DevHub Live 33 – AI & Blockchain Innovation with Trove Labs

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

1. 0:00

   want want want want want what

2. 1:54

   want want what you want you want what you

3. 2:59

   a hey welcome back to another episode of Dev Hub live I am your host Joe from devhub uh super excited to get back at

4. 4:03

   this after a few weeks off uh hopefully you've been following along and you saw some of the replays that we did we showed our some of our highlights from the past year uh some great projects that have been building all the way through our hackathons and everything else but new year new content new guests although familiar faces today um so I'm really excited about my guest today uh I have over the course of different shows

5. 4:29

   that I have done around near and blockchain I think I've had them on three or four times now uh they are longtime builders in the near ecosystem and like everything else um they are now building with AI and uh this is Trove Labs um if you didn't see the title screen originally uh they are part of the uh Nico project they have been doing stuff I mean all the way back to the initial like nft days of of near a few years back to now uh there the stuff

6. 5:03

   they're doing with AI is really fun and exciting and I think for a developer uh if you are even a newer developer or you're just trying to dip your toe into AI I think this is going to be a great conversation for you to be part of because uh it is not about just being super Builders it's about leveraging Ai and uh Brandon and and David are gonna come on and talk about this and I'm really excited to have them

7. 5:30

   so with no more delay uh let me bring them up and let's talk about uh what they're building here on near hey guys uh thanks for joining today Morning Joe Happy New Year my friend yes happy new year thanks for having us my pleasure uh I feel like every time you know we talk you guys are up to something new but uh this latest project really is uh pretty exciting because you guys are really doing a lot with it and you're moving so fast and you're building public which uh I'm a huge fan of uh you're making it really

8. 6:04

   approachable so let's let's just start at the very like high level of this like what is I guess aggregated app and Ai and like where did this idea come from yeah so really everything started towards the time frame of the second near Horizon cohort uh so we uh joined the the second cohort for near hor and that's really where we hit the ground running with aggregated AI um the basis at first was we wanted to bring all of

9. 6:39

   the crucial llm Services together into a single mobile app and then we we believed that the data and usage by users on our app could yield well to crypto rewards so we were saying hey you use our app we can give you crypto you get to use all of the best llms all of the best Frontier models open source models under a single platform interface that was where it all started that was

10. 7:03

    like what we were building during the near Horizon cohort but that has definitely evolved um since then uh much more so into agents um but the yeah the the Android app was sort of the first um to launch and we were able to scale that to over 10K app downloads uh pretty quickly after launching which was super great we proceeded to build out the Apple app as well um but then after the Apple app went out there were you know

11. 7:35

    there's already perplexity and perplexity is just a super super good product that basically did what we were doing so eight months ago our Focus trended towards agents and that's kind of where we've been super dialed in um moving forward yeah that that's awesome and it sounds like that this whole process for you guys isn't even just a like a pivot from like what you first had an idea for it's really been like this iterative Pro

12. 8:05

    process I mean just been following along from like your tweets it really seems like as you were trying to build one product it led to the next idea which led to the next thing yes yeah I mean everything's super experimental right now you you're just figuring out what works I mean the the entire space is is literally everyone's realizing this together which is really fun to be a

13. 8:28

    part of but I would say that everything that we're doing is sort of just to extend the services of what it means to provide aggregated AI um when you're looking at all of the different pieces out there everything's so fragmented all of the different Frameworks and workflows and so to be able to start bringing those things together orchestrating the best tools the best workflows into a single interface has always sort of been the goal it's just evolved from Individual llms to now

14. 8:59

    agent workflows and more so our focus is empowering Builders to build agent workflows very very easily I think that is like where we're most focused we're we're building lots of consumer experiences on the technology that we develop to Showcase it but really we're doing that to extend our offerings for the course product site which is aggregated AI and if you if you look at the agent framework that are out there I

15. 9:29

    mean elizza has exploded bid is doing great but as soon as you introduce a line of code or a GitHub repo um you're gonna lose a lot of people maybe not no so much people in our Niche but globally the only one% of the world knows how to code and so there's so much knowledge and expertise that lives inside the other 99% of people and we want to help make the agent building process as easy as possible so people can take what's in

16. 10:01

    their head and put that into an agent as simple as having a conversation picking some tools and hey if it's good you'll earn some money when your agents used in the aggregated orchestrated workflow I yeah no there's uh so different directions I want to go but I want to kind of start with one part here so uh you guys started with when you made this change you started with the near chat is that right yeah uh exactly so near AI chat was is our proof of

17. 10:32

    concept for our AI pods our agentic pods um and the idea for these pods is for every Niche every online community just similar to subreddits to Discord servers we want we're we're approaching agents in that way I think that if you want to have good agents you need to they need to be very hyperfocused like you have workflow agents you have swarm agents they sort of can uh function differently but for workflows workflow-based agents

18. 11:07

    are require you know an expert to lay out each step and if you don't have that expert it's not really going to be a successful agent workflow with swarms it's a you know there there's more Randomness and and that sort of a a feature not a bug there but for structured workflows you need those experts and I think that we are just trying to empower those experts give them a very easy stack so that they can build these agent workflows so that people can go from playground to

19. 11:38

    production in a few minutes and so we're in the near ecosystem so we built the near pod and we've built out the wallet integration so we can begin to add onchain agents and then a few simple functionalities like a a storage contract uh a trading contract and then you tap in some data feeds and and a schedule contract and you're able to set up DCA Bots you're able to set up event

20. 12:03

    driven trades and people can take their ideas take these little data streams and say hey I think whenever Ilia tweets about near I want to buy a near and maybe that'll be a positive trading strategy maybe it won't but I'm gonna I'm gonna let it Let It Go in the near AI chat and see what happens so that's the idea for near but we want to have ai chat rooms these pods for every domain

21. 12:28

    and we want to empower people to build those out and we can talk more about the tech stack as well if you want yeah I was gonna so I think right now where people is and I'm talking to both like just developers I think and regular people that the concept of Agents right now does feel a little bit ambiguous I think it's a great Point you're making here of like you know the ones that are more Super refined and based in expertise seem to do better and seem to

22. 12:56

    be more useful for people but to to that point like you have to have a path of like how does this agent get made and deployed and for you guys you must you had you had to also then start somewhere too of like how do you build this platform so maybe like just I think it's informative for everyone to get a sense of like what went into building this original platform even just this uh uh what you call the you know the proof of concept you started with

23. 13:26

    before yeah the guess I guess the first um big idea was realizing that we're spending like 14 hours a day in front of llms I mean that's kind of embarrassing to say but it's true between cursor Claude chat GPT sometimes still uh it's a a large portion of of our day has spent talking with llms and it's kind of uh lonely in in some regards but more so as soon as you get stuck you have a problem you're starting to go in Loops when when the llm doesn't have the

24. 14:03

    answer you're kind of uh screwed um you're now you need to go to stack Overflow or you you need to uh actually go onto the internet and and figure out the problem for yourself and so the idea for these AI chat rooms is saying well why can't AI be collaborative why can't we introduce subject matter experts within communities and start to have both parties you have these ecosystem these Community participants for their

25. 14:35

    respective domain and they can be another layer on top of the AI agents they'll be building the agents with their expertise but they're also in the community answering questions and almost like an evolution of of where we think Fiverr is going where you can have these supporting roles sort of on demand that can really help enrich where the llms fall short and so that was sort of the the initial big idea but it has certainly um you know continued to change since then Brandon I don't know if you want to

26. 15:08

    add anything as well um yeah for sure I mean I think that was the initial thing is trying to gather all of these um llms on the platform and then being able to have these agents that can help you with everyday tasks like um utilizing AI leveraging it for helping you with coding Hing with all of these different things that we were doing daily trying to get the most efficient use out of AI

27. 15:35

    as it's been developing and everyone's been using it and yeah and it kind of been I think one of the things we've realized is that getting people to create their own Bots and then providing them with that tooling getting this collaborative effort to like just improve the agents overall is a something that everyone's taking advantage of as well Bida and all of that and I think that's kind of what we wanted to eventually start adding to the platform

28. 16:00

    yeah for sure so when you started to build this platform was that something that you know you guys went totally from scratch and we say we're going to build this from the ground up and I think if I remember reading your tweets correctly like the two of you pretty much did the bulk of this work you know with the help of AI like it became like this like cyclical thing of like we're learning how to use AI better and we're building this product that's going to help other

29. 16:26

    people use Ai and it just became this cycle right yeah absolutely it's uh all all built from from scratch if you can consider working with cursor building from scratch a new toolkit yeah for sure yeah so uh you know I would say eight or nine months ago we started to to dive super heavy into the the existing agent Frameworks like you have llama index you have Lang chain Lang graph uh autogen from Microsoft we started just to study

30. 16:55

    those to to learn those and that it's it just became okay so I can build out these Python scripts on my computer and okay now I can like plug in a database and okay this is what the this is what context caching is okay I need to retain conversations like all of these little bits and pieces and it's just there's so many different little pieces even if you're a technical expert I'd say there's still a certain amount of complexity involved with bringing all of

31. 17:25

    those elements together all of the subscription management and and the time it takes to just put all of those things together so why can't there be an outof thebox managed fully managed solution for agents and that's sort of the direction we went building this realtime websockets chat room that has everything you need so if you're a developer and you just want to experiment with an like

32. 17:50

    just a new workflow maybe you want to see what a near agent's like instead of spending all of the time building out your own tools make all of those connections all of those little things that are required to build a production application you can just use this stack and come and take advantage of the fully managed framework but yeah it certainly was initially just learning the agent workflows and then how can we bring

33. 18:16

    these into production and make it easy for people to do the same or just use production agents and I say near AI chat was it's just been an evolution of that yeah um piggybacking really quickly off what David said um yeah he spent like so much time building out this framework we spent a lot of time researching it using it ourselves troubleshooting and I think one of the things um it's it's it's mainly a big threshold it's a roadblock for new developers who want to skip all of that part they want to go straight

34. 18:47

    into innovating making something new and I think the realm of AI just moves so fast that you know you can have an idea and then it's already popping up other places so being able to skip that step and just um as he mentioned earlier be a subject matter expert test your ideas and just keep shipping things I think that's really important I think um that's something we wanted to enable the

35. 19:10

    users so as you started to build especially as you you know the leveraging near first as this uh testing ground and proof of concept for yourselves what were some challenges that you ran into as you ran started to do this was it more around probably I mean maybe this a two-part question I'm sure there's AI related challenges but then those are also like near related challenges as

36. 19:35

    well as you integrated that as your first like kind of Niche area yeah uh absolutely um for I I guess the the initial roadblock was just navigating all of the deployments and making it making it easy for these pods this near AI pod to be very scalable in that we can take it and add another pod add another pod add another pod so that was probably the the biggest development challenge coming from you know just over the last couple of years putting the skills together and

37. 20:07

    amplifying our our development skills through through AI um but the the much greater challenge of course is the onchain near Integrations like rest development is a whole another Beast it's a beast that I am am not at a level of comfort with compared to Python and and node.js and stuff like that um which you know that that makes it harder to harder to do things it's harder to you can't move as fast and what what I've learned and maybe this is just because

38. 20:39

    I'm a Noob with rust but it seems like cursor is not so capable at delivering good rust code especially near related uh development and that was um sort of a a source of the first agents that we offered on near AI chat where this this full stack team of near developers um which was a hilarious undertaking um because we we it was like just so such a big Monumental task that we that that like didn't go great uh at at the end of the day um we took every single near

39. 21:20

    native code repo um that we could find and collect them all uh performed some some really solid R strategies um using like code splitting it we you know the the best practices were followed but it it it's still the end product wasn't super great and uh we we wanted to offer near agents that could help build near code and unfortunately that was not the outcome that that that happened and so then it was like okay well let's go back

40. 21:51

    on to the first principles of why we're building this in the first place which is to empower subject matter experts to build agents for the things that they're really good at and so then it became well let's let's do that let's build the agents that we're subject matter experts at and empower the people that do know what the heck they're doing to then fill in the gaps and um I I guess there's a

41. 22:17

    lot to be said about how you engage with people too um because I mean just building an a place for people to go chat like that is a big ask right so sure that that is another probably key challenge that I'm really excited with how we're how we're navigating um but yeah I think that maybe when people see your AI chat they might think dude No One's Gonna Come to your chat room like we already have telegram we already have Discord and and Twitter why why would I do that and

42. 22:51

    that's a that's really the right the right question to ask um and so what have we been doing we've been building for the the public interfaces where people are at um super focused on Twitter on telegram on Discord so that while people are building agents inside of their pods all of that is accessible to people outside of these rooms these rooms exist as like the back rooms for the agent information and so when you're on the Agra Mobile app all you do is ask a

43. 23:24

    question we route to the relevant agent flows whatever pods that entails um similarly for Twitter we have you know built out a a no code ex agent Creator um that that we can talk more about but that allows for any of your agent workflows that you create inside of these pod environments to distribute the information where your community is actually at where people are actually at and that was probably a a big Learning lesson to to navigate sure sure I mean

44. 23:55

    it and it sounds like you know again you're you're fighting a battle in two fronts and and this is something I think is important for developers to understand as well that like if when you're dealing with a new technology like Ai and again like AI is like not new new but a lot of stuff that's happening now is is new and for most developers this type of interaction is very different because like you said yes

45. 24:20

    it can spit out all this code but when you start to actually tailor it to very specific Concepts like a near smart contract maybe it fumbles a little bit so you still need some of that expertise there the other side of it too is you know again that user audience like who's going to use my product how am I thinking this through and when you start to say um I'm both trying to push the

46. 24:44

    tech forward so I can be more productive but at the same time I'm trying to push my product forward uh Brandon this must be something that really probably you guys go back and forth on of like where does that time go because this is only so much time in the day it's only so many of you guys right they only have so many hours how do you start to balance those two things yeah yeah well um I think one of the things about the platform that we're really trying to push and establish is

47. 25:14

    that being able to have incentives not just for um products that people build but for using AI so encouraging our users to create um agents create with the tooling that we provide create the agents that people want to use and then awarding people for creating those agents the more your agent gets used maybe it's something that's really useful you find something that people are putting to use in their daily lives some an agent that's really Innovative

48. 25:45

    that's great being able to incentivize people to create these things that the audience that the user base wants to use and then providing rewards for that I think that's something that can help us find U you know that Target user base find the next AI Innovation that people are actually going to use that's going to draw users to not just our app but to near to the blockchain that we're integrated with um it's definitely an

49. 26:09

    experimental process some of the agents that a uh David has created um are amazing being able to automate Twitter being able to automate posts and things like that it's just I think it's a really it's an exploratory process so yeah yeah absolutely and Brandon uh you know has sort of hinted at something that maybe we'll we'll ask you for some advice on down down the road maybe even

50. 26:37

    in this call today but the the pace at which people can build now is insane like a a 10x developer is now a 100x developer THX developer right so people can move so fast we're able to move fast if we can move fast people can move faster so how do how do we how do we meter that how do we navigate these conditions and it is amazing that people can take their ideas people can take what's in their head and bring it to life um but how can you activate people

51. 27:16

    and and get them involved in a way that pushes things forward and you're able to keep up with the pace of growth and so that has been you know the Eliza framework that has exploded and it's open source it's open source developers can take advantage of it and build with it and so for us we believe that that is the direction we're going to go as well so we want to open- Source this pod architecture and tell everyone hey you can build with this pod we're going to show you exactly

52. 27:48

    how to set things up not only that if you establish your own pod for whatever it is you're good at or you care about we'll incentivize that we have token allocations to support that growth just like we have allocations planned for Tool Builders and the the actual agent Builders as well and so it's sort of this sliding scale of technical sophistication that we want to capture and reward people for taking part of and

53. 28:20

    I think that the the crypto audience is pretty good at at being early to things and wanting to position themselves so that they can take advantage of of big Tech growth and blockchain enables that through tokens through this communitybased um incentives so we want to kind of bring that let let the crypto community be sort of these Architects for that growth give them some skin in the game and say hey we're we can build easy agents that anyone can take

54. 28:52

    advantage of that Joe the plumber can build out his this is how you unclog a sync agent just like there's a YouTube channel for every single category out there and when you have a problem sure you you go to the average guy the average girl that has been working in their field for 20 years that is you you walk past them on the street they're not a movie star but they have hundreds of thousands of of subscribers because the information that they provide is trusted and it's correct

55. 29:24

    and so content video content prove that that is successful role we want to do the same thing for agents we want the average person to be able to feel that opportunity that exists as content creators and make it even easier just take what's in your head spend a couple minutes deploy your agent and hey if if it's good you'll earn money you'll earn passive income I think that's a dream

56. 29:47

    that a lot of people have and I I think the crypto Community is a really good place to help broadcast that and be sort of the key um Visionaries to to help bring that to life so that's sort of where we're hoping to take things um I think that's the most exciting part of AI right now and what we're trying to capture with our product and as you mentioned um the things that we're focusing on right now the the

57. 30:12

    biggest agents we are have our eyes on are you know trying to tap into social media with the marketer and the V tubing um agent and then also trying to provide expertise um development help and transactions with the near pod but I have a question for you Joe what agents are you excited to see people start building what do you want to see on near and just in general from all of the tech that's available today so I will say you know I am

58. 30:42

    excited for the day when there are quality agents to do the things the like the other things inside my life that are outside of work um like having kids and having a family I I need an assistant but I I can't afford a personal assistant um and so you know having that stuff where uh I can you know again all that stuff of like you know hey you know here's what we want to eat for the week an agent does all my grocery orders and

59. 31:13

    things like that you know hey you know when it knows when I need to do the dry cleaning or whatever like all those type of things when those start to happen and that automation comes in like that's where for me as someone who is technical is like that's what I need like I don't mind doing the work because I like work but all the other stuff of my life that takes away from it because like I'd

60. 31:37

    rather just like work and then be with my family I don't want to go do all this other stuff and so when those Agents come around that's what I'm super excited about I I saw a funny post that was like right now ai uh does your creative work and um does your creative work so you can have more time to do your dishes but I want AI to do my dishes so I have more time to be artistic and my creative work 100%

61. 32:01

    that's exactly that's exactly me yes that's exactly where I feel it you know and and I can see that path like you know because of the acceleration of things that came out of like covid of like delivery and everything else stuff like that uh you know like yeah I save a lot of time because I can order my groceries and I can do all these different things but I still have to go through and do that so that's still time it's still effort right it's still all

62. 32:24

    these different things now we can start to automate get smarter make better choices for me I don't want to have to think about you know all the different stuff that's that would be amazing I think the days of a virtual assistant are are getting closer especi even one that's multimodal that you can speak to I think uh David's actually been building uh made quite a bit of progress

63. 32:46

    in that Arena as well that'd be amazing right just like I could as I walk my dogs tell it you know a few different things just talk to it hey get these tasks done and I know it's done that'd be awesome yeah um we that's that that has been a kind of a trajectory that that we're following can't order groceries yet but the the personal assistant um has been such a fun build um you know just thinking about the the polished and end user experience like how can you know when we're building these Technologies

64. 33:19

    how can we provide sound examples of of how you can take advantage of it and in doing that you're you're able to build consumer products and you're able to build infrastructure and that that's like I think a really fun place to be in and with with Ren you we've been able to make substantial progress on how how our Twitter stack needs to operate for for the no code Twitter agent Creator but

65. 33:41

    also the entire multimodal Suite um right you know we at this point we have a containerized build for an automated desktop companion and so I don't know if you've seen any of the videos that I've posted of Ren but there's a few pretty cool functionalities um just while I'm working I can just talk on my mic and it'll it'll turn my speech into text through whisper routes it over to our AI server and then Ren will will talk back her speech uh she has her own you know

66. 34:11

    custom tuned voice model uh and then she'll she'll talk with her voice it's all all animated the the mouth movements are there uh constantly pulling from Twitter for specific categories and um crypto near Proto call AI the the virtuals ecosystem all targeting specific creators that have Alpha that have very good information collecting all of their tweets and then every week automatically

67. 34:40

    on a Cron job for each category or in processes all of that enriches it by filtering out the the noise taking the important tweets that are labeled as important in performing Pretty robust research and and web searches at the end of it it's a video script that's created and and like the from like two weeks ago to now the quality of of the the video content that's being posted by Ren fully automated is like really really cool to see um but then you know at the same

68. 35:12

    time it's just a companion that you can talk to and and eventually you know it'll it's very task oriented and and to I think H having the ability to have them start ordering your groceries and stuff like that is a pretty a pretty cool thing to explore but I I know how how the approvals would work for that yeah well so that's and that leads me to I think my next set of questions which is a lot of the stuff that you are doing

69. 35:37

    in theory does not have to be crypto-based like it is right it is purely the traditional kind of web 2 stuff you're using you know these type of models in Ai apis and everything else but you have are experimenting especially with Ren using a token and everything else like that why do you guys in your opinion see the cross-section of crypto and AI being an important one I think there's like a there's a dreamer answer and then there's a realistic answer I'll give my

70. 36:08

    realistic answer and that's just trustless immediate payments between two things two entities um the democratization of of data and and everything else is a probably significantly more important but I think there there's bigger and brighter Minds that are are fighting those battles and um and we will hopefully help to contribute as much as we can um through having tokens where people can share ownership in in things that are created but I think the

71. 36:39

    true blazing utility for crypto and AI is that just immediate form of payments that can occur anywhere in the world any sum of of assets can be transferred I think the unlocks that happen there are very very pertinent to agents and niai as a whole yeah no I it's uh it's really interesting because one of the things that you know again what we're trying to get a better sense of as you know you see near near AI is coming out and

72. 37:14

    they're building out more product based type of stuff too is that like can all this stuff be simplified but it sounds like from what you guys are learning and correct me if I'm wrong that it is right now now it's really about can you have these things work together it's not about necess one tool ruling them all but can these all actually work together is that where you're seeing it yeah 100% um making things work together building products and services

73. 37:46

    that can stand on their stand on their own as you know the the noode Twitter agent Creator that's a that's a standalone web app but it's also immediately integrated into everything that we're doing uh through the mobile app and and just as a core service the same goes for uh a super easy image generator that that we've developed and we'll soon uh start showcasing that connects to basically fine-tuned um uncensored AI models for for image gen text to image image to image um all you

74. 38:18

    need to do enter some prompts click some buttons and you have really good images posted you don't need to be an expert prompt engineer to get quality images so stuff like that you can run some ads and have them Standalone as Standalone products but at the same time you're enriching your core feature set and making it so you've done the leg work to build out these tools now let's just make them immediately available to anyone else that wants to build with

75. 38:43

    them and just add them on to whatever it is that they want to achieve through their vertical agent workflow um I think that it's there's an importance in trying to capture just maybe more like the general audience or those who aren't as familiar with AI or as interested there there's a lot of things you can do right now with AI agents that might be a little bit simple to to those people it's I think it's

76. 39:08

    harder to kind of hook them into the capabilities and potential so they they they kind of want to see these more advanced workflows things that maybe you don't just let's say um execute a transaction or make Twitter posts like like a reply guy type bot but they want to see these Advanced things that you know really draw them in and as we've kind of said that's an iterative process

77. 39:32

    you kind of have to build out all of the tools and then as you create these tools orchestrate them together in those Advanced workflows and then you'll really wow them and I think that's that's something that's really required and important to kind of captivate this audience to really bring people like the amount of traction that you want over to your product over to your platform for AI so it's um yeah it's extremely

78. 39:55

    important and yeah and Brandon made a really good point there just so building the no code Twitter agent creator that was in response to how difficult it is to actually deploy a stable Twitter agent that is respecting the Twitter rate limits that that works within the bounds of of the Twitter API requirements and the first first step was just getting all of that infrastructure set up but then okay we have we have these personality based agents they can do certain things like check check prices

79. 40:28

    search the web or whatever but we want to automate the Nico account we want to automate the jump account so if I'm a business what are my requirements there and since we're actually building the consumer experiences on the products that we're developing we're able to see exactly what is required to achieve that and so what did we do well yesterday I dug into a cash augmented generation

80. 40:52

    which I think will probably take Google Gemini into a top spot in usage for Frontier models for Gemini 1.5 basically you have these Giga large context Windows now through Gemini we're able to load in document sets input your your link to your git books or whatever we route that through fir crawl crawl your dock set add it to our mon our database and now whenever someone asks the jump defy account or the Nico account a

81. 41:26

    question that's relevant to the doc set we retrieve that that Corpus of data from the database inject it into Google Gemini into its cache window for like a five minute um expiration time frame it's able to answer perfect responses to documents and that that becomes okay well now this isn't just a Twitter apply bot this is a a very capable business tool that that could you know in time help def itely support a social Med media manager and in giving people

82. 42:01

    immediate information that can be trusted to be correct so that you know that's an example of why I think it's so important to actually build on the stuff that you're building yeah I know that makes a ton of sense has it been a a challenge to start to combine some of the systems where you have this permissionless side of crypto and then you have this very permissioned side of you know Web Two and you guys are obviously trying to take as much off the user so like what has that been like as

83. 42:32

    a builder to put those pieces together I think um especially with the direction we're planning to go on the open source side like a lot of a lot of that is is kind of changing but I think if if if you want to use I I think we want to use the best technology that's available we want to give users access to the best technology and that is going to be at this time the clads of the world the GPT 4os um whatever of the world that you know they're they're

84. 43:05

    close Source but open source models are kicking but too and and we use them all of our personality based agents use uncensored fine-tuned llama based llms in conjunction with Frontier models but yeah it's it's a very interesting topic and I don't I don't think that it's a bad choice to just always go with the best technical the the best technology that's available and still democratize the value that's created from it if you can do that I think you're still abiding

85. 43:38

    by at least my philosophy of of what the value is to be a Community member in web 3 that's interesting when you think about what are the goals of right of building a web 3 and and if you treat it like it is another set of tools it is another experience put in there and you treat it like that and not everything has to be decentralized not everything has to be fully permissionless you are able to probably iterate much faster

86. 44:11

    than you would have number one but two um as to guys who are also worried about product and something I'm trying to talk more about to developers of thinking product focused that you can actually see what is working within your product because I feel like right now if you're purely just in crypto only you do run the risk of butting up against like is this a crypto issue or is this a product issue because you're trying to do too much but now if I start to Mel these pieces together like am I having a

87. 44:46

    different experience for my users all that kind of stuff yeah 100% And you know I think that that if if I I'm if I'm near AI um I would be I think I would be just most focused on those integration points like how how can because I mean that my experience has been that's that's the most challenging part is the the smart contract side and and bringing those elements together um I think that that resources there are so valuable and and I think a lot of I think other chains

88. 45:20

    are are starting to focus some efforts they building you know these these core technology Stacks that make those Integrations very easy but yeah I think that Focus there and resources spent on improving that integration process will go so far because it is it is pretty challenging and when you're looking at maybe the the tool set that Builders are using now through cursor if if it's really hard to write near smart

89. 45:46

    contracts with with those tools with those toolkits well then there's probably a barrier there that we as the near development Community wanted want to help mitigate as much as possible yeah that's a great Point too I mean it's it is it is in service of of two different groups while you know I guess Al all we're all trying to maybe go towards the same direction and it sounds like you know you guys are trying to serve both parts of both some developer Community some end user stuff as well

90. 46:19

    and kind of putting those pieces together if you look ahead and you look at like what your road map might be you know where where do you see this going in I'm not going to say lot no one could break down way down the road but just like in the next like three to six months how do you see this progressing for yourselves yeah in the next three to six months um we will have much sooner than this we'll have the open- source pods so

91. 46:46

    having that as a very easy tutorial-based process where people understand every step to to deploying their own pods the Val that is possible just through the technology itself but also by participating in this ecosystem that we're working to grow the value potential there and hopefully activating people to help us really reach a critical velocity in a in a much more sustainable manner than than we can do uh you know as a single team and looking at the success stories of these web

92. 47:25

    three agent Frameworks that are going full open source we want to yeah follow follow those strategies continue to provide the manag services on top of that so there's still you know really straightforward business uh you know business strategy involved but yeah I think that is probably where we're most focused is just making it as easy as possible for people to start using this

93. 47:48

    stuff and all of that is content cookbooks and um you know just getting our ducks in a row for for that lift but we have lots of PODS that are already like live like there's the salana Pod we have a creative writing pod um Chad bot which is a kind of a funny one um but have just a few more key tools that will actually enable those agents to activate and and do things Beyond just having discussions between back and forth once we get to that point then yeah it's it's

94. 48:22

    Full Speed Ahead of how can we bring the subject matter experts to this product and how can we Empower them to build out this growth cycle that we're working to achieve and integrating with near AI like how can I've I've spent the I don't know maybe couple hours a day for the past four or five days just going through the different agents that exist in near Ai and seeing all of the different things that are available well how can we start to just natively

95. 48:54

    integrate those into our platform maybe if there's these really awesome onchain agents that are getting built by these incredibly brilliant smart contract developers how can we take their hard work and make it make easy agents possible from that so that that we can sort of meld the that that last little part that makes the average person still not really you know it's not really feasible for them to to take full advantage how can we meld that and and help um you know make that a bit more

96. 49:28

    finished and and smoother for people those are sort of the the areas that I'd say we're most focused on and and also Brandon is just all in on the mobile app so everything that everything that we build Brandon is just incorporating that into the mobile app concurrent yeah I'd say on the longer term definitely focusing um on the end user so that we can provide all of the great things that we've been building all of the great um

97. 49:52

    agents Builders are able to build with our tools that we're integrating and then providing that to the mobile app allowing people to use a lot more of the no code agent Builders um on their phones things like that and then access all of the cool agents people are creating being being able to explore those agents and then that really ties in with the experience of earning rewards for creating these agents or

98. 50:15

    using these different agents just making it all seamless all of these great things that we've spent all this time building out bringing it together in like one platform nice yeah um couple things that I want to just hit on before we wrap up here so you know you talked a little bit about some open source stuff how much of the stuff that you're building is actually open source and what's your decision like how are you making the

99. 50:38

    decisions around that yeah current currently we have not opened any of this so we we'll be open sourcing the the core pod um for the first time um and that will come with you know all of the detailed information on on how to start using it um but it it will be this what what aod is is it's like a you know nodejs web app that has real-time Communications through websocket so you're you're immediately conversing with people very very scalable and it

100. 51:10

     has rdus for context caching mongod DB like all all of those bits and pieces are together so all you need to do is sign up join connect your wallet start chatting with agents build your own agent and and you're good to go um so that's sort of where we're starting and then you know just we have also our AI our AI agent server that that's where Ren lives that's where our oneclick

101. 51:33

     graph rig tool lives um all of the Twitter agent components that's a a separate server and so figuring out the the best strategy for like bringing a portion of those tools into the open source kit as well so everyone has a foundation for the key components that are are required for successful agents which you know there's there's few things that like web searches and you know just those those critical standard components will will be made available as

102. 52:04

     well nice nice and then I wanted to kind of pull all this kind of back together around building on crypto and everything else so uh with Ren you launched a token how has that been where do you see that and I don't mean by any means this is not Financial advice but I mean purely just from a product and development standpoint like how how do you see that token being part of both what you're doing maybe today and then obviously going

103. 52:37

     forward yeah for so for Ren um with her multimodal stack which you know we'll will have um you know some potentially some strategy for open sourcing that and just having it as this containerized build um but for for her as an AI virtual streamer well look at probably the number one number two market cap a crypto AI project is virtuals on base uh where anyone can create their own vtube streamer um the barrier to that is like $300,000 worth

104. 53:12

     of virtual token in order to deploy your own virtuals agent which is like pretty insane but regardless virtuals is on base they have this really powerful ecosystem well we wanted to get exposure to that we wanted to bring the near ecosystem exposure to that and be a vessel for near base and virtuals and so Ren is is a token that enables that it what it's minted on near so the the ren tokens created on near whitelisted

105. 53:44

     through Wormhole so bridged 50% of the supply to base and so they exist basically one to one in terms of the the token Supply half of half of it there half of it here and for use um looking at these live streams that we're um hosting where we're actually really excited to demo this concept uh of stream battles where I'm not sure if uh you know you guys are familiar with Tik Tok but basically live streamers will be battling each other for um likes and for

106. 54:17

     like donations from their audiences and over a period of time whoever gets the most donations they win well we're doing that but instead the audience having to pay they actually are the ones that get rewarded and so we have another V another AI vtube streamer that we're going to be showcasing for the first stream and they'll be on the same stream together all of the chat messages that

107. 54:41

     are rolling in from the audience those are being converted into onchain trades through our chaos trading agent and so the background is this really exciting flashy chart that changes color every time a new trade is placed placed and um basically the audience can pick which side they want to win or you know they that they want to support before the battle and all of the sentiment all of that will will basically dictate the battle it is to

108. 55:11

     some degree pretty random it's very chaotic and that there's no Surefire answer on who's gonna win uh but that is that that is using your in token and so these battles will will result in in you know the tokens being traded back and forth and whatever side wins that battle the the the other opponent all of their tokens will will go to the community so that's like the our activation for um integrating like crypto V tubing in in a pretty big Way Beyond just having the AI

109. 55:43

     agent on you know on there where you can donate to them and stuff like that so I think that's like one of the the primary use cases for Ren um as David touched on earlier we really want to kind of showcase um the products type of products that you can build with the Agri stack and we were you know this is really interesting because you were able to create this AI agent and there's

110. 56:05

     they're able to have a model and Avatar and we think this really can apply to a lot of different crypto projects and communities being able to kind of provide this example of the different gamification and like activations you can actually have um I think that can provide a lot of value to communities and really just explore what is possible um in the AI and crypto space it's really really cool they've been working on it a ton and I'm really excited for the first stream was been looking really great yeah I I I love that you know it's

111. 56:38

     thinking outside of the normal Fiat donation kind of stuff right and that's I think that's what we're always trying to get across is that when you the difference between being able to use crypto versus just fi everything else that is it does open the door to these type of things it does allow you to more properly align incentives across everybody both the content creators whether it's virtual or not as well as the viewers and everything else so it becomes a really interesting Dynamic and

112. 57:11

     I don't think we actually truly know what's going to happen but it's great to see people experimenting yeah and how that starts to end out absolutely yeah very cool um I really appreciate you guys coming on this time uh before we go where should people go how should people get involved how can we start building with what you guys have uh what are the details there yeah absolutely um well we're I'm I'm constantly posting video content on Twitter uh for demos of products so um

113. 57:45

     following me at uh dealer defi Brandon's got his Twitter account it should be the same uh as his username on the screen here so for for following us individually we're we're posting video content but of course for the actual platform and product you can go to aggregated doapp um we have a huge website refresh coming so what you see here is going to be kind of entirely different within the next one to two

114. 58:08

     weeks um follow us on Twitter aggregated uncore app and then also for a seeing our Twitter agents in action you can see Catgirl on chain um there's a few other Twitter agents we have Bas Ilia that's running running around out there um posting uh anger translator tweets for for Ilia so yeah you can you can check those out um but all of that uh is kind of made available through the aggregated

115. 58:34

     apppp website awesome very cool very cool well uh you know as you guys continue to keep building definitely love to see your progress here and really appreciate your time this morning uh to kind of come on and talk through building with AI and year and blockchain everything appreciate it guys cheers Joe thank you so much for the opportunity yeah bye guys talk soon bye bye uh awesome stuff uh you know this is

116. 59:01

     hopefully as if you ever call any part of this whether it was the beginning or the end or all of it um you know this is the experimenting that we're hoping to see come out of the intersection of AI and blockchain there's just a lot of opportunities there to still be explored and we love to see projects trying different things and you know as David and Brandon talked about that there are

117. 59:25

     opportun here to just try stuff and you don't have to fully write code from scratch you can start to build on top of other things and just iterate and see what comes of it and kind of go from there so that was really awesome uh we have some really great stuff coming up uh for the devhub live programming so we are really trying to heavily focus on showcasing more people building not just in the near ecosystem but along with AI and other things uh so each week we are trying to bring as many

118. 59:56

     Uh current developers on uh for for example uh next week we have a a def5 product coming on as well which we'll announce and we'll talk through that and going forward you know if there are workshops other things please reach out to us we we're happy to kind of uh put those on uh Below on the bottom of the screen here you can see uh please subscribe to the newsletter if you're not already uh so many uh great opportunities there we're always keeping you up to date with all the stuff's happening in the near ecosystem with

119. 1:00:27

     both you know uh near and near AI that will all be in our newsletter if you need support for building on anything check out dev. n.org or jump into our Telegram and you can more than welcome to ask questions as well uh we will be here every Thursday same time 4M UTC uh with hopefully again more great developer talking about all different things uh but we'll always have a show here talking about different stuff and lastly

120. 1:00:56

     the biggest thing as a new year with you believe in setting in New Year resolutions or not uh if you are a developer Builder even if you just have ideas you're not really a programmer uh this is the year to try things uh there are so many great tools especially around from AI to just kind of get your feet wet uh experiment try things you never know what might come of it uh we just want to see people keep building and we want to thank you so much for checking us out this week and we we will be back the same time next week thanks

121. 1:01:28

     very much everyone and uh see you then thanks all bye