Fluence: Truly native Web3 apps with P2P decentralized compute

Transcript

foreign [Music] in Berlin and I've been going around talking to people about projects they're building on and discovering some of the things people are building in the gnosis chain ecosystem and I met up with Guinea how are you doing it's been great great conference yeah the weather yeah the weather is pretty good I mean we're doing this on a boat right now which is really cool um so yeah if you work at fluence you're the co-founders of fluids tell me a little bit about what you guys are doing and uh well it's interesting yep we we are working on a decentralized compute Network which is basically the

decentralized technology of traditional clouds providers like Amazon AWS Microsoft and Google and basically we are providing the infrastructure and development stack for developers to build and deploy applications to such Network and do not depend on centralized like vendors and don't have vendor login and avoid like avoid this link or dependency on centralized infrastructure providers yeah so I mean in the crypto space over the years there's been a lot of projects doing decentralized compute so we have things like you know I exec uh Golem I think was also an early project that was trying to do something at least I think with the similar Vision how is fluence

different from those and perhaps you know how does it build on some of the cryptographic Primitives that have come since then like how is it how is it better in the next thing like as opposed to some of the earlier examples yeah I think the the main thing here is the basically like two two things is security model and the developer experience and like every project two different choices in terms of how their security model would work like for example you can create a network where people can run nodes and and be the providers of computational power but they would not provide you any like verification of the computationally they run so you need to rely on the reputation particular node and this is

the choice that you can make when you're building the like the digitalized computer project and I think like earlier projects like they followed only this direction of like a reputation-based networks and at this other the other side of spectrum like for example there is like a blockchain networks where all the nodes are the same and they have consensus for every transaction they also computational networks but they verify every transaction every state transition that happens in in the network and that puts a lot of constraints on amount of data they can put amount of processing power that they can have as a network and this basically makes the um the cost of computations work quite

high so what we are trying to do is is the flexible peer-to-peer Network which is like off chain there is no Global consensus or Global state so every every note is different and every node can serve different applications but the the application developers that can ask the some nodes where they deploy their code to to verify each other work for example if you if you want like you can ask the nodes to um deploy some some consensus algorithm if you need verification or you can select the nodes that are most trusted at the same time so like we kind of combine this verification way to like the the ability to verify the

computation and the reputation based system and we're going to have more of this but it's also applied to how our economics works so if I understand correctly the the verification mechanism in order to verify like the computations are actually done or not like part of the stack but it's something that you can add on and it's more modular yeah you're sort of modularizing the decentralized computation stack by offering the sort of way for for people to build a network around decentralized computation and then get to build whatever stack they need in terms of consensus in terms of verification in terms of economics is that yeah so so the verification is pluggable and

customizable so let's say for example you want to have an oracle for your like you're building a smart contract and like some some on-chain application but you need some data to come to your on-chain application from the all chain world and you basically need to to have some of chain Oracle and you can build this influence by saying that you want to create a subnet deploy your your job the computation job on several nodes and then you can basically program whichever algorithm they should follow in order to verify each other for example you could program this simple you know polarization of like deploying tasks to several nodes and then asking them to find the average and you know this data

you would just put to to the chain as a result of your Oracle job or you can say that you want like 10 out of 20 to get exactly the same result of the computation and we will have like a 10 signatures under this result and this you would you would think that this is the like verified computation when you submit this on chain so the on chain contract will see 10 signatures that it would say okay this is secure I trust this data okay what ecosystems are you guys building on is this a product that's easier to integrate for evm chains or are we also considering other ecosystems like Solana or or the inter chain or yeah so the network itself is like chain agnostic so like a soft chain

it's a vertical Network you can think of this as like ipfs as a security pair Network to store data fluency security network to run compute and you can plug different data basically sources to this network and you can plug blockchains you can plug decentralized storage like ipfs falcoin RV and stuff like that um yeah and on the Chain like if you take use case of when you want to process some on-chain data compute on it and then put result back on chain or store it into storage we have currently built-in adapters to ethereum mainnet to near and working on like adding a few others but like potentially it's like it's possible to plug fluids to any layer one or Layer Two like any chain

yeah uh it's just like the question like how how fast we build this adapter so Community builds these adapters to change okay so a contract you could have a contract that is doing its thing it can call it can call influence to execute a complex computation that you say couldn't do on ethereum just because it's too computationally hard influence comes back with a callback and Returns the data or Returns the result signed by say like a consensus of nodes that have verified the data the contract moves forward once like these parameters or like these criteria have been established and then like it keeps doing its thing yeah so this is the Oracle use case and this is basically one of use cases that's possible but there's also

could be like a vice versa use case when you have the fluence code that falls events from the chain and like if compute you know just you know you have a current job or influence yeah yeah it executes like one once per second and then it queries data from the chain and then when it it got the throne event it runs the computation and then it's for example again submits the result on chain or you know this place or sends a message somewhere in the notification to user so you can have all these Bots living off chains reacting to unchainable so you could build like some sort of a um like decentralized data layer or it's like Computing Unchained data uh something similar to like the graph or

okay yeah like you can do like a you know some data processing or indexing or uh you can you can like have some trading Bots for example build with it or there are some use cases like where in defy you need like some liquidation protection like yeah you need to look where is your like where is the price and then when when the price is close to the liquidation price you need to put additional liquidity for example so you need to have off-chain bolts that monitors the the on-chain activity yeah and you know you can do this for analytics purposes like you can build some you know uh local indexes of some smart control set of smart contracts or

some subset of data and basically like all the things that you put right now in the cloud in centralized Cloud you potentially uh can do here cool and what are some of the things that people are are you live or yeah so building influence so the network the the off chain network is live already there's only a thing that you cannot pay right now the network so in the cloud your your sign up then you put your credit card and you pay and they provide you available service affluence Network right now works like ipfs Network so it's not incentivized yeah that's why you cannot really pay to some node and and to provide you some available uh resources and that's why you basically can either rely on some notes that vs balloons Labs running for Community or

you can you need to run the node for yourself which makes sense when you developing the application but you know when you go to big production environment you you want to have something like really provided for you by professional computation providers so yeah we are not there yet the payment layer gonna probably come live next year uh but the like the off chain nonsense device network is is live and you know so so like to analogize here like it's the it's the ipfs and file coin like the economics yeah it from the computation computation Network okay yes so economics is coming the the computation network is like cool it's what's on your roadmap like other than this here like

what kind of things should people expect from fluency yeah we we actively researching and looking into different use cases like for example this Oracle use case is only one thing that you can do the other thing that you can do for example is you can create like decentralized if you're a influence for example yeah you can uh that's cool you can basically like have some resolvers into RPC requests like load balance in RPC requests or like failover you can have failover if like the first node that you query doesn't answer you can fill over to the other node so you can build this distribute algorithms that you can have in the cloud as well with their load balancer tooling and then routing and and API Gateway you can have

it there here but here they work in a decentralized fashion without any centralized coordination place so you should check out our developer tooling it's okay yeah so the biggest steps for us uh one of the biggest steps is we want to launch the Dow you want to start the dial that would govern the whole protocol and after that we probably launched the like the payment the payment and send to fail okay um and on the like we constantly do some new stuff on the development stack uh level so we we shoot updates like you know a lot of stuff so what's the website where people will find more about uh it's fluents.network okay cool thanks for being uh on this uh short interop

interview thank you for having me and testing out this new gear for the first time it works I hope the sound is good foreign [Music]