Agents without masters hackathon

Captions as YouTube publishes them, usually machine-generated, so product and people names can be misheard.

1. 0:00

   Mhm. Mhm. Tell me what you want. Tell me what you want. Tell me what you want. Hey. Hey.

2. 1:11

   Hey. Tell me what you want. Tell me what you want. Tell me what you want. What you want? What you want? What you moo?

3. 2:27

   Tell me what you want. You want you want me. Want you want you want me. Tell me what you want. You want you Heat. Hey, Heat.

4. 4:04

   Hello everyone. Um, yeah, I think I'm on air. Hi, my name is Maria and I'm the hackathon head mistress from Hackbox. I'm very excited today because I'm about to fly to Berlin and host an agents without the link right there. uh we already almost closed the registration. So you can still still register only if you're referred by a partner or by the team member who registered before. Um so today we're going to talk a little bit about this hackathon and a little bit

5. 4:37

   about near updated near tech stack uh specifically about shade agents that are actively shipped by proximity and our guest Owen Hassal. I think Owen is already with us. So he's going to do a quick demo like a live coding session later today. Uh but before I jump into that, I would like to uh talk a little bit about the hackathon. So it's going to start on uh June June 15th uh which

6. 5:03

   is this Sunday. It's going to be a part of uh Berlin Blockchain Week, one of the biggest web 3 gatherings in Europe and just one step ahead of uh ECC uh in Kodour. I'm sorry if I'm butchering the pronunciation in in the south of France.

7. 5:21

   Uh I really like Berlin. Last year I've been there with our team um at East Berlin. is Berlin is host is traditionally hosted every two years and this year unfortunately won't be hosted but luckily we are there to host our hackathons dedicated to agents without masters. I'm just going to show really quick what uh the schedule of this event looks like. Uh we have this page on Luma

8. 5:45

   that you're seeing right now in the bottom of the screen and uh the hackathon will start uh in the afternoon on June 15th. uh we prepared some snacks. We prepared some interesting swag for you. So around five we will uh start the lightning talks and workshops.

9. 6:04

   Around six we will start hacking. Um and we will allow people to stay overnight obviously with late night pizza and optional shenanigans. And then on June 16th um after coffee and breakfast and the morning hacking session, we will host some final workshops to make sure that uh your Dora hacks page which I'm going to be talking a little bit later and your pitch are looking great. U we will have uh a demo session at 12:30 with our respective judges and then by uh 2 p.m.

10. 6:37

    we're going to announce the winners of the hackathon. So, about the winners, I'm going to talk separately because we already have the bounties published on um uh one of our favorite hackathon platforms, Door Hacks. Uh we love Door Hacks because it's such a great platform for hackers. It's allow it allows you to submit uh a project and a project page is going to look great because it will

11. 6:59

    have all of the necessary details like a demo video, GitHub link, presentation of your team, a description. So all in all I think Dora hacks provides great hacker experience. So about the price pool um we offer 25k in bounties. So we have if if you want to go to to the hackathon pageora hacks uh you will see the prices breakdown. So we will have a main track uh up to 10k offered in prizes by near foundation. And also I would like to mention that um the the winners will be

12. 7:34

    offered a place in the protocol rewards program something that uh near horizon is managing for people who are really successful in building stuff and want to keep building a new ecosystem. So this is there's that there is monetary players and there is also a lot of mentorship a lot of opportunities uh and a lot of um networking chances with new horizon. Then there is proximity labs.

13. 7:57

    Um, Proximity is offering up to 2K in prizes to two of the best teams who built their first non-custodial verifiable multi-chain AI agent. So, Owen is going to talk about it later today. He's going to showcase what agents can do and how to ship your own agent on Near. Then, we have a bounty from Aurora, agents of onchain intelligence. Aurora is offering up to 3K in prizes. uh two up to four projects maybe a little bit more because we have this wild card pool. Uh the main mission

14. 8:28

    for the Aurora bounty is to build AID apps chains or projects using Aurora tech stack. And then we have a beta bounty defy agents and more. Beta is offering up to 10k up to up to 10 best projects and actually beta has four bounties to choose from. A lot of ideas here. So, if you go over to uh the Dora hacks page of the hackathon or if you already registered, you already have uh the link and all of the bounty descriptions in your email. We have um a

15. 8:59

    document with with bounty descriptions that uh you can you can take a look at and I think today is a great day to actually take a look at it uh to come prepared. Uh uh so we are expecting up to 100 people at the hackathon on Luma.

16. 9:16

    I already see that over 220 are registered but you know how Luma is some people don't show up. I really hope that not all of them will show up but yeah let's see we we prepared two uh floors of a beautiful co-working space W3 hub uh a beating heart of everything web 3 in Berlin and and to an extent in Germany. So uh it's it's a very special venue for us. We already hosted one near

17. 9:41

    meetup in that venue and uh this hackathon will be probably a beginning of a beautiful friendship with uh German community of web3 and AI builders and near is also present all over the Berlin blockchain week. There is a lot of highlights. For example, there is a conference called fund in the commons.

18. 10:00

    Uh but there is also um an AI agent summit right after the hackathon on June 19th which we're sponsoring. Um we are offering a volunteer at the hackathon uh I believe at the AI agent summit. So if you're hiking on agents without masters and you're still staying in Berlin after that and you still got some juice to go to another hackathon, uh please use the link down below to register. I already

19. 10:26

    requested uh some vouchers for our hackers and I really hope that we will receive some so you would be able to register for free. So without further ado, I would like to introduce our guest my good friend Owen uh who used to be a hackathon winner and now he is full-time employed as a de proximity. How did that happen?

20. 10:48

    Uh it just it just sort of happened. I I want to pick up one thing. You said you said good friend, but you really butchered my name there at the start. I did. Yeah. Well, to be fair, I can't pronounce your last name either. So, we're we're either I will I not pronouncing it to not to confuse people, but you know, web three is a very global community and and um yeah, people come with all kinds of last names. It's Hassel or Hassel?

21. 11:17

    The first one. Hassle. Hassel. H I I was kind of, you know, I was adding like an Arab Arabic floor to it, but yeah, you're British. You're going to have that. So, it's going to be all our our respected guest and the judge of the heck. So, you're in proximity now. How does it feel? Do you like the team? Are they treating you well?

22. 11:38

    No, they have me they have me locked up in a basement coding 247. I They won't even let me out. They're letting me out for Berlin, thankfully, though. Um just great uh it's great team to be a part of but I think it's also great to be like working on amazing tech um I was yeah around there near obviously for a while and really one of my big interests was chain signatures which is one of uh

23. 12:10

    proximity's main things and then also now shade agents and yeah I'm really glad to be working for a company that is working on something like pretty damn groundbreaking and it's something fun to work on as well. Um, yeah, with my startup I was I was planning to to pivot to actually using shade agents. So, it's great that I can like get paid to basically build build on top of them.

24. 12:42

    Um, yeah. How how find a way that your hobby would pay you? Something like that. So that's that's actually Oh yeah. Never work a day in your life or whatever. Yeah. Yeah. But we see that that white board is actually still white. So probably you haven't seen the other side of it.

25. 13:02

    That's got all my all my quantum physics on the other side. All the quantum physics. Yeah. I always forget that you're you're like Renaissance man. That's that's what they call you. Yeah. I don't know what that means to be honest. That means that you're talented in in in more directions that a person should be allowed to be talented. No, I I I just I just pretend to um Yeah. So,

26. 13:29

    should we should we get into it? Should we get into shade agents? Yeah, shade agents. So, the main thing about Mir is the night shade algorithm and then there's shade agents. U and I really like this naming tendency. is very mysterious, very very intriguing. So, what's shady about the agents?

27. 13:47

    Um, I don't actually I don't really know what shady. I think it's just a cool name. Uh, let me Very cool name, right? Uh, could you put my screen on here? I'm sharing. Thank you. Uh, yeah, it's I don't know. It's It's Shade Agents. I don't really know how this has come aart uh come about. Maybe the shade is is to do with there can be some like privacy elements in there. Um, but basically what they are, they're the first proper

28. 14:19

    like web three agents that can trustlessly serve like multiple users. So, previous agents, they could only really s serve yourself because um you you need to trust other people basically. Um but with shade agents, how they they operate is that they're operating inside of a TE. So, they have like a key inside this TE, which is a trusted execution environment. So, you don't need to trust the system. Um, but the problem with TE's is they can like go down and if

29. 15:03

    they were to go down then this private key would be lost. Uh, but the great thing about how shade agents work is that you're using uh these chain signatures decentralized key management. So whenever the uh TE goes down, you can just like boot up another one and have access to the same accounts. uh which has not really been seen before. It's been seen before in like ZK I believe

30. 15:31

    but uh ZK I'm just I'm just gonna translate it to the to the curious folks who are watching us. Not everyone is is so uh well experted in in blockchain. So yeah. So yeah, zero knowledge. Um but these proofs incur like a pretty pretty big overhead. But with TE's it's only like 20%. Um so how these are operating right we have this uh we have these worker agents that live inside of a TE and they have access to like LLMs and

31. 16:10

    external APIs and storage and then the agent is actually composed of a smart contract as well which lives on NIA um and you can have multiple agents be spit up spin up and interact with this agent contract. So when this worker first spins up, what it does is it registers in this agent contract and what it does it says what code hash it's running. So like what code it's running in the TE uh so you can verify that it's it's the

32. 16:46

    expected code so it's trustless and then also an attestation to say hey I am running in an actual TE I'm not just lying about running in this trusted execution environment and once it registers as a worker then it is able to use this this function here on the agent contract uh which is like sign transaction which through this function you can just like call any blockchain basically. Um so through this function

33. 17:20

    you can custody assets on like sui native bitcoin salana base ethereum uh like 95% of blockchains which is really cool and actually blockchains that will exist in in the future. Can I ask a question? So uh we call the hackathon agents without masters which kind of implies that we're working on autonomous agents and near has been shipping this whole concept like blockchain for AI AI should be decentralized. Why should we decentralize AI in your opinion? Why should we ship autonomous agents?

34. 17:58

    Yeah, I think in terms of shade agents, let's say that they did have a master, then you would have to trust that master basically. And if I was to store these funds in in these accounts that are centralized, then at any point, this master could just rug me and take all my funds if they wanted to. But with this system, there's they physically can't do it.

35. 18:28

    uh they're just not able to. And then coming out of like shade agents, I think another reason to have users owned user owned AI as we call it is that you don't know what these LLMs are trained on. Um, for example, they could be trained a lot on like one political party or one opinion or trained on a load of data to do with like a certain product and it's pushing you this product in all of its responses and you don't really know that or you don't want that. is just serving

36. 19:04

    the company that is uh like running this running this LLM, but we really want this user owned AI where its priority is is the user. And yeah, a great way to do this is is through TE so you know uh exactly like what code is running. So if you want to read a little bit more about how shade agents work, you should go on docs.near.orgaishade agent. So before we jump into how to actually deploy something on shade agents, what can you build? What kind of agents and what would they do? Yeah,

37. 19:45

    good question. Um, so we actually we actually have a page here which is a list of my favorite ideas. So yeah, with shade agents, you can actually use smart contracts with LLMs. So you can make decisions uh based on an LLM. For example, you could have a DAO with like an AI president that decides how uh how like the treasury should be spent and how the protocol should evolve based on like aggregate user opinions uh and the

38. 20:18

    public and it just optimizes things um according to this like sentiment and there doesn't have to be like any users involved. Um I think a big thing we are pushing is these what are now called agentic protocols. Uh because I think when like everyone came into web 3 we first heard of like smart contracts and we're like what are they? Why are they smart? And I don't think it's really an

39. 20:54

    appropriate name because a smart contract just sort of lives there on a blockchain and it doesn't do anything by itself. It has to be like acted upon. It just sits there. It like serverlessly. I don't know if that's a word. It's like serverless functions where you just call it and then it does an action. But with these agentic protocols power powered by shade agents, you can really have uh like actually smart contracts where

40. 21:25

    they're acting they're actually proactive and autonomous and intelligent. So they're actively trying to optimize things and do what is best for like the protocol and for the users. So an example of this is yield optimizer. Uh there's actually a team building one of these. Um it's like a an Ave rebalancer. So what it does is uh balances like USDC deposits on different instances of Ave depending on uh where has like the highest APR and when that

41. 22:05

    changes like every 24 hours or an hour or whatever set by the agent, it will rebalance these funds to make sure that the user and the protocol is always getting the best uh like staking rate and the best like money back on their money. So yeah, you can do some really interesting things and I'm really excited to see these these agentic protocols and have web 3 become actually smart. Now just before we dive into the

42. 22:38

    technical part of the of this workshop, I want to remind everyone that proximity labs is offering two prizes for people who build on on sheet agents. uh specifically uh MVP, agentic protocols, workflow agents, data oracles, and private agents. And if you want to see more details about this bounty, uh if you want to see more dev resources, you should go to dorhacks.iohackonagents

43. 23:06

    without masters. So, back to you, Owen. Let's dive into the technical stuff. Yeah, I think to to add that add to that um I want to see some like interesting agents as well because these are really flexible and you can do things on web 3 that you can do like on web 2 now. Uh so like an example I put in there was a protocol that gifts people NFTTS based upon like in-game points. So, I think

44. 23:39

    one of the bounties I put up is like gift people NFTTS depending on what town hall they are on Clash of Clans because with Shade agents, you can really get off-chain data like into your smart contracts really easily. So, I want to see people interact with some like funny funny uh APIs.

45. 24:03

    And this is sort of like part of what you can do with oracles. Um, okay. Yeah. So, we can actually like deploy an agent now. So, here is a template repo we have. So, I'll just I'm going to clone this. Let me pull up my uh Yeah, you can see my terminal, right?

46. 24:25

    So, I'm going to clone this. Uh, let's give it a shorter name. Call it demo. Um, so I'm going to step into this demo and open it in cursor. Cursor is the best IDE. Um, yeah, if you have any questions, just just shout up and I'll try and answer. So, this is like the shade engine agent template and it's actually, you can see it's a next.js product project. These agents can obviously be scripts like a NodeJS script or like a Python script and it

47. 25:07

    can be looping if you want. So it's always running. Um for demonstration purposes we have it as Nex.js. So you can just interact with a front end as you'll see later. But the actual agent part lives in this API route of the agent. So we have a few exposed API routes. The first one is like get ETH account uh which I'll explain later. There's this get worker account because each of these workers

48. 25:41

    operating in the TE's have their own account to interact with the smart contract. And then there's this send transaction where the real the real meat of the the agent is happening. So this template is actually just like um a simple price oracle. So you have oracles like chain link or or pith where they have multiple nodes that verify like the same data and then push that to a smart contract. What we're doing here is just having like one or two tees sourcing this data and we don't have to

49. 26:28

    have multiple nodes like 20 nodes or whatever because it's running in a trusted execution environment. we know that the price is actually what the price is. Um so to go through this we have this function get get Ethereum price which is actually just a script that gets the price of Ethereum off uh Binance and off Coinbase and then just takes an average of that.

50. 26:59

    So first we get the price we uh we actually take this price and get a transaction payload. So this is basically creating an Ethereum transaction and squashing it into a payload which will be signed. This is the payload is the hash of um the transaction. And to do that, we're getting the the Ethereum address, which this route does that we're going to use to send transactions. Then we set up like this contract instance and set up a

51. 27:41

    function call. So here we're calling this function update price and submitting whatever Ethereum price we fetched up here. Um and then we prepare the transaction with the sender, the contract address which is like the oracle contract per se and then the data which is the the actual price and what function we're calling.

52. 28:09

    Uh so once we have this payload we basically sign this payload with the worker. So the shade agent SDK provides this function sign with worker which takes a path which determines the address that uh it's going to use to send the transaction and then this this payload the transaction payload and this is going to make a function call to our agent contract which will then propagate

53. 28:41

    it and use this decentralized key management to sign it. and then we'll send it. So we get this this signature from the agent contract and then we basically add the signature to the transaction and then here we broadcast that transaction to the uh Ethereum network. It's actually like um and then once we have that transaction hash we just output that from the API a API group. Uh, and that's

54. 29:16

    that's everything that this this shade agent is doing. Any questions about that? I don't see any questions in the comment sections right now, but I would like to remind the folks that are watching us on X and YouTube that you actually can send your questions to the comments and I'm going to read them out loud to over here.

55. 29:42

    Sounds good. Uh so yeah, now we know a bit about how this was working. We'll actually go ahead and uh deploy an agent. So the first thing you're going to do is copy this uh example file. And I'm going to rename it um to remove the example bit so it's actually using it. And what we need to do is get a near account ID and a near seed phrase which is going to deploy the agent contract for us. Uh so I'm just going to

56. 30:23

    do uh oh that's not how you spell near account. I'm going to create a new account. Get the faucet to sponsor it. What should we call this? Um, DevHub demo.est. And then auto generate a new key pair. I'll just print it to the terminal and create that. So this should give us our seed phrase.

57. 31:00

    You can steal this if you want. It's just test net and my account ID. So this account is going to be deploying this agent contract and it's also going to be funding our worker agent. And then we have this contract ID which is going to be the um the agent contract that we talked about earlier and that is going to be deployed as a sub account to the account we just created. So you have actually

58. 31:34

    like two different contracts you can deploy. You can do this uh agent contract proxy which is just um a contract that takes payloads and submits them and signs them and then uh we send them and it doesn't do anything else but that. And then we have this sandbox thing which is the actual production agent which has the verification of the workers inside of it. But we have this like proxy so you can just do this

59. 32:10

    locally because if you're trying to send transactions to this sandbox outside of a TE, it won't work because by definition like it has to be running inside of a TE. And then we still also have some like code hashes which is the code hash of the code we're running which will be used to like check against the agent contract that we're running the right code. But we don't actually need to touch these. These are just going to automatically update when we build

60. 32:42

    um when we deploy it. And then you can call your project whatever which is going to save. This is like your Docker username and then the image name and that's where your code is going to be like deployed and and verified to. And then the last thing we need is this this Valor API key and Valor is the the product we're using to host these TE's.

61. 33:17

    So, I'm going to actually I'm going to hide this off screen because I don't want people knowing my API key. But I'm just basically going to go to Fer network and create um create a new API key. I'll just call that uh Owen's key. So, I've created a new key there.

62. 33:43

    Uh, I'm going to actually hide this off screen as well for a second. Um, okay. I think that should be good. Yeah, I've just inputed a new K down here. Um, and then I'm going to close that. Yeah. So, we're set up for the sandbox here.

63. 34:03

    So, what I need to do, well, what you'll need to do when you first start with a shade agent is install this this shade agent CLI. And you can do this with this command that's specified in the read me somewhere. Uh, yeah, somewhere in the readme. This will download you the CLI, but I already have it. So what you need to do is just once downloaded just run shade I think it's just shade agent CLI

64. 34:38

    and this is going to Oh yeah you need to insert your like system system password because it's using some like admin pseudo commands. So I'll enter that. And what this is going to do, it's going to take this code with the API routts. It's going to package it inside of a Docker image. Then it's going to push that image to Docker Hub so it can be accessed by by anyone really.

65. 35:08

    Um, and then once it's once it's deployed to Docker Hub, we're going to deploy some contracts. This one step like CLI will do it all for us. And then it's going to push the image to FA cloud which will uh which will be the deployed worker agent.

66. 35:30

    But this all takes a bit of time. Uh it takes a while to build this and then push it. So as all things go here is one I made earlier. So as you can see you have this this pre-made front end for you and this is actually being pulled out of F network.

67. 35:55

    You can see uh F network up here and as I said earlier it's just a simple price oracle. So it's going to be setting the price of ETH in the contract. This is making a few call to the contract to like keep up to date with what it currently is. Um and yeah, you have this this worker account which is the near account that's making the calls to the agent contract to like first verify well register in the

68. 36:26

    contract but also um like each sign request it's making it's using this this account ID. Um, and then we also have this a Ethereum Sapoleia account, which is uh the account that's going to be making the call to the Ethereum contract to set the price in the contract. Um, and we can actually just now this is deployed. Click to update the price of ETH in the contract.

69. 37:01

    Do do you like the little the little animation I made? This took me ages to make. So yeah, you can see the price was actually updated there. And that's going to give us a transaction in the explorer. Uh things take a while on Ethereum, but we'll see in a second that this has come through. Yeah, there we go. I'm not too familiar with this explorer. Maybe I can see it down here. How do I not JSON arguments?

70. 37:37

    So, if I decode this, yeah, I can see I set the new price of uh what is it? $2,858. The current price of Ether is uh and yeah, that's that's pretty much everything. Like, it's pretty simple deploy to deploy a shade agent. Um, I would show you the deployment that we did, but it's still taking a while. Um, actually, it seems like it might have been done. So, yeah, I just reloaded that. It's booting up, so we can access

71. 38:19

    that in a bit. Um, so like that's how you actually deploy a shade agent. If you want to if you want to like make your own shade agent, you need to go into this send transaction script and basically just simply edit this script to do whatever you want to do. So yeah, we were making a function call, but you could to this Ethereum contract, but you could do some other function call to uh a Salana contract or a a TON contract.

72. 39:02

    Or you could have one that just simply like transfers funds. Or you could have a transaction well an intent to be signed via like near intents which will allow you to swap tokens across different chains. One of the examples that have been being deployed already of a shade agent is this uh mind share agent which is really cool. Um, and it's basically just trades tokens with near intents based upon

73. 39:35

    uh Kao Kao Mind share. So whichever token or or protocols getting talked about like the most on Twitter, it's going to buy more of that token and then sell ones which are being talked about less. And that's all powered by uh shade agents. And theoretically, you could like buy shares into that mind share agent and it would trade on your behalf and you don't have to you don't have to trust the the system.

74. 40:09

    Uh let's see if it's Yeah. So you can see my my agents actually being deployed now two minutes ago. So, we can go in here and view all the logs. Everything looks good. Um, and I can actually just test the API by doing it in the browser. So, we have um we have one called actually I'll do it here.

75. 40:44

    So if we come back here, we had an API route called get worker account. So I'm going to test that. Uh so if I do slash API slash uh get worker account, that should hopefully make a call and get me. Yeah, that's my worker account. Then let's try uh get ETH account. Taking a while.

76. 41:22

    Yeah, here's my Ethereum account. So, I can actually fund that with my MetaMask. Let's just send it a small amount of ETH. Uh, that should be enough. I'll let that settle. Let me actually check this in the explorer. Um, it's balance should hopefully yeah balance is increased. So I could do my send transaction which we saw on the front end before through this API route. And this should hopefully give me back the transaction

77. 42:16

    hash and the new price of ETH uh like we did in the front end before. Yeah, there we go. So the new price has changed. I think it's gone up by by like $10 since we last checked it. And then I get a transaction hash here. And then I can put that in.

78. 42:38

    So you can see that you can interact with the agent like however you want. It doesn't have to be through a front end. So you could have it looping um you could have it looping in a system. So it does an action every like five minutes or or every day. So this is really where the the autonomy comes into it.

79. 43:01

    And yeah, that's that's pretty that's pretty much it. You can see that they're they're not that difficult to deploy. They're not they're not. But then I have a question. So we will have 24 hours with the hackathon. So as an experienced hacker yourself, you you won a couple of them and I believe you participated quite a few. How would you suggest to spread your time over the course of

80. 43:27

    those 24 hours? How much time should actually be dedicated to coding? How much time to brainstorming? How much time for presentation? What's what's your suggestion? And then how much for like eating and and drinking considering special dread also? Let's not forget that.

81. 43:44

    Well, it's it's Berlin, so there should be a I think we should allocate some decent amount of time to drinking. Um I think like this event is not just a hackathon. We're doing it quite a lot of like a learning experience as well. So people are going to be able to come in and learn like go from zero to sort of hero in near and learn a lot about N and then actually hack something. So I suggest at

82. 44:17

    the start of the hackathon you just absorb as much information as possible. So learn about shade agents, learn about chain abstraction, chain signatures and intent and then go on and actually hack something. And I think whilst during these talks you'll probably have like an aha moment and be oh that's what I can build. and then then you can spend however many hours uh implementing that.

83. 44:48

    But it's it's a reasonably short hackathon. So make sure you try and wrap things up reasonably early. Don't as with all hackathons, don't leave everything till last minute. Uh, I think something short and sweet, especially in this time limit, is better than something that you give me and I try and interact with and it it it doesn't work at all. Yeah. Uh, thank you so much Owen. And with that, I would like to introduce our next guest,

84. 45:22

    uh, Shai from founder success at Near Foundation. Hi, welcome. We just we just wrapped up a technical workshops and then now we are talking about how to win a hackathon. So, what's what's your suggestion here? Yeah, no, pleasure.

85. 45:35

    Pleasure to be on. Thank you for for having me. Um, I think the the key success I've seen builders have comes really when they spend time uh eating and taking time to plan out what they're building as opposed to just running into jumping and building right away. U in a 24-hour hackathon, usually we see people spending about an hour and a half to two. Uh, and I've seen that be really successful. And if you think about it, if you take that hour and a half to two at most to make that decision, what are

86. 45:59

    you building? How are you going to build it? kind of go through some of the architectural pieces, make those kind of decisions within that first hour and a half to two. You're you're less likely to have kind of pivots and uh changes down the road. Uh you also want to make sure that during that first uh period of time, you I would say usually carve out about an hour first. Uh lock down your idea and what you're trying to finalize and do and then spend that next half hour to hour after that kind of planning

87. 46:23

    the okay, how are we actually going to do this and implement? when you're spending that time during that second hour planning the how, think about um incremental milestones that you can achieve along the way. Um I've seen many lofty goals of we want to do this big thing and if the teams don't achieve it, they end up cramming at the last minute and it's never really turns out great quality. Um whereas the teams that kind of plan out, you know, hey, as a as a

88. 46:45

    very bare bones MVP, you know, what's the bare minimum I can show that solves the problem I'm going after? Uh and target that for like the first 12 hours of the hackathon. and having that done and that really gives you the next kind of 12 hours just to sit and refine that and continue working down that path. I've usually suggested kind of 12 milestones having the first one let's say at 12 12 hours the next one at 16 hours and the next one at 20 hours. Um and don't think that

89. 47:11

    you have 24 hours just kind of in your head tell yourself I have 22. Um, and then that way if you get to that 22 hour mark and you have things done and maybe you've hit that second milestone, uh, you're not necessarily as behind, but you can say, "Okay, how realistic is this in the next two hours that I'm going to be able to hit that third milestone and kind of give myself that stretch goal." Um, so that's usually I found really works well is just kind of time yourself as you're going through.

90. 47:34

    Uh, and then don't try to push yourself to the very last minute. Um, because a lot of times I've seen things break the last minute uh, when people are trying to make a last minute change there too. So, it's unfortunate when you see somebody working really hard throughout the hackathon and they just unfortunately break something in the last hour and then they they don't make the submission because they they broke it too far. Uh, you know, use your repository control. Uh, make sure you're doing frequent commits. Uh, keep track

91. 47:58

    of what you're doing. Yeah, that's just my take there. I wanted to jump in on that timing one. Thank you. Great suggestions. So, both of you will be judges at this hackathon. Uh, Owen and Shai probably started with Owen. I would like to know what judging criteria would you be using um except the ones that we are providing like what's your internal judging criteria technical and non-technical what you would be looking in uh looking for in a winning project.

92. 48:25

    Yeah, I think I wrote down some before they sort of evaporated from my mind. Uh, but I personally like to see projects that obviously it's not going to be a full project by any means, but an MVP of something that could actually be useful. Like, don't use I don't want you to use shade agents just to use shade agents. like leverage the stack for what it's worth and build something that is going to be useful and could

93. 49:00

    evolve into a product. But I also I like to have fun at these hackathons and it's really interesting when people see just like come up with crazy ideas. So yeah, I listed a few on the page. Um just make some make something creative as well, I think. Don't just do the norm of what uh of what everyone seems to do at a hackathon, which is just like an AI chatbot that interacts with like an RPC.

94. 49:35

    I mean, I I've seen, whilst they can be cool, I've seen like 50 submissions go down this route. Try and build something different. I think we're offering a tech stack here to actually do something different. Um, and then other things in your in your deck. Uh, like a lot of people come with a deck to to the demo. Um, a lot of people spend time doing like technical diagrams. I think I think whilst good like people normally just skip this slide after like 20 seconds.

95. 50:12

    Don't let me look at it. So there's no point really just having that. just talk to me about a bit about how it works, but also I want to hear like what is the business idea behind this as well like how could this become profitable or how can this operate as a product in the future?

96. 50:32

    Take it from a person who won two major hackathons right here. I think it's solid advice. Uh Shai, what would you say? But didn't go anywhere with those wins. So maybe don't maybe don't listen to me. Never good. Well, let me let me let me jump in and I'll capitalize on what Owen was saying there, right? Like uh in order for the product really to get to the next stage and one thing I'm I'm looking for in particular for me, right? Anytime I've I've been part of

97. 50:55

    these events is really around the the problem itself and are you are you working towards solving that problem? I I like to to call it a marriage, right? get married to the problem and really identify what that problem is and then as you're the milestones I was talking about before it should be clear to us as kind of the you know judges, panelists, educators, whatever you want to call us um that at the end we really want to see that you're you're honed in on that

98. 51:20

    problem. You're you're at least starting to solve that problem and you're making a progress down that path. You might not solve every single angle or permutation of it. um there may be a very complex problem that you're trying to solve but ultimately even within a complex problem there's simpler ones you can solve along that way there. So the the better you can articulate how you're approaching the problem then how you're solving that problem and what you've done to make

99. 51:43

    progress on that problem that makes it easier for us to really understand your your progress and your ambition and how how likely this is going to be to turn into that product. Um, you know, I think that's one of the challenges in the space is that there's too much of a rush to build a novel and cool solution. And and I I I agree with Owen on the get creative part, but getting creative part is how you're solving the problem. It's it's it's not getting creative on, hey,

100. 52:06

     what is your solution that we haven't seen anybody out there because then you're not working backwards from what that problem is, and you're just building something and it's cool, but it doesn't go anywhere, right? And and uh we've seen that many many times in many hackathons where those cool ideas don't go anywhere because there's not tied to a problem behind it. So, uh, focus on

101. 52:22

     that problem. Um, don't don't spend a I think I you call that greatly on the on the architecture there. Um, don't spend a ton of time going over the the nitty-gritty, the details and how you implemented something specifically and why you did this versus that because likely again that solution is going to change and you should be okay with that changing. So, for me, when when you're talking about the architecture, I want to hear more about how you're building

102. 52:46

     two-way doors, how you're building a modular architecture, how you've componentized things. Uh, and I don't have to again I agree that doesn't need to be a diagram, but you should at least be able to talk through that of how you've compartmentalized what you've been building out. Um, and how you understand that pieces of it can change over time. Uh, and that to me is indication of a of a strong product that

103. 53:04

     could have product market fit because you're able to be nimble and agile and change the pieces that you need. Um, I mean, finally, as you're as you're presenting and you're you're talking through your your problem, um, it's it's okay to be nervous. It's okay to stumble over your words. It's okay to not find the right words, right? We know that there's a short period of time here. Um, but what's important is that you get through why you're solving this, right?

104. 53:27

     Like why is that important to you and why you're passionate with it. So, I love seeing builders that start off the presentation with a referenceable story, something that's personal to them. And and that really ties home why why you're building this product and ultimately why you want to take it to market because you're also solving a personal problem maybe for you or your family. Maybe it's a friend that you know that ran into an issue once. Um, and then that creativity is stemmed from that that pain of that

105. 53:51

     problem. And and when you see that connection, that's usually when you also see the continuation of somebody building that too because you know that that's part of it. We want to see in this hackathon that you're setting yourself up for success from the very beginning and you're not sort of boxing yourself into one particular area and you can continue growing from here and other events that we have and on your own and through other funding and mechanisms that that are available within the ecosystem. Uh so again get married to your problem not the solution.

106. 54:17

     Thank you. Yeah. And I think I think a lot of problem I think a lot of what you have said and about how to approach this hackathon is mirroring sort of the startup life cycle like think of that problem. Yeah. Get married to it.

107. 54:37

     think of the solute like what is the best solution to solve this and set what that MVP looks like and don't complete that MVP. Don't keep trying to get something that is awesome but will take you years to build and then I think like this is how you should approach a startup as well. And then at the end of this hackathon, we have the demo which can sort of be the same as like raising

108. 55:07

     in front of a VC. We don't like just want to see your your technical ability. It's also like how are you as a founder as well? And especially if this hackathon is going onto something like the protocol reward scheme, we want to see like how would you or your team, how are you as like a founder, not just a dev? And I think this is like a good work experience for that.

109. 55:37

     Thank you. Great suggestions. So, we're going to wrap the session up. Uh, and I'm very very looking forward to meeting you two and all of our hackers um at W3 Hub in Berlin this Sunday on June 15 starting uh 3 p.m. and you can also join our customers where you can talk to Owen where maybe talk about talk about pretzels and and beer uh to me because I'm the one organizing the hackathon and I'm in

110. 56:08

     charge of bringing pizza to the builders in time. Uh so do we have any final questions on social media? I don't think so. If you do have any questions to try uh everything about agency without clusters, please drop your comment on YouTube and ask. Otherwise, please join our chat down below on Telegram and ask your questions there. Uh we already have our members from your foundation um and also from Aurora. Everyone has

111. 56:40

     joined chats among you are missing. If you're missing and uh just a quick reminder that we're at capacity uh builders wise, but if you still want to register, please check if someone you know already registered ask them to refer you. We still accept registrations by referral uh from partners or from the community. So uh closing remarks um I mentioned the protocol rewards try reward that the horizon offers. Maybe you can talk a little bit about what

112. 57:11

     what's next after the hackathon. What is there for the builders except the monetary bounties? Yeah. Know I I I love the way our team approaches builders, right? Owen made a really good point there about, you know, you're think about the product that you're building down road and think about the go to market and how you're going to take that down the next step.

113. 57:29

     Um, you know, one of the things I love about our team is that we are here to help you after this hackathon. We are here to help you turn that idea into an innovation, deliver it to market. Uh we have a really good well-rounded team that understands go to market, marketing, business development, community building, and the tech side.

114. 57:44

     And ultimately, we're going to look out for you and help you get through that process. We have really awesome programs like the protocol awards program that was mentioned. We're going to be launching different programs throughout the year to help accelerate you in different partnerships with different ecosystem partners as well. So there's going to be lots of support to help you

115. 57:59

     graduate really from that builder mindset to the founder mindset uh and continue to deliver your product to market. So this is very much a an initial planting of your seed uh and we are here to help you nurture that tree and help you grow it out and eventually also help you prune that tree to make sure that you're staying focused on your final product. So uh very much uh enjoy this opportunity and thank you all all the builders are out there listening in.

116. 58:22

     Uh, and just remember, we're always here to help you. Any questions that you might have on any of the, you know, pillars I mentioned before, tech enablement, uh, marketing, community, and business development, we want to see you successful. We want to see you grow out, and we will develop the programs with that in mind. So, please do share any feedback as you go through the process. If you'd like to see something that you don't, we're here to make that a reality as well. So, please stay in touch. Uh, thank you very much. Very excited. Thank you, Shai. Thank you,

117. 58:47

     Owen. And see you in Berlin in just a few days. I'm here waiting. What you want? What you want?