came up to World

computer but it's made of other computers in terms of l2s um and and the sequencers um many ledgers UniFi to one master Ledger rather than just having one Ledger like Solana um publicly verifiable um like basically you have to publicly verify the master Ledger and anyone is able to participate in verification of the mar Ledger but this is basically what ethereum is today in just plain layman's terms now the issue with this is the fragmentation I think we can see like eight logos on uh on the rollup side there's like five logos and side chains this this image seems to be quite dated to me but I probably think there are like I don't know some 3040 rollups uh

20 30 side chains some 30 40 channels and the point here is this is the this is the way you can scale ethereum but is also the way where you have a lot more fragmentation um fragmentation in terms of everything which we'll Deep dive later on but this is the CRP of the problem etherum is facing right now um and we're just going to Deep dive into like how we can use AI to solve this well there are a lot of problems but sub problem number one um maybe some of you technical here maybe not but um every chain that choose to go non evm these days alt alt VMS are like in the trend everyone have a new new new coding language there's solidity there's wiper there's move these days stock stocket

has its own Kyo coding language um there are like I I think I came across a project that was trying to unlock def5 use cases using python there was a project trying to unlock D5 use cases using other languages like rust and just so if if I'm a developer who just comes into or like let's say I just graduated from University and I just know like Python and maybe JavaScript I'm expected to learn I'm expected to pass through like I'm expect to pass about 10 20 el2 figure out which um language they use to for me to build them build upon them um learn that language spend like three three months in a boot camp and then uh I may end up building like a small PC and that and if if I if I find out that

this chain is not what I wanted to not what I expected to be I have to move to another chain I have to learn coding again I have to learn their own language and then again learn learn coding and then go upstairs it just just feels like a very broken process um well how is it solved today well today you have like uh Team spend huge amounts I mean we Shak also do this because we have like Cairo which um is at least in our opinion a better look better coding language but hey we boot camps courses big numbers of devils tutorials hackathons mentorship programs all these are non-scalable ways basically non-scalable ways in terms of onboarding developers and you're you basically forced into this Zero Sum game of hey um these are like I don't know

some 100 developers all of these folks are new to crypto now 10 layer tools are Behind These 100 developers trying to onboard them to their ecosystem learn my coding language um our coding is better our coding language is better our coding language is better it just doesn't make sense it's like a zero sum game when you should be pushing for getting web to folks in crypto rather than just fighting for crypto people um so that's so basically so problem number one very inefficient way of solving it today sub problem number two broken ux um very cliche and very used up term these days but that's the reality I mean by the way these images I use Strat GPD for these images um but yeah this is this is exactly how at least I felt when I

started off in crypto I I had to a metamask a phantom an Argent a braavos um I had had like I still have like I think six seven wallets in my laptop and that's something which I don't don't want and I also don't know which wallet supports which chain I have to become like used toist spend like a couple of months in the ecosystem stuff which I don't want to deal with and stuff which basically has a high drop off R so if anyone from web2 comes to web 3 they see all these hurdles if you're a developer you see those hurdles you see if you're a user you see these hurdles I don't know I think the drop off goes up to like 70 80% and that's that's not something which either of us want either Which Way um coming to like the broken ux and how

it's solved today you have wallets that are multi- chain these days you have Bridges you have cross indexes you have things like hyper layer zero which have some sort of traction but the crasp of the matter Shield remains in terms of the broken ux I guess you go on and on problem number four five six like lower efficiency in def5 is zero some environments these days you have W Gardens in terms of like polygon a layer or like um ZK sink layer or like having multiple chains Under One Roof um and then Walling them off as like beautiful gardens um all these have Arisen from these problems of scaling eum which is the layer to ecosystem which is fragmented well there must be a better way right um as I said in the beginning I I've strongly believe Ai and

artificial intelligence is the way to go about this um it basically if if you look at even uh non crypto why is AI such a big hype is a question well it's such a big hype because now uh my grandmother who's 80 years old doesn't need to do a Google search you can just go to chat GPT and type in natural language this helps us interact with computers in natural language why is it any different why is the world computer which is ethereum any different from a normal computer you can you can interact with it either which way using AI That's essentially my thesis um so what's listed like two bull cases uh like the absolute top which I feel uh make my case more compelling uh number one is abstraction you have you

basically if if I can use something like chat GPT but it's for crypto um transactions I don't need know which chain I am on I don't need to know Pro I don't need to know programming languages I can just tell it to code for me I don't need to know which defa protocol gives me highest yield I can just put my money in say you know what give me 12% yield uh on a risk adjusted basis on this much amount of money the the the agent or the assistant can go out and do it for me find out which chain has which protocol which is the most uh least risky and just deposit my money in there research like I can probably know um which K or Twitter have the have the highest mindset these days stuff like that um so basically abstracting all these complexities is

like the biggest thing um basically being able to interact with crypto in a natural language uh example um this is actually one project I'm currently following and working with um as you can see we just gave it a prompt which is Let's Make a token where you decide what it's called what the supply is code and deploy it entirely on your own give me one% of Supply do it on the base that's it um I don't need to know which wallet is active on base I don't need to know how much money I need to have on my wallet I don't need to know which bridge to use what the hell which coding language is used in base I don't need to know um this is one example and by the way this example is live uh the demo video I I'll probably share if anyone is interested it's live

and this this bot is currently working it deployed this on video um bcast number two automation productivity I think um this is like the second one which is all manual process like search uh analytics trading um I mean we all use trading bards uh since like the past one year or two years um when when ethereum hit certain price buy this much for me blah blah blah but like even taking it a second level or third level deeper and saying and having complex logic um and like automating that logic is like quite quite simple using AI um but am I just retrofitting uh this whole AI thing to like make you feel that I am smart no um I think the the best proxy

is just look at web to um clae 3.5 Sonet it's able to solve 64% of coding problems this was I think 6 months ago um I I hear rumors CH is coming new version is coming out in next month is I think probably going to blow past these records as well so I don't know I mean web2 example seem pretty solid so why is web 3 Left Behind um Cleo well basically uh it's Cleo is a defy app uh is Cleo is not a d app Cleo is a web 2 app that helps people manage their wealth um it's a British company closer to become a fintech it's been using AI for like I think 3 4 months 3 4 years now um these guys have been just killing it again

another good use case in terms of having assistant that helps me manage my money uh we use we have to us as it in traditional world we can use it in D5 World why not uh perplexity I don't know how many of you know perplexity but it's basically going to replace Google very soon is what the founder claims I use it in fact I use perplexity I think to prepare half of these slides um very very helpful uh lot of automation happening there um AED content and media I think this is a post which Sam wman put out uh 3 months months ago uh but you can basically create kast videos using chat GPT I think the the version I forgot the name um but this is all generated content I don't need to be um

a video editor graphic designer whatever I can just prompt it on jgpt um hope that convinces you now the next question might be well what are the barriers barriers are I think the first and the biggest barrier is Talent which is um crypto looks at Ai and says at least I look at and say there's a lot of potential which I can use AI to solve for crypto and that gets me excited I am I've met a few people but not a lot of people who are actually excited about who are actually in Ai and looking towards crypto and want to work in the in the industry and uh this is something we have to change I have no answers but like the biggest barriers are Talent the fundamentally big biggest barrier is

Talent everything else is follow on like on onch intelligence like security smart contracts having like piping in uh offchain and like getting Z to prove it pretty easy uh proving Shak itself has like the one of the best provs in the world so like the follow on like sort of issues but not the biggest issue like Talent is the biggest issue um well I I'm pretty open to like suggestions on like how we can solve it but this is essentially what I'm beating my head towards for the last couple of months uh whether this is just going and meeting people in AI conferences or is it just um talking to them and understanding and going through a referrals um well what is snet doing about this whole fragmentation thing and like how are we improving our ux um and our CH

abstraction well um firstly like we've put together a build station currently active online hackathon uh using Brian Brian is also integrated with stocket um helps do make API calls to agents on the back end um trading um depositing money put spinning up a wallet pretty easy um Bas as a contributed bunch of people are contributing here we should if are a builder come and contribute um second one is Agents uh I think this is our agent which left went live four months ago um this helps code developers uh pass through documents uh pass through our Kyo Playbook our Cairo book and stuff like that where you can um you don't need to go through

documents like perplexity but for stocket um Giza um so Giza is one of the very very first AI ml projects uh in the crypto ecosystem if you can read in in in in the small letters there it says uh web3 and from Giza plans to bring autonomous Bots onto stocket by end of June June as in June 2024 so that's like 6 months ago so we have we have bots on stocket since like the past 6 months and Giza has been a very very big contributor of that they start off as like a zkl project trying to prove AI models of chain and bring them onchain but have now also started focusing on agents and enabling those use cases um we've partnered with ritual um by the

way all this is not a shill this is more of a this is what we've done see if this makes sense for you and like try to copy it like why not copy things or like why why why try to do noral things when there like low hanging fruit so we partner with ritual this is very recent like I think last week or like a couple of weeks ago um ritual basically is like a co-processor where um competition done offchain can be brought onchain um in a in a trust minimized manner they have like Z proofs optimistic proof whatever uh I can basically run llama 3 I don't know if I can R run a Lama 3 but I can at least run a billion parameter llm on the back end and uh get the inference on chain using Giza in a very cost minous

way uh IO um hio currently uses um Giza and it's basically a D5 application which helps in reversion strategies um yep proving well I don't have an image for proving but we basically have like a bunch of provs at shockware like Stone as of now which is currently in production uh St uh is experimental but we I think we've hit like the the maximum amount of transaction for the lowest amount of uh money possible it goes live in Q2 of next year is uh it's going to be like game changing in terms of having chat GPT on chain like how do you get t on chain which basically

extremely extremely expensive to prove it uh to prove the competition happen off chain and bring it on chain hopefully still Bridges the Gap you not charity at least like Lama 3 or something like that um like other stuff like no code tooling chat assistant user generated content for games stuff like that I mean sh is very big on gaming and so one of our cases is around uh user generated content in terms of um not only like written content but also like more in terms of images and videos um the result is very much a 10x Improvement in ux at least at least we we are on the path there we haven't achieved it it's shockware at least but at least this is something which I hope to happen aing abstraction improved

Capital efficiency no Cod tooling games with complex logic I can I could have gone on and on I was just too sleepy so I just um put it down um but yes that's about it from my end I kept it very short try to you know make it a bit more non Technical and just give your folks aot lot more impact uh happy to take any questions okay