Skip to content
by Unity Nodes
NEAR Protocol

[REDACTED] Online Hackathon Wrap-up

about 59:23 long

Nothing is loaded from YouTube until you press play, and then without tracking cookies.

About this video

NEAR [REDACTED] online hackathon wrap-up with Devhub.

Live DevRel and Community Support: NEAR Dev Support Channels (Telegram): t.me/neardev DevHub Announcements: t.me/NEARDevHub

Transcript

8,403 words

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

  1. 0:24

    you want want want want what you want want

  2. 1:28

    want hey hey you want you want you want he he

  3. 2:34

    m oh n

  4. 4:06

    hey we are live again here at devhub uh I'm back your host Joe with along with Owen uh and we got a really great episode here today but let's start Owen you are live in Thailand right now how's it going it's uh it's going well I'm having a bit of audio issues no problem one second ah I know what was happening I had the live stream on my other window so I was hearing you twice I was so confused uh yes from what I

  5. 4:43

    caught hello I am in Thailand um you might have to repeat what you said again no just say just saying hello you you are coming live from Thailand we have the redacted event that is this weekend uh that you will be live at attending is should be uh pretty great I know a lot of people already out there yeah I'm uh I'm really looking forward to it um it's going to like you don't get to see people often that you work with in this this remote setting so

  6. 5:16

    I'm glad to see everyone in person not just because there's an event just nice to see people I've worked with and I know their profile picture more than their face yeah of course of course that's always that's always the struggle there right everyone's Avatar um but we got a great episode today we're going to talk about the redacted hackathon that we've been uh going through so much for the past couple months is over all the submissions have been submitted uh we are just lining up to kind of eventually

  7. 5:48

    announce the winners which will not happen on this episode we are not announcing winners uh anything we showed today is not a winner this is just we're just talking about projects uh we're going to highlight a few different things that we've seen just to kind of show off some of the work that's being done none of this is like an endorsement like oh this is definitely a winner like it's none of that I have I am not a judge I have not seen the judging so

  8. 6:11

    this is purely just like uh my own opinion uh Owen participated so he doesn't know what's happening either so um how was participating how was it being a hacker ow it was pretty good like a lot of a lot of effort to be honest because we we're building something quite big um and well in the last few weeks as it always is in the last few hours there was it was like a mad rush to the end

  9. 6:45

    you always think you're prepared and going to have everything done uh like halfway through and you think you'll be able to do do like double as much as you end up doing um but yeah big rush in the end but it was good like for me it's not always about trying to win money or whatever it's just being able to like be in the right mind and think hey for this next two months I'm hacking and that's like my number one priority and I can

  10. 7:17

    also because I was working on a pre-existing project it's also nice to it's like an easy way to get people involved as well I can just say hey we're doing this hackathon um come come build on our project our pre-existing project because we already have like the setup and the mechanisms and like we know what we're doing and we can just we can just hit it straight away like don't really need to fath about so like even if if we don't win any prizes it's nice that we got some

  11. 7:50

    work done and made some progress there but yeah in general in general was good and I like like it's nice to well I I was at the hacker house so there were people building alongside me but it's nice to build alongside people like online as atmosphere of of everyone building uh and everyone's in the same mindset so good vibe good building got some stuff done awesome yeah I think it's that's a feedback I've heard from a lot of different people so far and I I do feel like though the

  12. 8:31

    amount of time really allowed for more people to kind of jump in oh here we go good reference to the US election nice uh I think I think it really allowed people to kind of like you said like jump in who maybe maybe wouldn't have jumped in before Stu like that did you though feel like a times maybe was too much time like where you you find yourself still like waiting till like the last week to really make a push or

  13. 9:05

    is this a nice was this a nice Pace because you could also work your job as well oh I think he's Frozen oh well he's he's looking Frozen all right so I'm gonna talk to myself for a minute um we we'll see if uh Owen comes back let me bring to the stage here here I'll hide Owen for now um until he's ready to come back so uh if you haven't seen on devpost uh you can go to the project

  14. 9:36

    Gallery you can jump in and uh check out all the different projects that are right now they've all been submitted these are actually complete projects uh you can see their video you can see you know what they've actually wrote up um they're not any particular order I'm not sure what order these kind of go in guess you could sort them in here by like uh name or recent stuff like that so uh but what I wanted to do today was

  15. 10:08

    talk a little bit about again some projects that I've I've kind of just got a sense of that we're building and again these are not endorsements nothing like that I just wanted to highlight a little bit more of like what actually is out there and available so let's just do a quick little recap and and reset here so couple things uh went on here so if you missed the beginning part of this uh there's over $700,000 in prizes uh we had over, 1600 participants uh we had

  16. 10:42

    hundreds of submissions and then again some of these participants became teams and things like that so uh there was a nice mix here of of all of that what we then got from there is again you can go here and see is you know everything that's in the project Gallery is the stuff that was actually submitted on time things like that oh looks we have Owen back here he is there he is sorry about that that's the problem with traveling is always terrible Wi-Fi but I I'm back on data I

  17. 11:12

    think it should be fine nice nice yeah before you you froze I was just asking about the the timing of it you know with all that extra time was it almost too much time or was it nice balance you know did you find yourself still scrambling the last 24 hours what was that like yeah like I said in the end it was a bit of a scramble because I think to start with I was like oh yeah it it was somewhat too because I was like hey I've

  18. 11:44

    got ages to do this I'm gonna take holiday for like a week and a bit and not do any hacking and then when I'm looking back at it at the end of the at the end of like the so I was was thinking what why was I not doing anything for that whole week like we could we could have got a bit more pushed out the door but it's um yeah ignorant um yeah you're yeah you could always do more but we got done what we what we could um but yeah it was nice to have

  19. 12:22

    more time in general and build thing proper yeah no that that's awesome yeah I I I hope um you people kind of felt the same way uh this was definitely a different approach one of the things that I'm really curious about and I don't expect you to answer this necessarily but just kind of in general of with the amount of AI tooling that's available now is that gonna like change how hackathons are done where like I can rely on AI to write so much of my code it's really almost I feel like less

  20. 12:54

    about like writing the code and almost start to becoming more of like the idea in the presentation and can I actually execute on it because I can now have chat gbt write most of my stuff what what do you think about that my take on it would be that Things become a lot easier but like implementation is Not Dead by any means in in my opinion like sure I I I don't think the tools are there yet like this is very good at hey just build me a

  21. 13:34

    web app but with a lot of the models they're not fine tuned to building their smart contracts by any means like they are still using the old um SDK AI will take your price um and and when like when I'm writing my contract I think AI is great for front end and great for productivity in general but it's very very difficult to articulate exctly what you want because there's like a thousand different approaches you could go to building something and you may want to do it a

  22. 14:14

    certain way for whatever reason and it's difficult to explain it to the AI and I've not really tried to build a big project just through like prompting AI with chat GPT or whatever and maybe that's something I should try um I don't know have you ever done that I've done like like this week creating some script for some like an example I was writing and I just got AI to do all like all of that 100% of that but I don't know I don't

  23. 14:51

    think it could do well enough for a big project with something that's so specific yeah I I would agree with that I've never never done the whole big project either but just from my experience I can't imagine that actually really going well because you like you said you have to be hypers specific you have to know still implementation parts of it it's one of the things where like

  24. 15:14

    it it really is just a tool right now I don't I think if people are using it as the thing and not as part of their tool set like that's where you run into issues uh currently but uh it's going to be really curious to see how how hackathons continue to go uh going forward from here but uh there was a lot of AI available in this hackathon uh there was you know some of the prizes

  25. 15:38

    including were like best used of AI uh there were multiple AI sponsors in this as well so really should be interesting to see kind of also where that blend fits in here and what we actually get out of it so uh definitely excited to see some of the things but let's jump into some of the different projects uh and if you're just joining us again like these are not winners these are not

  26. 16:03

    necessarily you know endorsements of anything these are just some different projects that we we I kind of jumped in I went through the project Gallery which is fully live on the devpost site again you can go the project gallery and you can go see all the different submissions I definitely encourage you to go check out uh what different people built uh there's some pretty cool stuff in there

  27. 16:27

    uh but I wanted to again take some time today and just kind of go through some of different projects that I saw that I thought were uh just kind of noteworthy to me all right let me yeah there's a lot of uh I've done a bit of a scroll through and there is a lot of AI projects and and you can tell there's a lot of AI generated image as the the like what do you call thumbnail or something yeah I think go going back to what we were talking about before I

  28. 16:57

    think it was very good that you pointed out like most people and it's best to use AI as a tool at the minute I think rather than it's like leading and that's how I've been using f for the majority of things like like I say with writing a script it took the lead there but how I'm normally using it is just and um C or or like get will will help me out like in line and when it get stuck it might help me as well but doing like the whole implementation so yeah yeah yeah yeah I

  29. 17:41

    it's I mean one of the things definitely is people got definitely comfortable with different AI content generation uh you know that's going to be a big piece of it I I'm sure some of these videos that you're going to watch people where AI generated uh and so the the creativ ity pieces where I think all this kind of fits in but let's jump into one here and let's we'll start I have a

  30. 18:05

    few here I want to watch um so this one I guess we'll say this as scaven scav I I don't know how I don't know how I'd pronounce that I'm not gonna attempt yeah so we we'll just leave it um but here let's let's let's watch and and see what what's all about oh I'm clicking on the wrong thing there we go hi everyone I'm a today I'm excited to introduce you to SC an AI powered smart contract auditor designed for near smart contract as developer

  31. 18:37

    myself we know how important to maintain the security in our application especially in the smart contract to maintain the security we usually do an audit however manual audit are timec consuming and also expensive that's where scaven comes in it automatically scan your smart contract for vulnerabilities security issues and provide you with actionable feedback to help you bu more secure smart contract

  32. 19:01

    let me take you through some of the key features that makes scan special first we don't just scan the code it leverage AI model and combine with knowledge base of near common vulnerabilities and also best practice this means you get a context aware analysis not just a generic Security checks second SCA integrates seamlessly with your workflow you can run it as a get up action and then check it for every comit or PLL request or use it locally during

  33. 19:26

    your third is a highly cost isable you can choose your AI model adjust your parameter and then adjust your prom and a get report in a different format and the best part while scaving is free to use you maintain full control over the cost by using your own API and in the future I think with these with these AI tools uh I've seen a few smart contract Auditors yeah and I think a lot of them

  34. 19:52

    for example like the last hack th we in Singapore there was two submissions Auditors and I think what's becoming obvious is that it's not necessarily about how good of a job like the actual audit is doing because you that that's the part that's really difficult to be honest because you can you can just prompt a pre-existing model with a um like near contract DET or actual near contracts and like the documentation but I mean you probably you're probably not going to be training

  35. 20:34

    your own model on this right so no I wouldn't think so yeah so I think right now the most interesting bit to me on this has been like how is this linking in with a developer's workflow so I see here that it's linking in actions which I think is really cool so I just whenever you're making like a PR or a you like push your code to GitHub then it's going to run there for you when you don't really have to think

  36. 21:09

    about doing it yourself and so I think that's really cool and it's just like low overhead right you just see that the audit went well or didn't go well um other integration we saw in Singapore was someone doing it through um cargo n which is like the cargo tool we have for deploying and building contracts so it would like do an audit whenever you go to deploy oh yeah that that that is cool

  37. 21:42

    I I do think here the piece that is most appealing is the one the custom customization piece of that but also the other part of that is like you said here being able to like Drop it into your workflow so like through GitHub actions and things like that which is really clever I do always worry a little bit about some of these ones here where like I just don't feel like there's enough

  38. 22:06

    proper data out there where you know if you ask let's say one of the like chat gbt or Cloud to like write a near smart contract it's not going to be great you know and so there's going to be problems and I am constantly finding bugs as I'm just even not having like do code completion right like if you just turn on like co-pilot and you're doing code completion like like bugs will show up

  39. 22:29

    and so I'm a little skeptical of like where that fits in but I like the idea here of like fitting this into um your workflow you let's watch a little yeah I I was um just before we do I was actually quite surprised um when I was using pilot and I was writing some contracts like nearby asynchronous I'm I'm not sure how it works in ethereum I've not really touched like solidity contracts but it recognized when I was making a writing a cross contract call

  40. 23:00

    and a call back that should revert the state um the if the cross contract call was to fail which is something like people Miss actually a fair bit so I was surprised it it called that actually oh that's interesting yeah that's very cool for the local model support so without further Ado let's rise C all right here's how easy it is to set up SCA in your GitHub workflow first you

  41. 23:33

    just need to create a new file and then copy this yl code to your repository because we use a remote model don't forget to add the API keys in the secret of a GitHub letter here I already already did some changes to the code notice in the pool request or the action tab we can see the sca running its analysis immediately once the scin finished scan left comments directly on our po request as you can see here here's the

  42. 24:19

    with detail for abilities found in each files that's cool see what's behind the scene can we like give it a pause I want to see see what it's picked up on yeah let's see we again like I say I think this bit is very difficult to do right but I'd still be interested to see high severity I can't really see issue the mint function allows a limited minting of without any access control I know that seems pretty reasonable right it doesn't it's not checking for um like storage deposit

  43. 25:02

    right right right yeah no yeah yeah for sure for abili found in each files okay now let's see what's behind the scene scan uses retal uped generation or R to improve its quality and the accuracy of the smart contract Audi how we do that we maintain a database of known vulnerabilities Security based practices and common issues in your smart contract all right for the process during the audo process when analyzing a specific part of smart contract we perform a simility search in the factor stores to find the most relevant entries

  44. 25:39

    from our knowledge space and finally after we find the retri information the this information and is then provided to the model all right so what are you waiting for head over to our guub try SC for yourself and let us know what you think thanks for watching cool yeah I mean look I I I like the the approach uh you know definitely you know again leveraging near in in a different way um yeah I I think

  45. 26:12

    that's let's try another one here yeah I think those Auditors are really interesting um and like I said at Singapore we had a few of these submissions and I saw that we were going to have submission so before um the judging happen whipped up a purposely terrible smart contract to see to sort of Benchmark these Auditors um and most weren't catching the errors but th this one this one seemed quite good I I could see they were like training it specifically even on the

  46. 26:53

    Nate like the SDK SDK s SDK that just didn't sound right for some reason yeah SDK types yeah we'll forgive you it's late there so we have a AI crawler next so another AI one here uh so detect the semantic structure of the website user visits parses data on the Fly users reward for just browsing interesting okay let's uh see what they got got no sound okay no audio all right so we just you can do a talk over it if you

  47. 27:39

    want to yeah uh yeah looks like they're scrolling a Twitter here all right got a little little scrape data scrape okay jum we get it maybe using like Master Maybe seems to be is like scraping each yeah website you're going on so here let's let's get a little B more information uh impl is a web browser extension unintendedly scraping websites to user visits once created once for the website

  48. 28:18

    uh interesting so where's the near part of this I'm guessing payments paying people to browse the web I guess payments well I think in this uh haathon it didn't even have to be near specific as long as it was conforming to one of like the Bounty sponsors so I suppose if they were using something like Massa it could be under the um under that Bounty yeah the I think it was called like the chain agnostic Bounty or something yeah yeah but yeah this is this is interesting how

  49. 29:04

    it's sort of like passively scraping the websites you're you're visiting oh I see some seems to be some like wallet up there or something yeah I think it's probably like it's you're getting credits for rev viewing and stuff like that you know um yeah it's this you know again I I think people I like that people are trying different things here uh trying to monetize you know and scrape in a different way uh I do wonder how while

  50. 29:37

    that goes over I could see sites blocking that uh Brave the brave browser does something similar with like how you see ads um you know and they were rewarding through a token so kind of the same concept there um yeah I think these concepts are like sort of going back to n's routes right because we know that well if you don't know they started off in AI originally and then they wanted to do I think it was like a six month detour to build like a

  51. 30:10

    blockchain for payments right and they ended up being being a bit longer than six months uh so using using with like Ai and the blockchain for payments is a good good use case in the original one like proposed by me so here's a new one so this one's called web3 content Hub this one is leveraging one of the the Bounty sponsors uh Masa which we had a workshop with them here uh on devhub live and it looks like what they're doing here is uh what

  52. 30:49

    thing we host events frequently without little notice scram make quality social media content on time uh so here here's this see the got I am the pr of Club in my college we conduct a lot of even sometimes on a super short notice so there is hardly any time left to create content for social media last minute captions Rush post and some random images so for this we needed a smarter approach so I found about react hackathon but there is only 10 days

  53. 31:28

    to do my project with only 10 days left I have divided the project into three segments first one is preparing the AA model and second one is the front end and back end and third one is a deployment use masa's prescribed tweeted data thanks to Masa they have provided the data in the CSV format after converting those five CSV files into a single file user enter a query it passes

  54. 32:01

    to the database it gives top 10 results according to the query and serves as a context for the llm to produce a caption for that so how this caption workflow Works firstly after sending those 10 tweets a agent is responsible for converting those 10 tweets and understanding the context and other agent is using to converse those tweets into a p passage where agents know how to what the context is about so first off as we continue through this video

  55. 32:40

    production topnotch like brought the video production here we got the we got music we got Fast Cuts so I mean he's he's on top of it uh initial thoughts so far for myself I just you know uh I like the use Sasa uh that's obviously nice putting that stuff together uh though the question always becomes and I'm curious to see if he answers it in the second half here is like how does

  56. 33:12

    this outperform the more the current generic tools you know why can't I just use I think your your first comment about the video yeah it's an it's an amazing demo I think stubby and a well we'll see how good this project is in the end but maybe if it doesn't work out you can be a like a video content producer yeah my demo was was nothing like this it's just my monotone voice going over it yeah just ripping through I think that's like 99% of the uh the

  57. 33:52

    videos here right are probably going to be just people doing like screen share and like let's let's go but uh yeah and another agent is defined to write the caption and also the image description of how it should look now the air model is done but I have never built a website I have zero experience with Nick JS and backen but I know what exactly I want yeah thank you too much is this AI generated yeah that is feels like

  58. 34:50

    it good this left and I hit rock I have zero idea about the deployment I have have two paths to choose from either Rush the development and break everything or tell its story and bring it to it life sometimes the Gen is more beautiful in destination and here am I with my project H interesting interesting well I mean Jo jokes well maybe I guess part of this is like I feel like if you most developers would just be like I'm just going to break it I'm just going to like let's just break it it's 24 hours before I'm going to break all the all the things and I'm

  59. 35:36

    going to scramble until till we're we're ready but um I love the video I like I like the the what went in there um yeah you know I mean to be honest I I I'm not quite sure what the the the project is is doing to be honest but I felt like I was along for the journey there yeah yeah for sure yeah definitely not long for the journey and I think this is also too where like for this is where I think using the partner piece of it is interesting in a hackathon when you have this larger hackathon you have all the different

  60. 36:19

    partners like I think you're going to probably see a lot of submissions that are like this that are probably more focused directly on like hey is is really like being submitted to Massa right it's not really necessarily near project uh it's an interesting idea it's solving a problem for him you know so you have all those pieces there but from like a larger like near project doesn't quite fit into that but again it still fits into saon because we have these partner prizes

  61. 36:49

    so yeah yeah it seems to have it mentions n in here I suppose it it might be able to be applies really yeah all right let's try another one here next one Flip Flip is an AI powered platform that makes Smart contracts and near trans protocol transparent accessible Emer GPT for analysis with oh Dow intelligence all right did you dropping the dword here um let's did you pick any the on AI I

  62. 37:29

    think everything is AI I just I'm just picked through I think I think everyone just everyone used AI this year it was like you know what like I I guess I'm I'm willing to bet anyone who didn't use AI you know like it probably still feels like use AI or maybe everyone just used a I don't even know but yeah my project did not feature the letters a and I together at all well you just lost so sorry probably I'm I'm

  63. 38:03

    behind gota get with it all right let's let's take out this video here you got to mention it in your in your like pitch deck if you're pitching to VCS even if you're not using AI yeah you just like yeah you know we're using AI to make sure that we've optimized uh you know beverage orders on the third hour of coding development like that basically FR is an AI powered platform that makes Smart contracts on near protocol transparent and accessible we

  64. 38:38

    merg gbt 40 AI analysis with Community intelligence to help users understand verify and safely so when you get into our application of Frip near so you're going to see this UI the first search bar here and you just start with connecting your wallet first in this case we're going to use M your wallet but we also support your wallet bit wallet senders as mentioned in the near wallet selectors

  65. 39:09

    okay and let's try in this case first start you need to know the smart contract of those protocol on near protocol I mean on near chain or by searching every I mean every dab or defy or call on near there will be they'll be showing the smart contract on that documentation you can search them and then just put it it in here just as as you like yeah in this case we going to use our our own protocol I mean r near to show the C okay once done just

  66. 39:48

    enter okay our AI will do we perform the analysis to check whether what is all about yeah even you are not you are not a programmer all right so let's I think we can kind of summarize little what's happening here so it's looking at just what are the actions that you can take on the contract so it's not doing an actual evaluation of the code um and the reason why um people aren't aware is because so near compiles into wasum and

  67. 40:21

    WM is not directly viewable you can't just like look at a wasm file um and see everything in it so so there is there's challenge there so this this is just looking at the methods that are available to the contract um interesting to take the AI approach I do Wonder like if you are how much fit is AI yeah and like also on the same side of that like how much are you really getting out

  68. 40:53

    of info out of this like where the AI probably can recognize right like a basic like token contract nft you know um right maybe some like def stuff but then like if I made something truly novel um how would it know like how would it know because it has no context around it so I be curious do you mean about like trying to like pass the bom or no just like all right so like it's clearly just looking at the methods right so like what if I

  69. 41:23

    added a method called like I don't know get mango right like what how would it know what get mango means right I mean it's like but where where is it pulling these methods from like the ABI or it must be it must be similar to like how like near blocks works I'm assuming right because near blocks let brings in all of that so must be just from the AI yeah uh unless they're using like something like a sour Sky

  70. 41:55

    uploads the whole C then you would have a good chance recognizing what it does yeah yeah more not coder or SM contact developers you still understand it because it's it's going to use simple word to interpret all those complicated smart contract so you can understand how many functions this this contract uh functioning and also what is all about in this case in this cont for FR testet it's about social interaction contract

  71. 42:28

    that allows user to comment to add comment lies dislike to the post so it it help user engagement and sentiment tracking through comment and rating yeah moreover our AI we also do the evaluation for the contract analysis as well so what is what would be the risk or what you need to like uh to see before or what it's all about how so in this case there will be some risk that the contract rely on user

  72. 43:02

    generated content which may include in appropriate or misleading comments additionally there might be the potential of manipulation of light disly rating through coordination action here okay and we also uh our AI you you check toly in detail and then put the unverified contract because of most of the contract on near protocol have not been verified so our FRN going to help this issue because you cannot check the code it's not that really transparent so it's quite really hard okay so they're they're

  73. 43:43

    acknowledging that you right they can't see the actual code you any vulnerabilities um hoping that at least from a descriptive standpoint you understand what's going on uh yeah definitely need a warning in there where like this is not a a method of security necessarily right like it's just giving you Insight but um someone could still put malicious code in a contract around that I'm gonna jum ahead a little bit here I think with

  74. 44:14

    these um with these demos uh like I know with my own demo I did the demo is a lot better than actually interacting with it like there's there's some there's some jump Cuts in there and there's some like uh yeah there's jump C to make it a lot better and you're doing very specific things a lot of the time so I would like to interact with a lot of these um like submissions I think

  75. 44:49

    what it is that offering here seems actually like something I would use because I'm a I'm a big documentation head and I like to write out exactly what every method is doing so if I come back to my code at a later date or someone else does then you can easily see what it is doing and I think this would save me a lot of time I'd be interested to see if it works for my own contracts really yeah so that's a great point so I think like like the use case

  76. 45:20

    here maybe is even a little different than maybe they first kind of thought about but um I I always app appreciate if someone's trying to again put some tools together trying some different stuff um again I you know the the part here that is I think a little bit novel is that they're trying to make it again uh easy for anyone not being necessarily technical um the Dow based governance

  77. 45:50

    system I guess of like probably verifying the contract say oh this is a okay contract that's interesting um to some degree of like a social layer of saying you know Yep this contract is safe or this contract should be blacklisted things like that um I always wonder how much people though would ever participate in that kind of stuff um but listen you got to put it out there to to see first so uh definitely seems interesting you

  78. 46:20

    know uh again the challenges here like again it's just you know leveraging GPT I want again I can see projects being matched together a little bit you know people who are making their custom things and stuff like that might be interesting too all right yeah I think I think you see like a lot of similar submissions um like I was doing a betting based one crypto betting yeah it goes very hand in hand so there was a lot of

  79. 46:53

    betting submissions and I've been around trying to think hey how can we collaborate with these other projects because my view is like I'd rather something be ex be successful um and like lower my my token allocation or whatever and have it actually happen then never happen yeah yeah yeah yeah I think that's uh that's a good point I mean and there is only so much I mean it it is hard to be creative you know like a totally new

  80. 47:31

    novel idea but it's where I think execution can certainly make a difference ESP hackathon a little bit longer here like this one where it's a couple months um you know hopefully you start to see differentiators from there uh so let's I check out this one this one seems interesting dead project sniper or snipper sorry not sniper snipper uh I thought it was sniper as well yeah yeah S I just know well in America we love our guns so I always think gun first

  81. 48:05

    um uh so uh you track social activity of projects that received grants to see if the projects have met an unfortunate end I kind kind of like that I like the uh the thought this one seems unique yeah let's see what's going on here welome to dead project snipper record for projects that have met on unfortunate end this tool tracks projects that receive grants from poock our goal is to identify challenges early on provide insights and help make data risent

  82. 48:35

    decisions to deal with the triing projects so let's start by taking a look at here where the project from polock list is listed currently we don't have everything on here only a few selected top projects which have the most donations in if if you select one of the project like this it will show the graphs and the scores and the reports like this and note this is a d dat we are still wrangling with it and what we

  83. 49:12

    focused on is to build a pipeline and yeah which we used with Masa and neblock indexer to retrieve social and launching data so uh right off the bat I I understand the problem they're trying to solve uh totally makes sense to me uh they're intera inter integrating multiple uh available projects and tools on near uh really this certainly seems interesting and this is actually again

  84. 49:42

    where it's like this is a pretty novel approach in my mind where it's it's fairly straightforward like it's just using again like the Twitter data from Masa it's using commits it's using uh contct updates and stuff like that but uh they made it look really nice that the dashboard looks great um you know even though probably it's a it's a beta dashboard yeah it's it's pretty pretty sexy dashboard I'm wondering a bit more about how it's

  85. 50:14

    working because I saw they put in like a near account ID but then it's giving information for like GitHub and and Twitter so I wonder how it's getting that data that was like these socials were submitted as part of the or do you think it's so Advanced that it can go from like one link to another and like uh scrape that and find other links and and like navigate down this tree yeah that's

  86. 50:46

    a great question I I wonder I definitely wonder how that's being handled here and I could certainly you know it feels like they have a lot of room to kind of run with this a little bit and see how it goes I I I don't know if that data probably exists today of like having those connections but uh it would certainly would be pretty cool to to get to that point yeah yeah let's watch watch a little bit more like and a half and this shows the the Sumer core

  87. 51:20

    driven from the activity data which is shown like this and for scoring R report we utilize llm as a judge to assess whether a project is effec effectively dead or still progressing the assessment is backed by a combination of textual descriptions and the visual data representations here uh these insights should help stakeholders to make informed decisions regarding future actions so this is more like a report

  88. 51:52

    View for the stakeholders and we have another UI that helps the experts uh decide on which metric should be like worth something like this experimental dashboard is designed for subject matter experts to quickly iterate and experiment with different data analysis approaches you could just press buttons here and then it will run the code here uh we will end up uh giving reports like this so you could experiment with different prompts and down here it's not created yet but if

  89. 52:25

    you could create create plots and just uh use different kind of comparative analysis down there but that projector snipper on aims to raise awareness about dying projects and provide early signals for intervention uh we did not Implement yet but we are planning to alert stakeholders and initiate to initiate review processes and furthermore we want to go build some more features for the agent and further the AI funding agent

  90. 53:02

    initiative interesting um you know I like the notification piece that makes sense to me uh yeah I mean there certainly solving a a problem here uh it would be neat if you could like link this into a couple of different features like um you know like in the chat saying you know can can can it be botted still can you fake it um so how are you evaluating that can you do some bot detection uh

  91. 53:35

    could Milestone payments be tied to some of this could you start to build this you think of scale right so uh certainly a lot of opportunities uh here I think is really really pretty neat this seems like a pretty decent project uh overall and I like where they're going with it at least definitely well we had plug from partl previously on here on live streams and it seems like this is exactly like one of the components he was looking for and you could really see the when he was

  92. 54:14

    talking through the AI funding stack and the components to build this AI funding model that like there was a lot there was a lot to it and you could down the rabbit hole um really make the whole process so it's like can get really complex and really complete and I wonder like what stage we can get to right now and and how much can be automated by this but this seems like a step in the right direction for sure yeah absolutely so uh really

  93. 54:55

    excited to see you know again people trying to think a little bit outside the box and just trying stuff that's what hackathon is for right you don't have to have perfect project uh but definitely try to get out there and try Ser stuff uh overall look I I I it's really neat to see um what people are building uh I'm sure there was so many submissions I'm sure there's a lot of great ones and we just randomly kind of grabbing some uh there a lot of AI I

  94. 55:22

    think it's be a big theme I'm really excited to see some of the winners that come out out of this uh looks like there's some pretty good competition all the way through uh if you didn't get a chance to participate uh on this one but you want to keep building un near uh the down below we've been scrolling the whole time dev. n.org uh you can get started there uh there is always time to to build and try different things and if you are going to be live in in person in

  95. 55:54

    Bangkok this weekend for the redacted event uh it should be great there is a small challenge that will be there as well it's not necess a hackathon or anything but there's a kind of a puzzle that people can participate in uh and may win some prizes as well so it should be should be a good time overall yeah and if you're out there and spot smot a little tiny me and come and say hi it' be great to see you in person

  96. 56:25

    and we do yeah if you didn't get a chance to participate we are also sponsoring um e Bank should be good um we'll have some some more bounties there which should be like Ai and chain abstraction around there absolutely absolutely yeah it's uh again I I don't know the the status of like any kind of uh tickets whatever else but again if you don't know the irel event uh a lot of great speakers including uh Edward Snowden doing uh a remote speaking engagement he obviously won't be

  97. 57:04

    there so uh one of the things here that again if you want to see the schedule of everything that's going on this will all be live streamed so if you cannot make you can still watch all the great videos hear all the great talks uh and just you know see what's going on near I been told that there's going to be a lot of great Alpha at this event a lot of new stuff coming a lot of announcements that

  98. 57:28

    are happening at this event directly so definitely check it out uh it should be should be really fun I'm sorry I canot be there so but it should be really great yeah I'll make sure to take a lot of pictures to make you jealous just as I did with uh when I went to the F1 appreciate that yes yes it should be fun but I I do not miss I do not miss the traffic by any means I've seen some pictures of that and I've seen some pictures of people trying to cross the road and I am

  99. 58:01

    all set with all of that so enjoy that um again uh yeah how many near Foundation employees will get run over during the weekend right let's hope with zero but maybe there's a poly market for that um over under on that uh look this was this was fun this hagon was great we're going to keep doing workshops and stuff here right near devhub uh so devhub live will continue uh different weeks to do stuff we actually have one coming up near the end of November uh around privacy

  100. 58:34

    building uh taking that a little bit step further there's some great stuff really coming down the pipeline just because the hackathon is over the event's going to be over after this weekend we are not done there's so much more stuff and there's a lot of great stuff that's going to be being built here on near that we want to make sure everyone's a part of subscribe to the newsletter down below uh Owen personally curates it each week so so make sure that you're you're following along and

  101. 58:58

    you see all his great uh effort and work that gets put into that and again uh any questions that you have or whatever else you want to get started building dev. new.org is the place to get started uh and we hope to see you there uh until next week uh if you are in person and you are at B got redacted have a ton of fun be safe learn lots of stuff send me prizes and I will see you all next week

  102. 59:23

    bye folks bye-bye all right by now still live and bye