NEAR AI Ecosystem - What Did You Ship This Week? #16

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

1. 0:17

   near AI. What did you ship this week? Hello everybody. Welcome to episode 16 of What Did You Ship This Week? My name is Cameron Dennis. Uh we do this every week to give ecosystem builders an opportunity to share what they're building. Uh think of it as a bit of a open decentralized engineering standup.

2. 0:39

   Uh a lot of it is demo oriented. So if you're not an engineer, please bear with us. Uh sometimes uh there's a lot of code displayed on the screen, but I do my very best to try to get the founders and the teams to talk about what they're doing in terms that everyone can understand. So, um, without further ado, we're going to kick today off with, uh, Jay from Ping Pay. Uh, Jay, how you doing?

3. 1:04

   Wait, let me add you. Jay, welcome. Hey, Garin, how's it going? Uh, I'm I'm awesome. Thanks for having me, man. Super stoked for you to be here. Uh, just real quick, I'm going to ask everyone else to everyone else uh that is speaking, please, when you join, uh, introduce yourself, introduce what you're building, and then unpack what you're going to ship this week. And if you're going to share your screen, please be sure to only share the tab of the screen. If you have any problems, let me know. Yeah. Awesome. So, I'm Jay.

4. 1:31

   Um, I've been in the ecosystem since the back end of 2021. And the project I'm representing today is Pink Pay. Um, we're currently aiming to be the payment layer for the future of commerce with the goal of making crypto payments as simple as possible for users, products, and agents. Um, which is across kind of a variety of different payment options.

5. 1:49

   But what we did ship this week was some pretty awesome updates to the ping onramp SDK which we've made some good progress with. Um which I would like to share my screen with full let me just do so um hold that should be fine. By the way, everyone has about eight minutes for their slot. Just FYI. And I will pull you off the screen if you're going too long. But you're doing great. Yes. All good. Can you see me yet?

6. 2:18

   the screen. Yes. Yes. Awesome. Okay. So, for the um the on-ramp, it's a chain abstracted on-ramp which has a lightweight SDK for easily integrating fiat to crypto um on-ramps which where users can basically purchase near intense support currencies with fiat which is ideal for any teams wanting to add this to their applications.

7. 2:37

   Currently, it's powered by the coin ramp uh Coinbase onramp API and near intense where users can deposit fiat to receive native USDC on base with zero fees which then gets sent to the near intense network. So the user can receive their selected assets on the preferred blockchain. Uh just quickly I do have a user demo which I will run through. Um so let me just play this now too. So

8. 3:00

   this is the user flow. So the on-ramp button is here to basically mimic the popup or click from the application it's integrated within like a wallet or a DEX. Then the flow from there is pretty simple. Um so essentially once you're connected this would be automatic through the wallet flow or the DEX flow but just for the sake of the demo it's there so you can see. Um but yeah the flow is pretty simple. You just go through select the asset um that you want to receive the amount you want to deposit and the destination address.

9. 3:27

   Then you obviously then click proceed. The next step would be fill in any of your information. But this starts the USDC on base deposit through the on-ramp which you can see the wallet is waiting for currently. Um so right here this is where the USDC on base would be transferred and then you can see here once that's received the swap is in progress to the desired asset that you selected. We right now in testing have

10. 3:50

    USDC and near. Um and then once the swap is complete, which should be any second now. So just so I go ahead. Uh yes. So once that's received, you can obviously see the confirmation of the transaction and then view on the explorer or you can see in the wallet here that that's been successfully received. Yeah, go ahead.

11. 4:23

    Yeah, so I just want to make sure I'm understanding this correctly. So, um, a user is taking fiat, putting it through the Coinbase on-ramp to get base USDC into their base account and then or sorry, yeah, their base wallet address, sending that base USDC to near intents and then you're able to withdraw any asset on any chain. Is that correct?

12. 4:44

    Yeah, correct. So, the USDC deposit address is picked up in the back end. So, they don't necessarily need to have their own. that's handled by near intense. Um, so it's essentially just you directly deposit your fee. The on-ramp handles that. Intense handles the swap from USDC on base. The reason we do USDC on base is because that's zero fees. So it's like a pretty awesome way of being able to get minimal fees for the on-ramp um from fiat directly intense because intense has low fees as well. So that's a kind of super awesome

13. 5:12

    unlock that we've got through the client base API and your intents. That's so cool. So just so I kind of put this all together like you're pretty much creating a 0% fee on ramp besides the fee you guys might charge in the slippage fee but like actually turning dollars to USDC on base is free. Yeah.

14. 5:32

    Yeah. So USDC on base is free via Coinbase. So that like irrespective unless they change things which I believe they're not going to for some time anyway. um that is going to be free and then we want to be able to offer teams like if in their own business models to set their own fees on like the near intent swap side. So wallet dexes can have their own fees on top of what they want but right now it's like as low zero fees as you can get. It's just

15. 5:54

    purely on the near intent side that's any fee. I mean that's so cool because then at that point you're able to then go offer this to any ecosystem saying hey stellar foundation do you want to go get USDC into a stellar native wallet use ping pay really it's driving demand for near intense you know they provide liquidity on their native asset you're essentially pushing the the fee

16. 6:18

    structure to the foundation saying hey your users are going to be able to you know really pay as much as you're willing to provide in liquidity. So it's like this is really cool. Yeah, it's like that's super exciting and that's kind of especially as near intense grows with more supported currencies, we can then further on forward additional um additional ecosystems that way.

17. 6:39

    Obviously right now we've just got near and USDC but in the next like week or so we're going to start adding in the additional currencies uh that are supported by near intense. So then you can kind of go fiat to any intent support currency which is super cool.

18. 6:52

    And this is amazing because I needed USDC in a near native account. Like that's a there's a gap there. And so now you have an API to do this. Yeah, exactly. And we want to make it as kind of as super and easy as possible whether it be for integration or kind of the user flow. Um which kind of brings me on to the next thing which was just quickly we do have an on-ramp form um for like interest. We're going to run essentially we're going to run like a small beta for

19. 7:18

    any projects, ecosystems or anything like that that want to integrate this on-ramp into their applications. Um we're looking at probably like a few weeks time, so um two to two to four weeks we want to do kind of a small beta for projects to use this. Uh you can scan the QR code that's on the screen there if you wanted to fill that out. Um you'll also see at the bottom that there is a demo link which is on this tab

20. 7:40

    here. Um, essentially with this we want to do some feedback on the overall flow of things. So we've removed the step for the on-ramp. It's purely just you can go through the flow. You can send the USDC on base yourself to the destination address that we'll show you and then you can go through the on-ramp flow. Give us any feedback if you want to do so. Um, just use small amounts. Um, but yeah, that's kind of something that people can

21. 8:04

    do if they would like to. Otherwise, scan the QR code if you're a project team, agent, anything like that uh that wants to use a chain of strike on. Okay, this is so cool. I'm excited for a bitcoin use case. If you can get a dollars to bitcoin into a non-custodial bitcoin account for whatever fee that pingp pays plus the charges plus the slippage. Yeah. And if you can get a lot

22. 8:28

    of liquidity on the USDC to Bitcoin pair, this is like the cheapest way to get Bitcoin in an accustodial way. Yeah, exactly. And that's that's why I find it like super exciting and kind of the core focus of ping is going to be expanding this on ramp to be as streamlined as possible for the user and projects.

23. 8:45

    Dude, this is so cool. Um, awesome. Thank you. Amazing. Yeah. Anything else you want to add before we move on to the next speaker? No, that's kind of everything for me. just the beta interest form or demo feedback would be absolutely amazing. You rock, keep it up, man. Thanks. Thanks for joining. Appreciate it.

24. 9:03

    All right, moving on to Kent from Agora. Kent, how you doing? Hey, how are you doing? Great. Living the dream in. So, uh, curious, we have not met before. you're one of the few people who I normally I work with all these projects and so uh I'm a part of the house of stake so I'm quite familiar with the work that you're doing but uh for those who might not know you and myself I guess please introduce yourself introduce what Agora is and what you're doing in your ecosystem awesome so Ken

25. 9:30

    Fenwick co-founder CTO here at Agora.xyz XYZ. We're an onchain governance and uh enterprise workflow company. So we are working with Near on building the house of stake uh v1 and v1.5 v2 hopefully uh for a long time. We super excited to be working in the near ecosystem. We come from of course the the EVM side and yeah we work with some of the biggest dows on the EVM world and are really excited to be here for for near. That's I'm going to show you today some in progress how a stake v1 work. Awesome. Yeah. All right.

26. 10:05

    So, I'm going to share my tab here. Make sure I got this. Okay. Can you all see this? Yes, we can. Great. Okay. So this is what our uh so we're going to be taking a group of the house of stake through this at ETHCC. Uh honestly we just pushed some updates like an hour ago and I have not tested so this will be a real demo where I'll I'll test some stuff uh that uh that could break. Um quick tour for those who don't know House of Stake.

27. 10:41

    So house of stake really was this great proposal that gauntlet and the near foundation worked on for what governance should look like what the future of governance should look like for near and then an amazing team the fast team which I'm sure you all know um came in and built the v1 contracts in near and what we are doing as our first project with the near foundation is building a UI UX

28. 11:05

    over top of those uh v1 contracts and then are going to continue the v1.5 the v2 2 contracts and making sure we've got par there. So, uh this is what the current uh beta is looking like. So, these are all on test contracts. So, I'm logged in here. You can see these are different proposals that might be going through House of Stake. So, if I click into a proposal, of course, we don't have uh much going on there, but we can

29. 11:30

    go ahead and vote for this uh test proposal in the way you would imagine voting for your your onchain proposals. Uh you probably can't see my transactions here uh with the stream, but I just went ahead and voted um here. Let's make sure that settles.

30. 11:49

    Perfect. There you go. So, it shows here that I voted with my uh 1.2 uh ve near. One of the most kind of fun uh parts of this project has been this assets page. So, uh there's a couple bugs here. I'll let you find them. Uh but the idea is making it really easy for you to see which assets you have and how to get those into our VE near token. This is an accumulative um uh governance token. So

31. 12:16

    you're going to lock this token into a locking contract and you're going to earn vear as you're locking your near. It also works if you're locking your LI near or your ST near which will be coming. So here uh you can see that I have 10 near that I could lock up. So, I'm going to go ahead and and lock and stake some near here. So, let's say I want to lock five near and I'm going to get my VE near. So, I'm going to go

32. 12:39

    ahead and lock those. Again, you're not going to see my my transactions here. Oh, and of course, I did run into an issue. Let me try one more time with a different lock here. And I will do my five here. All right, there we go. That's looking good. If you can't see my I'm just signing some transactions here. So there I've locked these five near. So now I've moved those into a lock contract uh that is going to act on my behalf. If I go next now I'm able to then choose am I

33. 13:12

    going to uh stake and get uh LI near liquid stake or ST near liquid stake. So I'll say I just go here and I I had to deposit two as part of the lockout process. So I've only got three available here. So I'll go ahead and liquid stake these guys while I'm at it.

34. 13:31

    Wait for that to transact here. Oh, and transaction error. Too bad. This is what we just shipped. So, this will be working soon. Uh, let me just refresh and see if it actually went through. No, it did not go through. But that's all good. Uh, still a work in progress. You can see my initial locked worked. I've increased my ve near voting power and I can see how that's going to evolve uh over time as I'm the more I'm locked in.

35. 13:55

    Um, and yeah, that's uh that's our current status here. Well, we've also got our our voters page, so you can click in and see who the community is, how much veer they have. You can see your total supply, votable supply down here. And yeah, so we're racing here to get this um all done for uh for ECC.

36. 14:15

    Everything's pointed at test net. We'll be moving to mainet shortly after that. And super excited to get all of your feedback. No, this is awesome. Um just real quick, why are people locking their NER for VE near? What do they actually do with their VER? Ah, great. So, VER is your governance token. So, uh in order to participate in governance, we want to see that you're willing to put some skin in the game. So, you're actually um

37. 14:42

    locking up your tokens and you you won't be you will lose liquidity on those tokens. You can still liquid stake them to earn those rewards and there will be a reward program for those that participate and lock in governance but it's essentially using a model of a skin in the game style governance model the vote escrowve model. So uh the goal is to find really aligned near users that

38. 15:07

    want to participate. the more you lock, the more stake you're going to be accumulating as time goes on and the more influence you're going to have when you're voting on these proposals. So, if you want to really be involved in how the House of Stake is going to evolve over time, uh you're going to want to lock as much near as you're willing to lock so that that stake grows over time.

39. 15:28

    Got it. And so, but you're also voting I believe like you're you're locking this near to vote on the people who vote on these proposals or do you vote is it like liquid democracy or is it more representative? Uh both. So, um the contract support delegation. So, you can go ahead and you could decide to delegate your uh stake to somebody else.

40. 15:48

    Um you could you could go ahead and vote yourself. there will be a suggested uh endorsed group of delegates that the near foundation is is hoping you delegate to. Um but uh like in anything those dynamics could uh could change over time too. Cool. So summarize is like we're applying liquid democracy to a group of people to decide on what to do with the treasury uh to accelerate the your ecosystem. Is that right?

41. 16:15

    Absolutely. And as the as the uh house of state continues to evolve um you know more and more uh could move under governance. Um so yeah uh really exciting stuff here for the for the new ecosystem. Cool. And I'm kind of playing dumb because I'm a I'm a I'm one of the delegates. Uh and I'm super excited about this because this is sort of take two of an attempt to decentralize your

42. 16:39

    foundation. And I do believe if we're really trying to build the open web, we're going to need a group of community contributors actually deciding on, you know, how money gets spent. So, um, I think this is super cool. This is amazing work. I'm excited to test everything, uh, probably when the all the bugs have been kind of worked out, but really appreciate you coming on in this sort of early stage because, uh, you know, it takes some guts to to show off a a product that is not quite tested. So, really appreciate that.

43. 17:05

    Absolutely. Every week that's what we're doing. Thank you for having me. Cool, man. Yeah. Thanks for joining. All right. Up next, a familiar face to the show. We have Charles from Pure Folio. Charles, are you ready to rock? Yes, sir. All right, welcome back. Thank you. Yeah, thanks for having me. Of course. So, what's going on with Pure Folio? What's new? Yeah, last time I was

44. 17:26

    here, I think we're really focused on uh sharing this survey that we sent out to get some feedback on what preferences people have with onboarding, how they feel about giving trust to autonomous agents. And we built that and kind of took from that and built that into our onboarding flow. And over these last couple weeks, especially this past week, we've really been focused on like, okay, what do we need to do to be able to move beyond this internal phase that we have

45. 17:50

    where everything we're shipping? Right now, it's only in use by our team. When can we move this out to an alpha stage? And at what level of features do we want to make available for cryptonnative users, the early adopters that we would want to invite to be alpha users? And then how do we transition from that to our ideal instate where we're really targeting mainstream consumers and it's a very different uh experience. So we

46. 18:10

    had a design jam and worked on features kind of moving towards that and then spent a lot of time kind of testing out the complete breadth of features that we have remaining and putting together scaffolding for how we're going to implement these deepening our integration with your intents and planned out our entire uh roadmap taking us through the next couple of milestones through September when we plan to

47. 18:33

    complete uh the overall app. So I want to share today just showing the details of what that but hey Charles real quick for the people who might not know what Pfolio is and this might be their first time tuning in can you just give a quick TLDDR on what exactly you're building?

48. 18:47

    Sure. So, we're building an application to make help mainstream consumers feel accessible to including digital assets in their savings and investment practices like stripping away the tech, stripping away the wallets and bringing everyday utility to um a AI agent assisted acquisition of assets and then make it easier to spend those every day with smooth offramps.

49. 19:10

    Cool. I I see a collaboration think pay in the future potentially. Yes, that's right. Yeah, actually tested that out a bit and u looking forward to to integrating with Ping Pay again as that as that project moves along. This week we we focused on trying out the the native Coinbase on as well. But and now I definitely have a need for leveraging ping pay to really, you know, just remove some friction so we can go

50. 19:33

    directly to USDC on year and make it easier to get where we want our users to go. Uh but this is our public roadmap and last time I was here we were probably referring to the application a bit a bit more stealthily and we've kind of landed on our name peer folio really grasping on the social aspect. You're doing this with your peers and you were using our agent to help people who especially people who are not techsavvy

51. 19:56

    or interested in crypto um to have some guidance towards including different sets of tokens um and and acquiring them with with guidance and doing that in a social way. So, we're this is what we're working towards. Um, we're kind of into or leaning towards Q3 coming up next week and we're just expanding to build towards having an alpha launch where we can make the application available outside of our internal team to cryptonative users to get some feedback from you, our audience, and and help us

52. 20:22

    make this experience as solid as it can be. Even though we will be targeting the enduser experience for uh uh normies for people who are not interested in crypto who maybe have Coinbase but don't use it because it's overwhelming they don't know what to do or who people who really would take advantage or benefit from an experience where the wallet aspects is completely um hidden from the UI as much as possible. Uh, so that's what we're

53. 20:46

    going to be working through this next quarter, expanding our um expanding our support for multiple tokens, really nailing in the user experience and kind of navigating this line of how things work on desktop versus the experience that we want to provide for for mobile.

54. 21:00

    Um, and aligning these different road maps of things we want to build over these milestones to how we can group them in such a way that we can get have a more efficient security audit and have a really longer lifetime for after we do our audit. on the set of features that are included there as we work towards our general launch uh around uh September and further integrating different features there. So this is like the bird's eyee view of everything.

55. 21:24

    So excited to get this included on our near ecosystem roadmap and we share even more details on things that we're shipping on our Twitter at the Nebulabs. So see more updates there and I will be joining these in the future when we have more more demos that we can share a bit more kind of hands-on feel. But yeah, just wanted to jump in real quick and share our road map for now. Man, I love a team that publishes their roadmap publicly. Uh, good work. Curious on the

56. 21:50

    feedback that you got in your survey like how how comfortable are people with agents trading on their behalf. Yeah, that that was interesting. I should have included that slide here. So, the there was like two camps that we saw. There was a group of people who were comfortable with we basically had three levels. No autonomy, no trust. I want to manually approve everything that this agent recommends me to do. Um, and the

57. 22:15

    next level was some autonomy. They wanted agent recommendations, but they wanted to set limits on the percentages of their tokens that an agent had access to. And then there's complete do everything for me. I'm happy with the risk and have the agent make decisions on what tokens to trade within the parameters that you set for conditions like when uh your entire portfolio reaches reaches a certain USD value. let the agent decide on its own which tokens

58. 22:39

    and what quantities of each to swap to USDC to help realize that capturing that USDC value and we found that most people landed in that some autonomy place and then we also allowed them to further customize by changing different different aspects of the permissions. So for example, people would choose some autonomy, but then they wanted to enable the agent to continue to trade over the course of the future even after doing a

59. 23:02

    accomplishing one swap that met their conditions, but to continue to trade and recommend um changes to their conditions so that they can adjust their goal as their portfolio changes over time. Uh that was the main area that people landed on that some autonomy and then there was the no autonomy but then choosing some options where they did want to see what recommendations the agent would choose but users wanted to receive a notification to be able to

60. 23:25

    trigger them on their own. Um those are the the how people kind of landed in the areas what we the selection areas that we provided. But one of the most interesting pieces of feedback that we got in addition to like wanting to see security audits and how you approach that one was a suggestion of I want to have a sandbox environment where I can set up all of these conditions interact with the agent on test net and then see it work end to end and before moving on

61. 23:48

    to the full autonomy. Um so that was really interesting and something that we've added to our road map that that wasn't there before. So really appreciate the feedback. Yeah, that's awesome. And it kind of reminds me of uh like I know of like some schools that do like uh sort of simulated stock trading games. And so before like you give a kid money to to lose, you essentially give them simulated crypto or simulated

62. 24:11

    stocks. And so I I I love that test net play. That's a that's really feedback. Yeah, I like that. I like that idea, too. Cool, man. We'll be back. Yeah. Thanks for joining. Hope to see you soon. Cheers. All right. Next up, we have Bill Shei from Pond. Bill, how you doing?

63. 24:32

    Good, good. Yeah, thanks for inviting me. How are you? I'm living living the dream, man. Um, quick for those who might not know what Pond is, please give a quick overview of what you guys are building, uh, who you are and your role and then what you shipped this last week.

64. 24:48

    Yeah. Um, totally. I think that that's actually what I want to present right now. So first of all you know I'm uh co-founder of P billshi and you know I what what we have shipped uh is what I want to talk about which is you know what we do because it's a whole chapter for us whole new chapter for us so let me share my screen.

65. 25:17

    All right so yeah let me see if you you know cannot hear me well or can I see yes we can we can hear you. Awesome. All right. So, yeah. So, what we shipped this week or in the in the past two weeks is a is a it's a whole new uh is is like a total revamp of our platform.

66. 25:32

    So, we're now a uh all-in-one platform for AI builders to deliver their dreams or build their dreams. So, this what you see here is the um is the platform architecture. So, things everything starts from uh hackathons. So previously we were known as a model competition platform. So we have worked with many projects including uh you know Ethereum Foundation, Vitalic, Gitcoin, uh Coinbase etc. to host AI model competitions. So now we totally expand

67. 26:04

    on that and build a hackathon platform to be you know more general more broad for AI builders and then winners from the hackathon platform will be uh will be able to enter our markets. So, so um in a nutshell, the market is like a you know a token launch pad where the projects launch tokens uh and use the token to um you know to get the resource they need initially to build but it's not just another you know memecoin launchpad. So actually we're actually

68. 26:35

    very selective. We want a uh good projects, high quality projects to create a real value behind those tokens, right? That's why we have the hackathon as a screening phase in the beginning, right? And then uh and um and secondly, we have uh we encourage the community uh to be engaged from day one. Uh so so you know, so so we want to foster this genuine community project relationship

69. 27:02

    in the very beginning, you know, even in a hackathon phase. And then thirdly um we have adopted a very uh I think a very new token unlock schedule or mechanism. So you know unlike unlike you know the the memecoin launchpad which is you know completely fair or or random here uh we take a approach suggested by CZ you know from finance um that initially the token to the project team are locked until they deliver their products. So we want to make sure you know this is not you

70. 27:37

    know a rock pool game. So the project actually to work hard to deliver and then start their token unlocking schedule and then the actual unlocking is based on you know price milestones, time milestones so that um everybody you know all the parties are aligned on the long term right so the the project team you know the community the market are all aligned on the uh on the value of

71. 28:03

    those tokens in the long term. So, so those are the two parts you know are just the first step and then after that we actually help the builders to build because we're you know we're supposed we strive to be a allin-one platform not just you know a launchpad. So here we're building a AI studio which is um which provides things like you know GPU computing of course in collaboration with many of the DPM projects. Uh we're

72. 28:32

    also providing like agent tooling all this kind of things uh data sources data integration to help those projects to build their uh their product and then once they finish building um it's time for them to use our accelerator to help them uh distribute their you know their their product to do marketing to do operations etc because what we think we feel like you know in this uh new AI era like building itself um it's not that hard anymore, right? So the the barrier

73. 29:04

    the difficulty is really lowered by the uh by the you know um AI powers but then um what's what's still challenging to many builders is you know where where do you distribute how do you distribute your product you know how do you get the initial users how do you do marketing all this kind of things so we have built this accelerator to take care of that then um this accelerator will also help

74. 29:28

    them to deploy their products to our integration hub if it's for developers, right? If it's a API or things like that or to our app store, you know, if it's a consumerbased product, then this is like the whole end process. Um, we're trying to, you know, build on our platform and also for the builders. Uh, okay. So, so that's I think a very long answer to the question, you know, what is Pound? Uh, it's a whole new pond right now and and and we just shipped this week. uh is uh let me see if I can switch my tab. Okay.

75. 30:08

    Yeah. So we so this week we just launched the first three projects uh on our you know platform. So Omnic style critic refinery and and it's um the result is very promising like you know the first project uh they are their you know their their seating phase ends in less than 20 minutes. So they reached their target in less than 20 minutes from community contributions and for the you know for all the three they all reached their targets in less than I

76. 30:40

    think 12 hours uh on day one which is uh very good news. So now they're in the building phase. Uh so we were you know dedicated to helping them fulfill their dreams, deliver their target and also we're building our whole platform to power them through their journeys.

77. 30:57

    So that's uh that's what I have today. Bill, this is super exciting. It's uh seems more of like an ocean than a pond. You're doing a lot. Um it's it's pretty significant. Um kind of what what led you what were the learnings from like doing model hosting competitions to hackathons? Like what were the interests from the AI engineers that were training models and like what led you to then go focus on hackathons?

78. 31:24

    Mhm. Yeah, that's a great question because we, you know, we we believe that um you know, first of all, we we don't limit what type of projects people will will build. You know, it could still be a be a model. It could be a agent. You it could be a totally something totally different. Um the reason we want to make it more broad is is that um we think that eventually, you know, if you want to build like a one million ARR company

79. 31:46

    or whatever, uh you need like a consumerfacing product, right? So if you just have a model, maybe developers will love them and it's a viable product. But if you you know want to really build something um that's uh like you know consumer friendly then a model might not be the best uh you might need actually wrap something around it like you know most of the application today are repres

80. 32:12

    for example uh it's not like you know totally new but you need to build a product around it. So that's why we made it more broad. Got it. Cool, man. Well, thank you for joining. Is there anything that you want to uh say before we move on to next guest?

81. 32:28

    Uh, no. So, yeah, feel free to check out the website and if you you are interested in hosting hackathons, you know, feel free to reach out to us as well. Awesome, Bill. Thank you for joining. Thank you. And up next, we have Arto and Matt from Asimov Protocol. I'm gonna bring you guys up to the stage.

82. 32:47

    Matt, welcome. There's your screen. How's it going, Arto? Hey, camera. GMGM GM. So, last week you presented a ton of stuff. How the hell did you ship even more stuff than the last week? It's like, yeah, Positron, Matt, I think you're muted.

83. 33:12

    Sorry. Yes, unmute. Right. Okay. Um, so yes, we did ship a lot of stuff last week and um, probably even more uh, kind of this week. So, perfect segue from hackathons. Okay, so for those that don't know us, Azimoff is all about uh, verified knowledge, right? Verified uh, trusted knowledge uh, especially around AI and LLMs. So last week we talked about Posatetron which is our first uh

84. 33:39

    kind of enduser product that allows you to actually use an LLM and presents an MCP server that can interrogate your own personal data. So your Telegram, your email, all that lot. That's all that's all great, but trying to do demos of that and research on that is a little bit tricky because it's personal data, right? And the whole point about Azimoff is that it keeps that personal data local. So we needed something to kind of

85. 34:03

    test okay more public data. And speaking of hackathons um we then took the opportunity to use a Google hackathon actually two Google hackathons around AI that were going on uh to build something. So as you know I'm based in Barbados. My eldest daughter Katie is back in the UK. She's an aspiring developer. Um moved over from education to tech a few years ago. And so we took this opportunity to work together on building something using the concepts from Azimoff. So we created a thing

86. 34:38

    called youam. And what you hear them does is it takes all of the hours and hours and hours about 8,000 I believe it is hours of video recordings from the Barbados Parliament sessions and builds a knowledge graph of them using all of the concepts that we are doing in Azimoth and gives you an AI in which you can then uh converse with them. So you can go to the site it's live at the moment you.com and you can ask it a question. So, um, what's happened recently about the economy and it will

87. 35:11

    come out and give you information here and even a direct link that you can click on that will take you to the YouTube video to the exact point in the video where the point was made. Now, one of the great things that we sort of learned out of this was as you can see here when I've asked about recent stuff and it's come back and it's talking about 2023. So one of the things was is that it is hasn't got that good a appreciation of sort of temporal

88. 35:35

    awareness and also I can actually there's a sort of a hidden command here where I can bring up the knowledge graph behind the scenes. So the next iteration of that we are actually using something that's going on in the background here which is doing dduplication of entities in the knowledge graph using AI. So it's going through here and saying that uh for example independence is the same concept as constitutional independence.

89. 36:02

    Previously it treated them as two separate things whereas here it's saying actually no those are the the the same concept together. Right? So if I go and ask exactly the same uh question on the uh version that we have now in in in development uh then it'll come up and show here um as it works away here much more recent stuff. So it talks about 2025 and like I said I can click on there and it'll bring you straight up to the video in which they talk about it.

90. 36:37

    And not only that, the knowledge graph is much richer now and much better connected because we're dduplicating. So we've gone from an average of about one and a half relationships per entity to about three and a half relationships per entity on average here. So that's what we've been working on. Launched that this week. Uh actually launched it on Father's Day on on Sunday, Father's Day

91. 37:01

    in in in the UK uh with my daughter. So that was uh that was really cool. In parallel to that, Artto's been working on a whole bunch of other stuff. So I will pass on to him to talk about that. Thank you, Matt. Indeed, uh the the way we split efforts is that uh Matt is our pioneer. he's out there in the jungle with his machete and and uh you know making that path for us and and then I and our engineering team we follow and

92. 37:32

    we lay down the road. So so that's maybe as a metaphor useful. Now um we have a big week coming up next week. Uh we are going to uh roll out the Asimo platform. So I'll give a bit of a sneak preview uh just for our community on on this stuff.

93. 37:51

    Um there is a um so there's a GitHub, there is a a Twitter handle you can follow that's platform uh there's a website uh in the in the works that will be uh available shortly. Uh hope that comes through and um we have began to roll out things like for example the specifications that underly uh the protocol and the platform. Um, so this uh this is already uh live at asimo-specs

94. 38:24

    github.io, but not all the not all the specs are uh there except the stubs. Um, but some some specs are already uh fleshed out. So this the specs will uh describe um everything that we're doing and why it matters. Um, I could show here uh a bit of a a brief live demo uh preview since live demos are the name of the game. Uh, so let me just get the terminal um hooked up here.

95. 38:56

    There we go. So here we have the um Asimov uh uh CLI and um using this we can we can say for example you know what are the what are the modules uh that are installed that are uh enabled or disabled and we can go ahead and um we can install for example the near module here. So the the near module gets uh fetched from the binary build on GitHub.

96. 39:24

    Uh we can install other modules like um you know data data providers um that uh would uh would let us pull a great variety of uh data sources um and uh then we can uh uh then we can basically uh do what um you know we talked about last week came which is we are resolving um particular URLs into which module will handle this URL based on the uh URL protocol. So in this case of course near but also also based on things like URL prefixes and patterns.

97. 39:59

    So with this we can we can check uh uh that uh you know we have modules to handle you know some some variety of uh of data sources here. Um and in one case for example the the Google search results if you wanted to get those we have even two modules that can um two two that can handle that is for me I think you're good I think there was feedback on that end second.

98. 40:40

    Bad time for a styling glitch. Okay. Am I back? Yep, you're back. Cool. Cool. Uh all right. And um uh once we have modules installed, that then means that we have uh um we have facilities like in this case we can just uh see what we have. We have the uh near fetcher here and we can put in uh put in that and pull in you know block data from uh uh from near and and similarly for anything anything else that we might uh might wish to to obtain as a knowledge graph.

99. 41:15

    So so that's a that's a bit of a sneak preview. Uh this all will be rolling out for everybody next week. Um and yeah, how we doing on time? I I can always present more, but uh maybe Yeah, you're you're good. Uh you have about Let's see. You have a couple more minutes. Two more minutes. All right.

100. 41:34

     All right. Oh, see you. It's kind of funny. So, Matt, uh quick question. So, assuming you have like higher fidelity knowledge knowledge graphs, like what does that mean for model builders exactly? um basically much more accurate information for the LLMs to actually reason over. And that's that's the the big problem is that in all of these cases with these knowledge graphs, you're doing a sort of an an MCP uh rag type sort of um fetching of data,

101. 42:09

     bringing it in, and then giving it to the context. But you can't fetch the entire thing. I mean, in that example I showed you, we have nearly 700,000 statements from that from those recordings, right? Um, and like I said, there's no way you can pull that all into a uh an LLM in one go. It's it's just going to blow it up. Y, right? So, um, it's about fetching the relevant pieces and pulling them in. And the better the knowledge graph to start with, the easier that is to do and and return that to the LLM. So like I said,

102. 42:42

     the whole dduplication um you know it's it's something that's some of this stuff has been studied in knowledge graphs for for a long long time. Um but actually having the ability to do this now with LLMs makes this kind of stuff a lot uh more accessible now.

103. 42:56

     Um you know so we're actually prototyping you know and actually dog fooding going from like raw data through a knowledge graph to an LLM which is what Azimov is is kind of all about. Um and having a like a lab in which we can sort of play with that data. Got it.

104. 43:11

     your one, just so I'm super clear, one of your main like target partners are model builders. Like you want them to use the Asimov knowledge graphs or add it into their existing workflow for RL or something to further improve their model if it's getting demand like it's going to improve it's going to want to improve itself.

105. 43:30

     Is that the target market? That that would be you know ultimately that would be amazing if like when you go on to you know claw.ai AI or you go to, you know, chat GPT that one of the tools down the bottom is ASMO. That would be that would be perfect. Um, but you know, we realize there's a long way to go to get to that point. And so, you know, we're working on a sort of an incremental strategy with that. And so, one of the things is this tool um, Posetron that lets you uh,

106. 43:55

     work on on your own personal data as well. So, we're kind of approaching this from several different angles. Got it. Cool. Arto, welcome back. You are muted. Thank you. Thank you. Um, yes. So, maybe just as a as a call to action here, uh, you can sign up for the, uh, Positron beta at get positron.ai and, uh, we'll, uh, we'll ship you an early early version uh, sometime next month.

107. 44:23

     Amazing. Well guys, thank you for joining yet again. Um, hope to see you next week or the following week and we will do our best to start driving more attention to Posetron. I actually part of my widget ship this week is I hired two interns which is great. So they're going to help me uh manage the accounts and everything and so hopefully we can drive more more signups uh to everyone's alpha tests. Yeah Cameron the the

108. 44:48

     actually the point here of everything we are doing is to uh supercharge your interns. It's not to replace them right. So all about intelligence augmentation uh more than AI as such and tell them tell them to go to getposetron.ai AI and they can sign up for the wait list.

109. 45:05

     Amazing. Thank you guys. And somebody who is also going to help supercharge my uh new interns is Zahed from Post. Zahed, how you doing? Hey Kevin, I'm doing good. Thanks. Thanks for the invite. Of course, man. Um this is actually part of one of their first assignments is to uh write entire threads about this exact uh what did you ship this week? So, I actually mand mandated them to start looking into post and start using it. And so, super

110. 45:37

     excited for you to give the update on what you've been working on. Uh, it's been a few weeks and I know you're constantly shipping. So, what's what's new, but for those who don't know what post is, quickly give them a quick lowdown on who you are, what post is, and then what you ship. Okay. So, hey everyone. Uh, I'm Zahed, founder of JSU, and today we I'm going to talk about POST. So post is a AI u social media

111. 46:01

     manager for LinkedIn. So think about u if you're a founder or if you have a thousand things to do like uh most of us. So then post is a way to kind of a delegate some of your task to a agent that kind of a uh write social media content and post it for you consistently in LinkedIn. So we are starting with LinkedIn but in the future we are uh probably move to X and other platform.

112. 46:32

     Uh and today uh my plan is to show one of the feature uh that we are building more for housekeeping. So this uh today's demo will be more for the founders and the builders who want to kind of a who have a project now uh trying to get more people trying to attract uh uh kind of a more users. So there are many different way and most of us are kind of a engineers. So we don't uh we don't

113. 47:04

     know much about marketing but like a uh one of the thing we are doing is like we are exper experimenting with different technique that a professional marketer will do and then uh kind of a baked into our product. So that's something I'll show and I hope this will help uh everyone. So let me share my screen uh share screen.

114. 47:29

     I'm going to share everything. So, okay. So, uh most of you uh if so quick demo. So, this is post and if you want to create a uh LinkedIn post, you come here, you give a topic. There are uh writing styles. So you can create your own writing style or if you have a preferred uh person and you want to kind of imitate their writing style uh then you can add them uh in in this uh you can create your own writing style from

115. 48:02

     here. So think about writing style is a adding a persona to your writing. Uh and then if you give topics then click generate it will generate uh uh a LinkedIn post for you. And if you have no idea then there is a small like this idea icon where it will give you some basic idea of like where you can get started and these are based on uh based on the uh brand kits. So brand kit is a way to train a post how it should write about your uh LinkedIn post

116. 48:32

     and who it should target, which industry it should target and your brand kit if you have a brand kit uh uh for your company. Uh but again today I'm going to talk about uh the growing the uh kind of the your project. So there are many ways to grow the project. One of the thing that we found uh is kind of instead of hiring a salesperson, why don't we incentivize people to share uh and kind of share this project with uh with

117. 49:04

     others. So you probably a lot of you probably know or use some sort of uh referral program. Uh but unlike most other team, we always love to build our own. So we end up building our own kind of a referral program. So uh like if you are if you go to your uh kind of a uh settings then you will see there is a referral code and then this is the referral code uh if you share with others and when they sign up so they get one month uh post free and you

118. 49:38

     get one month free and uh you can see uh and we internally track all these things. So this is something we built internally uh to figure out or to explore how to build a great referral program. The other thing we did uh so referral program are great for uh kind of a um user sharing with other user but sometimes some of the product you need uh micro influencer or influencer to uh share your product and then lot of cases

119. 50:11

     influencer are not happy with just getting one month free. Maybe there is a monetary incentive. uh for that uh we added something called coupon management where coupon management is basically uh let's say if I want to uh I'm a influencer and uh if I want to share uh share this link with with this community and want to track like and and allow them to get let's say 90% or or 80% discount then I'll be able to kind of create a coupon and then internally

120. 50:46

     we'll kind keep track of how many people are uh kind of signing in, how many people are subscribing. So this is also a great way to kind of a u people who are managing let's say their uh their intern and then they want to kind of a more quantify like okay so uh they want to derive let's say your goal is to drive 100 people to this platform and then now if you if you create a a kind of a coupon and then let them uh share

121. 51:18

     this link and then you will be able to know and track like okay how many people are kind of coming into the platform. So, uh if any builder want to kind of a uh learn or interested about kind of a how do I create a referral program in in our platform uh just uh send me a message.

122. 51:39

     I'm happy to kind of a uh uh sorry there is a uh alarm testing. Yeah, man. You should go. Is it testing or is it Is it a testing? Yeah, it's a testing. Um I can assure you it's a it's a testing. It's since morning they're doing it.

123. 52:02

     Okay. So uh so this is like a uh for referral program. Uh, and then the next thing I'm Man, God speed. I mean, I'm assuming you've been dealing with this all day long. So, yeah. So, okay. So the last thing I I would say like what we are building is um I think I discussed with you uh Kevin so can you hear me or like yeah I can hear you it's a little distracting but I think it's more distracting for you than for me okay so the thing that I'm uh

124. 52:45

     kind of a we are working right now is uh we discussed with you Cameron where uh if you are a organization and when you post something then you want a predictable way for other in the organization to share amplifying this message. So we're kind of a building two things. One is a AI simulator where before you post something you will be able to see like how it will influence different group of people based on the

125. 53:14

     persona because the AI will have agent will have a different persona based on different profile and then you get an idea of like a how AI kind of a see your post. The next thing will be if you have a team when you post something and they will get a notification and they will they can delegate but like a the point is they will be able to quickly comment and likes. So a lot of time what happen is

126. 53:45

     like even jutsu we post something our people don't get a message in the LinkedIn so message doesn't really amplify so it's a way for organization to amplify their message without paying so much money to LinkedIn so that's something we are working on um yeah it's great I'm I'm a huge fan of like community or like team marketing is I'd say something that's really important ecosystem led in the you know the example here. But this is super exciting. Zahed. Um is there anything that you want to close with?

127. 54:19

     Uh yes. One thing like if you want to use um just go uh this link and kind of a oops um so if you see this link yeah so scan this link uh to sign up for post and kind of yeah give us feedback. Awesome. Thank you. Well, thank you as always. We'll have feedback for you. And that concludes this week's episode of What Did You Ship? We are going to be back next Thursday, 11. Actually, wait, next week. We're actually going to uh

128. 54:57

     move it to Friday. So, we'll need to talk to the team about that. Um, but I won't be able to do it next Thursday at that exact time. Gonna have to postpone it for the following day. But, thank you all for joining. Uh, we do this every week. Uh, my name is Cameron Dennis. I help lead integrations and partnerships at Neurai. If you are interested in contacting me, uh you can message me on Twitter directly or on Telegram, Cameron Dennis, no space. And uh yeah, hope to see you next week. See you everybody.

129. 55:24

     Thanks for watching. Subscribe to Near Week.