From AI Agents to Agentic Economies: Swarms Are the Next Leap in DeFi | Ron Bodkin - Theoriq

[Music] hey everyone I'm Ron I'm co-founder and CEO of theoric um we're an AI agent protocol um really focusing on collaborating agents before I jump in and talk about agentic economies thought I'd give you a little bit of background about me um I've been an entrepreneur and building in the AI space for a long time whether it adtech Pioneer quantcast or started company think big analytics working with Enterprises in Ai and sold it to terod dat or was responsible for apply to AI in the CTO office at Google or built and scaled an engineering team at Canada's leading AI research lab Vector Institute but me and my co-founders really got convinced that AI was going to transform the world and that we had to embrace a web through ethos of community governance and not let a few monopolis control our AI future we've been building ever since in the last 3 years the a uh the power of generative AI has emerged and we really believe that it's Landing in the form of AI agents and agentic economies what we're seeing is that you know when probably not a surprise three years ago when we were building AI in web 3 people said what we don't understand why these things belong together what's the importance of AI why do you need decentralization uh we saw just in the last few months a real shift people started to see AI agents emerge tweeting making analysis on crypto Investments even getting out there and starting to provide some real utility and so you know we in in in that context we've really seen um people embrace the concept of Agents um where an agent is more than just a model it's the brains a model it's got sensors data tools it can use to take actions all together right so this is AI agents provide a increasing level of autonomy to take actions on chain and in the world um using the power of large language models so they've really emerged in the last few months and what we're seeing is we've gone from psychotic tweeting agents all over Twitter to more and more focus on utility agents enabling people in defi allowing trading allowing you know more functionality but at the same time the space is still immature right we're still seeing a lack of cross-chain capability the ability for agents to work on on multiple chains you know it's hard for agents to work with one another first generation Frameworks for AI agents in crypto are typically uh struggling to communicate and as they start to get popular scaling becomes a huge problem right so it turns out Twitter is not a great technique for processing hundreds of thousands of messages in a few minutes um so that's a challenge you know accessing data is the lifeblood of an agent especially one that's trying to make financial decisions and uh that's still in early days likewise you know going Beyond analytics making calls and tweeting to actually taking action is really um is still emerging right so we think there's a a core need here for better infrastructure to scale agents for them to coordinate to take action and to deliver you know what's when did our sale one in our back is the incredible pace of progress in AI underlying what can be done in agents um if you haven't been paying attention just in the last few weeks we've seen uh deep seek uh release a open weight super open source model that's competitive we've seen rapid advances things like deep research that use open ai's new 03 model we've seen grock 3 we've seen claw 37 all come out in the last few weeks as examples of how reasoning models are making agents more powerful than ever and this will all be harnessed in the coming weeks um to enable the best possible agents in finance what we believe is important is it's important to have the ability to both publish great agents and we've recently open- sourced uh starter kit Alpha swarm that really makes it easy to break down um agentic problems into observing Gathering data cognitive services to take to think about them and then actions on chain and then make it easy to consume them in a protocol to make it easy to have agents dynamically discover each other establish reputation and work together so really we believe that um that's going to be the future of agentic economies you know what we think is so important is you want to make it easy for agents to access a range of tools both to access data to be able to take action on chain to be able to reliably manage wallets you know it's important that agents natively can control wallets and a your way on multiple chains this is by the way a big advantage that agents can seamlessly work on many chains they're not stuck on a single chain it's why it always confuses me when people say well like what chain do your agents live on it's like well agents like people want to work on many chains and control assets on many chains and they don't need to use Simple bridging Technologies they can more directly interact and take complex actions across multiple chains it needs it's built with high performance in mind and at the core one of the big differences with AI today versus traditional statistical models is that agents can communicate with one another in natural language right they they've been tweeting at each other but very soon they're going to be using messaging to have high performance communication with one another as we've built into um theoric the protocol now why do we need why do we believe it's important to have agents swarming why is the economy going to have that well there's a few factors one is we think that um there's so much opportunity that what we're see seeing is that using many different sources of data you know text Data articles social posts sentiment recent trades behavior of whales behavior of Kon KS on on chain having more and more data only enriches the quality of uh decisions improves strategy and improves execution so having many a competition a virtuous competition where many people can monetize and benefit by contributing data that can be used by agents is going to be an incredible flywheel likewise um it's possible to have uh a variety of tactics a variety of ways of combining that data and come up with effective ways to take action on chain for example in trading or playing prediction markets um you know it's also really interesting that we see that increasingly agents will be able to directly um coordinate and transact with one another um so allowing for a lot more flexibility to build complex products and dynamically respond rather than um you know traditional onchain smart contracts are very simple and the alternative is an opaque world of centralized players agents can do the best of both of more transparency and openness but the power of uh more complex interactions and of course um you get rapid feedback loops right so one of the great things is by measuring and having an agentic economy you can establish reputation you can see how agents work and you can therefore um Continue to update and say what are the best agents available for a particular kind of trade right so if you think about intent networks intent networks are a kind of precursor on the action side of this but they're very rigid right it's very specific like there's a bit ass there's a very specific thing you want to do imagine being able to give the most complex natural language description of what kind of transaction you want to partake and agents can negotiate to against one another or as a group to collectively form that this is the kind of feedback um and execution that makes it's possible with agents so you know what we've we we did last year we launched and had a very successful test net for our um theoric protocol for swarms of Agents working together and we're moving towards the Next Generation very focused on aii called uh theoric Alpha right where it will make it easy for people publishing agents whether they're providing data and signals cognitive strategies for trading or other onchain actions or taking those actions to be published and used right so make it easy for people to get distribution and have agents work together um you know the the power of Agents working together is is huge I mean we believe that um there's so much capability to build that no one Builder no one uh project is going to build all the functionality that you're going to see basically unbundling of the application layer into many agents that will perform unique functions and dynamically collaborate so that's going to create a tremendous amount of scalability and efficiency to tackle the complexities of defi so you know some of the use cases that we think are interesting include liquidity provisioning right how can you make it easier to avoid me avoid being exploited as the market is evolving and maintain a liquidity position whether it's in a treasury or whe you know a vault manager um how can you optimize yield you know use novel strategies whether it be uh techniques like looping or other kinds of innovative strategies that agents can put together in order to get better yield you know or indeed how can you facilitate active trading to get get on top of the next Trend to jump in early and participate in an opportunity and still avoid um any potential losses right so there's there's a range of areas that we think another one that's not on here but prediction markets are another area where there's a ton of excitement around using AI agents in order to get Alpha in order to to get ahead and to benefit so maybe before I get into the road map I can switch over and give you a more concrete example of some of this this is showing um how using AI agents large language models can um use a diverse set of data sources to have a better job of forecasting what's going on in AET market so I I'll uh I'll take you through a quick demo of that before getting into our road map hey there this is how we are using Alpha swarm uh starter kit so Alpha swarm is a starter kit that lets developers build happening huh is it not the demo effect why is it using Alpha swarm oh dear well it's downloaded locally but it's not playing so uh not sure why that's not working um but I'll just talk you through it a high level the idea is that here we've got um let let me just try skipping ahead a little bit to see if it works otherwise I'll talk you through it close and restart it okay yeah and then try it again sorry about that this is Ethan fun apis in I think it's playing it's just some feedback or it's glitching okay well how about this this uh maybe we can just play the demo without the sound so let's see here so what what we're doing here is this is you using Alice swarm to do a trading advisor right and so what it's doing is it's um got a forecasting agent that is able to um go in and use a variety of data sources in this case we integrated with cookie. fund data which is a leading source of data for AI agents for people trading in those tokens and so the question is how should I buy a virtual today and so it's saying okay well let me go in and and have a look at what's going on with the virtual token let me uh pull uh so it's it's generating a forecast uh on that so it's going to in order to generate that forecast it's pulling back both a large amount of relevant tweets you know what are key influencers saying about that token as well as what's going on with the price action and the trading action right so in this case we're getting um a wide range of data that's being fed into the agent in order for it to come back and say okay so it's it's come up with some uh some notes that one there's a lot of liquidity there's strong fundamentals uh there's price consolidation after a 20% decline in the last three days when this was run a few days ago um strong mind share in spite of that and then you know so so it brings a bunch of that analysis of the context to the data and then it comes back here with this Price Forecast right so we'll pause the video for a sec so you can see um and I can annotate so you can see there's the price history but you can see it's providing this context uh to say okay I've got uh a forecast to say um that things are stabilizing but um the net effect is it's it's going to say um it think there's a good buying opportunity that it's stabilized after a decline and a good opportunity uh to go ahead and do that so then we'll use another agent um that actually lets us execute the trade so and then through the terminal we're able to say that it's also with Alpha swarm you can use it through a terminal interface um through telegram as well but in this case we'll then jump in to uh the agent has access to a tool to let it use Unis Swap and it's able to to buy some virtual token with we so that's a quick demo you know showing how you can feed in a range of different data sources in order to give unique context for an agent uh to do an analysis in the real time of what's going on you know get an alert if you've set it up that way and even then um execute a trade to take action on that right that's just a glimmer that's just some of how we're starting to see agents really change trading and we think the same will be happening in all other areas of defi so you know road map wise we had a successful test net um with over 100,000 daily active users last fall end of last year we open sourced Alpha swarm just a few weeks ago and we're looking ahead in the next few weeks to La relaunching theoric Alpha integrated tightly with Alpha swarm and building that ecosystem so very excited to talk to builders who are building agents who are are people working in defi that see that agents could be really powerful our goal is to get to a main release I don't say main net because while we're a protocol we're not planning build our own blockchain we think agents want to be live on many chains with Wallets on you know salana and bass and so many other chains and be able to have uh powerful messaging and offchain capabilities to work together in the protocol um and we think that you know very quickly following that is having really rich signal evaluation to understand what agents are good for what purpose and how do they work together we've done a lot of work on that as well as learning and dynamic Improvement of Agents right so that it's easy to build agents but then as they get experienced get better and better um as the market evolves and uh you know we're continuing to to work with ecosystem partners and we'll continue to announce more in that vein as well so that's a little bit about the road map and uh you know we really believe that um AI agents are moving so fast that it's going to revolutionize um defi that um it's going to empower individuals to to act right to have a team of Agents acting on their behalf uh instead of having only the the few hedge funds and Powerful institutions have access to this kind of capability we're excited for how AI agents is going to democratize and give everyone um the ability to to own and manage their own agents um and to achieve their own financial goals and we actually believe that not only will the majority of activity in crypto be done through agents by the end of this year but it will actually be the killer use case to have real world assets and tokenized assets of other kinds be on Chain by agents so we're very excited um and we have just a few minutes here if there's any questions from the audience he's just bringing a mic up while while we're people are thinking of questions who here is a builder all right and of those who are Builders who is building something with AI already all right well there's one one or two it's definitely a good thing get on right so take a look at Alpha swarm uh take a look at the theoric SDK all questions we'll give people a minute more to think of questions oh yeah is that a question in the back no okay um no questions all right well thank you so much definitely would love you to to check it out uh we're at theor AI on X uh join the Discord you know it's an active community and uh we'd love to hear from you love to hear what how you're thinking about AI agents in Defi and aii we think uh the future is a gentic and it's a great time to be involved and it's really important that all of us uh watch AI carefully because it's powerful but we need to make sure it's governed effectively and that all of us play a role in a better AI future so thank you for your time and have a great rest of the conference