David Minarsch: Autonolas – Autonomous AI Agents. Ep. 532

Transcript

welcome to epicenter the show which talks about the Technologies projects and people driving decentralization and the blockchain revolution I'm F an and I'm here with Mah Roy today we're speaking with David Minaj who is the co-founder and CEO of Valerie and founding member of oronas and oronas is a funny project it kind of it's an AI blockchain crossover and why that is interesting uh we'll dive into in just a second let us tell you before um about our sponsors this week though this episode is brought to you by nosis nosis builds decentralized infrastructure for the ethereum ecosystem with a rich history dating back to 2015 and products like safe cow

swap or nosis chain nosis combines needs-driven development with deep technical expertise this year marks the launch of nosis pay the world's first decentralized payment Network with agnosis card you can spend self- custody crypto at any Visa accepting Merchant around the world if you're an individual looking to live more onchain or a business looking to wh label the stack visit nosis pay.com there are lots of ways you can join the nosis journey drop in the nosis Dow governance form become a nois validator with a single gno token and lowcost Hardware or deploy your product

on the EV compatible and highly decentralized nosis chain get started today at nosis doio cars one is one of the biggest node operators globally and help you stake your tokens on 45 plus networks like ethereum Cosmos celesa and dydx more than 100,000 delegators stake with corus one including institutions like bitgo and Ledger sticking with course one not only gets you the highest years but also Al the most robust security practices and infrastructure that are usually exclusive for institutions you can stake directly to corus one's public note from your wallet set up a white label note or use the

recently launched product Opus to stake up to 8,000 e in a single transaction you can even offer high yield staking to your own customers using their API your assets always remain in your custody so you can have complete Peace of Mind start sing today and cor. one hi David it's a pleasure to have you on yeah pleasure to be here thanks for having me absolutely tell us a little bit about uh yourself and uh your background sure um so I came to crypto from a sort of uh background in maths and economics I did um maths undergrads

and then uh really uh got into applied Game Theory um there were some fantastic courses at UCL where I did that and one thing led to another and I ended up doing a PhD there um and then that um you know if you fast forward quite some time uh LED for me uh to discovering that I really like this intersection of Game Theory and um machine learning which I done a lot um together with um an interest which had sort of grown steadily in crypto and blockchain and so I'm working in that space now for over five years

and particularly at this intersection of crypto and AI yeah super interesting sounds like uh applied economics and maths uh and yeah sounds like the ideal background for getting into crypto you as we said in the intro you are co-founder Valerie which is a cool contributor to oronas um you co kind of co-founded with um someone else also called David um and what kind of what motivated you to kind of co-found this project kind of what's the problem you were setting out to solve yeah it's a it's a great question so Valerie's mission is to basically

create uh open- Source software for people to co-own primarily a gentic AI um and we'll kind of uncover a bit what we mean by this um my I actually have two co-founders um one David Galindo um has a background in cryptography um what's cryptographer and S um and the other co-founder um has a ponm called Oaks BR batan he um has a product background and the three of us uh really kind of um in different ways um were excited about a tonomous agents and this General pressure which you see in um AI towards a

gentic kind of AI systems um and we had different experiences with this topic different um insights and we came together to basically build a substrate on which you can as groups uh kind of co-own these agentic AI systems so I think that's sort of the driving force is to provide this kind of software set which allows people to do that and also kind of create applications which people then can own in that way you've used the term agentic AI system um quite a few times there what is an agentic AI

system the way I think about it is that if you look at the sort of dominant forms of um AI then you can sort of see maybe fre WS so you have like in the earlier parts of the last century like this dominance of rules based systems um and then and in day basically say you know you have hardcoded rules often extremely sophisticated which allow you to build sort of um certain types of AI systems and by no means has this kind of um part of AI research and uh applications gone away but at some point you then had more these kind of learning systems emerge where you have like neural networks and deep uh learning and other forms of learning reinforcement

learning where you effectively use data to uh construct part of the um algorithm effectively right so the the system learns from data rather than all the rules being prescribed and if you look at what's now happening is we have these very powerful large language models and other types of um uh powerful AI models but they by themselves um are certainly now not not atic in the sense that they you know they some some data which sits somewhere and then you effectively query these models you instantiate them you query them and then

you get a response and that can be a very sophisticated response but that's it um what what's interesting is once you think about um effectively uh systems that are having agency and can sort of autonomously uh act and there's an enormous pressure towards the system from a pure optimization point of view and evolution point of view like if you think about it where can you make the most money where are the most exciting applications it's well it's autonomous systems uh which can take actions by themselves and then are you know maybe instructed by some third party some human or or some other

system and so yeah these kind of agentic systems we can uncover a bit what what they look like conceptually um but that's I think where the train is headed um where where a lot of focus is uh going towards across like the AI Fields so David maybe to kind of let me repackage this somewhat would it be fair to say that an a gentic system um is something where you kind of give an AI agent in a goal but don't perfectly specify how it should how it should go about achieving that or is that too simplistic yeah I think um that works

and in particular I mean what we're interested in are sort of these autonomous uh agents and so we can briefly Define that sort of conceptually so usually it's a sort of software system which is placed in some environment um from which it perceives certain information um you they could be blockchain events literally um or they could be things from an API um they could be something from a sensor um which it has locally it then uses that information plus whatever it's internal architecture looks like to then take uh action again in its environment and that environment again can be like a blockchain another API uh

another agent some actuator of any any form so this is what we would call like an autonomous agent and effectively what I'm saying is that what we're seeing increasingly is that there's more Focus to basically create U models which can act as subsystems of such autonomous agents or even like almost like subsume and autonomous agents as a whole right and so there's this kind of pressure um towards these kind of systems so as an example of an autonomous agent maybe we could think of um so imagine there's a there's theosis Network and then there's the code base

of theosis network and one could imagine like a coding agent of some kind where somebody opens an issue on against the code base of diagnosis na or against the code base of the network I want to add this feature to the code prot call of diagnosis net workor then an agent could be something that kind of as a first step isolates the pieces of code that need to be changed as a second step creates code uh making those changes as a third step does some form of testing so that could involve kind of like static analysis but that could also involve like runtime analysis then gets feedback from the

environment and then makes another set of changes and comes up with a draft like a draft change a draft pull request yeah I like that um I think that's that's a good example um also speaking ofos chain um so one uh autonomous agent which is running there one type of autonomous agent which is built on the autonomous stack and is running there every day is a agent which trades in prediction markets and so if you kind of map that into this model which I was just describing it might be quite help for so um again here what it's observing are sort of uh basically new markets um opening so it

adds us to the list of markets which it um kind of has a look at um it might then uh fetch information um pertaining to the events which are referenced in these markets from really anywhere in the web um so a search API or just like um crawling itself almost it then um uses that information that context basically on the event um as well as um various AI models at the moment most of the agents use some form of large language models um to basically uh prompt these models with that kind of information and then once it arrives at

a um prediction for that event um together with an accuracy and other kind of information wiet estimates then um sorry confidence wi wiet estimates then um it will construct a transaction and then sort of act in that market I kind of take a position in that market so for instance if it's a buy new market yes or no um by the relevant um tokens which represent these events and so here what you then have is this environment being sort of the smart contracts and the information end points were which are pertaining to these markets and then the actions are these POS taking these positions in no markets and then some time passes and then the agent um might

actually um make some money oh okay but that sounds primarily like kind of like automation technology right so basically people wouldn't necessarily know that I run this sort of software to do things just like kind of I run for instance say trading scripts right so um H how how do we know that this is I mean I assume to some extent this is already happening um but kind of like where it gets really interesting is kind of when you kind of design um systems where several of these agents kind of come together to kind of um in in a game theoretic um way to kind of figure out

um you know something to do or some conclusion or something right so yeah a couple of things so firstly I think um you're right that like there's sort of Automation and then there's different levels of autonomy and like if you think about a self-driving car they have these sort of levels um and it's a bit similar to think here like you have different kind of levels and you can be closer to what people might describe as Automation and then there's also this thing where as time moves on um we tend to prescribe things which maybe we s is more autonomous towards automation because I kind of get wrapped behind like an agent for instance and then I can just sort of see the act of

interacting with this agents where the agent is actually autonomous as for me from from the perspective of the user it's almost like just like automation I'm just calling this API which then goes away and creates an outcome for me and so I think there there's always um that um but but you're right and then um in this system which I was describing actually actually the way it's practically implemented already is that there's already three types of Agents um today so um the trading agent itself doesn't actually come up with the prediction it's other agents who specialize on that um and now we're even like picking apart that role because um what we basically see and is that from a

practical perspective um if you can sort of specialize your agent um that has its benefits like the same way we specialize as humans um um but also um from a sort of practical users perspective of running the the agent that can have its benefits so for instance if I had an agent which has to have all sorts of um let's say open source model which need to run alongside it um which it uses then this can become quite um a beast to actually run this thing um like quite impractical whereas if an agent can use use other agents to get something done then um it might be as simple as making a small crypto payment to for instance

get a predictions and so that's the case here and then you have to obviously trade that off um with other design considerations of the system so like to State fre the's question in like different ways so any standards taking company would would be running for example price oracles right so in in in a price Oracle it's fetching the price from somewhere and submitting the price to the blockchain and it's getting paid in crypto to to do that and the one could imagine that entire so it's like the the code of like a price Oracle is is highly mechanical

it can it is specified entirely in in in a in a programming language the input to it is uh very very structured it is probably coming as like Json files that that are structured in a particular and it output is also very structured it is producing transactions that have these fields and Etc perhaps that is like actually like an agent itself except it's like a very dumb agent and the kinds of Agents you are thinking of are like AI agents where we are trying to climb the hierarchy of well the inputs no longer need to be that structured it me we it may not come as Json or or protuff or any of these protocols it

might come as a English language and it could be anything that comes in so the input becomes unstructured then the processing logic instead of being structured in the form of code you could have processing logic where um the agent like d noo comes up with how to execute on a certain input and its execution path is kind of like invented for that particular input and it might be different from what it was uh what it was previously and then finally on the output side its outputs could also be unstructured meaning it's it's producing output in the terms of

English language which has like of course English also has structure which has like but less structure than uh a programming language output or a Json output would have and so maybe the oneing of the AI agent is we trying to generalize the input the processing and the output of what what is already kind of like uh a traditional crypto agent so validators price oracles we might think of them as traditional crypto agents but we are trying to kind of push their boundaries in like what they can do yeah and there's these different dimensions which you're kind of pointing at right so you have like the levels of

autonomy um and then the levels also of the kind of how Dynamic is the decision making um how open-ended um is it how structured and unstructured can be the input and output um and basically like if you look at it from our perspective the way we look at it is um our St kind of allows you to build across a whole range of these things so uh we have some uh products which are very very structured so they're basically rules based of the kind which you know an oracle is actually one example we we you can build an or on our stack it's not like anyone is like majorly focused on it but we have some

demos of the sort and then all um you know you go bit to the right on that Dimension and then you add this um prediction agent where it becomes a bit less structured because yes some of the flow of it is entirely structured in the sense that it will always sort of do certain um actions in a certain sequence I'll get to that in a second as to how that's actually done on a code level and then inside of these states actually let me explain it right now it's like we structured this as sort of finite State machines so basically we say okay the the overall agent is described as this graph likee structure where it transverses through these states and then in some of these states it might um

sort of dynamically choose which path to take going forward but sort of the Rails are given right so it can't just sort of totally go of the rails and suddenly say I was a prediction agent now and now I'm kind of doing this other thing um shopping clothes or whatever um and this um we see this as um basically a a pragmatic approach and B also a big Advantage because um you know obviously these kind of AI enabled agents autonomous agents is something relatively uh new in that form they were stuck in the sort of D DRS for a long time where basically nothing much happened for for decades in multi-agent systems research

I mean no no sort of big um move forward and um on the other hand you now have these sort of AI Asian models based mostly in large language models where it seems a lot is happening but then when you dig a bit in often if you leave them too unstructured nothing it's an interesting research exercise but practically not too much happen so The Sweet Spot is still in between is what we say where you provide a certain degree of structure and then within certain States the agent can be uh dealing with unstructured input or output and and and and could can do what you were just describing um and that I think it's um you know how long we will be in this

phase where it's so um an in between I don't know um you know there's suddenly attempts to build sort of of like almost like a large language model but for actions where people sort of train this sort of um in in into the model itself um we'll have to see when when they're actually I think usable but if you want to use off-the-shelf Technologies today um then you're sort of limited to still providing some degree of structure the other way to look at this is also from an efficiency point of view so once you actually know that your AG is meant to be an autonomous agent in prediction markets that it's meant to make its money there and that's that you want to

use it for that then it's kind of pointless is if every time it's running it has to figure this out from first principles that's a very dumb approach right the same way in in in in in in programming if I write an efficient program I might not generate everything dynamically I might have like sort of hardcoded you know lookup tables or whatever where I just um pull values out because is way more efficient than if I were to generate them on the Fly even if I can and so the same thing is here sometimes you might want to apply an agent actually at the building stage so going back to what you were saying earlier mayare um applying agents to build agents is also something we're focused on so we have like some internal tooling now where we are able

to basically prompt um our tooling and then it generates um sort of how of the Asian like not all the C is finished there's still like some software developer engagement needed but it generates a lot of it um so there's this angle as well from an efficiency point of view where you don't necessarily always want to figure out everything at run time you might want to sort of ahead of time build a better agent which is then forced to um act Within These bounds given by um that design I'm still a little bit confused kind of as for the Asian terminal techology so um I think kind of there's these cases where kind of I can imagine

you kind of you have large games that you kind of you optimize for and kind of that means kind of you don't you don't have to do so much onchain because kind of you can optimize it offchain and kind of Agents can keep each other in check right um and that to me is kind of like the multi-agent um system um at least in kind of my lay understanding um but kind of in in your description now it's it like um what I would have conceptualized as one agent you guys often think of as different agents that are somehow Amalgamated together into kind of like a super agent um as you said kind of like

there's the prediction there's kind of the research agent and the prediction uh making agent and so on maybe you can kind of um delineate the the terms here a little bit for us let me zoom out even like a bit further so one of the you know uh core things which I guess like the idea of multi-agent systems is is that you have multiple potentially different types of Agents which generate some sort of emergent outcome um so if you look at any individual of those agents then they themselves wouldn't bring about this outcome and then it's only the collection so this is the example I was G giving earlier where you have these

fre typ of agents and then they kind of coming together and the outcome are sort of AI driven prediction markets where no human ever participates in now if we look at our stack specifically um it gets um a bit more interesting still which is that we basically say okay going back to this idea of co-owned AI and cod uh autonomous agents like what motivates us there well what motivates us is that we're a bit concerned that as there is this tremendous pressure to build better AI models and as there's this tremendous pressure to build better agentic AI systems that

ultimately they will be owned in a very centralized way and also operated in a very centralized way and so the question is can you create basically a substrate where people can own them in a decentralized way so now one obvious answer is if you somehow can make a SM contract smarter and there's a lot of exciting projects which are kind of trying to do that with ckl and other kind of technological approaches where um effectively you just use a blockchain a public one and you run some code on it uh which might have been sort of verified off chain um uh sorry verified onchain which might have been proved off chain um

now in our case what we offer is basically okay if you want to build an autonomous agent and you then want to run that code as a decentralized system then you can do that so in the Ola stack you can develop this Trader agent which I mentioned earlier um and you then can run it as a multi node system so what basically happens is that the trader agent is like the the whole of all these agent nodes and here it's a bit different because these agent nodes are effectively like blockchain nodes that's sort of replicating um the the work and also the code um they're you know often quite identical or can even be fully identical instances

of each other and then they work together to effectively become the trader agent and so on chain they're represented as a multi-c and off chain there's this couple of nodes which have like State synchronization between them so very practically what they use is tender um at the moment as a consensus gget so that all the nodes in the system agree um on the actions this uh agent should take next so in the field of llms itself right like and now I'm referring to let's say like the non- crypto part of building on top of L llss which is

probably a thousand times bigger than the the crypto part there's like lots of different Frameworks that are kind of like building agents um using llm so Lang chain is probably the most commercially successful but then you'll go and find like Microsoft autogen which which is a multi-agent system uh in in how it's constructed but there are but there are like loads of others in fact in fact the problem is it's a problem of Plenty rather than a problem of uh problem of scarcity so maybe to start with in terms of like the agent framework You're Building uh what is like really

different about your agent framework from the things that might be happening outside crypto as a as a whole yeah I think one key thing is that we always um want system which are sort of able to take action on chain like any other users um so we see like autonomous agents as this sort of the active users of various protocols and we can talk about this later in a bit what benefits that has for the protocol but that means that um in our case sort of the crypto wallet and also the onchain representation of the offchain

agent are like first class citizens so we think of this from the design and beginning and that has implications for instance when we come back to this Trader if I want to co-own like let's say a longchain agent um well you'd have to basically build what we've built um because you need some way of basically sharing ownership of let's say an onchain wallet like a a safe let's say multi um with these offchain instances of Agents so our our framework lets you do this that's that's one way to look at it so it's just a sort of native um crypto support I guess um the second thing is this if you go further to co-ownership there's sort of

two extremes there again um so if co-ownership can be achieved entirely on chain so for instance you have like a Save which has some assets and now you have a lot of let's say land chain agent or Auto GPT agent or whatever one of those framework agents uh all kind of holding a wallet and then um being a signer on the safe then this could work right because they don't necessarily need offchain consensus depending depending on what the application is but actually once you look into the interesting application turns out that almost always once you go beyond like simple things which are done onchain uh you need offchain consensus because often it's like things like oh like even

an oracle needs to agree offchain potentially on the data it wants to put on chain um certainly efficient oracles want to do that offchain um and then if you imagine this offchain system wanting to act upon something else offchain then for sure you also need offchain consensus so there it then also again helps to have a stack which gives you this out of the box which I always does now a third way to look at it and this is sort of purely on the independent of crypto and more sort of on the structuring of the agent is that um coming back to our discussion earlier automation versus autonomy and like sort of fully um AI based and dynamic uh um

um um uh agent systems rather than those which are maybe like sort of based on hardcoded rules the reality is if you want to build like really use cases which can use PE which people can use today and which are actually meaningfully and securely achieving something then you can't go yet uh in these sort of fully unstructured models where you just basically repeatedly prompt an llm like you can do it but it's it doesn't work I mean you need to provide structure and then if you look at the Frameworks you know L chain is an interesting example and and um you know I have no um no nothing bad to say about it but it's an interesting example

they're moving towards graph structures as well because it's pretty obvious that a chain won't cut it like your decision making is almost never a chain that is the most basic kind of application where it's like a b c d and then going back to a right the reality is you're GNA have even in the most basic application you're G to have the happy path which might look like this but then off the happy path you have all these Arrow paths which need to sort of loop back to different states and so you're basically in a graph structure um and so that's where we started our journey we we basically like 5 years ago said well if you're building uh autonomous agian systems then it's unlikely that we're going to have in the short term these

sort of fully open-ended um sort of just models which we need to Trend there somehow the agentic uh system pops out but um instead we still need to provide some rails and then use um models um alongside those rails and these rails in our case are these basic graphs along which the agent has to travel and now if you put it back all together I think one of the benefits you have our sack is that you can go and say okay I have a a use case where there's some um States in which the agent is very free there's other states where I want the agent to just travel along this track then I can do this and now I also want this agent

to take um action on chain um ever so often then it already comes out of the box um so we obviously from the beginning when we built the framework were really um heavy users of the safe so we had like ethereum support with safe since basically day one since the framework is usable and now as we're sort of expanding it um to other sort of types of um uh uh blockchain ecosystem um we're always kind of having the same design Paradigm again where we pick like a multi um which is dominant in this ecosystem and then build um the compatibility of the stack around it okay I think I'm I'm now less confused

about the agentic part um but I'm still confused about kind of like the protocol as a whole so kind of like if you look at the stack now now we kind of have some understanding of what these agents can and can't do I can't just give an agent I don't know I can't just say here you have a 100 die you make me some money um and basically the the agent will go and kind of like uh either kind of like B you know build an arbitration bot or kind of um make Saucy pictures on Middle on Mid journey and kind of put them on only fans and I mean so basically it's like it's like you have to give it some structure I understand that now but how do you put this all into protocol and kind of where do the

Cod ownership come in because this is something that in principle with an llm model and like some uh deaf background I could just do on my own right I don't need alteras for that yeah and um that's a great question so basically I think for like one of our core insights is that it's not about building individual agents it's about building effectively um many agentic systems which can interact um because ultimately we're like big Believers in the specialization and even like from a very practical point of view if you want to build

better agents people will build very different agents so the framework will have to cat a very different sort of use cases so the protocol was always designed to um enable basically entire Asian economies and enable their bootstrapping so there is a couple of mechanisms which facilitate that the stack itself is open source so when you have an open source stack there's never a forcing function to tells them oh you have to use this protocol um so you have to basically create like a a reason on top where why it would make sense for people to engage with this protocol so one thing which we noticed is if you want to have these basically autonomous

agent use cases really grow then we need obviously a lot of uh development um you know developers who build uh on the stack why do developers have the benefit of building on the stack well there's some of the technical reasons we mentioned before but there's also one of composability so we basically have created um a very composable framework um where it's not so much about composing arbitrary python libraries which is a focus of a lot of the other Frameworks but where it is the focus of this to compose business logic itself um and that's particularly with autonomous agents of the current

generation if you think about it it's very important so if I have like for instance this Trader um at some point it's going to settle a transaction um you might say well that is just a matter of sending the transaction well this is actually not true there's around like 20 or 30 uh states in the finance State machine which takes care of settling the transaction because there is like on the the happy path where is things like you need to you need to sign it from all the agents you need to then submit it you then need to wait for it to be settled um and if anything goes wrong in any of those States the resolution looks different um now you don't want a developer to reimplement that similarly if you think about things like

um uh interacting with these prediction markets that might actually be like something which you might want in another agent being able to kind of compose these things is is is very very interesting and so one big part of the protocol as a result is this focus on creating a developer mechanis developer incentive mechanism whereby developers get rewards for contributing these pieces of agents and entire agents uh into the stack um so that's this code side of things um they can do that permissionless so very practically um you you know you develop the stuff you uh uh registered on chain as these nfts

and then there's a sort of reward system which works sort of on a on an Epoch basis um on the other side of this is um is the question of capital so um obviously the developer rewards come partially from you know emissions but over the longer term they will have to come from productive agent systems which the Dow kind of operates I'll get to that in a moment um but even to get you there um basically you need a bootstrapping mechanism whereby the people can actually use this um Olas token and so that's where bonding comes in so whenever the protocol is deployed on a new chain then

effectively um there's a bonding mechanism in place whereby um people who use or believe in the protocol can provide liquidity um in this uh token and the chains token and then return that LP Tok into the protocol and receive uh effectively Olas and what this does is that basically you have a very decentralized way of bringing that utility toque into more chains why do you want it on more chains because that's the third bit which is staking so once I obviously have like code which does something useful um now I want to be able to actually operate these agents and as we

noted before you can operate them like a decentralized system so it ends up looking a bit like operating uh a blockchain so you effectively then have the staking system where the operators of the nodes in any given agent um can basically earn these uh uh staking rewards um and in order for that to all be smooth it helps when the token is basically accessible on that chain directly so that's sort of the free mechanisms like stake being the last and then the the code Capital sort of um uh pair and then we can dive in there if you want yeah so actually that's a lot of things right so

let's try to recap so right at the layer of building the agent um what you're saying is like okay your framework in a sense like there are many Frameworks which we can can be used to build agents but the the differentiation of the autona framework is that it contains the differenti is is on two Dimensions the First Dimension is it contains components that would make blockchain integration and blockchain transaction uh creation easy so this could it this could involve things like

okay a blockchain an agent if it needs to interact with a blockchain it needs to store a private key so maybe it it needs some components for the securing of private Keys it needs components by which it can read blockchain data it it needs components by which transactions can be sent and it can it can figure out that they were confirmed or not so there are like some standard pieces of uh logic that are used in a lot of different places maybe even exchanges use it in the heart polet or things like that and you're you're going to build like the standard versions of those components and integrate them into your framework so a developer doesn't have to worry about those aspects then the second thing your Frameworks providing

is it is providing some kind of cognitive architecture what by that what what we might mean is that um you want the agent to basically apply its intelligence but you want it to you want to constrain its intelligence in a certain manner which is like you know for for for a particular problem always think always create a tree and reason through the nodes of a tree or in this particular problem um create a line and like there are nodes and reason through all of these nodes so particular problems might have particular ways of thinking that if we constrain the agent to think in that particular way it

will produce better outputs and so you are you providing a way to develop against some of these um of these like constraints right like so a developer can put these constraints into their system and then they could um they could use it those are on the framework itself and then what you're saying is actually like the the network itself so so now we jump from like the framework deals with the problem of how do you build a single agent or how do you build maybe two or three agents and they coordinate with each other but then you jump to the network level where it's the problem of ultimately you want

thousands of agents to be to be built and there the kinds of problems you're trying to solve are how to provide developer incentives for the Improvement of your agent framework itself yeah so this is so if we zoom out a bit what we ultimately want is a machine to well even if you zoom out a bit more so the co ownership of autonomous um agents and gentic AI ends up being I think what if you think about like d like decentralized aor organization is sort of almost like the

end State um of that so if you if you think about this concept of like some organization which we own which is in itself autonomous and which has the highest degree of decentralization we can achieve then ultimately this will be using forms of AI and and be agentic right by by its definition and so the different angle at which to come at this is to say okay how how can you basically coordinate all the actors which need to make that happen right because if it's just on chain then you're always constrained by what you can do on chain right so if you just have a smart contract then there's always someone who has to call that

smart contract for something to happen and by necessity you will always be limited to what's possible onchain which I think will always be less than what's possible offchain and so in a way the the other way to look at it is to say how can you create basically a protocol aonomus which allows the creation of these kind of co- ownable autonomous agents and then this means you need to coordinate a bunch of actors you need to coordinate those who are the developing them that's why you have the def incentive mechanism you need those who uh operate them which is around staking and you need those who basically uh provide this um liquidity for the whole system to exist at any given point which

is the bonders and so that's kind of what the roles of the protocol is is to coordinate all these actors now obviously it's highly uh complex so we should make it a bit um concrete if you think about um what we had earlier um discussed quite a lot the the trading agent use case with the prediction markets um then there's the system called uh the mech inside of it um which is this third type of agent which basically just specializes on making the predictions and these kind of um agents um are basically something which you can imagine running as this decentralized system which the auton sty itself then can own so so you effectively then have

a situation where the autona style can provide on an ongoing basis this kind of offchain system with configurable degree degrees of decentralization which offers these services to other agents in the uh auton ecosystem and then that allows you to sort of um bootstrap this over time if that makes sense can I think of it like this that so today we have a few different chains that are trying to build what I call like puppet accounts or delegated account control uh those are those are two like two interchangeable words but the essential idea behind it

is so so the ne network is trying to build this okay so so ne's idea is that okay there's a blockchain with a sort of with the set of validators and what if this blockchain itself quote own a Bitcoin address not only a Bitcoin address but another ethereum address and so from from the perspective of Bitcoin it's like a normal address with a private key but the private key is actually split into the validator set of near by some really smart cryptographic protocol so Bitcoin thinks this is like a single it's a normal address a single individual but

in reality underneath it is actually the validator set of near that controls that account and in a sense that you can say that okay that near Network by itself is kind of like owning this owning this address on bitcoin and this other address on on ethereum if you start with that point and then and then kind you kind of you layer on the idea that is it possible that okay that there be a way by which a network could own not only an address but an address plus a piece of like running code and that running code is one an an economic agent which that running code

is an autolus this framework agent so it has an address and it has some kind of like structured and unstructured logic so you can actually message it give it tasks and expect responses and so auton is is trying to do that ultimately like how do you have a dow that can own an address plus some kind of code and so and it it it owns both of those components together and then it can also sort of like make money make money through it that that is what you're what you're seeking to

achieve well this this is yeah so we would call this like a protocol and server so it's basically if we go back to this um uh concept which you said earlier if you have an existing validator set so basically you could say okay well let's just do this all uh on chain you know like let's just somehow modify the chain so it can sort of run long running tasks and then you will find very quickly that there's all these arguments as to why that cannot work like you need an application specific chain in order to have long running task because if you have a public chain and it becomes an immediate um basically attack Vector for for denial of service um distribute denial of service because you can just sort of preempt future blocks indefinitely by sheduling task

for future blocks now so effectively whatever near is doing there I don't know in too much detail but there's limits to um kind of putting too much on a block public blockchain which is meant to run repeatedly um or scheduled basically so you need to do it on some sort of application specific layer and now you could say okay well we can just run some sort of Layer Two or layer three or layer n or whatever um and ultimately there it's mostly about having sort of again an architecture where you can basically then inherit some degree of security right and in uh execute some of those instructions and in a way um I

think ultimately you know in in in the future one day an autonomous service will look uh quite similar similar to an specific roll up potentially because it will basically have a lot of degrees of verifiability um and it will potentially even inherit some of the security um as a result from the chains on which it acts but um it will have these more autonomous I long running um tasks here um which is executing which is different from like a public blockchain where I always need to basically at any given time offer these blocks which accept a certain am amount of basically bidding into them and then once they're full

they're full right I can't guarantee you that I'll execute you whereas an autonomous service can do that it can say well I'm application specific uh ever so often I'm doing exactly this thing okay so um I feel like this has become super abstract um maybe let's kind of make some examples right so one of um the main topics that kind of you posit this will be used for um in the short term is um optimization of dows um can you give us some examples how kind of things work in Dows today and how you see them improving by kind of putting

these um autonomous agent systems on top of them or kind of in meshing them yeah so actually there's a no answer to this question even in the sense that originally we thought when we started out with the stack that like Dows are this primary customer for that right they have various offchain processes which are often quite centralized let help them make them more decentralized um and more autonomous um both things which are in the name turns out from a good to Market point of view um it's not particularly great because a lot of dows have actually a lot of things to do and they're maybe not the best organized uh entities always and so it takes a lot of time and you you're

not getting uh to the goal very fast you also need to coordinate a lot of actors by the definition of it um so actually what we noticed is that what's um whilst we still believe in this and I'll talk about an example is that it's better to focus on problems we see in our own Dow and make them as autonomous and decentralized and or just build basically users for other decentralized protocols so what I mentioned earlier the use case these autonomous agents are basically users of omen users of gnosis users of safe you know they have done um around 70% of all safe transaction on glosses since summer um like on a on a

on a weekly basis basically they've done hundreds of thousands of transactions which basically benefit these protocols on which they're deployed and obviously themselves as well because they're they're profitable um now an example which I like um because it's very easy to understand um which can apply to many Dows and which they can adopt quite easily is governator it was a bit of a joke project which is sort of slowly maturing basically it's built on the uh autonomous service stack which which allas offers autonomous um what it does is it's it basically replaces a human delegate in a DA so if I obviously have

tokens and I don't always want to vote I could delegate them to someone I think will vote more or vote in my favor like um with my intent and so on and uh we implemented that in code so basically there's an autonomous service which continuously watches those STS for which it holds delegated tokens and then when it sees those proposals either on Snapshot or onchain it can then vote um in those proposals and obviously in order to do that it needs to use um a large language model to actually read the proposal and reason about it um it also needs that in order to make sense of the preferences it is given and sort of bring those two things together to arrive at a voting decision but the

actual voting coming back to the structured versus unstructured is a very structured process there's zero point in having the Asian figure this out every time because it will probably fail most of the time instead you just have that part hardcoded right so basically this is a nice example of what we were discussing about earlier in very abstract way you have these sort of structured bits which are defined very well and then you have these unstructured parts of of the logic where uh you're looking at this proposals making sense of it and so on are there new attack factors that are introduced here so basically if if I kind of um if I kind of trust um an autonomous agent to kind of make voting decisions for me

I kind of I Rely heavily on the fact that um this autonomous agent actually um will act in the way that I would act if I where to look into it right so H how do you make sure that the agents actually do what they are meant to do on the face of it it's a great question so the there's two parts to this well many but I would split it into um one is like the preferences so that's where the governator St fall short it doesn't actually allow you to express very um Rich preferences at all and that's just a matter of our time and

uh effort which has gone into this part of the application but one side where it exceeds on is the um basically certainty that it implements the decisions the decision logic which it is meant to implement so if you think about a human if you delegate to them you basically have no clue right it's all reputation based if if you were to think to delegate to a single L chain agent or to GPT agent well it really depends on the developer who is running that are they even running it if they're running it are they're running the code they told you they running right all this kind of stuff whereas with an autonomous service which has multiple nodes operated by different operators you then start getting into a similar basically threet

model which you have with like a you know like your Cosmos chain basically or any other uh sort of um bantine fault system whereby you have to reason about okay how many like operators are there how decentralized is it and then um is the majority of them honest if the majority of them is honest um then you have uh very high security guarantees because you effectively what happens is that each one of them has to agree or the majority of them has to agree and each one of them uses these models so you're not even relying on a single model instance which is another issue with l language models they're not necessarily deterministic at all they sometimes can be configured to be but

like um some of them can't even uh be configured to be deterministic um so then having multiple agents each come to independent valuations and then sort of pull that decision making and then agree is actually like um a massive Improvement so on that Dimension I would say governate is already better than a human because a human could you know do whatever and here you have like a node system implementing that decision logic yeah so I think kind of what we often try to do in these episodes is kind of we try to understand how exactly things work and I think this was more an episode about kind of talking about why it would make sense to have something like this so kind of I kind of I know I

want to change gears a little bit here um and kind of um ask about concerns you may have about this so kind of like um if you look at AIS the way that they have improved in the last couple of years at least kind of like in in the popular um mind I know that kind of it's been a long time coming and so on but it's really impressive right it seems absolutely certain that they will kind of surpass uh human Ingenuity and you know capacity on all kinds of axes in you know the very short term and if you talk to AI safety people um it kind of often they will tell you um they're not

so concerned because you can always switch it off and now kind of pairing it with the technology that by definition no one can turn off does that worry you yeah I think it's a a good topic to discuss and one we will will not be um obviously s about um I think the the first thing which I strongly believe in is that it's very very very unlikely that there will be sort of one model which kind of runs away and like takes over and that's just um even in like very favorable cases to the sort of um super intelligence arising and being able to consume a lot

of resources um there's like geographical physical sort of constraints which make it unlikely um I think what's much more likely is that it that we'll have a situation where certainly a lot of centralized players will own very very powerful uh models and so I think actually what we should be most concerned about is um the economic impact of this kind of change in technology on people rather than these hypotheticals where some software slays us all I think it you know it's not it's important to kind of keep it in the back

of our minds but and and like with every technology be mindful as to when these dangers become more apparent that we kind of think about them but like the the much much bigger concern I think is you know economic on the economics if you if you hit hit Listen to Someone Like Sam Alman it's this naivity of the economics which really rils me up like they all go around and say you know I mean they you know by all means like they're great like you know entrepreneurs and great great products and so on but everyone has their we points I think here it's like this kind of naivity around just because I create better technology everyone will be

better well that never worked out that way the reality is that it's always a distribution question and if the distribution sacks of access to these kind of models and um people's ability to use them for their lives and improving their own situation then it doesn't matter how good the best model is then there will still be even bigger disparities in sort of income health wealth uh around the globe and I think that's what we should all be really worried about and that's kind of the mission of our entire business and the mission of auton is about creating these kind of systems which can be co-owned so that there can be groups can share these uh systems that doesn't mean that all

problems are solved because now you know these groups could again be better off than others and you still have these kind of distributional issues but at least it's it's a start so I'm worried about the economic impact of this much much much much more than these kind of hypotheticals and which I think are interesting for dinner conversations but really don't kind of miss the point mostly um having said that I think you know let's say we fast forward there's like multiple generations of advances and like um even like models which are basically agent in the model sort of you know like some call them large action models now I saw and others call them differently then you know uh open AI has their um reinforcement learning merged

with um a large language models there's different attempts whatever it will be in the ultimate State um if you imagine that to run in a sort of blockchain like way um where it sort of has a bad intent and we can't turn it off yes I think it's something we should keep in the back of our mind that and and think um about Solutions but I think the flip side of this is um again that if this model is used for good then having transparency and um uh kind of censorship resistance can bring many uh Goods as well so I think let's take it one step at a time basically and focus on the problems

which we for sure know will happen which I think are distributional I feel like we've touched on many many things if people want to learn more or kind of build um their own Ag gentic Systems um for a tonas or kind of just use um systems that are already there um where where should we send them yeah so we have this thing called The Academy um that's a great start um so that's for people who want to basically have more like support as they're building we have the docks all of that can be found on the website so all us. Network and then if you uh follow Aon us on Twitter as well there's

like weekly updates um where I think those two places are are the best perfect um I am so curious to see how this is uh going to evolve I think uh we should cancel in kind of a followup uh soonish um to just to see kind of like what people build and kind of um H how it actually changes things because the opportunity space here is absolutely enormous yeah let's do that yeah it's been a pleasure to have you on thank you very much was a pleasure being on