DevHub Live 19

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

1. 0:00

   oh o

2. 1:15

   what so we believe to stop marketing starting D that's what we're doing over at devhub is that now the focus is no longer just about let's sell you something on why near is the best developer experience which it is right we all know it that's that's why we're here but we want to actually start to engage you as a user and better understand what are you need and how do we help you be successful instead of us trying to put our product to

3. 1:50

   you and welcome back to another edition of devhub live I am your host Joe from near devhub uh we have a really special episode today uh we're kind of trying to change things up a little bit if you don't already know we have the redacted hackathon going on right now uh it is over I think $600,000 in prizes right now uh get involved you can you can still join you have until the end of October to submit your project fully online tons of great categories and so

4. 2:22

   what we wanted to do was kind of uh change how we're using devhub live a little bit we always have these great interviews talk about different things happening in the ecosystem but because of the hackathon we really want to dive in and get people acquainted with how do you actually start building on some of these different you know uh partners that are part of this how you start building on near all these different things that are happening and today uh

5. 2:46

   we're really excited to have uh B to come on because they are sponsoring uh some bounties there uh through the reacted hackathon and they're going to come and kind of we'll do a new segment called Teach Joe and they're going to come and teach me how to use this stuff which in turn teaches you uh as always you can ask questions in the comments we will try to bring them up and address them if we can uh but otherwise uh let

6. 3:10

   me bring uh Nate up to the stage from B neate hello how are you doing great great thanks again for joining uh you are now multiple time uh participant on devhub live so uh congratulations I'm just here quick and we'll be leaving and exiting as soon as Mark gets on here he is the one who has been sort of helping out micro micro's been leading the way on the whole agent building system uh but Mark has been

7. 3:42

   implementing a lot of the sort of cross agent Communications so he I'm quite excited when he actually joined let's see ask me anything yeah while waiting for for Mark um you know let's just talk a little bit about like I guess what are what's the highlight level of what's capable right now on bitte so high level is let's say we launched so let's do a little timeline right we built our first chat BT agent a year and a half ago so building these

8. 4:14

   things doesn't just happen and all of a sudden magically you talk about building the best transaction building system that exists right so a year and a half ago uh we built the two chat PD plugins realized we couldn't execute uh main net transactions through those plugins so we then put it onto ai. mbas XYZ uh which then forced us to build a wallet uh so that we can make those transactions easier and then hey Mark uh just going over where we're at today uh but just to

9. 4:45

   finish that let's say uh then the wallet now started doing its own transactions and we basically just wanted to solve all of our nft problems right so create a mint uh create an image and mint uh buy all the nft on this contract list all the nfts on this contract once we did all of that uh micro did it a step further and said let's make defi swaps so when he built that agent that that was our first external agent uh we kind of had this aha moment of like what if we open this up to everyone to build their own agents

10. 5:18

    and we focus on the cross chain signatures the security and maybe a little bit more of like the the visual elements like reporting like show me a nice chart of eth over the last two months and these sort of things while folks started building on top of that nice so that's where we're at now the very first implementation uh we thought 20 agents could all talk with each other and we could figure that out

11. 5:41

    that wasn't the case uh so we took a step back and now we're reimplementing a new system where only two or three agents can talk at a time and you can select these two or three different agents uh together to make those transactions happen and focusing on the playground to create this world where developers can easily create an agent within less than five minutes so Mark's going to take us away a bit more on that side uh go for it cool uh Mark thanks so much for joining us before we jump in

12. 6:13

    though with everything I want to level it a little bit around the hackathon uh but so let me just I'm going to share my screen real quick here and then Mark I'll let you show share yours so uh if you haven't already uh check out the dev post here I'm going to put the link here at the bottom uh check out redacted.ch you how to actually use this stuff uh any kind of little tips and tricks that we should you know be aware

13. 7:05

    of as builders for this bounty uh I think Mark's going to cover the majority of these uh we're going to update our docs today uh we basically wrote this long dock of how to bootstrap your own agent um we've now launched this new thing called make agent which is going to cut those docks down in half if not 34s uh which is going to make the the process way easier uh and the cross chain mechanisms so right now we are currently building a cow swap uh agent

14. 7:35

    behind the scenes so don't build a cow swap agent please um but anything else goes right like being able to have conversations with Dows that's would be a huge win uh and just uh opening up the world also the idea is don't just build an agent uh you can extract money by building these agents so what's the reason to build an agent if I build the best a better cow swap agent maybe I pull a better fee from it uh you as the

15. 8:04

    agent Builder can make that happen maybe you up the anti just a little bit more than the Swap and then the uh ease of the creation of that uh works so we'll be publishing the cow swap agent uh so if you want to create a Unis swap agent um and use that as the base that's totally fair game uh is that fair to say Mr Mark yep exactly cool all right well Mark uh thanks so much for joining uh can you just kind of

16. 8:35

    introduce yourself to the audience a little bit of uh you know what your role is at bit day and and what are you going to teach me today yes uh I'm Mark um yep at betza I'm doing uh basically a lot of the AI stuff working closely with lu on that and yeah we've been focusing a lot on the playground and the building these agents making it public for everyone so that's what we're going to jump right into and show you guys how how to

17. 9:00

    basically uh set up uh your own agent test it in Dev live and be able to test it on the wallet and then also be able to register it so anyone else can see it publicly and use it as well on awesome awesome yeah so feel free to share your screen whatever you need to do uh the floor is yours awesome uh we're on the docs here let me know if that we're good to go on the screen share yep we're all good awesome so yeah you're going to want to start just jump right to our docs and right from here we can see building

18. 9:32

    agents and this is pretty much a pretty good guide on you know uh start to finish what you're going to want to be doing and what we're going to walk through in this example we're just going to do uh we're going to use coin gecko's apis so they have pretty simple apis that a lot of people might be familiar with and we're going to work to grab uh cryptocurrency prices and explain basically how we can

19. 9:57

    just take any API and uh make that possible for an AI agent to understand it and also to combine that with our functionalities on the wallet so perfect uh essentially uh what we're doing is once we have our uh ref swap agent here that you can look at this is a great example and you can go ahead and Fork this and get an idea of of where to work from and I've essentially done something something very similar to that here

20. 10:30

    where I have this coin gecko demo which is also a public template and I'm just going to walk through basically the main differences between what we have here and a typical next project so it's very basic next project uh but in our API route uh we're exposing this route AI plugin and we are instead of uh we're just writing uh an an open API spec here uh which is very common like a Swagger spec and this will be then served with a

21. 11:03

    dynamic uh server URL and using our make agent tool which is a new tool a new CLI tool we have uh this makes it very easy to test this live just by having this uh one route we can serve our a our spec as well as test it live on our um on our wallet and can I let me just pause you real quick for a second um so interesting approach so how come you're taking the spec approach for that just for people's edification uh yep so the open API spec

22. 11:40

    is uh you know very popular and used for a lot of uh a lot of apis um across a lot of you know services so we just added a very small addition to it um which is this uh beat extension here and this is some of the information we need to create your agent and the idea is to just make make this you know as easy as possible to turn any API into something that works with our system so you're not in because of this setup you don't have to actually do next project but next is the recommended because you have the

23. 12:11

    templates but it doesn't have to be in next exactly yes yeah we're using uh yeah should definitely made that clear thanks for that we're using excuse me we're using next for the ease of uh you know deploying and testing uh quickly but right essentially you just need to serve uh on an end point and op an open API spec so you can use anything you want for that um if you actually follow our ref

24. 12:38

    uh agent example this is uh using Elia it's it's it within next but it's using Elia which is very uh very much like Express and you can just do that with any kind of uh Service as well nice very cool uh yep so what we've done um we have our spec defined here uh now importantly what we're giving it is our new account ID and then we're also giving it uh some info about the assistant such as the

25. 13:09

    name we want for it a simple description and then instructions where we're going to go more in detail about how it should function with our uh our apis especially if you have any like caveats or or anything like that ours is quite simple we're just going to use the we're just hitting the price API so what we've done here is just use following the open API spec we've given a uh a description for this endpoint and to get that and derive like

26. 13:41

    how you're going to write this out um we we're just looking at the endpoint basically seeing the requirements for the endpoint so this is going to need uh the ID's parameter this is going to need the currency we're basing against and from there we can uh basically give it per in the parameters we're telling it that we have this parameter IDs and we have this parameter versus currencies describing these parameters and basically also given an examples which help make it very clear uh for the

27. 14:12

    AI how we're going to work with this nice and once we have that all set up uh this is essentially this is all we're going to need is just this route being served on any kind of endpoint um we've created a tool that will help you during devel ment which is the make agent tool and uh basically what we're doing is just by just by running this it's going to uh help us to serve using local tunnel

28. 14:42

    um uh your you're um open API spec on a public endpoint so that we can also connect it to our wallet playground directly and we're going to jump in and just do that right now so the only thing that that we're going to need to be adding uh for that is this Dev script here so we're going to want to install make agent and then add this Dev script and if you want to you know make it even more simple you could actually

29. 15:16

    run this dynamically this uh tool so that you don't even need to install the make agent but there will be some other things with make agent that will help you during development cool and if we were to once we run this uh script this going to launch our Dev server as well as forward our uh Local Host through local tunnel to a URL and we we're also given this link where we're passed into the uh smart

30. 15:45

    actions playground with our temporary uh agent and we can see that already our agent um is giving us you know some responses based on our based on our open API spec and descriptions MH and we can test the price function and this excuse me this will call our agent uh which will then know which route to use based on our spec and the functions provided and now that we have this live this uh make agent tool some of the cool

31. 16:27

    things are is that any Chang that we make to this uh AI plugin uh route now it's or the spec in general it's going to automatically be um refreshing and updating our agent on our playground and one thing that I forgot to mention that's actually very important uh specifically for next the way that we're serving this because uh based on our docs we have to be serving on the well-known plugin route let me just find that really

32. 17:05

    quick yep so this is yeah so this is where we explain the where the plugin manifest has to go so on your URL on this route we need to have the um the AI plugin Json and to do that with next we're just uh and to build it dynamically we are just doing in our next config uh a rewrite so that we uh actually are serving that API route to this URL and that's how we're able to find it

33. 17:38

    um through our registry and just to give uh now like a walkthrough of development we're going to add a new we're going to add like a new uh functionality and we'll keep it simple um we're going to do this this route here for example um so we have an API route to Ping the server and we want to basically add this to our ai's uh functionality so first thing we'll do is we'll build that route ourselves in our

34. 18:20

    project and let's just set up something basic and what we're going to then want to do is we're actually going to fetch out we're going to fetch the uh URL provided here so we're hitting the coin gecko endpoint within our route and we're going to return the result of that basically cool so yeah very simple just hitting this we're just hitting this and getting the result and we can just test to make sure that this

35. 19:13

    works and we have our ping route let's see what we're doing here oh yeah oops just need to do get the actual data and we see we're getting that data there so now we have a simple route that's just giving us whether the API is uh up or not and we need to now actually add this to our um spec so that we are basically explaining to the AI that it has this this functionality to call this route

36. 20:00

    so in our paths same as we did for the price we're going to add one basically for just making this ping and we are going to use cursor for help here because writing open API specs is not very fun no definitely not and basically we can just go through and this is not what we want people you know spending a lot of their time on obviously so right uh we could go through now we have this ping we have the location which is in our API ping we describe it very simple endpoint and

37. 20:50

    just by Saving this we should have gotten an update to our uh a or to our plug-in Json m and we should be able to now ping from our playground and this should have our agent calling that ping route and we get this uh request that we wanted basically nice let me let me ask you a couple questions here that I think people maybe will have so I think the important part here is one when you create the new endpoint you have to make sure it's in the spec just because you made it if

38. 21:32

    it's not in the spec it's not going to load in the agent yes exactly yep okay yep as you go and as you you build uh and you just showed here so people are aware this is just like any other next app as you are saving it is updating and you didn't have to redeploy you didn't do anything different right we didn't miss anything yes exactly and that's only because we're using the uh make agent

39. 22:01

    tool um which is serving to this URL and it's watching uh this file it's it's basically watching our our AI plugins file nice uh you mentioned early on the so the initial directions that are generated so that is generated based off of what's here in the specs you're not feeding that anywhere else correct uh yes yep exactly yep uh so this will register an agent uh on our platform using this information and then

40. 22:39

    when you prompt our system uh there's some additional context added when when using your agent as well but a lot of it's derived from this uh spec here right so there's no there's no other I don't have to as a developer I don't do anything else as long as I'm detailing out my specs here with that information then it's going to do everything else for me yep exactly yep uh super easy are there any restrictions around different data sets or endpoints

41. 23:11

    like can I bring in static data can I does it have to be something that has to go fetch uh it can be anything so uh yeah we can return whatever we want here and that's what we end up end up uh giving back so yeah you could do your own logic here you can take from an API change some of the logic um you know if you want to take a fee anywhere you can do stuff like that um so yeah that's the idea there basically by take a feed you mean like charge the

42. 23:42

    user yeah or if you want yeah basically like if you're uh if you're if you're providing a service that's going to build a transaction um maybe you want to add within the list of transactions like a small fee or referral uh something like that that's a possibility can you just show like an example of how that would go together um yes we can uh we can play around with it it might be a bit Yeah that's what we're here for awesome Yep so uh to do something

43. 24:15

    like that let's jump into maybe the ref example um before before that maybe I'll just wrap up uh registering this as an example yeah we do some more like experimental stuff yet so we've been working on this plugin um this is all on dev so as soon as we close this uh our agent is actually going to be deleted and it's not you know accessible anymore in the playground so um once once we're done with that uh we're going to need to be we're going need to deploy this on a live service

44. 24:50

    and I'm using forell here for example and I've done that where I've deployed this to the service here and all we're going to need from there once we have this service deployed and if we visit this uh the most this spec is an optional thing to you do you can do for users but the most important thing is that we're serving our uh open API spec which we can see um here that we are so our URL has to serve the spec that's the most important

45. 25:24

    thing and we're just going to grab our URL it could be the full URL or just the domain and we're going to head over we're going to follow the docs and head over to our uh agent registry endpoints and here this is tentative to its own UI and stuff of course but right for now we're going to use this API ref and we're going to register a new plugin and you can also do this you know through post minut or whatever you may

46. 25:57

    need to do it with mhm give it our uh URL and this will register the agent uh and if it's successful it'll give us a debug URL and it'll give us our API key so you're going to want to save this API key because we need this uh for updating the agent or deleting the agent and you're gonna you can take this debug URL and just start testing it right away in the playground uh but the difference here is that this

47. 26:30

    is the persistent playground where I can share this link with anyone and it'll always be up and running got it so all right that's cool very nice when does it become like Fully live like outside of the playground so uh we're not doing that just yet but essentially you're going to want to make like a verification request on your agent and from there once it's approved we can make that open to anyone through like the UI or through uh a

48. 27:02

    registry otherwise it's just going to be like uh you know like a link where you can test and share with people if you want them using it through the playground yeah so for the sake of hackathon that's perfect right so they just as long as they deploy it they can they can show it through the playground uh that's that's all they need exactly and that's the requisite for the rathon is we want it to be uh available with a link on the playground that we can test and use

49. 27:24

    nice and so yeah that we'll wrap on that and we can jump in to uh taking a look at the ref agent for example yeah and we can see some some uh maybe some cooler stuff nice so so yep so this is our ref agent example mhm and so here we're just serving our spec uh raw we're not using the make agent tool yet on this example and one of the one of the important things now that um just following from the docs so that we don't

50. 28:20

    get too lost um see so so we don't get lost on how people can find what we're talking about um so we have this tools field on the uh beta extension and if if we're going to want our agent to have the ability to generate transactions we need to specify that we want that Tool uh available and so if we don't have this for example we our agent is just going to you know use our endpoints serve our and provide that to the user if we also add some of the

51. 28:56

    beta Primitives uh like generate transaction or submit query it'll also have those functionalities as well nice okay so here we're specifying that we want to be able to generate transactions and then we also have our uh our end points here for our ref agent and if we jump in to what our swap route is doing so here we have a swap route that's using the ref SDK and actually getting uh the best uh swap

52. 29:31

    route for the provided inputs and in here uh we can there we can either add a ref a reference to the uh to the ref uh transactions I believe they support uh adding like a refer referral um or otherwise we can just add to our list of transactions because we're returning here we're returning a list of uh near transactions on this endpoint MH and uh if we wanted to add something else there we can do we can do something like that

53. 30:10

    so it's like a batch transaction at that point yes yep cool yep um yep so basically that's that's what we're doing there um further there's one more thing I definitely wanted to share uh I don't know how much we want to do with like you know testing this live I don't know how much time we have or if we want to jump into that we can do some stuff like that but we have time yeah even more let's uh yeah I mean the goal

54. 30:49

    here is we want to also like Inspire in people to try different things and so I think it'd be great if you could just show just different ways that we can kind of be creative with his uh awesome okay sweet so let's let's then uh work on let's get this using make agent first that's one of the things we're going to want to be doing um so we're going to want to be serving this let's add a new route here I wonder

55. 31:23

    if this is going to be fine to do a H yeah actually let's let's see I'm thinking because the only thing is this is using Alia which could be a bit of a funk show feel AF free to make mistakes we all do as developers we all always experiment so okay don't to be perfect don't worry about that all right let's see what we're let's just jump into the we're working with then so this uh example is basically serving a open Avi

56. 32:17

    spec and well we are in order to test this live we're going to want to make this part Dynamic so let's see okay so we can test this live at this link the thing is this is going to now use the uh assuming this is live let's

57. 33:23

    see okay so what we're going to need is a a way to serve this was I just using pingy Okay so this will be the live I keep doing that I need to set up the prompt URL okay so now we're calling this project basically the live deployment yes but we want to be talking to our local one so what I'll do so some of the stuff is what make agent is doing I just don't want to convert this whole project right

58. 34:22

    now to that so let's just do we're going to serve to we're going to set our server URL to be Local Host and we're going to set up a tunnel which is what we're doing a lot with the uh with the make agent and now okay so with make agent when that's running do you actually have to do anything manually or is it is it actually doing all as long as you put the right parameters in does does you have do you have to do anything else uh no so it's going to be doing all

59. 35:09

    that for you um essentially what it does is yeah it'll start this local tunnel for you and the URL of the tunnel it'll dynamically set this field here in the servers got it okay and that's how we're able to uh just connect to it directly from the playground and let's see so if we jump into this should be able to connect to our local instance and set the prompt to hey okay and and yeah and then from there we do need

60. 36:03

    to actually H we do need to actually register this plugin as well so that's why we actually this make agent tool is actually very nice so for people who are watching let's just clarify a little bit so the first example was set up a little bit differently so you didn't have to you didn't have to link it through pingy right like you could just go and then this is this is set up a little bit because the way this is set up uh you have to now actually do one extra step is that correct yep exactly yeah and

61. 36:43

    this might be may be a little confusing for people but um essentially if if you're using the make agent tool a lot of these steps are handled for you if you're just doing this more raw uh this is kind of the flow that you're going to need to follow so let's grab this URL and it's not finding the plugins back which it should be there I believe okay oh there we go okay I had the URL in there should have caught that but

62. 37:43

    let's save this API key for now in case we need it and we have our debug URL here and now uh as we make changes to this we can update our plugin and then test it live in the playground for example awesome so so basically here what we're doing is uh getting token info is one of our routes and then we also have making swaps and let's see like estimate

63. 38:45

    swap what what could we do for example um maybe we can add like a referral fee or yeah let add a yeah we add a referral fee let's see see what that looks like okay so or we actually might already have that here so we have this referral ID being passed um if we wanted to update this so let's say we're going to lower the slippage and change our referral ID something like that mhm uh let's see okay yeah then we're going to want to just

64. 39:41

    update our uh plugin and again I think this is great to see this more manual approach but again if you're using make agent you do not have to go through any that extra step to update it will just update yep and maybe we'll jump back into a quick make agent example after this and see building from there so when you're when you're making an update or deleting um an agent you're going to need your API key that you got when you registered it

65. 40:19

    um so this will update which will refetch all the plug-in um spec and everything from our from our API and now let's jump in and let's give this a test so let's say like let's let's try to generate a swap transaction for near interesting let's see what we got let's just see metadata see if this is actually working from the endpoint what was nice there is that it

66. 41:25

    gave a pretty detailed description of like what possibly was the problem it didn't actually just try to you know Force anything or just fail yeah right so it's giving you yeah great yeah this is like oh you know what this is this is um this is not right here that's why yeah that's why we we encourage us to make agent for now but the reason this is not going to be working is because we actually need

67. 41:52

    the we need this pingy URL to be here I'm not I'm not great with uh Alia so I didn't want to just try to serve this dynamically um but if we were to serve this dynamically we can update this route to to use our actual deployment which would be this got it so yeah so okay I mean this is this is great experion to developers see I'm going to run into this probably during the hackathon as they try different things whatever else so again a couple of things that are I

68. 42:30

    think we're seeing here is one obviously use the make agent U make your make your life easier but there's obviously you can again you can experiment and do this any way you want um but they're just make sure you're going through the steps and really being clear about how that all goes so it's not going to just manually update even though yes next is manually you know is automatically updating you need to actually go in and

69. 42:55

    en force that update each time you make a change yep yeah let's jump out of this because essentially This Server URL should have been the whatever URL we're deployed on um we had it as Local Host I was assuming that it would uh forward to the local tunnel but that's not was the case but um if we maybe I could show something else that we can do that's um that's a bit interesting go for it yeah so here's an example

70. 43:31

    of so we like the coino API is a a huge API and it's well documented so if if we actually look for the um we can actually find the API uh reference for coin gecko so they have they have their own open API spec so uh what we could do is we can actually take uh any existing open API spec and extend it to use uh our our uh our beta extension and instead of having to write all these stuff and and everything we can actually try to make something like

71. 44:12

    Dynamic that can just use an existing API and this is uh just walking through like an already done example but by using uh like a proxy like this what we're doing here is we we we just took that whole uh AI plugin I mean uh open API spec and pasted it and host into our well-known folder so uh I didn't write all this obviously right and then from there the API route that we serve uh it's just going to

72. 44:52

    forward to the actual uh coin Geo API and we can basically then have descriptions for the whole API that our AI can uh understand and still use the existing uh services so that's another option if you want to uh make something like more Dynamic so you're not specifically writing every single API route uh you can just forward the requests so here we're replacing the path basically with whatever parameters the user's giving

73. 45:26

    and just foring that directly to the coin gecko API did you still have to so in when you just kind of drop that in you still had to though add the mint base spec though in there as well right yes exactly yep good point so this is the the only thing we're changing is basically uh these fields here so we we want to update the server URL this is wherever the live deployment is going to B

74. 45:59

    um and then we want to update the uh the mint based uh metadata and basically uh the cool thing with this is that so we're using um we're using like a Swagger um API ref generator here for example and we can see that this is actually um going to expose all of our endpoints so we now have all of these end points and the AI can use all these endpoints and this is like a more you know experimental kind of approach it's

75. 46:40

    not going to it actually does quite well the only issue is if there's an endpoint that returns too much data there could be issues with with that um okay but this this is one way to with just one API route um that acting as a proxy to serve like a whole uh existing API um and pass it into the agent so really cool I think there's a couple things that instantly come to mind again around hackathon you know so existing data sets this existing open apis uh you certainly start making

76. 47:16

    agents about uh it sounds like there's room here to combine you know apis so maybe again it's not rewriting everything from scratch um I love this idea that you can just forward it really it sounds like the way you guys have structured this one if you follow the basics of like your guys automated tooling uh it seems like a breeze I mean this is it I mean that first example is super fast manual yeah it's a few more steps but you really have a lot of flexibility where if you wanted to put

77. 47:48

    this on top of your project you could um so that's certainly there and then on with all of this really is long as you have a data set already from like from an open API you can be off and running in I mean few minutes and at least have maybe it's not a winning level agent but it's the start of something that hopefully maybe can inspire people to kind of keep going from there yep

78. 48:14

    exactly yep very cool um great last point I think from trouble shooting standpoint sure there'll be questions of like why isn't returning stuff so too much data being returned can be potentially an issue I assume that's because of the way it's tokenizing it is that the issue yep and just some limits on a lot of the AI services on like uh response data got it so good to know for devs out there uh if someone needs help as

79. 48:45

    they're working on this during the hackathon where should people go so uh you can jump right into our Telegram and you know you can message us from there um uh just you know just reach out to the team in in any way telegram is usually best and we can we'd be happy to jump in and help you out from there awesome awesome uh here I'm going to actually I'll put that in the in the thing here so people can actually see so this is your your developer uh Telegram and

80. 49:19

    people wondering why it says mint because they because it's part of mint base uh but now the this side there the AI while side is P A Min Bay still exists um so yep yeah awesome uh you have tons of templates looks like yeah with that quick yeah just these are like the the main templates that you can work with um ref agent template and then we have these two uh coin gecko templates um that and these are basically where

81. 49:49

    you can see some examples of existing agents and work from there very cool awesome uh great stuff uh I mean this is a really exciting this is a really uh easy way to get involved with some AI stuff uh if you know viewers if you saw the call yesterday if you're going to go back and watch the call the AMA yesterday about AI uh we really want to stress that like you know you can jump into using AI in your project in a really easy manner this is one of them this is kind of exactly what we were

82. 50:22

    hitting on yesterday uh you can there's no reason why you can't start including it right away and actually have something relevant to your project so again with the hackathon you can you can mix and match bounties so you can have a whole uh project and you know you have it let's say you have an API structure build an agent on top of it right integrate it like start start putting these pieces together uh there's a lot

83. 50:49

    of cool things that can come of this uh awesome uh Mark any last words uh but Ian this this was a really great great tutorial thank you uh no so you know looking forward to any submissions uh thanks for uh thanks for having having me on Joe and um use make agent that's that's the rules sounds great I really appreciate your time today thanks Mark awesome thanks take care so uh that was awesome guys listen you know we're going

84. 51:18

    to keep stressing here at at devhub at nearer that this hackathon is going to give you a ton of opportunities to really build and to dive into the whole entire ecosystem and this is another example of one of those where you know it's it's probably easier than you first thought right building an agent is you know that first agent that he showed using their automated tools up and run

85. 51:43

    in a few minutes and again even if you want to go a step deeper and manually integrated it into what you're already doing uh the steps are all there they have extensive documentation definitely check it out if you haven't already signed up uh the link is here here at the bottom redacted.pdf us so our Dev support uh just for the General near ecosystem telegram is here down below uh jump into that ask

86. 52:34

    questions not only will the dev RS from near devhub be part of that but the community is also there to help answer any questions uh we will be back next week with more tutorials we're going to keep doing this and going you know and showing you some more different projects of the ecosystem if you have questions if you want to learn more please reach out to us uh but otherwise uh I hope you

87. 52:56

    learned something I certainly did and this has been Dev Hub live we'll talk to you soon bye