NEAR AI Ecosystem: What Did You Ship This Week 6

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

1. 0:14

   near AI. What did you fit this week? Good morning, good afternoon, good evening to everybody. It's good to see you here. Um, Frederick, do you mind making me bigger if possible? Um, this is the sixth episode of what did you ship this week? Uh, my name is Cameron Dennis. I am just here to quickly introduce everyone and then I will be giving Near AI's update at the very end.

2. 0:39

   Um, just for context at Nearai I'm responsible for all uh external integrations and partnerships. And this is just something we run every week. It's essentially a decentralized ecosystem standup. Um, and the whole point is for people to share what they built. It's pretty straightforward.

3. 0:54

   Everyone has around six to eight minutes. I pop up on the screen uh when you have about one minute left. If you're going to share your screen, please only share the tab that you plan to share. We've been through this before. Um let's not make that mistake again. So, uh without further ado, I'm just going to quickly introduce the the agenda for today. uh given a quick intro as I'm doing now and then we're gonna pass it off to Mohamd from Growthmate

4. 1:20

   and then Nico from Nillian, Zahed from Jutsu, Andrew from Doppelgangers, Peter from JS and Rust, uh and then Jivvon and Ulana from Instant Apply and then at the very end I'm going to be talking about some of the stuff that is cooking on the Near AI side. But uh without further ado, Muhammad, I'm going to bring you up to the stage. Uh you have about eight minutes to talk to everyone

5. 1:44

   about what you shipped this week. Um just want to quickly flag introduce yourself, introduce what growth growth mate does and then talk about your your recent developments. Okay, awesome. Hi. Um yeah, so I'm Muhammad Ali from Growthmate. Uh we build recommener systems for web 3 users. Um and so what we are trying to solve is the discovery problem. um at any given time there is so many cool things happening in web 3 and it's

6. 2:10

   usually hard uh to stay up to date with everything that's happening um and so here let me say maybe I can share my screen and I can show you uh how does this work let's see um I guess Now you should be able to see my uh browser, right?

7. 2:41

   Yeah. Okay. Awesome. Um and so it's usually like traditional that every ecosystem has uh like um like some kind of uh directory where they list all their projects. Um and it's usually like um great to find out which projects exist. Um but then it's limited to just like giving me a project logo, project name and uh some short description and for me as a user knowing that okay this is a DEX or this is like you know linear I can stick but

8. 3:14

   then it doesn't expose the opportunities from within um and so it's just like okay good to know that the DEX or lending platform is there um how can it help me? Um and so we go a step further and uh we have our own bots that collect um all opportunities from within uh the projects. Um and so for instance here you can see we have like um by the way we are multi- chain um but for near yeah you can see that um it's aware that from rhea you can provide bitcoin liquidity

9. 3:46

   today um you if you like uh you can lend your SDC for 7%. Um you can do like a bunch of stuff. You can trade me coins here. This is how much you would get for lending your die. Um you can earn points on mistake. Anyways, we expose the opportunities from within um each project. Um and so we use this to uh provide personalized recommendations to the user and to tell him like hey uh based on your onchain data uh here are some cool matches. Um

10. 4:17

    and so we recently um added this to uh as a near AI agent and I think now you should be able to see uh yeah our agent it's under a growth mate. Near discovery agent um and so for instance the agent is aware of the user's um transaction history um in like a human readable format. So it knows that yeah um I sent um some uh tokens from growth mate to splut um yeah it knows I received some tokens

11. 4:50

    and yeah it's like aware of my transaction history. It's also aware about uh my uh portfolio holdings. So um like I have some near um I have some USDC um and I don't have any D5 positions uh in this uh specific account. Uh and so based on this data, it's able to give me like cool matches from the opportunities here. Um and so for instance, it uh like saw that I have both USDC and near. And so the first recommendation was hey by the way you can get like really high yield by lping

12. 5:21

    into the USDC W near pool on uh REA. um or because I transacted with the Shih Tzu DAO uh I know that okay uh it could be relevant to go and uh trade meme coins on meme cooking and yeah this is uh basically it um I expect this to be like really interesting to um other agents um because many like assistants would be interested in providing such uh discovery features to their users um and

13. 5:53

    yeah I would love to know if you have like um any questions Oh, I think one last thing is you can you can configure like here by by creating on on our platform and so here you can have access to some filtering options. So you can say I want I want to protect users from gambling. So you can just uh like um black gambling and so all gambling things will not appear in the first place. How do you define gambling?

14. 6:25

    Like that's such a you know a lot of a lot of crypto is debatably gambling even if it's not you know traditional gambling. So um how do you how does the agent actually define that? Yeah. Yeah. Well well there is there is definitely measures that go in here. Uh but from our point of view we always try to uh narrow down like to the closest category. And so for instance, if something is a DEX or BP, then it will

15. 6:56

    be labeled as BP or if it's a memecoin trading, then it is it will be labeled as memecoin uh trading platform and so so gambling would be you know literal. Got it. Do you manage that taxonomy like of of all and everything? Yeah. Yeah. So, so we we are getting the data set data set from providers and and merge them into our own schema and whenever we see that we need to create a new

16. 7:35

    category NF subcategories. So that would be like AMM per mecoin trading platform. Um yeah, super cool. Awesome stuff, man. Um, what's the best place for people to learn more about what you're doing and stay updated? Growthmate XYZ. Uh, also on Twitter, Growthmate_XYZ.

17. 7:57

    Amazing. Um, are is there any final thoughts? Uh, the parting comments. Well, please reach out if you think that, uh, this could be helpful to your project. Awesome. Who's your target market exactly? Um for the near AI agent, it's mainly other um agent builders uh who want to expose this uh through their assistance.

18. 8:17

    Um we also have like a standalone chat interface under chat growth made XYZ. And here you can see everything is like more retail focused. It's uh you know the formatting and display is more refined. Super cool. Awesome. Um I'm going to DM you. I have a bunch of questions uh that shouldn't be asked in public. So I will uh DM you. And next up we have Nico.

19. 8:39

    Thank you, Muhammad, for joining. Hope to see you soon. I appreciate it. I appreciate it. Yep. All right. Um, Nico, can you hear me? Hey. Hey. Yes, sir. Yes, sir. Sweet. Um, it'd be awesome if you can quickly introduce Nillian, yourself, and what you ship this week.

20. 8:58

    Yeah. Uh, yeah, my name's Nick. I've uh been work with working with Nillian for about two and a half years now on the engineering side and we built a whole bunch of really cool stuff and um primarily what Nillian's focus is is privacy-enhancing technologies. So what that means is uh creating composable solutions that can work well together to add privacy and other you know secure as aspects to your guys' project as builders. Um so one of the projects I've

21. 9:30

    been working on uh lately is on the agent side. Um we have um different initiatives happening with um like TEES and um secure inference and and things like that. So stay tuned for those kinds of uh projects and you can always look up um what's happening on our uh GitHub page and we have a super active community where you can learn more. Um today though I'm going to talk about the secret vault and um let me share my screen here.

22. 10:03

    See it's giving me options I will do I'll start with this one over here. Okay. So secret vault um what our secret vault is it's a it's a distributed database um as a service but it's distributed amongst trusted um you know infrastructure um runners to like each one is going to contribute hosts so that they can uh together you can compose a cluster and your cluster would be you know m of n or you have um certain number of nodes that you select

23. 10:43

    from these different providers that have particular characteristics that you like, for example, geo um latency and and things like that. And so you can pick which operators are going to be your um your Nelian secret vault. And then when we're um creating you know when we're encrypting um and saving data to those hosts um they get secret shared across the different hosts which means

24. 11:08

    that not one of the hosts um has data in their on their own to be able to they can't view it like even that that operator themselves can't actually expose the data leak data um without total collusion from you know the um enough of those particular parties. So that's a a a really cool uh thing and what we think that this is um valuable for agent builders is that it gives us a way of uh using distributed storage or

25. 11:40

    and or using a a database service that's distributed that's uh web3 friendly and and very much focused and um to to provide you with with some extra privacy enhanced um technology there. So the demo that I wanted to show today, I was using the um the lench chain um extension that that the um that Near team has built um to wrap around the the Coinbase agent kit. And so what that's done is it's given me this orchestration layer um in Near to be able to activate

26. 12:19

    the the lang chain workflow. And that gives us um some really cool things in the agent. And I think I need to stop my this share and start another one. So let me do that really quick window. We'll do this one. Okay. So on my screen um I'm just going to walk you through some quick codes so that you can see what I think is the ease of using these these patterns of the of the agent kit um and the or I should say the lench chain tooling that is um created that's going

27. 13:01

    to be an interfa like an agentic interface maybe a low code maybe vibe coding interface to using our secret vault. So, um, let me close that guy. So, everything's pretty boiler plate plate here, but I'm going to show you where I started was just by giving this prompt where um I'm defining what that Oh, let me back up. What am I doing? So, I'm creating a trader. So, maybe it's I

28. 13:31

    don't know if it's similar to Muhammad, but I was really interested in what he was doing there. So, I created a a fully autonomous agent that's going to take a daily report that's generated from a scraper that I made and it's going to um pull like I'm using the secret vault to store my trading preferences. So, it's got my risk tolerance. It's got any of my, you know, information. This is the

29. 13:55

    what you're seeing on screen here is how I'm describing what that risk profile is. And so, this is my schema definition. All it is is just natural language, right? So you're able to describe I say create a basic schema in the secret vault and then I describe what that schema is. Now the secret vault like what what the schema actually is. It's like a MongoDB's style schema where it's a it's a JSON schema and the

30. 14:24

    way to designate which of the fields are encrypted um has some extra like uh structure and so this agent prompt like or this tool will handle that for you which is really nice. It'll it'll decide based on how you describe it whether or not it's private. You could say this field is private. You could also say this field is financial. This this you know like whatever kind of terminology.

31. 14:48

    There's some built in to the tool itself to do that for you. So after that schema is created then you could do different operations on it. So in my case here I say you know here's here's my risk strategy and I'm saying that I want to be an aggressive trader. So the aggressive trader I also described in these terms so it knows what I mean by that. Right? So if I want to update my my profile to say that I'm balanced or that I'm you know conservative, I can

32. 15:18

    easily update that record and make it so. Then the interesting thing happens. Um let's make this a little more readable. So here's my agent task. My agent task says fetch the data from the the risk profile. So the the agent is already wired up to the tool that knows how to find my risk profile and decrypt it and present it into the context of the um the inference. So it just handles like this secret data. I don't want my

33. 15:49

    profile stored locally. I don't want my profile accessible to others. Um I don't want them seeing because maybe I'm a whale. I don't want them seeing what I'm doing, right? And that's my that's my own private I you know my my ID u my desire there. So the the agent context here is it's just all um prepared with the data. So now I'm I'm just giving it kind of instructions on how to work with that profile, how to use my uh crypto

34. 16:19

    report to create a strategy. And then it's executing trades. And you could see here, I mean, I've been running this for, you know, several days. Um, I just have it running, you know, just kind of in a loop. And you could see that it's been busy doing stuff for with my wallet that I have attached to this. So, um, this is all real output. It's really happening. It's really it's trading on mainet. So, it's it's interesting. you know, like um I see that there there's

35. 16:48

    also on the agent hub um there's something kind of similar which I I think is pretty cool. Um but adding in that secret vault piece uh gives you like that should of a a primitive right a data primitive that you can have where you can dep you can deploy service and have this you know um software service but preserves privacy and you know in in a in a new way. All right I think that's all I got for now.

36. 17:22

    Well, privacy is an absolute prerequisite for any of this agent stuff to work. Um, I we have X-ray on every once in a while on this on this call as well. Like this is like the thing like if whoever solves this problem like actually can you know uh personalized agents and no one's ever going to want to use agents if it's not actually personalized to you. So I think this is like really really cool. Um yeah, anything you want to add?

37. 17:49

    No, I think I think you're you nailed it. It's um it it is our it's our passion. It's our mission. Um it's what we've we've started out uh to to achieve. And so this is one of the areas that we see that's relevant to to today's you know the the excitement of of using agents and and even vibe programming. So it's it lowers the complexity of integration. Uh but it also offers such a significant piece of

38. 18:20

    uh of privacy technology. No, I'm I'm so stoked. Um looking forward to actually getting this integrated in the hub as well. So thank you Nico for coming on. We do got to get moving on to Jutsu from uh sorry Zahed from Jutsu. Uh Zahed bring you up on stage. Zahed is building a lot of stuff. Um and so Zahed I'm going let you kind of like manage what you want to talk about today. But first

39. 18:47

    um intro yourself intro Jutsu. What did you show last week? Okay. So hey everyone I'm Zahid uh founder of Jutsu and in Jutsu we are working with Near AI to kind of build a uh Typescript experience. So since we started the project, our goal is to build a system that uh a developer can come in and kind of a start building a near AI agent and deploy it in minutes not hours. So we made some progress in in the TypeScript SDK and uh I will show

40. 19:23

    you a demo um and then I'll show you another uh project that I'm working on. Uh but let's let's start with the TypeScript demo. Uh I'll share my screen. I'll share my entire screen. So make it bigger. So uh the way you you will start using near uh AI TypeScript SDK is just simply type uh npm install the near a ts cli. So that will uh kind of a install the npm package locally. And then to create a project you will just say npm uh near cli create. That will create a

41. 20:07

    local project. You don't have to do anything. And then you can go inside your project and you'll see the familiar uh things like type uh python project. So you have a agent.ts and metadata uh JSON file. And to run a agent simply you just run u near aits run. You don't have to do any other things. So it's as simple as it and then you get a uh command line interface that you can you can interact with your

42. 20:43

    agents and uh you you can update your a uh agent typescript uh code uh add your own functionality. uh it is not one-on-one comparison with Python yet but like we are working hard on kind of making uh these two framework uh work the same way and all the functionality. So now we are asking something and then when you are done uh when you are ready to deploy uh all you do is just uh near AI uh TS upload and that will upload your project to the uh

43. 21:21

    near hub AI hub and then when it deploy you get a link that you can go and check your agent. So one important thing here uh today you can upload your TypeScript project to hub but it's uh you cannot use the hub UI at this moment but we are working on kind of a uh making that one work. So hopefully by next week when I come back it will be end to end where you can deploy and you can use use the

44. 21:52

    hub uh and you can interact with your agent. So I think like a the goal since the beginning is to kind of from zero to one in in a minute. So that's kind of a what we are working on and hopefully next week uh we'll we'll ask the community to start kind of a uh using the uh TypeScript SDK and give us feedback and it's for you and we want we I love to work with community and then kind of a build the best uh SDK possible.

45. 22:26

    LA next few minutes I want to show another project uh that uh that was part of like a near uh useful agent hackathon and uh thank you uh Cameron for running that hackathon. So that useful agent is becoming more and more useful. So I will show you a little bit of demo. Um so this is uh this is the agent that we build you uh in the hackathon. But now uh since after that like we improve the

46. 23:00

    uh media the generation of the images and video uh quality probably 10 times better. Now uh like typically like a this for example this is a generated image but it's almost indistinguishable from uh if you would record a video uh by yourself. So that's that's a uh probably 10 time or more improved than the last version uh that we had. And the other thing we added is lot of um small and medium businesses already use Canva and we heard lot of requests from the

47. 23:36

    people like a they have all the assets for their businesses in Canva. So if they could use Canva with post so now you can uh we are now in a process of getting approved from Canva um because it's in a Canva marketplace so we need approval. So hopefully by next week it will be generally available where people can start kind of a bringing their asset from Canva and then uh make their uh social media automation and save lot of time. Uh so that's yeah

48. 24:10

    that's that's kind of the update uh from the post side and I would encourage like a go and check out like a app.post.ai AI and give it a try like okay uh if you're not posting today you are probably missing out uh lot of exposure that you would get as a founder or as a as a engineer or as a builder. I think one thing I learned last one and a half year is it is great to be a great builder but you also need to build in public and as

49. 24:42

    a engineer like I don't like to be in public but like that's why this kind of automation and AI agent is there for kind of a like be a digital twin of me and then do the social media for me. So yeah give it a try. Uh thank you. Yeah, this is amazing. Um, I'm using post these days. It's working. Uh, and speaking of digital twins, we have coming next integrating with external web apps like what exactly is that process like that? You got to get approved like how do you even start that for a lot of the

50. 25:21

    builders here that you know might not know. Can you repeat? Yeah. So like how do you h how do you uh you know you want to go integrate with Canva like have people like using their Canva accounts like interact with post how do you even start that process like you just use their API you have to get approved like what does that actually look like? Yeah. So it's uh the integration process is like you have to create a uh kind of a sign up

51. 25:47

    with the developer uh account. So you have to build a app there and then the API integration. Um you have to build a third party API integration and then when you have it ready then you have to submit your app. It's more like a uh when you deploy a mobile app in app store. So it's similar kind of process.

52. 26:08

    uh and then you get they will do the review which makes sense uh and then uh they'll list your thing in in their media and then the benefit is like as a user you don't give us any p password or anything so it's like a third party integration so it uses uh u yeah open like it uses their API so you are safe we are safe and for builder I would recommend another thing is try to build those integration

53. 26:38

    because it become a distribution channel. Canva is one of the largest uh design tool today. So if you're building something else maybe partner with uh other ecosystem uh and that gives gives you more exposure and also validity right. So like a um so yeah so the process is in short process is like app store if you ever build a app have to go through the review then get listed in their ecosystem. Super exciting stuff.

54. 27:08

    Um, love to, you know, hear how that process goes with Canva for next week. Gonna bring on Andrew Zah. Is there anything else that you wanted to add before we hop off? Give it a try. And and if you have uh TypeScript SDK, just reach out directly. DM me. I'm available and I'm happy to work with you.

55. 27:28

    Beautiful. Postt.ai. Go check it out. Thank you, Zahed. Bringing on Andrew. Andrew, how you doing? Uh, you've been here before. Introduce Doppelgangers yourself. What did you do last week? Hey Cameron. Hey everyone. Uh, my name is Andrew. I'm a founder and CEO of Doppelgangers. Um, at doppelgangers, we are bringing we are building user owned AI. So, uh, anyone can create their own personalized AI assistant that, uh, well do stuff for

56. 28:04

    you. Um we have uh recently added uh a voice feature. So now you can not only chat but you can also talk to your uh personal AI. And uh on top of that uh we are creating digital twins. So think of it as uh personal AI versus public AI. So uh your digital twin is basically a representation of you on uh you can put it on socials. Uh you can uh um make it uh go to a podcast or to Twitter spaces

57. 28:42

    for you. Uh it's uh basically um your digital twin. And uh on the other hand, personal AI is uh secure uh stored on your device uh and uh well 100% I mean both of them are 100% yours. You are the one who controls what goes in and uh yeah um user own AI is here.

58. 29:11

    Awesome. What did you ship last week? Uh voice. Cool. Do you have a demo? We have a voice feature within our agent. You can talk to it. Uh uh we also can create a digital twin of your voice. Uh the more you talk to it, the better it becomes. Uh so uh check it out. Uh uh app store and Android are both live.

59. 29:38

    Can you show us? Unfortunat mobile sharing feature um and all of them are on uh like on mobile app. So uh I would love to Can you can you pull up your phone and show us? Sure. Sure. Thanks. Let's try it. It's okay. The demo gods I'm feeling they're with us today. So it they're with us.

60. 30:06

    Let's chat and fine tune how it responds. Hey, how are you today? Feeling great. Thanks for asking, Rocket. How about you? Anything fun or interesting going on in your world today? Uh, we are actually on the live stream right now. Can you say hi to everyone? Can you say hi to everyone? We are on the live stream right now.

61. 30:46

    Absolutely, party popper. Hey everyone watching the live stream waving hand. Hope you're all having an amazing time. Let's get ready to rumble and dive into some awesome crypto and blockchain discussions. LFG flexed bicep sparkles.

62. 31:03

    I love the sparkles. The sparkles at the end really made it. Uh that's great. So uh what's next? Um video but uh we are mostly working on integrating all these features like voice into near AI. Uh so uh more integrations on the near AI side are coming. Uh basically within our app uh your personal AI agent will be able to use any agent in the near AI discovery tab. So, uh eventually it will be able

63. 31:35

    to book you a flight uh uh buy you buy you head headphones like IA presented last time and uh get you an Airbnb and uh generally speaking any uh integration that will be live via uh near AI is going to be available on your device with your personalized agent. Uh uh we have a wallet integration near wallet integration already in place. So, you'll be actually you'll be able to just ask

64. 32:04

    it to uh buy stuff or get you stuff uh or book you stuff and it will just go ahead and do it. I want it to manage my LinkedIn account for me in natural language. I would pay money for that actually. So, uh this is great. Love to see it. Um, is there anything else that you want to I guess say before we hop off to Peter Salomonus who is next?

65. 32:28

    Sorry, Solo Mon who is next. Uh, yeah guys, uh, download our app. Uh, we require more feedback uh from our users. We are constantly improving the user flow and design. So uh uh double doublegangers AI available on app store available on uh Google Play. Join join join us. Cool. Uh before we hop in 30 seconds or so, any advice for current agent builders to get their agents actually available on the app store? You did it. I mean

66. 33:04

    uh it's very straightforward. You just uh follow the policies. To be honest, for me, App Store was easier than Google Play cuz Google Play in introduced a whole bunch of new policies where you do when you need to run close testing uh with a bunch of people uh which is weird. Uh and App Store, you just go through approval. They going to tell they it's a two-way communication. If

67. 33:29

    they don't like anything, they tell you what's wrong, you fix it, and you just go live. Got it. Cool, man. Andrew, thank you for joining as always. Cheers, guys. Yep. See you soon. Peter, welcome. Welcome. Thank you. Great. Great to hear. Thank you for having me.

68. 33:48

    For those who might not know Peter, he ships so many really compelling things primarily in the web assembly space. He's written books. He's constantly on the cutting edge. So, I'm really excited to hear what you have to share today.

69. 34:00

    Um, but before we get going, just as a reminder, introduce yourself any further. uh what web assembly music is or anything else that you're working on and then exactly what you shipped last week. Demos are preferred but in a per you know demos are preferred.

70. 34:15

    Yeah. So I'm Peter and uh as Cameron already said my first intro into the near ecosystem was the web assembly music project which I also had been working on before then but the purpose was to be able to publish music on the blockchain. So actually a full p piece of music is stored in a small web assembly binary with all the instruments all the sound generating and and it plays music uh this uh web assembly music executable. So um uh so there are

71. 34:48

    some NFTTS on that but today I'm going to talk about something completely different. I'm going to talk about uh the JS in rest uh project and um now um oh this is I just have to uh allow my browser to to share. So uh let's see uh sorry about this. Um uh oh no worries at all. It happens to every single person. Yeah. Yeah. Let's It says I have to uh uh open systems quit and reopen. Oh, okay. I will be right back.

72. 35:33

    Sorry about that. Oh, good. I can I can actually do move on. Do you want to come back to you, Peter? Yeah. Yeah. Next. Yeah. Yeah. Perfect. Okay. Okay. Sounds good. We'll we'll hear from you soon. Yeah. Yeah. All right. Uh, Jivvon, are you ready to rock? All right, cool. And Ulana, are you here? Yep. Great. Welcome, welcome. It's good to have you guys back. Um, really excited to hear the update, especially

73. 36:01

    after your conversations with Silverstream over the last week. Um, I've not gotten the update yet, so this is part of my update. So, uh, yeah, would love for you guys just like reintroduce yourselves. What is Instant Apply, and what have you guys been working on over the last week?

74. 36:18

    Uh I can start off. I'm Jin. I'm a CS student at Minerva University. I'm doing CS and Econ and we instant apply in the last useful agent hackathon. Uh same at Zahed and we got an sponsor award from Silver Stream. So we have been working with them uh to build the app further.

75. 36:37

    Um yeah, I'm also a CS major at Manurva University in my third year. And um the idea behind instant apply is as students we struggle with applying to internships every year and a lot of our graduates also struggle with finding jobs postgraduation because you have to send out like thousands of applications literally not even exaggerating. And so we decided to build an app that would allow you to do exactly that in just one click of a button. Um, yep. Joan, do you want to go

76. 37:10

    about the Okay. Um, okay. Perfect. Uh, so this the timeline. I think uh, we started with the hackathon in 14th and 13th March. uh literally 2 days ago or like 1 days ago we got uh 20 beta users for the app. Uh and today in yesterday we're completing applications for our first user and while also completing the application like based on their experience and the conversations we're having we're also extensively changing the app both in the

77. 37:48

    front end and the back end. Uh and one of the major things was our resume parsing. Uh so when they upload their information we were using pi PDF and not really working well. So we are now using Zeli uh to pass their resume which is doing much better for like all the sections. Uh and we also changed our job recommendation system. Uh so if you want if you are in CS you might want to be a

78. 38:11

    software engineer or an ML or go like multiple places. So based on all of that it recommends job and with a much better match. And we also change our front end uh since the last time I think we only had like one HTML button nothing else. Uh so we change a lot of our front end.

79. 38:29

    Uh so this an example on like the job recommendation. Uh we generated a PDF for our first user uh just to show them what kind of jobs we are applying for them to. And it also gives you a match score uh and a recommendation to apply or not and like what skills are matching for you and the URL for the job. Uh and for jobs that do not really match uh and like what skills you need to develop uh to be able to apply to the jobs or if you have other experiences you can

80. 38:54

    change your resume or the information you have on our profile uh so we can apply better and it recommends that you skip and don't apply to the jobs. Yep. I'll now uh present to you our better front end and parsing. Okay. Um just a moment. Okay. We'll start with a login. Um just you log in. I created a freshly new user. It's completely empty. So you go to profile. You don't fill out anything yourself. You add your resume.

81. 39:45

    Um, it parses the resume. That takes a bit of time because it does the calls to AI, but um, the results are really nice. So, we get to actually see all the okay, all the details. I did not type any of it myself. This is not hardcoded or anything.

82. 40:03

    name, summary, um the text of the resume itself just for convenience because you know when you upload a file and it has been a month since you don't remember what's there. Um professional experience, personal projects, uh description of the personal projects, relevant skills, all of that, certifications, demographic information because it's often asked in the job applications but not so important. If you want to change anything, you can edit it manually like

83. 40:33

    I don't know, right? And you save it and it's changed. Um, so now for those of you who have been here for our previous shipment, um, I am proud to present to you a prettier button. Now you can just press this, which looks significantly better than the previous button we had, and it will apply for the jobs.

84. 40:59

    Um, yep. I'll not be doing that right now though. Um, I think that's it. Jan, do you want to add anything? Yeah. Okay. I think for now we're Oh, I have one in. So I think uh one of the other updates we wanted to share uh like in our progress what we did last week was a lot of applying for VC fundings uh collecting the applications making the slide deck and uh sorry uh and uh preparing the demo uh that we can uh share with like potential VC funings or summary grants.

85. 41:35

    Yep. Yep. Yep. Incredible stuff. Um how do you guys manage uh credentials like you know you're logging in like how are credentials managed? Uh so the first thing right right now is this database the login uh a top problem we are facing right now is when we apply to jobs they often require us to login on behalf of the user and so we might need like their email uh like the secondary is something like a codes and stuff like that. So for our first user uh for all of those we created a new Gmail account so we can like get those

86. 42:14

    like codes from them but I don't see that being a more uh sustainable thing. Uh so the likely option is we when they have the tool all to themsel they might need to upload that for each of uh those website we ask them for them.

87. 42:29

    Uh but yeah that's something we we are like thinking of. Yeah. Well maybe you can help. Yeah. Go ahead. Oh yeah, I just wanted to add quickly in terms of how our like login works. It's just a flask uh login. So I'm not sure if that answers your question, but they back largely last flask.

88. 42:49

    Got it. Yeah. I guess it was more on like the if you are logging into other sites that like you know needs to like do you see those you know passwords and plain text. Um Nico might be able to help you there. So, uh, more to come on that, but is there anything else, guys, that you want to add before we pass it off? Pass it back to Peter. Where can people learn more about the projects? Where can people learn

89. 43:13

    more about you? Um, love hungry young builders. That's always, you know, it's this is the one of the first projects I had to say like, you know, another project, another project, another project. You guys can build billion dollar companies. So, uh, yeah. Where can people learn more about you and stay in touch along your journey?

90. 43:31

    Uh so yeah the app right now is open source on GitHub uh it's called instant apply I'll share the link as well and for both of us we are in the near uh the telegrams uh and we're also on LinkedIn I'm like just gven see us on LinkedIn I'll share the links uh and yeah happy to cite about this or if you can have more suggestions on that I'll also share the link to my LinkedIn and while I'm in the same telegram channel I think it's one for me on telegram Um, yep. Happy to chat.

91. 44:02

    Awesome guys. So good to have you here. I hope to see you guys soon and yeah, best of luck for next week. Peter, bring in your howdy. Howdy. Yeah. Yeah. New computer, but now you can see the screen. So, so I will go back to the where I left off. And uh what I'm going to talk about today is this project called JS in Rust which is uh basically letting you add JavaScript on top of your Rust smart contract. So uh the contract we're going to look into today is a web for and funible token

92. 44:39

    contract uh which uh uh allows you to create a website. So uh but um uh I will first of all uh it's it's this um uh web application here where you have a client to uh to this tools uh that you also have the source code for in this uh project here. So you can find all the source code here and um this particular feature that we shipped last week was the uh was not a very big one. It was just about getting this funible tokens that we have here uh for being able to

93. 45:20

    uh use also this chatgpt feature uh which is going through a proxy because that allows you to kind of charge for the AI conversation by using this fungeible tokens. So we can say now that uh I want to buy our rice tokens and we can ask the near AI for that and it will then uh when it uh and we have to confirm it and we actually have to confirm it again and that will if you say okay now you will be taken to the wallet and and confirm uh so um and that

94. 46:00

    will uh give you these tokens for uh for using chat GPT but uh let's also try the other interesting part which is about uh running JavaScript and also creating a web 4 uh site. So what I want to do now is to to create this new uh web 4 account. I want to create a new near account really uh I would like to create a new web 4 account and let's uh it can be named uh web 4 uh 34 demo

95. 46:39

    uh near let's uh remember that name and we can uh ask the near AI for doing that and uh This will of course require some storage for deploying the contract. You will have the key so you can delete the account and get those near uh back.

96. 46:58

    But let's just create it and confirm with a wallet uh and approve. You can see now it will actually call this web 4 factory and create this new account and we can approve that transaction. And as that uh thing said uh we now have a new account and we will also soon get a website up here. Uh but we can see now the account was not found. But if we now reload it, it should be there. But we can see also that it's nothing there yet. Uh so um and also in the local storage now we

97. 47:41

    have the keys for for that particular account. we can see it here. So, so that you can uh save and and use if you want to delete the account or access it in any way. Uh now we can also try to use since I already have some rise tokens, we can also try to use chat GPT for uh deploying some content to that web for site. So let's um uh do that. uh create a web uh for page on that says hello world and we can ask and now it starts the uh chat GPT

98. 48:21

    conversation it will take some Frenchible tokens and we can have them refunded those unspent after and now it's deploying that JavaScript so what it actually done was to to create some JavaScript here that is uh having some uh HTML payload and let's go to that web page and see what happened. It says hello world. Uh and can you also uh display the current time on that page?

99. 48:54

    Uh and we can again see how we see the tool call response here. uh and um and we can see that uh it should be updated here right away. Yeah, we we got the current time and and there's also a in browser simulation here. So, uh can you uh also add the current account ID uh and run the simulation here first and let's uh ask for that and uh and we can see now it actually runs the JavaScript that it created and it posted that to the contract and

100. 49:48

     Um this is how it will look like. And uh now it asks us to deploy it uh because it's not updated here yet. But uh let's uh say yes to this. And and again it will uh run the deployment um and it's deployed successfully deployed. And we can see now if we we have the current account ID and we can continue like that. Uh uh so so actually the JavaScript uh can you show me the web page JavaScript code

101. 50:27

     just to have that also. It looks like this. So it's actually this web for get uh function that is implemented that you're a uh a able to send to the uh JS in Rust contract. Um and uh yeah so so that's uh basically it uh that's uh I have for this that's the most recent development on this uh project here. The plan is to integrate it also with this arise portfolio uh report so that you can actually uh talk to this application when you are uh tracking your uh near account

102. 51:07

     activities. So that's the plan ahead. Um and now I'm not sure uh yeah you're already there Cameron. Uh yeah, I mean I I'm don't mind this taking much longer just to kind of I want to better understand like your whole vision here.

103. 51:24

     Like you've been you've been programming on your for quite a while and I kind of see all these projects slowly kind of coming together. Yeah. So I just take a couple minutes just to kind of explain like I mean I'm sorry to start from square one but like why web assembly?

104. 51:41

     Yeah. uh you know you wrote a whole book on this so I'm assuming you have this um and then also like some of these other projects that you're building and kind of how you see the vision kind of coalesing. Yeah. Yeah. Uh so web assembly is of course uh uh uh this compact uh binary format and you can and you can create uh uh code in many different languages. you can create this very efficient binaries that lets you execute this in the web browser on the server in in uh and we also see on uh

105. 52:15

     that was what I was looking for when I found near actually I wanted to see a blockchain that used web assembly as the smart contract binary because I thought that has to had to be the ultimate thing and then I I found near for that and um uh and I I've already been many years working in the web assembly space uh with several different uh projects uh and uh when it comes to uh what I really want to do is is this is this music

106. 52:46

     project that's how it's all started that's been focused around that all the time uh uh it's about um also using web assembly for live coding music and uh I would really love to show you that also one day there are several videos on my YouTube channel about it but the thing is you can live code music. You write your music in code and it compiles to a web assembly binary on the fly and and and you can just uh evolve your music,

107. 53:16

     your instruments uh and adjust your code and you can hear the results instantly uh in the music and uh and now that we have got AI uh this is perfect match because uh then you can actually use the AI for writing that JavaScript code just as we saw here with the web for page.

108. 53:35

     So, so this is kind of just leading up to my next uh where where I want to go next also is to to have this AI generate music. And uh one thing that is also mentioned in this written about this this book is to to write um instrument plugins uh which all the music producers use in all the music you found there on on stage wherever and and and um you can write that in the browser. you can use the AI to express what kind of sound you want

109. 54:08

     and and and then you can use NFTTS for kind of uh selling the unique sound uh to to music producers and you can and they will have ownership to it. So so I think that's a great use case for for the NFT and and the web assembly binary is compact. you can ship it. Uh you can have uh this also very tightly connected to the NFT and um and then uh and then you can also uh actually use that in the music production software that is uh common

110. 54:45

     today. So yeah, as you understand my my uh kind of uh vision here and my passion is is a lot around the music but being into the near space also lots of other uh things has to be solved. So so that's why I I'm into this accounting system.

111. 55:06

     Uh that's one of the things um web for we want a web website for it. Uh the JS in rest was also just part of fulfilling this um um yeah the way of being able to use JavaScript to customize and and JavaScript is also part of that music project. So it sounds like you're building a new internet like you're kind of addressing like every piece of the stack here sort of all using W bosom. So yeah, it's incredible. Yeah, that's true. Well, so cool. Uh, Peter, what is the best place

112. 55:46

     for people to follow up with your progress? Um, I understand you're most of the time busy shipping, which is why we want to host some of these spaces is just like give these builders who frankly may not even love posting on Twitter and all this other stuff, the platform. So when you're not here, thank you so much for this to to present here and and yeah, like you say going on Twitter and and post that's uh always,

113. 56:12

     you know, uh taking time from actually building and shipping stuff. So So um so uh but I I love this previous talk here about this uh thing that can help you with the social media. I think so. We all need it like we all um Well, yeah.

114. 56:29

     Where can people learn more? Uh, so my my Twitter is where I posting, but I I would actually recommend going through my YouTube channel where I have uh videos showing all of this that I also just showed with the web for the JS in Rust, but also the web assembly music.

115. 56:46

     There are some videos of the showcases from this book that I have and there's also a chapter in the book about near protocol and also about this JS in Rust thing. So, so, so that's also uh if if you look up that book, that's also a a good place to to u uh yeah to to learn about what's the name of the book? Uh the name of the book is building and uh deploying web assembly apps. So, uh and it's funny, I haven't got the

116. 57:20

     hard copy yet from the publisher, so I don't know. I I've seen people have bought it and they have it, but uh I I not even the author. That's kind of hilarious. Um so plenty of places for people to learn more. Um thank you so much for coming. We'd love for you to come on again. We'll do this every week.

117. 57:38

     So whenever you feel ready to give another update, you're welcome back whenever. Thank you for having me. Yeah. Awesome. Thank you, Peter. Taking you off. Um that concludes all the guess like presenters today. Um, I am also going to just kind of give you guys a quick update on what I've been cooking. Uh, the whole near AI team, by the way, is primarily focused on building the most useful uh, assistant in the world. Essentially turning every app into a super app uh, through a network of agents that are all

118. 58:10

     verifiably private. Uh, and the inference that those agent builders need is currently free. So, if you are an agent builder, you can get free inference. Pretty cool. uh that's one of your main costs and then through the assistant we will drive demand for your agents assuming they're useful and that is my job is to ensure they're useful and so right now over the last week been very focused on hiring a team uh this team is primarily just going to include a director of ecosystem engineering think of it like all external

119. 58:43

     integrations and a marketing team just like one marketing manager and maybe a social media management intern um but But right now it's all me. So I need a scale team because that's ridiculous. Um, next thing is really working with Nillian and Next Trace on the crypto memory stores. This again is an absolute prerequisite to personalize your agents.

120. 59:03

     Uh, critical. We already talked about that. Um, we're also just on the kind of internal and your AI side uh formalizing all of our success metrics, all of our KPIs, OKRs, all that really really fun stuff to hold each other accountable. Um and then other thing is we are exploring adding more data sets to the near AI hub. Uh would love more feedback on what kind of data sets you might be looking for. Uh we're right now on the hub it's

121. 59:28

     primarily just like near rust contracts and they have been used by teams building like auditor agents and other things. Uh agents that can sort of streamline the auditing process for your smart contracts. Uh but I'm actually really that's cool. Don't get me wrong, awesome stuff. want those people to keep building, but I'm really interested in having an agent that can audit other agents uh to check for any security

122. 59:51

     vulnerabilities, any data leakage, anything else as a really critical piece of the discovery process. So before an assistant is able to go connect with another agent, I want to make sure that that agent has been verified. And so we did post this request for proposal uh to get AI security teams to build this agent. uh we reviewed a lot of them. Uh we had like probably like eight really

123. 1:00:14

     great submissions and uh now we're sort of moving forward with u I guess like two of them, one and a half. Uh it's kind of a weird thing because like we need more useful agents hosted on the hub. It's like security is always this sort of like secondary thing to a lot of teams but it is a primary thing for us because we will be managing keys and money. Uh so very critical. Um so hopefully in the next coming weeks

124. 1:00:39

     there'll be more information. I don't want to prematurely share too much about the architecture of this because I want the team that is actually building it to do that. Um so more to come on that but making progress on the auditor agent side of things. Uh next thing is the house of stake. Uh for those who might not know the near ecosystem is very near foundation is very committed to decentralizing the near foundation. Over a year and a half ago, we started an

125. 1:01:04

     initiative called the near digital collective to essentially run an election to determine who should be who can actually decide on how money gets allocated. There was about 10 million USD uh of near allocated to that treasury. And this was a crazy experiment that didn't end great uh because there was a lot of vote buying and a lot of crazy things on the governance side. But guess what? That's not stopping us. We are doing version two. Uh and this version two is a little

126. 1:01:34

     bit more top down than version one and is going to need agents that help with the governance process. Governance is an incredibly difficult thing. Humans are inherently biased. I can say so myself like I am biased to proposals that I can actually read and understand. So English and a little bit of Spanish. If you send me a proposal in Chinese, I'm biased.

127. 1:01:58

     I'm not going to be able to read it. And so how can we actually create these agents that can streamline governance uh for the people who are deciding on how money gets allocated? And so uh this is really important, really big. This is all going to be done through the house of stake initiative. And I'm working with house of stake and also metapool. I have a call later with Claudio, the founder today, about how to streamline governance inside of Metapp, which is an

128. 1:02:23

     amazing liquid staking app that is very committed to to decentralization uh through a DAO. And so they have a very active DAO. How do we make it more active through agents to help people make decisions more quickly? That's on the governance agent side of things. Uh the next piece is agent price discovery.

129. 1:02:43

     So a big problem right now is like all these you know AI companies really struggle to identify how much their solution should cost. Um and it really is dependent on how much value that they're providing. And so in the onboarding process of the near AI hub uh we would like to add a piece for you to monetize your agent but to monetize it properly you need to discover the utility and price of that agent. And so,

130. 1:03:08

     uh, working with a cool team that, uh, it's called Nevermind, uh, to to help address this problem. I've been talking to them a lot this week on how to do so. More to come on that. Another really hard problem to solve. Uh, fortunately, I'm not the only one to solve these problems. There are amazing teams out there with many more employees that are highly more qualified than I to do so.

131. 1:03:27

     Um, my goal is to just get them integrated into the near AI hub and, you know, made available for all the builders out there. So that's on agent price discovery and I have a long list by the way very it's it's been a crazy couple weeks. Uh next is uh really want to add uh near intents to a Coinbase the Coinbase commerce on-ramp uh because right now when you want to on-ramp uh from Coinbase USDC from Coinbase to an app or anything else your wallet um it's only available

132. 1:03:59

     through base USDC. In a perfect world, we can add a near intent network where when you want to, let's say, on-ramp USDC to Salana through base uh sorry, Coinbase on-ramp, uh you can actually add the intent network in the middle of it, essentially swap that base USDC for your Salana USDC or near USDC or anything like that to essentially enable interoperability across every chain, which is pretty cool for an on-ramp. And so instead of right now my main interest

133. 1:04:28

     for that for the near side is so we can easily get uh near agents access to an easy USDC on-ramp with very very low fees. But it just so happens that doing so will actually enable USDC on-ramp support through an app like Coinbase if we can figure it out and work with them to make it happen. I don't want to overpromise here. Uh we're still like cooking on the architecture um for everything. So yeah, USDC, Coinbase,

134. 1:04:53

     On-Ramp, all chains that support ECDSA signatures. So that's uh Bitcoin, Dogecoin, Ethereum, uh every L2, pretty much everything that's not Salon. So moving on, there's more. Um setting up posts for the new AI LinkedIn account. Uh I hate LinkedIn, but we need to post there because the AI community is on LinkedIn a lot of times. So, uh, setting setting that up with Zahed this week, which is exciting. Um, so expect more content from the Near AI LinkedIn account. And then, uh, not really

135. 1:05:27

     related to Near AI at all, but maybe in the future, who knows? I run a nonprofit. We start blockchain development courses at universities. Been doing this for about seven years now. Uh, they ran an amazing hacker house in Argentina. We're very focused on the Latin American market. And, uh, that was just a huge success during Crescent. Uh shout out to the Crescent Mento team to of uh you know bringing crypto education and adoption to Latin

136. 1:05:50

     America specifically Argentina. Uh really like what they're doing and you know this is an evergreen market. So um if we're building AI agents and assistance to streamline governance and everything else, this is a really great market to do so. And uh that's it on some of the stuff I'm working on. I also went to the doctor for the first time in a long time which was great. So be healthy. Uh go to the doctor if you can.

137. 1:06:12

     And uh that's all I got. So, thank you all for joining. Um we do this every week. Make sure to like the Near AI Twitter account. Uh this is where everything gets live streamed. And for the future, anyone who uh is joining, be sure to stream to your audiences. This is a great way to get more uh eyes. So, thank you all for joining and hope to see you next week. Thank you.

138. 1:06:35

     Thanks for watching. Subscribe to Nearweek.