Restaking Revolution: From Jake Hartnell, Sreeram Kannan, Sydney Lai & Moderated By Alexis Johnson

[Music] awesome all right uh excited to be here today at e Denver it's my fourth e Denver um excited for this panel reaking Revolution we have some amazing panelists here I figured we'd start off by doing some quick intros my name is Alexis Johnson I'm the co-founder and CEO of light node Ventures we are Venture studio and advisory firm based out of New York Sydney hey every everybody my name is Sydney Li I am one of the co-founders and CTO of Gaia what we do is we provide developers to build AI applications that you can have portable context in memory for your AI applications hi everybody I'm shiram uh founder of Igan lay uh Igan lay is a mechanism for Building verifiable Services hey everyone my name is Jake I'm the uh co-founder of a project called layer the best name in crypto just layer terrible SEO uh we make something called waves or wavs ws. XYZ it lets you easily build decentralized offchain protocols it lets you easily make avss and we need more abss let's go hey awesome okay so our first question is a little bit broader and and and more intro topic but for everybody here and uh for audience members that are new to to the concept of reaking you know how would you define reaking in in in in simple terms and how does it differ from traditional staking methods that's for me okay I'll take it um reaking the idea is when you stake you're making a promise that you're going to run a network correctly that's what you're doing reaking is the idea that when you're staking for running ethereum Network correctly you're also promising that you're running a whole bunch of other networks correctly you choose what those networks are and you as long as you run them correctly you're going to earn rewards from all of them if you don't run any of them correctly you may lose your deposit this keeps the system honest nodes that participate actually remain honest but this massively increases the surface area of innovation because now you can build anybody can build new new and interesting Services we call these Services they could be blockchains but they could be much more generic than blockchains we call them AVS earlier we used to call them actively validated Services services that need active validation you need a group of validators to run and validate that Network we've rebranded this to autonomous verifiable Services autonomous they run on their own what do you mean run on their own anybody can come and stake and continue you're running that service verifiable they don't do the thing correctly the run the service correctly they're going to lose their deposit and be adjudicated correctly so you have autonomy and verifiability combined so that's the idea of reaking reaking is you're sing to run more than one service correctly that was a great answer I don't have too much to add but you know a word sometimes like alternate words you know restake a lot of people don't know what staking is maybe in this audience they probably know what staking is but I actually like the word commitments and I think of reaking and you use this language a lot as a self enforcing commitment system right and what are you committing to in this case in the case of abss you're committing to run some offchain protocol in a verifiable way awesome and so my next question is you know what are some of the known benefits as well as the potential risks associated with reaking I think that like the main benefit is that we can build full stack decentralized apps I think that we've kind of hit a lull in terms of crypto apps and what's possible with purely just onchain stuff right and what reaking enables is it allows you to write decentralized protocols that can do things like make Network requests or open websockets or store files or run things on gpus or do a lot of compute or run things in Trust execution environments like you know things that you need to like make real applications and this opens up a wide array of like a whole new generation of decentralized protocols this is a really uh interesting point you know if you go back to let's say 2017 2018 e Denver you would see lot of excitement around non-financial General applications oh can I run a uber onchain can I run like a um Twitter onchain can I run now ai on chain right like these kinds of use cases where in the air and I think what happened is in the last seven years you know if you go back to 2017 2018 2019 the infra was nowhere near ready you can't build an Uber onchain you can't build an AI onchain none of the stuff was even conceivable I think 2025 we're at the point all the infrastructure is ready you have the ability to anybody has the ability to build and deploy these autonomous verifiable Services application Builders don't even have to build these Services they can mix and match these services to get the properties that they want so this is really really cool and I think this kicks off the reimagination of what is possible with crypto crypto you know if you look at why crypto people say crypto because payment rails right like yeah not that interesting they say crypto because you decentralize it most users don't care about decentralization but they care about verifiable commitments if an app can make a verifiable commitment you Ma some commitments are so important that they can themselves expand the market as an example Uber right you want to compete with Uber you may want to make Uber is taking now 60% of the fees which go from the rider to the driver can you build an Uber which takes 0% and this commission or only 5% to just sustain the network this commitment is enforced programmatically on the blockchain so a set of things that you can build the benefits of reaking so to say or this kind of verifiable commitment system is massive I I was actually going to add one of the reasons why Gaia even looked into agan layer as one of our main Partners is that to be able to re stake is so important in the context of building AI applications and the reason is because you need to make sure that the model or the data sets that you're using is truly who they say they are and one of the main reasons we took this if you will more of a economic or reaking approach is because as of now one of our leading thesis is that something like te which requires more Hardware scaling specific types of Hardware it's not the it's too binary for being able to handle a lot of cont context and Nuance that requires for AI to be able to serve the end user which you talked about so even with an Uber that decentralized Uber application it might there might be some binary decisions to make AKA drop it here drop it there you know point from one point to Second point but then if you can even imagine there's a lot of nuance that happens even within your own Journey new data sets or new information might might be introdu because your friend says oh actually can you actually drop me as well off there and so if you have something that is uh too um I said I guess Concrete in terms of the settlement process then it it really is not Dynamic enough to be able to process these types of AI applications yeah just to add one one more thing on that uh you know some people ask me this conference why why crypto xai like what what is this what is going on in this intersection as you know all three of us here on stage are very Pro bullish on that intersection on and doing different things in that in in various parts parnerships here but to make that case let me just start with crypto has an apps problem we've all built a lot of infrastructure crypto has an apps problem there are no apps here there are four apps Bitcoin Unis swap a poly Market not enough to sustain our Market here crypto has an apps problem AI has a trust problem intelligence everywhere non- trustworthy is catastrophic AI has a trust problem AI agents are the new apps crypto is the new trust crypto xai is Unstoppable amazing um next question for you shiram you know igen lay has pioneered a lot of the reaking movement you know what lessons have you guys learned since launch and how has your vision evolved for igen layer yeah uh so we started igen layer the word igen in German means your own your own layer open Innovation anybody should be able to come and build services and applications this was the original vision and we named this even before we knew anything about reaking so IG L is not a reaking project I ler is a project to maximize the surface area of open Innovation so how do you maximize surface area of open Innovation how do you Ensure that so what we started with is we said hey when somebody wants to come and build a new application or a service they need trust where do they get it from they get it from staking so there is an already large pool of stakers in ethereum let's rely on it and get the network started but stakers are only the first side of the marketplace so we have like you know depending on each price between 13 to 15 billion worth of stake on the aanl platform the next side is operator who is running these Services we have 2,000 operators on the igen layer Network starting from regular home operators to the coinbase and block Damon to Google Cloud running first party operators so that's the operator side on agen the third side is people building these Services we call them autonomous verifiable services and we have 100 plus services that are being built the next Frontier is apps so you have stake you have operators you have avss right now the AVS developer Journey sucks that's why we have this guy here they're building one of the really coolest um platform for expressing and building very complex distributed systems building an AVS is kind of like building a whole new layer one it's a daunting task you need to massively upscale up tool build like very strong developer platforms so that's that's one major problem that we have that is being solved as we speak by layer and using their framework called waves and so that's one the next level is how do you build apps using these avss apps themselves could be avss but apps could use avss so that's that's where we are in fact one of the things we're doing to like make people reassociate us not with restating but with this idea of verifiable commitments we're calling Ian layer the verifiable commitment Cloud it's a cloud right like AWS but it's verifiable commitments where applications Services sters all make verifiable commits I will also say that verifiability is so absolutely critical because again if you go back to today's analogy right if you look at YouTube it's really hard for you to understand which YouTubers have the verifiable data sets you might decide by doing your own research you might decide by just understanding what followers they have what kind of content they push out and one of the reasons why I use YouTube as an analogy is that when you look at how the marketplace of AI applications need these very fast portable context for your applications you look at how when you go to YouTube you have uh a Content that you're interested in you're not going to search for General YouTuber you're going to go vertical specific you might be looking at information such as hey I want to buy you know a car a Tesla maybe you want to then look at vacuums or plants you're going to go into very specific domains of expertise and what why this is important is because what you've noticed even with you know the most previous models within chat GPT open a open AI chat GPT uh Google bard Etc when you're going through this prompting you will actually see hallucinations you'll see uh incorrect information and sometimes those incorrect information comes from their their models being trained on unverifiable data sources it could be blog posts it could be content and you know I've actually had arguments with my chat gbt where I said you've this is actually Incorrect and they said back to me yes I know this is incorrect but I needed to complete the task it was the weirdest thing I've ever seen and it actually freaked me out but I think that this is why having this verifiability layer um in a in a node network of different models and agents so that you can trust I think back to what you said earlier being able to trust Ai and truly truly trust AI so yeah you know that kind of brings up a big question of trust is liability as we make these more autonomous AIS as we make as AI like gets embedded in more and more crucial areas makes decisions that impacts people's lives or spends a lot of money how do you handle the liability this is actually a big question facing the industry and I think things like verifiable commitments and reaking are the mechanisms by which we can like effectively have insurance or liability Frameworks for this like increasingly like AI driven World awesome and to expand on shir's last comment on you know what Jake is building uh to help uh new avss launch quickly Jake can you expand on what your building at layer how is your solution unique tell us more about waves it's called waves or wavs some people call it wavs I like I personally like wavs um you can find out more at waves. XYZ ws. XYZ what it is it's a Wy AVS runtime I'll unpack that a little bit we've talked about avss Wy is web assembly system interface so what you can what we try to make it as we try to really improve the developer experience so that you don't have to worry about writing code that listens to rpcs from different chains right you want to listen to events on bass or arbitrum or a cosmos chain or salana right and have that trigger some action right you don't have to write that logic right uh doing things like offchain triggers or scheduling this is all parts that you would have to write previously to make an ABS we also hand like we basically make it so that you only have to write your application logic in these Wy components which you can dynamically deploy to the service without having to coord with your operator set on like an upgrade and then we also on the other side on the submission side we handle signing the outputs with like BLS signatures aggregating those signatures and sting them on on chain or potentially we also introduce a new thing called composition because these services are really great it's much more interesting okay we've got a we got a dkls service and we got an AI agent service and we got an oracle service Standalone they're very useful but if you can combine them together and compose them in verifiable execution flows that's when we get to the next level meta and that's what's going to enable us to build the next generation of great crypto apps and user experiences is these kind of intelligent protocols that's really crazy I didn't know that the composition framework so one of our vision for thinking about how one maybe metaphor I can start with is when you think about web application development imagine it's 1995 I know most of you are not old enough to have lived through that but some of us here are um so 9995 if you want to build a web application you have to either go build your website on some very stock application like geoc cities and then just like do a static web page or if you're building something very complex and dynamic you have to build your own server build your own database build your own um payments identity everything thatas in 2025 you absolutely just use the cloud use a oath for authorization stripe for payment mongodb for database mix and match them a typical application on the web application uses 15 SAS services in the back end and some people say oh you know when you have many modules in crypto is there going to be enough fees the largest Venture backed investment area in the last 20 years is software as a service and the reason is software as a service means you're hyp specializing in providing a useful utility for everybody and so that's what we think of avsr can we have ai inference which is a very fundamental core AVS that'll help everybody build new and interesting apps so but when these can be composed you get some superpowers and last question we have about two minutes left you know why should uh AI agent networks leverage reaking and restake assets I think that in in this kind of Journey that You' you've been painting about where we are in the development cycle of um applications autonomous applications AI applications one of the main reasons why we chose reinking for our node network is because something I kind of alluded to earlier uh there's there's two main reasons what I alluded to earlier was a lot of the leading um security and verification let's say it's te is still very Hardware it it requires a lot of Hardware considerations and it makes scaling very difficult by Design what we wanted to do was make sure that you are not um locked into whichever vendor I think you talk a lot about modularity so modularity is very important and being able to handle Dynamic context and uh Dynamic data and information that is input into your application another reason why is because we actually we don't need humans to what what we're looking for humans to do frankly in the process of staking voting you know kind of the operator set um is that we're still building out the application layer stack um if you're vot if you're creating AI for humans because at the end of the day what are we creating AI for it's to really improve the lives in our own workflows right and so as we're in this process we really believe that reaking is able to handle the Dynamics of security requirements and then number two there's human interaction to help design the right uh ux layer and the right workflow automation that is truly needed uh for us here real quick I think it's it's how you actually make decentralized Ai and why does that matter because maybe decentralization is too abstract of a concept for many people but it's really simple it's about the terms of service and if you don't have de if it's not decentralized then you're using some other terms of service right and I for one would rather live in a like a free world and not be a a techno feudalist surf like a centralized AI Marketplace and if you stop paying subscriptions they'll take it away and then you won't have your memory wow awesome so unfortunately that's the end of our panel but I'd like to thank all of our panelists here for uh contributing and our an our audience here for tuning in thank you everybody and hope you enjoy e Denver thanks everybody thank you so much Alexis thanks everybody