NEARCON Innovation Sandbox Review & Winners Presentation

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

1. 0:10

   So, thank you all for being here. The for anybody that doesn't know, the innovation sandbox was a 21day builder sprint. Uh, it was a hackathon leading up to the near con. The hackathon ended about one week ago. Results were uh announced about 3 days later. And so this presentation is about that review, what happened during the hackathon, and then what were the uh the projects that

2. 0:32

   ultimately won. So today we're just going to be pretty quick. We're going to talk about overview of what the hackathon was, the sponsors and the partners of the hackathon, the awards that we gave out, and then something else that's happening right now at this very conference.

3. 0:49

   So the near innovation sandbox, it was a 21-day global virtual hackathon. uh it focused on shipping productionready products that had to meaningfully leverage the near tech stack. That meant that it could leverage near private cloud near intents shade agents and a lot of near protocols unique capabilities such as its account model uh chain signatures and global contracts. So the tracks themselves focus a lot on decentralized AI, privacy

4. 1:18

   preserving consumer apps and intent driven commerce. a lot of the themes that you've probably seen in other talks today. The results is we had 485 participants, 85 projects submitted uh were submitted. We had over 20K in bounties coming from three ecosystem sponsors. All three sponsors are here today. Uh six workshops with over 250k engagements across social platforms.

5. 1:47

   And then we had our sponsors. So I actually see a couple of them. Hot Pay was here. Peter, he's going to be giving a talk later today if we can squeeze him back in. Ping Jay over there in the corner. Go ahead, give him a round of applause.

6. 2:04

   He'll be giving a talk here in a minute. And then also Julian with Nova who's also sitting over there. And then also want to give a shout out to our partners, Funding the Commons, who they were really instrumental in managing this hackathon and making it sure it all worked out. and also PL Genesis. So PL Genesis, I'll give a little bit of a a blurb about them, but if you are a builder and you are hacking right now, if you want to be uh developing and submitting a project, the PL Genesis hackathon is live right now

7. 2:34

   and you can submit to it. I think they have over 100K in in prizes available. Um and it touches on a very specific track, so I'll talk about it here in a second. So the hackathon that I'm talking about it is over but it was judged by near devub which is kind of near protocols development arm um and then also near AI which I think we're all familiar with.

8. 2:59

   So how projects were evaluated projects were evaluated on the technical execution. So how sound was the architecture? What was the implementation like? Uh was it reliable and how it worked? The user experience. How well did the app uh work? Especially if you were a consumer. Does it look good? Does it feel good? The impact and usefulness of the app. So, does it address a real problem? Does it have a credible path to adoption? A go to market strategy? And then the completeness and the functionality of

9. 3:28

   it. So, how ready is the product at the end of our program? And I also added another one here just because uh it's pretty relevant to the projects in the tracks. We judge projects as well on the relevance to the tracks that they were submitting to. The prizes we had, it was $3,500 per track. There were three main tracks. It was a $2,500 only on near bonus and then two $1,000 cross track

10. 3:56

    bonus prizes. And then our ecosystem bounties also had their own prizes. Nova had 3,000, I believe. uh Hot Pay had 3,000 and Pink Pay at 15 uh 1500. Okay, so here's the meat of it. These are the winners at a glance. So with Nova, who we originally had Julian presenting uh about it if you want to find that live online, but in short, Nova is a privacy first decentralized file sharing primitive empowers userowned AI at scale with encrypted data persistence. It's a

11. 4:29

    pretty cool uh application. It it leverages TEES in order to kind of have like group encryption on on data. And so the projects that leveraged that were the winners. Private RAG. It was a vectorless reasoning based uh rag with end-to-end encryption. It uploaded these documents up into Nova and it was all kind of like a clientside uh memory layer in a way. Proof of pulse was another winner. So that was a biometric attestation oracle. we it uh

12. 5:03

    it would take your health data and then you would kind of create a proof out of it and then that proof would be uploaded into Nova uh in that encrypted data share and then also privy finance. So this is kind of like your rocket money equivalent uh that was built on near it.

13. 5:17

    You uploaded your bank statements for private financial portfolio tracking. all that data was uh secure and private by leveraging Nova for storing that encrypted data uh uh and then in the TE and and etc. And then um also near AAI so you could talk to it about your data and since near AAI has private inference you knew that the information that you talked about with your private uh financial data wasn't going anywhere.

14. 5:46

    So that round out the the nova. Next was hot hot pay. So, uh, Hot Pay is a non-custodial open source crypto payments platform for merchants and developers who want to accept crypto payments. Um, if you are familiar with the near ecosystem, you've probably seen it around. It's basically kind of like a Stripe in a way.

15. 6:05

    And the two prizes that uh won for that was near skills for OpenClaw. So that was like a uh AI skills for your open claw that could manage near. It could make payments. It could do swaps and bridge via t chat. You could talk to it and would create a hot link or a ping link. So that was one of the winners.

16. 6:25

    And then Neptune AI. Uh this was kind of like a um chat interface that used the private AI inference. You could talk to it. you could um you could pay for your subscription through it through hot which I believe is how it applied to the bounty and uh yeah it was just kind of a really cool uh concept.

17. 6:49

    Finally, we had the ping bounty which kind of rounded out the ecosystem bounties. Ping pay is building the orchestration layer of payments. They enable teams to accept and automate payments across any chain fiat rail and a single integration. You're going to be hearing from Jay I think in two yeah two uh talks right after me but it's a really interesting project. I hope that you all stick around and check it out.

18. 7:11

    But the three projects that won from that were Cricen AI. So this was a social media agent browser extension. I believe you had to pay for your subscription with Ping and that's how it kind of applied to this track. But that one was pretty cool. you like install this browser extension and it will learn from how you talk and what you do on social media so that it could keep on writing social media posts for you.

19. 7:36

    Again, the near skills for openclaw placed again in this track. uh you could you were able to create the hot links and the ping links and then Cabo which is a private autonomous defi trading agent on near it uh leveraged the ping onramp uh so that you could on-ramp your funds directly onto near in order to use it in that in that product. These were the main tracks of the innovation

20. 7:58

    sandbox. The first one was private web and private life. This was focusing on private consumer apps. Uh really leveraging the near AAI cloud. Uh focusing on an app that people would return to every single day. Something where privacy is kind of uh specific to how the app works rather than just like a fun little feature that you added on.

21. 8:23

    It had to be core to the the product that was being delivered. So this had a runner up that was called a test AI. This was safe and audible private AI for governments. It was this desktop app that essentially had this near private AI chat, but you could uh have your conversation and you could create a QR code that was able to reference your actual individual conversation. And so I could share that with you and you could reference it in your own private AI chat. Um, and you could verify that it

22. 8:54

    came from a verifiably private conversation. And so the the application there is like if you're working with governments or doctors or lawyers, you could essentially prove that any type of work that you did with an AI uh when completing what you had to do. Um you could prove that it was private, that the inference that facilitated it didn't leak anywhere. And then the winner for

23. 9:17

    this track was Argus, which uh I wish Aussie was in the room, but uh it's kind of it's it's like this focus on oh what's the word? Anyway, it was it's it's like uh a mission control boardroom. You can kind of see what's going on around the world. Um you could get news coming in from all the different uh all these different sources. It scraped a lot of sources.

24. 9:42

    And kind of the really unique thing about it though is it would scrape a lot of public sources, but it would also accept data directly from a private source. So you could log in as say a journalist or maybe somebody that's on the ground and maybe it's a little risky to be able to share this information to whistleblow. But by logging in with a unique pass key that leveraged a specific library that they built, you could kind of pifificate that your identity and you could be fully

25. 10:11

    anonymous and submit that data uh in a verifiable way. So Argus was kind of built all around there. Um this yeah crowdsourced crowdsource on the ground type of knowledge. The other track was AI that works for you. So this track, just so everybody knows, I mentioned the PL Genesis hackathon that's going on right now. This is the track that if you build for it, uh you can submit to that hackathon as well and get access to all those bounties. This one focused on

26. 10:42

    using the Near AI stack, including the private cloud, private chat, shade agents, ejective payments, but ultimately to build uh autonomous agents. So something that maybe did a job really well and could work around the clock, an AI that works for you.

27. 10:59

    The runner up for that was Nest Home. This one was pretty unique. It uh was a private home care kind of focused on like an uh elderly folks. So this one had a hardware piece to it. It was a Raspberry Pi that could have a uh a webcam hooked up to it and you could it had object recognition and facial recognition and you could like assign the caregiver, the patient, um maybe the friend or the family and they could identify and always like be running and watching that elder as they're, you know, they're doing their

28. 11:36

    day, they're maybe bedridden or they have to take their medicine and you could get daily updates on what they were on what was happening. uh with in that room. And the really unique thing about it was that because it leveraged near AI and uh I think it also leveraged Nova too, but all the data that came with it, which is very confidential, you know, of your loved ones, um it never

29. 12:01

    left the app, you could verify that it was completely private, not being leaked to anything else. Uh and then the winner for that was uh Clanker. Clanker, who's actually the winner right there, is is sitting is in this room wearing the yellow hat. He wore that in on purpose for this very moment. Um, but he built a he built this like uh really cool. It was like a chat room where uh you would say your intent such as I want

30. 12:28

    to buy a Coke bottle or I want to buy a bottle of Coke and you would tell that to your AI agent and then it would kind of enter into this chat room and social network to uh negotiate and talk to other AI agents on your behalf to figure out how to get that uh that Coke bottle.

31. 12:46

    So that's what you kind of see in that picture right there is that chat amongst agents that are kind of figuring it out and then each agent would have its own uh near wallet so they could actually transact and facilitate payments with each other. And then finally the last track was open society from finance to the real world. So this one really focused on uh real world solutions. So stable coins, real world assets uh

32. 13:12

    leveraging near intents. Um but there was also there was a uh really important emphasis on this one about oracles because on near you can build oracles really well using a type of tech called yield resume where you can like yield execution and uh on a block in order to be resumed and called back via indexer or something so that it calls the contract and it kind of completes um that execution. And that's really great for building oracles if you want to be working with an external data

33. 13:43

    source or you maybe even want to be able to uh pay for inference for example um via a smart contract call. So the winner for that one focused on these oracles and that was Nest who's also here wearing the uh red hat over in the back. Uh this one was optimistic oracles on near uh it was actually twofold. It had a uh decentralized data protocol where you could uh attest to these oracles and essentially claim whether or not uh this data was true.

34. 14:20

    And then the other side of the product was the market. So a prediction market that essentially leverages those uh you could create markets and you can say of something and then the decentralized data layer is what would actually attest to it to it happening. So I think that these are this is actually live right now and we may have gotten some markets some prediction markets out later about the talk. So if you want to check it out uh it would be kind of cool.

35. 14:47

    And then finally there was this only on near bonus bounty. Uh this one really focused on near native capabilities because if you guys are new to near protocol is really cool. There's a lot of like really native uh awesome solutions baked into it such as the account model chain abstraction offchain computation patterns like regarding shade agents and how you can use code hash atestation in order to prove this verifiable computation. Um there's all these really cool things and this prize

36. 15:19

    ultimately went to the project that kind of leveraged this tech the most and that was Delibera. So this one is uh this one's another level. It's like multi- agent voting on year. It's essentially like an autonomous DAO where you have all these different worker agents that can deliberate with each other and vote on proposals in the DAO. And uh this is leveraging yield resume in order to go ahead and talk to external services to kind of deliberate and and figure this out. but also um every single worker

37. 15:56

    itself ran in a trusted execution environment and then any of its reasoning would leverage near AAI private cloud. So all the workers deliberation was private in itself and um they would work together and kind of have their own opinions. Um oh man and there was also the perpet the persistent memory. There was so many layers to this one. It it also it used a uh some some uh I forget the name of it but it had a

38. 16:27

    shared agentic persistent memory across them all so that it could maintain the identities of the worker nodes so it could continue and vote uh realistically it was a really interesting project and I believe that they're they're still going to be uh uh building it. So I'm really happy to see that. And then I want to give a shout out to the community vote. So, uh, when all these

39. 16:48

    projects were submitted, they were uploaded onto a voting application where you could browse them and you could see them. If anybody here wants to browse and see all the, uh, the submissions to it, you can check it out. It's at vote.nearleion.com.

40. 17:03

    Um, big shout out to the projects that did win that community vote. Unfortunately, we didn't see a whole lot of overlap on those projects as what won the tracks, but nonetheless, it's great to get that engagement and it's great to see that this community, the Near Legion, uh we really like to use the products that are coming out of it and um I think that's a really unique unique piece about it.

41. 17:28

    Okay, so there's a QR code if you want to scan and uh see the projects yourself. you should be able to watch all the videos and all that type of thing. And then finally, uh there's actually another hackathon going on right now. So, there's a there should be some QR codes around the venue um that if you scan them, you can do the agent wars. So, the idea here, if you see them around the venue, is you can scan the challenge and you can give it

42. 17:59

    to your agent. you're just going to give that link to your agent and your agent is supposedly supposed to figure out the rest. So, this is built around uh near AI market which um has a skill file that your agent can read and it knows how to interact and how to uh do this stuff.

43. 18:17

    And so, if you scan the QR codes, there's four different challenges. that link also has a skill attached to it that your agent will be able to understand and try and solve the challenge. So this can be done with open claw with ironclaw via the deployments that he showed earlier. Um I've also tried using it with open code. If anybody uses that uh you could use that too. But essentially, it's it's yeah, it's a pretty it's a pretty cool uh take on hackathons where it's not really the

44. 18:50

    human that's doing the coding, but actually your agent and how well your agent is able to run autonomously. And then that's it. So, thank you very much. Uh those are the projects. And then how about a round of applause for all the projects that submitted and won in this acton. So, Cool.