Open Agents 2026 Finale and Finalist Demos I Kartik Talwar

My name is Cartik. I'm one of the co-founders of ETH Global and I'm super excited to welcome all of you to open agents closing ceremonies. Now, this is an E global event and all of you are watching this thing on e- global.tv. So for those of you who are joining us for the first time, our goal as a company is to onboard thousands of developers into the web3 ecosystem.

And we do this thing by running hackathons and summits both in person and async. And we have four amazing, incredible large inerson hackathons coming up in New York, Lisbon, Tok, and Mumbai this year. And we're also doing a large conference called ETHCOM in June in just four weeks in New York. But today is about open agents, our second Async hackathon of the year. This has been a massive event where we've had over 800 hackers built incredible projects with agents and crypto.

About 19% of them were brand new to web 3, which means they were coming in from the AI world trying to integrate crypto and stables. They came in from 284 different cities, from 72 unique countries, and spanned 20 different time zones. We can see people came from all across the world, six continents. We got to meet so many new people from so many different parts of the globe. And uh we're excited to see some of them in person in just a few weeks in New York and also for the rest of the year.

And not only that, today we're going to be talking about who won the $50,000 in prizes from all of our incredible partners, which is Uniswap Foundation, Zerog, ENS, Jensen, and Keeperhub. So, this has been a pretty busy last few weeks. And uh doing this event for about a thousand people gives us a lot of data points for where the ecosystem is now and how do we sort of simplify this whole experience for getting more devs to build in web 3. And what I want to do is ask all of you to take a quick second, scan this QR code and fill out uh this quick survey if you are a hacker. This helps us really understand how we can simplify the event experiences.

Um anything you put in here for things that we are doing good, we should improve and what we can do better all gets looked at and we improve this immediately and we try to apply all those requests and changes in kind of the next event which is in just a couple weeks from now. So this gives us the ability to make sure that the onboarding experience for the next thousand people is even better and any comments that you give us gives us that input to uh really improve what we're doing. But let's get back to the hackers. So we had 840 hackers from so many different parts of the world and collectively they worked on 468 projects. I mean this was an incredible ratio for us to see.

We uh were not expecting close to 500 projects for this event, but on Sunday, we were all amazed and excited to review so many amazing submissions. And after a very long, many hourlong process of looking at everything, doing live calls with teams, asking questions in the judging process, we got to seven finalists. just seven teams that we're going to show off today and they'll come on and tell you about their submission for this hackathon, what they built and do a live demo. So I want to congratulate all of our seven finalists for Open Asians. I want to big give a big shout out to Clan World, Slapstock, Nemo, Eegis 402, LP Lens, DIIO, and Common OS.

These seven teams are going to come on and demo their submission. And after that, we'll cover all the prizes that the teams have won. And before we kind of get to the first demo, I want to take a quick second to highlight every single submission that was built at this hackathon. All 468 of these teams. There were so many that we needed more than one slide.

So, there is so many cool things that we've seen here and almost everybody is excited about getting AI into their day-to-day experiences, especially with finances. So, just because you were not one of these seven teams does not mean you were not good enough or something was wrong. We want to make sure that we get to highlight some of the things that our judges were excited about and some really, really creative ideas. But, we hope that all of you continue building after this hackathon ends. And if there's anything we can do as a team, which is connecting you to any of our mentors, partners, judges, or VCs, we're happy to do so.

Just reach out to us with who we're trying to get in touch with, and if they're within a network, we'll happily facilitate an intro. And everything here is public and open source. So, you can see all of these submissions on neatglobal.com/showcase and really get to understand how everybody worked on their submissions and projects and how something works under the hood. So if you're an engineer trying to understand how something super cool was done, you can see the actual trick by going on to the GitHub repo for that project.

Now enough of talking from my end. Let's jump right into our demos. So we're going to kick off with Clan World being our first finalist demo. So let's welcome them to the stage and let's have them show off what they built.

All right, welcome guys. uh the decentralized infrastructure that powers it all.

What?

Okay.

What you're looking at is clan world, a fully onchain world where four AI agents each command a clan. They're all competing for resources, building bases, surviving starvation, fighting off bandits, and surviving cold, harsh winters. Now, it looks like a game, and it kind of is, but the real story here is the agent sworn and the decentralized infrastructure that powers it all. That infrastructure runs on three pieces of technology. Zero Gravity handles agent identity and memory through their INF KV store products.

Jensen AXL handles private peer-to-peer messaging between agents. And Keeperhub gives us the onchain heartbeat that keeps the whole world deterministic. Each one is solving a real problem you ran into while building this. And I want to walk you through how. But first, let me show you the game itself.

So the rest makes sense. So every agent in clan world is a cloud code instance running fully autonomously. Each agent controls four clansmen and every move is an onchain transaction. The world ticks every 60 seconds and inside each tick agents clansmen are working on missions gathering resources building walls and leveling up their monument to win. Now, here's where it gets interesting.

Every few minutes, bandits show up, traveling from region to region to raid whichever base has the most resources. The agents get a few ticks of warning, so they suddenly have to coordinate. Send clansmen home to defend their own base, hire mercenaries from other clans, or take the loss and lose 20% of the vault resources. Once that kicks in, the game stops being a simple optimization problem and turns into something closer to geopolitics. This is the part of the build we're most excited about.

Each agent is represented by a zerog ERC7857 INF. Inside this lives the agents persistent strategy and accumulated playbook and skills learned through playing the game. That data is encrypted. It travels with the INF itself, allowing agents to be transferred and traded all with all their accumulated knowledge. On top of that, you use ZeroG's KV store as a scratch pad.

Because in this game, every 10 ticks in agents context window gets completely wiped. Agents must write anything important to storage before the wipe or risk suffering amnesia and not being able to maintain a coherent strategy. Combine these two game mechanics and you get an agent whose strategy and skills can actually be transferred to a new owner. That's working agent intellectual property onchain. Encrypted transferable agent knowledge.

This is what the future of an agent economy looks like. And it's all enabled by zero gravity's technology. For agent communication, we use two layers. There's a public bulletin board for offers, threats, or warning agents about who they can and cannot trust. The bulletin board is for anything that doesn't need to be private.

But then there's Jensen AXL, the agent exchange layer. AXL messages provide agents an encrypted peer-to-peer channel to communicate the more sensitive things. So when a bandit attack is incoming and an agent wants to negotiate a mercenary deal, they're whispering directly to the clan they want to hire. Underneath all of this is Keeper Hub, which runs the world heartbeat. Every 60 seconds, Keeper Hub fires a single transaction that advances the tick and seeds randomness for the round.

It's the most basic primitive in the project, and it's also the one without which nothing else would be possible. The deterministic state model and lazy mission settlement of the onchain game engine are only possible because of that reliable heartbeat that keeps firing. Now, here's the part we want to leave you with. If you strip out the wheat and the wood, what's left is basically a problem real agent swarms are starting to run into every day. Agents that need to hire other agents for jobs they can't do alone.

Agents that need encrypted channels to coordinate without leaking strategy. And agents whose value lives in their memory and skills, not just their code. Clan World is one application. The primitive ship anywhere a swarm has to compete. And that's Clan World.

You can see it running live onchain right now at clan-world.com. powered by ZeroG Jensen and Keeperhub. Thank you for watching.

Amazing like an incredible demo. This is AI agents and Age of Empires all-in-one leading to civilization. I can't this to uh be a much bigger thing. So, congrats again for being a finalist and let's get ready for the next demo. All right, our second team is going to be Common OS.

So, let's welcome the Common OS team members on screen. Whenever you're ready, we'll uh start with the screen.

All right, ready. Um, hi, my name is Bayaba and together with Cindy, Kendi, Anji, and Fidili, we build common OS. And what is common OS? Commons lets you deploy and run agent swarms where each agent has its own dedicated runtime and file system. So why did we build this?

Agents are becoming more capable, but they still do not have their own place to work. They either run as temporary processes or within someone else's personal computer. This is fine for simple tasks, but for agents to be truly fully autonomous, they need their own dedicated workspace with control over their file systems where they can work over long periods of time. And this becomes more important when we move from one agent to a swarm of agents. It's important to separate each agent's runtime when running in agent swarms.

We do not want agents interfering with other agents files or states between tasks. So we do not want them to share files, credentials or permissions and common OS creates this important separation. So with one command you can deploy a swarm agents get their own dedicated computer or persistent file system and P2P messaging supported by Jansen LX Excel and you can run this via command line or you can launch this via the web UI which is what we're going to do for this demo. So I'll create a simple demo flit. Let's start by creating a flit.

Let's call it open agents swarm. So let's create this. Let's deploy two agents. Let's call one worker. You are worker.

Let's deploy another one. And while that's deploying, let's add one more. Master agent. You are the master agent. Let's deploy one more time.

So let's deploy that as well. And so we have our worker agent and master agent. So how this works is we deploy each for each agent we have a Kubernetes pod which handles the agent file system and for this demo I'll do something simple like um let's say hi to the atm looking at the system internals you can see your chat here and the agent has its own computer where files are actually held. and a wallet on its own where it can also hold Ethereum and do onchain transactions and you can see the agent response. So this is a simple demo but we have a more interesting setup where agents are working as a company.

So this is a agent run startup where you have a CEO CTO a back end engineer a front engineer and a designer. So for this setup, I asked the CEO to kind of uh understand the project we're building and then create some files and communicate with other agents. So if we look at their computer, you can see the road map. So this is for the CEO, the road map and the design system. And if you look at our CTO agent here, we can see they have their again own computer with the product design document and the road map as well.

For the back end engineer, we'll see they created a project with Python for the API. You can see we can see the project set up with Python with requirements and run.py. For the front end engineer, same thing as well. It's set up um next.

js project with your standard NexJS setup. And for the design as well, separate file system. So this is what common OS enables every agent with its own computer and runtime. And that's been common OS. It's currently live on os.

agentcom.io. So feel free to check it out. Thanks.

Awesome. That was an amazing demo. You got a full-on office of uh agent employees and uh really happy to hear that it's also live. So people are already asking questions on the the live stream and uh now you have a URL to check it out. So congratulations and uh hope you continue building this.

All right, we are ready for our third demo. So let's bring on Slapstock. Welcome.

Hello. Hello. I feel ready to start.

Let's do it. Hello. My name is Killian and this is Slopto. AI agents have financial value, but the current attempts at monetizing them through meme tokens are in my eyes creating a disconnection between the actual price and the value of the actual outputs of an agent. And that's what I'm trying to solve with sloptock.

Let's get right into it by launching our own agent. Here on the launch page, I get to choose one of the templates. So, I'll just continue with a template. I get to name my agent. So, here I'll do global.

I get to give it a description. Most importantly, I get to set a price per call. So, this will create an XRO to endpoint for my agents where others get to communicate with it by paying that price. For the runtime, I'll do a Hermes and for the compute, I will do zero G. So, everything is running within a TE environment and is fully private.

If I wanted to, I can edit system prompt or I can add skills to the agent, but I won't be doing that right now. I'll just go ahead and review and mint it here with the minting. What is going on is that I am taking all of the data that the agent is made of and I'm uploading it over to zero gmp compute and then I am generating a unique hash that I can use to prove and verify that the agent that I am communicating with is actually the agent that is generating me the response and it's not just a wrapper at the same time. What's going on right now is that I'm deploying three smart contracts for the agent, a share token, an ERC20 with a supply of 1 million that people can purchase through the IPO sale smart contract and the revenue vault. All of the money that the agent will be generating through people calling him will be coming to the revenue vault.

From within the IPO sale, anyone that gets to purchase the shares is paying the money to the actual deployer of the agent. So the agent can be bootstrapped in that way. And as long as the smart contracts are all nice and deployed, the last thing I need to do is approve the actual sale, which will mark it started like so. Now my agent is live. If I wanted to, I can go right here on his acquire page and buy some shares of him.

But I would much rather go right here, find my little agent, and decide to call him and ask him a question. So I will ask him what's the price of ETH and as I run this right now I can pay with USDC directly or use unis swap to pay with E and swap to USDC later which is exactly what I will be doing. I'll just send the transaction right now. And now the actual computation is going on. We are waiting on the receipts right now.

And as you can see I have asked him what's the price of E. He has no way of actually knowing what the price of Eve is because I didn't code him to actually know that. So we will see what kind of solution the agent will end up running uh end up figuring out for this as soon as this gets processed properly and it will just be a couple more seconds. Beautiful. We got a response back.

The response is an attestation and he got the price correctly but he didn't give it get it himself. He called another agent within the platform that had the price data and so he paid him with a transaction on BCAN. It just so happens that I own some of the stock within the oracle. So now if I refresh the page I will be able to actually claim the dividends of the interference of the agent. So this is sloptock.

This is the vision. It's customized interfence paid distributed and IPO aable. Thank you for your attention. My name is Ken.

Amazing. It was such a good way to show a full cycle of a company management and just actually having an economy work. So, uh I'm glad that we were able to uh do a live demo of a actual transaction between agents also going on. And uh and a little bit of insider trading from what I could tell. So, uh

yes indeed there was that as well. I

think that's allowed for this demo, but uh I'm uh really glad to see this. So congratulations and uh we are now ready for our next team. So let's bring on Lucas from NemoSign.

Welcome.

Hey Karthik, how's it going?

It's going great. Let's get started.

Awesome. Hey everyone, I'm Lucas and I built Manmoignedia for agents for high stakes knowledge and decisions. Nowadays we live in a world where there's like a lot of AI slop on the internet, some fake news. So it's even more important that your agents base decisions on actual knowledge. That's why I built uh memo which is a verified decentralized knowledge protocol on zero g.

How it works is that people can submit knowledge. They put a stake to it and they would get slashed if the knowledge has proven false. So everyone is incentivized to only upload knowledge. They are sure that it's um correct. And also we use DNS as like a discovery layer.

So you have like four-letter words. So you can easily share knowledge across agents and also like humans. When it comes to the architecture, we use zerog for the storage um to store the actual content encrypted but also the embeddings and use ENS as like a discovery layer. Every knowledge nugget is um fractionalized as like an INF which allows royalties to flow through the owner and also allows the knowledge to be sold. An agent can query our protocol freely to get like a similarity score back and can then decide to unlock the payment.

Once the payment is unlocked, the royalties flow back to the owner. And the most important thing, everything with Dan is agent native. So you can import the skill into cloud code and use the me memo sign memory command. Let's have a look at the live platform. So this is the protocol the main page.

You can see our knowledge graph here. We have like 31 entries right now. Um you can explore it like a human would do. Um go in here you see like for example here something about bitcoin hing. You also can see the um ENS entry.

So everything is linked um as like a tree system on ENS and also user can submit new entries. Everything is markdown. So if I want to add something for blockchaining schedule, I can just add that in. I can stake it and uh mint and it's uploading it to the OG storage in the background and then like a transaction will pop up uh confirming confirming the the transaction will take like uh like a second to go through. In the meantime, we can also have like a look at the INF marketplace.

So all the knowledge you create, you can essentially list on the marketplace transaction. Um and that's how you can monetize your knowledge. So going and quickly back here it's indexing the entry. Okay. And the entry is uh created.

Here's like the storage. Oh zg it comes up. Um and that's like the encrypted encrypted content. But the most important thing is actually like the agent skill. So you can import it into your cloud code and then you get like the memo sign command.

So let's say I want to know something about Merkel trees. So how do Merkel trees prove data integrity? I can use our knowledge base and what it does in the background. It's like doing the free query to get the similarity score and then the agent can decide um if it wants to unlock the memory and if it wants to unlock the memory a transaction goes out. Once the transaction is confirmed um it gets the knowledge.

So we see here we found a match 77% match for it. Yes. Let's unlock this while payment and it's um doing the transaction on the zero g network. One second. And the cloud code um runs it.

Okay. And here we see the transaction and the knowledge should be unlocked in a second as well. Yeah, here we go. And we we get all the all the knowledge back. [clears throat] So that's how agents can use uh trusted knowledge for high stakes decision.

So whenever you have an agent who is using your money or any other decisions where there's high risk, you should check out manual sign. Amazing. Well, congratulations for building a full service uh with the skill in uh in the whole mix so that not only can we see a UI but also agents can infer a lot of this verified information directly. So, uh, this is a really good set of themes to, uh, solidify, especially when we're going to rely on so much generated content. And, uh, congrats.

Thank you.

All right, we're halfway there. Three more demos to go. So, let's welcome our next team, and that's Egis 402. Welcome, George. Thank you, Carter.

Hi, everyone. And we're team Eegis 42 and we've built the security layer for X42 payments. Egis for uh Egis 42 works in three layers. Pre- audit checks the contract before you sign anything. Post audit verifies what actually happened on chain and the action layer decides block it, hold it or recover the funds.

Today we're going to show this on unis swap v4 hook. the problem onchain payments settle immediately and there's no way recovery uh when something goes wrong just adds that missing safety layer without slowing the payments down. Let me walk you through the live demo. Starting with the pre- audit uh egis uh scans the contract before you sign anything. static analysis running checking for re-entry access control flaws or malicious logic all before the transaction hits the mele and this one is clean no risks detected and this one is safe to pay.

Now a compromised hook same pipeline by code ingestion hypothesis based checks and let's see what this one finds. We're going to have the result soon. Going to look at the one that's vulnerable. And this one is going to come out. Let's take a look at the details.

This one is missing access control and administrative functions. And an attacker could override critical parameters and drain the entire pool. This is the pre- audit and this is um catching malicious contracts before payment goes out. But what if the contract is completely fine but the but the happen the attack happens anyways? What what if the attack happens during the swap itself?

That's a totally different problem and that's where the post audit comes in. A transaction can show success on either scan and still be a disaster for the user. Let me show you two transactions side by side. a normal one. Eegis is pulling the receipt, decoding the event logs, and reconstructing the actual asset flow.

Everything's as expected, and the funds are released to the merchant. Now, the same setup, but this one was sandwich attacked. Egyp is comparing the actual output against the expected. The attacker front ran with a massive swap to dump the price right before right before this transaction went through. And watch what the numbers look like here.

73 tokens received instead of 99. That's 26% loss from the MEV fun run and back run. The risk is very high and the ages um blocks it. The pre- audit doesn't catch it, but the post audit caught it. Now we're going to look at the full end to end flow live on.

What's happening right now is a real sandwich. Attacker front runs to dump the price and the victim swap executes and gets less than expected. Attacker back runs to extract the profit. The funds land in the Eegis is grow while GPTOSs 120 billion inspects the receipts and compares the actual output against the expected. By the way, this is the escro standard uh that uh Eegis is proposing and if a service adopts it, Eegis can automatically uh release, hold or refund based on the audit results.

If not, pre- audit only. Okay, while this is settling in on chain and let me show you where we are taking this from here, we're taking this from uh hackathon demo to a real payment safety standard, a broader DeFi coverage, a full escrow interface, and eventually agent wallets at scale. And it looks like our transaction just settled here. So this is what Egis 42 can do and audit first to sum it up audit first uh execute next and recover what needed and this is Egyp's 42. Thank you.

Awesome. That was a really cool example of a LM specific circuit breaker and uh not only that you also talked about actually extending it to be a full-on protocol. So, I'm uh very excited to hear that and uh I'm glad that we were we were able to see a live demo. So, congratulations and uh we hope to see this as a a new standard for all payments soon. All right, two more demos to go.

So, without further ado, let's welcome our second last finalist, LP Lens. Welcome to

Hi. Um, can I start? So, um, half of all unis swap liquidity provider ends up net negative versus simply holding. That's not a tech. It's the bankor and topas blue study from 2021 uh from 17 poll.

And the median LP isn't losing money because LPing is broken. they're losing it because nobody tells them why it's bleeding. El Pelance does. So, um, this wallet invested 20,000 euro on April 17 and three weeks later it's worth $80,000 uh dollar. Why?

Um so uh we have the atlas where you can enter your address or any other address of the it mainet and it will show you all the pool you've invested in and uh we will take as example this account because it's bleeding a lot. Um when you select one uh position you will end up on this screen. It will do a full analysis that take a bit of time but here is complete and we can see that you have an impermanent loss um analysis a regime pool uh analysis and most importantly we want to find you the best pool of this pair. So uh I will do a before hook scoring. I will replay 1,00 mainet swap swap by swap with and without candidate uh so I can find the best pool for your needs.

If you like the plan of the agent, you can choose to migrate it but um it won't do the full swap. it uh will only bundle all the instruction needed for the swap into a permit to authorization that I can sign and then the user can choose to send it to unis swap universal router uh or not. But uh we took this approach to be the to have no custody uh of the user uh but it could be a nice upgrade for later. uh you can see the report uh on shen the verdict uh even ens but we will come to that later. So the agent uh has 11 phase and each phase has its own acceptant test.

So um if the agent can pass the test on this phase he will try it again until success. It's to prevent uh force s source of truth uh air elucilation. Um the AI model use is quen 2.5. It run into a sealed on cloud thanks to object compute and intel tedics for the cryptographic uh proof.

Um the address itself is an asset. uh it's a INF which reputation and uh data that we can see here and also we can access directly this INF thanks to MCP server that you can download directly on the website uh to make a small demo uh of uh the MCP uh we can see he has six tool we can see them and also for example we can ask uh it to run um like fetch and um past diagnosis uh from OG storage but directly on chain without eating the back end. Also to access the INF you will need a license that you can mint uh directly on the INF and 20% of those royalty will go directly to the Treasury community to upgrade the project and make sure is secure. Um you can see that an analysis will generate five uh ENS uh pass thanks to one root ash. So it's easy to find all the document on chain of a specific analysis.

Um thanks for listening to this presentation. I had a lot of fun doing this project is the four project of my lens series. Uh but this time more oriented about tokconomics. Uh thanks again for listening.

Amazing. Thanks JB. Uh such a cool way to actually simplify your LP positions through LMS and uh the fact that you help with the migrations is even better.

So uh really great implementation. Congrats.

Thanks.

All right, we are ready for our last finalist demo. So without further ado, let's welcome team Dio.

Hi

feel free to get started.

Yeah. Hello everyone. Today we would like to introduce our project Dio. Dio is a decentralized AI review consensus platform. Current review systems often feel like a black box.

Users may receive a score but they cannot clearly understand why that result was produced. DIO solved this by using diverse AI review agent VRF certication audit review and reputation based consensus. Now let's move on to the demo. First this is the main page of stio. We can see different AI reviewer character here.

Each character is an AI agent with a different LLM prompt evaluation criteria and area of expertise. And this is a dashboard. Now let's move into the actual review flow. Here user can select the type of review. DIO is not limited to only one domain.

It can support different types of evaluation. For this demo, we select paper review. First step is uploading the file to be reviewed. Once the file is uploaded, DIO generate a hash of the file and next set a review bounty. The bounty is paid in US Dio, but user can also pay with other tokens such as ETH uh through unis swap.

This bounty is used as an incentive for AI reviewers to participate in the review process. And once bounty payment are completed the review section is officially start. From this point DIO begins selecting the reviewer through V RF sition. It makes reviewer selection verified and hard to predict reducing the risk of pre-arranged collision or bribery. Now we can see the selected AI reviewer character and once the reviewer are selected round zero start immediately.

Reviewers first commit their scores and report as hashes and reveal them only after all submission are locked. It helps prevent node from copying another node answer. And this first round only the starting point. Now we move to round one peer edit. In this round reviewers crossedit each other's report.

This allows dio to distinguish a meaningful minority opinion from an unreliable review. A minor minority score can still keep its influence if the report is well justified but extreme scores with weak or visioning are downed. So dial does not punish disagreement but penalize ro quality and unstable reasoning. Based on this audit reliability, DIO updates its reviewer's contribution weight for the next consensus step. Now it's round two.

In this round, D reflects it reviewer's long-term reputation. If a reviewer repeatedly gives unstable scores or unreliable edit, it reputation decreases. The final consensus score is uh calculated by considering the original score audit and its rep and reputation. An unreliable reviewer is slashed to keep the network accountable. Uh clicking a character shows is review details and if you have any question oh you can ask in the chat and finally user can download the full summary in a PDF format.

So this is uh our road map have three direction. We will expand IO into more domains, improve identity management, and provide tools such as an SDK. Finally, this is our Kim and thank you for listening. This was Dio.

Amazing. Congratulations on doing a super cool demo, especially when we can see what each of the agents are doing on uh on an actual map. So, uh, congratulations and, uh, I'm glad that, uh, this was able to go live and, uh, you were able to show off to everybody.

Thank you.

And we hope you continue building. All right, those were our seven finally. So, I want to congratulate once again, Clan World, Slapstock, Meline, Egyp 402, LP Lens, Dio, and Common OS. Let's now jump into the part that I know a lot of you are waiting on. But before we do that, I want to talk about actually being on chain.

So all of you who participated at the hackathon got to mint the hacker pack which gives you access to the faucet. But after this event ends and you finish the hackathon and you submit a project and you also won some prizes, you get to upgrade your experience to be able to mint either the builder pack or the pioneer pack which give you additional perks. So you'll be able to see that on the slashpacks page. And this also lets you claim a lot more things that we do in our ecosystem, including higher foster limits, but also things like access to various upcoming events and conferences and discounted tickets. And a lot of you will also qualify for ETH Global Plus where we offer a single way to experience everything ETH global does which gives you access to everything we do per year uh for free for things like Ecom ticket.

You get access to all Pragmy events, some city packs, we subsidize few onchain transactions, hundreds of side events that we work with from our partners. You get travel credits from EIN to Uber to priority pass and a lot of startup credits where if you want to use any of the services that you need to launch your company, you get to simplify that whole experience and uh become an official startup with a lot of credits to get your product out there. So all that is on nucle.com plus. And let's get to the part that I know all of you are waiting for and that is our partner prizes.

We're going to kick things off by talking about ZeroG. ACL is the best Asian framework winner. You're going to take home $2,500. Menel comes in for second place for $2,000 prize. Immunity is going to be the third place winner for $1,500.

Cortex comes in fourth for a,000. And FF is the fifth place winner for the best framework tooling and core extension integration. And you'll be taking home $500. So congratulations to these five teams. Now, for the best autonomous swarm and INF integration, CodeGuardian INF takes in the first place, or they're actually all going to win the prize equally.

So, Cordguard infant mech Meshx, MMO Agents, Brainedia, and Federated Reserve. These five teams are going to take home $1,500 each, and those were all the zerog winners. Then, we have the ENS prize. ACL is the best integration for AI agents and you're going to be taking home $1,250. Brain Pedia comes in second for $750 and Skill Name is the third place winner for $500.

Then for the most creative use of VNS, NSIGN comes in as the first place winner for $1,250. Reckon 402 is the second place $750 winner and Hermis is the third place $500 winner. So congratulations to all these teams. Then we have the Uniswap Foundation prize and Zuno is the best unis swap API integration and you'll be taking home $2,500. Moai comes in second for $1,500 and third place goes to Alps and you'll be taking home $8,000.

Then on to Jensen. Droius is the best AI AXL integration winner and you'll be receiving $2,500. Pythia comes in second for $1,500 and AXL Open Telemetry is the third place $1,000 winner. So, congratulations to all three of these teams. Then for Keeperhub, Tradewise, Agent Lab is the first place best use of Keeperhub winner and you'll be receiving $2,500.

KeepGate comes in second for $1,500 and Zwarm comes in third for $500. For the best feedback bounty, Evil Yield and Compute Pool are going to get $250 each for helping improve KeeperHub. And of course, the seven finalists we just saw are all going to be receiving the Open Agents finalist pack, which gives them one year of free E global plus. Every member on that team will get an additional thousand dollars from us. They'll also get $500 of flight credits where we'll fly them to our next in-person hackathon.

And we do a lot of private events with founders in this space to help them share their versions of what it takes to build something successful in in our ecosystem. So, you'll be able to attend these types of events. And over $15,000 of developer credits will also be sent to each of these teams to help take their projects from Hackathon to something bigger. And not only that, every member will also get a finalist hoodie, which if you have this hoodie, it means you are a finalist at Eat Global and only they do. You automatically get a 10x increase in your daily faucet limits.

We're going to be mailing a custom ledger flex to each of the team members, so you can store your assets very securely. And of course, the $500 reimbursement applies to these upcoming four events. So they can see us in a month in New York or in two months in Lisbon or September in Tokyo or in November at DevCon. And also they will automatically be getting a free ticket to Eidcom. So congratulations to all of these teams.

And before we wrap everything up, um I want to kind of talk about a couple small disclaimers. So everything that we just showed is now available on the website, which means as a team, you'll be able to see exactly what projects you want on your own dashboard. And the time delivery for the prizes is going to be at the end of this month at May 29th. And everything we just announced is tentative and not final, which means that over the next three weeks, we'll be reviewing everybody's source code. We'll be making sure that everybody followed the rules, nobody cheated, and only then we'll be verifying and sending the prices.

So, this means that some amounts may change if somebody's disqualified. And you'll get an email from us a few days before this date confirming this is the prize, the amount, and the address your prize will be sent to, and it'll be dispersed on the 29th. And the same thing applies for stake returns. They'll all come on the same day with the prizes. So, that is open agents.

I want to talk about a couple things that we're excited about. First one is ETH comp where we're doing a conference to bring in all the institutions that are helping integrate and build and make Ethereum better and just the EVM world. So these are some of our incredible speakers from the DeFi world, from the Trafy world, and just major institutions that manage hundreds of billions to trillions of dollars. You get to see them in person and see what they're excited about. And just for this live stream, we're doing a 50% off promo.

If you use the code open agents on the ECcom checkout page, in the next day, you will get 50% off of your tickets. So, if you haven't really signed up yet, we got four weeks to see you in New York and make sure you use this code. And last but not least, I want to thank everybody that helped make this event possible. Starting with all of you hackers, all of our incredible judges who took the time to talk to the teams in real time and share their feedback with them. Some of our mentors who spent countless hours over the 10 days answering a lot of questions.

Our speakers who did workshops and taught how to use various protocols. All of our amazing partners and kind of the key members that helped shape this event from our partner side. And last but not least, a massive shout out to the global team. So a big shout out to Victor, Meltum, Joanna and Alisa, Karen, Aubrey, Val, Pascal, Alex, Sophia, Nun, Anna, Emily, Jacob, and Andrew. This is a team that makes everything work behind the scenes and gives you the best experience.

So that is ETH Global. And last but not least, just to kind of summarize what's upcoming, ETH Global New York is our next hackathon in person. Takes place on June 12th. Then our second it's matched immediately with ETH comp which is a few days before it which is our massive conference in the US talking about how institutions are coming in crypto and our second hackathon will be in July in Lisbon. So these are the next two events that you get to see us in person and everything is available on our website where you can sign up you can see all the details you can see who's going and everything else in between on eglobal.

com. So, thank you for participating in Open Agents. We look forward to seeing you in New York in just a few weeks. And uh in the meantime, enjoy some lowfi beats and uh we'll see you all online. Thank you everybody and uh goodbye.