Innovation Sandbox [Workshop 3]: Streaming NEAR Data with Goldsky

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

1. 0:00

   Heat. Hey, heat. Hey, heat. All right. All right. We're live. Good evening, good afternoon, and good night. Near. We're back again with the second workshop in our series as we're building for these next two weeks. in the innovation sandbox. I'm excited for this one. Today, uh we're going to be talking with Gold Sky, an indexing and data stream solution for your near projects.

2. 1:05

   I'm joined today by Andrew. Let me bring him on. How's it going everyone? How's it going Andrew? Welcome on. Uh thank you. Thank you. Excited to be here. Yeah, we're excited to have you. How about uh you want to tell us a little bit about yourself and what we'll be doing today?

3. 1:23

   Yeah. Yeah, definitely. So today I'm repping the Gold Sky vest. Uh I'm calling from San Diego, so vests are like pretty ideal versus a big warm jacket, you know. So today we're going to be talking a little bit about, you know, what is exactly indexing and how it can optimize your workflow. Um I think that there's a lot of different ways that you can, you know, leverage onchain data to either do analytics or

4. 1:46

   power your DAP. And we're just going to talk about how you can use Gold Sky to do that. And a little bit about me. I'm the customer success over at Gold Sky. So I basically, you know, just try and figure out what the developers are doing, how we can engage with the ecosystems and just make it easier for all of you to build. So part of that is doing little workshops like this just to teach you, you know, some cool tips and tricks.

5. 2:09

   Awesome. Well, we're excited to have you here. Uh love the energy. This is great. I'm going to go ahead and bring up your uh presentation now and I'll come down and then take it away, Andrew. Awesome. Sounds good. Thanks, Elliot. So, yeah, this is uh this is basically Gold Sky 101. So, if you've never been building before with indexing or if you're a professional, you know, data analysts, either way, Gold Sky has a solution for

6. 2:35

   you where we basically make it really simple to get started if you're just learning how to index or if you really want to get custom and and get really into the weeds, then you know, that's possibility as well. But yeah, let's dive into it. So, as mentioned, I'm on the customer success team. So, yeah, just really excited to just figure out what everyone's building and how we can make it easier for them to do that. Now, just to give you a quick recap of what

7. 3:00

   we're going to talk about. So, today we're going to touch on a what indexing even is, b what subgraphs are, and then c what mirror is. And at the end, there's going to be a little sneak peek into some kind of a new thing that we've got that I wanted to just give you a little alpha on. It's our nextg uh pipeline engine. So stay tuned for that.

8. 3:20

   But basically just at a very high level um what we're going to talk about with indexing is essentially the benefits of indexing onchain data rather than just doing direct RPC calls. Um subgraphs just to give you really high level that's a customized graphql endpoint that you can query. This is probably what most of you have heard of and in general when it comes to indexing EVM chains this has been the standard approach. Um, and then mirror. This is

9. 3:45

   where we're going to be diving really into the, you know, the near data sets and and showing you how you can index near, but real time data streams directly to your database are at the very high level what mirror is. So let's dive into indexing. So essentially whenever you want to look at onchain data right now let's say you do an RPC call it can work for very limited things but often times there's going to be complicated block data and the RPC

10. 4:12

    call is maybe inefficient and especially as you're scaling your app you know whether you want to get to the size of poly market or you want to be the next big big you know trading platform defy you know whatever it is you want to do whether it's real world assets you want to get into stable coins you know when it comes to scalability you're going to really want to have your app and your analytics and all of your data just be really on point and, you

11. 4:35

    know, real time. So, especially as we're going to be bringing on, you know, all these people that are familiar with web two speeds and web two applications, when we're onboarding these new people to web 3, they're going to be expecting web two speeds. And a lot of the times you'll find as you're building to scale out your app that RPC calls cannot work in in a fast enough capacity to basically give these web two speeds that

12. 5:00

    people are familiar with. So indexing is in a sense a way to really optimize for speed and efficiency so that you can get the data you need whether it's for analytics or to power your app in a really efficient way. And to give a little overview of the differences between RPCs and indexing.

13. 5:18

    So when it comes to RPCs, the query type is going to be more simple lookups, whereas indexing you can have complex queries. And that's because in indexing, it's essentially these pre-structured. They're these data structures that are designed to be queried. So when you have complex queries or specific data that you're looking for, they're intended to be searched, you know, they're intended to be used and indexing basically makes it more more readable format for these queries. And as far as performance,

14. 5:46

    because of how this structures, you know, with RPCs, they're going to be often slow and a lot of them are rate limited. So, you know, even if they're fast enough, you get to a certain amount of um, you know, RPC calls, you're just going to get hit with rate limits oftent times. And indexing is fast query optimized. So, it really is built to have performance. And believe it or not, even though indexing is the faster

15. 6:09

    solution, it's actually lower effort. So RPC maintenance can often be a high cost for developers and indexing is really simple. You know, you just spin it up, it's quick to deploy and there's not a lot of maintenance. So when it comes to scalability, you know, effectiveness of decoding, doing aggregations, all of that is optimized and just a lot easier with indexing. So in general, indexing

16. 6:33

    is just going to be a superior solution to RPCs when you're really trying to scale your app. Now, traditionally when it comes to indexing, EVM chains have often relied on subgraphs. And just to give you a little context because it's it's helpful to understand the history of indexing before we get into data streaming pipelines. So basically the way subgraphs would work is you would index the network to essentially extract the data that's happening on chain and

17. 6:58

    then it's transformed into this more optimized structure so that it's got organ organized what we call entities to basically help you understand onchain events that are omitted or you know whatever's happening whether it's a mint or a transfer right it basically transforms the raw data into these organized entities and then with a subgraph in particular that data is stored on chain and then you get this

18. 7:22

    GraphQL GraphQL API endpoint that you query and so typically a lot of the time when people would build with subgraphs they would have this endpoint they would find the smart contract they want to get in information for and then they would query that now subgraphs although they are very powerful and you know an improvement over RPC calls they do have some limitations as well and so that's

19. 7:46

    kind of what was the you know the need and and the birth of mirror on our end is that as we were seeing, you know, all these projects hit new new waves of scalability and just growing really quickly, we we realized that we needed to create a new framework that can keep up. And so essentially that's what mirror is. It's like the evolution of indexing from subgraphs. Mirror is the next thing. So basically with mirror, it unlocks a lot of possibilities because

20. 8:12

    of the way that it's designed. So recapping with a subgraph, you've got this GraphQL API endpoint. You're basically calling that and then pulling the data. Whereas with mirror, you're basically getting this onchain data, creating this data streaming pipeline that real time pushes data and it's going into your database. So a database like click house, Postgress, S3, right?

21. 8:37

    So essentially what's different here is that the data is being streamed real time and you're querying your database directly. So that's going to be a lot more efficient and it's going to unlock a lot of very cool use cases. And some of those include off-chain and crosschain uh types of merging of data where you not only just can, you know, very quickly get the data from a specific chain, but you can now have multiple data sources. So you can

22. 9:02

    combine, you know, information from three or four different chains all into your database and then you just query the same table and it's all in one place. So a offchain or onchain, between chains, you can also do off-chain. So if you have, you know, your own data sets or your own database or just some kind of other off-chain source that you want to rely on, you can do all that in your database because of course your database

23. 9:26

    isn't restricted to onchain data. You're just basically using onchain as a data source to put into your database and then your database you can then aggregate all the different kind of data that you want whether it's your own internal data or another off-chain source but essentially mirror unlocks a lot of possibilities because of the flexibility you have and you own the data. So mirror is a very very cool new innovation and to give you some insights

24. 9:50

    on some of the benefits. So real-time streaming to a database, you know, that's just going to be very fast and that's going to be almost instant. So basically as events happen, they're just pushed directly to your database and so it's updated real time. You have the full control over the storage of the data because you don't have to query the endpoint each time. It's just in your database and once it's there, it's already there. So you basically deploy a

25. 10:15

    pipeline, backfill the data, and then once the data is backfilled into your database, you fully own it and control it, and then it's all just about maintaining, you know, the the stream where real time it's continuously updating, right? So it's a lot of benefit to have your own data, especially if you're trying to, you know, make it really, really fast querying where you just query the database. So that's a pretty cool benefit. And then, yeah, like I

26. 10:38

    mentioned earlier, the multi-chain and off-chain data. If you wanted to combine data from multiple different chains with subgraphs, you would have to do multiple subgraphs. Um, and you know, the that could get complicated really quickly.

27. 10:50

    But with Mir and with pushing the data into your database, you're essentially creating like a like a home base for all of your data. And so really what that means, all that together, it's designed for scalability and speed. And you're able to get, you know, historical data, edge streaming data. And one way of looking at it as far as the scalability comparisons subgraphs it's kind of like

28. 11:14

    a vertical scaling where you just kind of throw more compute at it and and you know it scales vertically whereas with mirror it's horizontal scaling. So it's in a sense infinite scaling and you can have a lot of just a lot of things happening at the same time and it can handle it no problem.

29. 11:31

    So now I'm going to go ahead and give you a little demo of how to actually use mirror in the Gold Sky app. And if you haven't yet already, feel free to follow along in the demo. If you go to app.goldsky.com, you can sign up just with Google. Takes like a second. So you can, you know, follow along with the demo. But I would also suggest that you go to docs.goldsy.com/chains/near.

30. 11:56

    So this is our dedicated page for near and it's a really really good quick guide just to get started. It gives you some insights into you know what are some of the benefits of building with near data sets how you can get started you know from as simple as installing the CLI to you know giving you an example of some SQL transform you can use a YAML file you can use for the pipeline config. So, this is a really,

31. 12:19

    really good page just to reference for when you're getting started, but I'm going to go ahead and show you a couple examples. So, when you are in the pipeline page in Gold Sky, creating a new pipeline, super simple. You just hit new pipeline. And you'll find that there's two different versions. So, for this first example, we're going to go with standard. And that's just like our

32. 12:42

    typical mirror pipeline. Turbo, this is that alpha I was telling you about earlier. This is our new product. It's the V2 version of pipelines. And I wanted to just first show you the standard version because we have a really nice aesthetic flow editor and the the turbo pipelines. That's going to be more like a YAML config file. So, uh the visualization of the standard pipeline, it'll help you guys just kind of understand what's happening. And I

33. 13:08

    believe in the future we will make it so that turbo pipelines are a part of the flow editor as well. But this is all still super new. As you can see, it's got the beta flag. So Turbo pipelines, they're pretty next level, but let's go ahead and get started with the standard pipeline.

34. 13:24

    So this is the flow uh the flow editor. It's really familiar probably to a lot of different editors you've seen. Um and so basically it's as simple as dragging and dropping. So the first thing you're going to want to do is take your data source, drag that on there.

35. 13:41

    And for this demo, of course, we're going to go with near. And you'll see there's a few different data sets. So, just to give you an understanding of what each data set is, it's basically types of data that we've pre-ingested so that it's even faster to push into your database. And the four that we have here, they're designed for mirror v1. And then these three, they're designed for Turbo. So, that's why they've got Turbo there. Um, it's just the latest version of the data set. But yeah, for this example, we're going to

36. 14:10

    go with execution outcomes and we'll start indexing at the edge of the chain. So basically, step one, super simple. You just pick your data source, you choose the chain, you choose where you want it to start indexing, and then you're on your way. So next step is going to be figuring out how we want to actually transform this data. So you have an option. You can just push all the data just entirely of the whole chain, you know, of of the execution outcomes data set directly into your database or you can get super custom and actually create a subset of the data.

37. 14:46

    And it's definitely encouraged to create a subset of the data because unless you need all of the data of, you know, the entire data set, it's just going to save you costs and it's going to be more efficient because you're going to basically funnel in only what you need to look at. So rather than you know ingesting all of it, you can just customize it to specifically what you need. So there's two ways you can do that. One is custom SQL and other is

38. 15:10

    HTTP handler. Custom SQL that's, you know, pretty straightforward. You just basically put in a query and it will filter it real time so that as the data goes through this transform, it's going to filter that before it goes into your database. Now, HTTP handler, that's a little bit different where it basically pushes that somewhere outside and then you can do the transform in your own way. Um, this can be really custom, but

39. 15:36

    basically custom SQL is the typical way to go if you're just going to say an app. So, I'm going to go ahead and click that option. And as you can see, there's this little popup window when you click into the query. And this is where you can pretty much modify it and customize it to whatever you want. So, and by the way, it it references source one that's referring to this source. So, just to kind of make it more clear, um I'm going

40. 16:03

    to rename this to execution outcomes. And then, as you can see here, that got renamed there too. So pretty much this SQL right now what it's showing is it's select all from execution outcomes. So the data set I picked if I do nothing it's just going to give me everything.

41. 16:24

    But I've got an example query I wanted to run. And basically for this query it's essentially just looking at any transactions that failed. So this is going to be a pretty cool way where if you wanted to use this example to debug, you know, what's happening on chain, maybe there's specific things you want to monitor and, you know, outcomes that are failed are an interesting insight that you can take a look at. And so this is what we'll do for this one. So basically, once you add your SQL, you

42. 16:55

    figure out, you know, what it is that you want to filter it by, you can do run, and that will just go through just to make sure that everything checks out. So, as you can see here, it went through each of the variables just to make sure that they're legit, everything is formatted correctly, and it gave me the uh the green check. So, basically, I'm I'm good to go there. Um, but yeah, when you expand it, it kind of shows you what

43. 17:20

    kind of data you're filtering for. So, this is just basically the blueprint for creating the subset of the execution outcome data that you are ingesting. And quick side note too, um you can do multiple transforms as well. So let's say you want to have um you know a second transform that filters it even further beyond this, you could do that as well. Um as far as sources, you can

44. 17:47

    have multiple sources too and you can even put all of them um through a similar type of transform and then all go to the same table. So this flow builder it's it's really really cool because you basically get to visualize how the data is moving through. So you know it's it's going from the source here funneling into this transform filtering here and then next stop is it's going to be going into the sync and

45. 18:13

    the sync is going to be the database that you choose. So, I already have one connected um for the demo, but if you wanted to create a new sync, basically what you would do is choose from the available options. And all of these, just for reference, they're all great.

46. 18:29

    It really just depends on what your use case is and what you want to prioritize. So, some databases are more optimized for querying, you know, a specific thing, whereas another maybe it's faster in this way or another, you know, you already have all of your data for your application built there already. So you just want to keep everything in one place. So really it just kind of depends, but the most popular ones we've seen um are usually going to be

47. 18:55

    Postgress and ClickHouse. Um so for this demo I'm using Postgress and specifically hosted Postgress where it's not actually my Postgress, it's a hosted Gold Sky Postgress and I just took that secret key and I put it into um this app called Beekeeper Studio. Um it's just like a really easy way to just plug in a hosted database. So that's what I'm using for this example. But yeah, essentially that's all there is to creating a pipeline. You know, you just

48. 19:25

    pick a source, choose how you want to filter that data using the transform and you know, you just customize the SQL. You choose your sync and then that's pretty much it. So when you're ready to to deploy the pipeline, you just want to make sure that you name it. So for this it can be like near demo and then once you save it looks good and when you deploy you'll see that there's uh different resource sizes available.

49. 19:53

    So just to give you some insights small is probably be going to be good enough for almost everything that you'll need to do. If you ever need to increase the size that's just going to basically um make it way faster to backfill. So like you know let's say for example you wanted to backfill all of Ethereum data and you needed to do it really quickly.

50. 20:14

    If you were to just throw in a large pipeline it would get it done much faster with the back fill and then you can bump the sizing down to small just for the maintenance right. So something to keep in mind though is that the higher the resource size because it's more literal resources being used it's going to cost more. So typically we do suggest starting with small just because it's already extremely powerful. it's capable for most use cases and it's

51. 20:38

    going to save you the most costs. So that's something to keep in mind. And then you deploy it. Now, I'm not going to deploy this because I actually already deployed a pipeline before. So I'm going to show you the one I already deployed just so that you can get an insight into what it looks like when there is data flowing through a pipeline. So this is one that I deployed earlier today. And as you can see, there's already records written, which

52. 21:05

    is basically, you know, data going into my database. And it shows you here which data set you used as the source, which sync you picked, and then the configuration. So this right here, this YAML file is pretty much the code version of the visual pipeline builder you saw in the flow editor. So I picked the execution outcomes data set. That was that first block. This is the SQL transform I did. This is the second block. And then this is the uh the post

53. 21:36

    the Postgress um sync that I used which was the third block. So really as you can see at the end of the day a pipeline is just essentially a YAML file and the flow editor is just a way to visualize it and make it easier and just a more simple way to get started.

54. 21:55

    So once you do have your pipeline live and you're pushing data into your database, I'm going to show you what that could look like in the database. So this is the the beekeeper studio I was telling you about. I connected my hosted Postgress and um you get the the the table basically populated with the date data from your pipeline. So you can look at the raw data just like this by opening up the table. But this is where

55. 22:20

    it gets really powerful is you can create queries to get specific information. So for this example um I am querying for recent activity and this query basically provides real time data of errors. So in a sense it's like a feed of you know the most recent failed executions. Um and I did add a note I wanted to mention. Uh so typically when you're looking at a pipeline, let's say for example with Ethereum, records received is most the

56. 22:50

    time going to be the same if not more than records written. What's really cool about near and it's just part of the way that the the functionality works is you can see here that based on records uh received there's actually more records written and that's because um near data it's it's not or it is one record per block but we actually get many transactions and events in a single block to emit. So although we get one

57. 23:17

    record received in that block received there's more transactions and events that actually get emitted. That's why there's more being written into the database. So, just a cool side note I wanted to mention. But this is an example query, right? Where you're looking for recent activity and this is what it's showing in the database. So, another example query, this one is DAP health where it's basically identifying,

58. 23:42

    you know, what are essentially the noisiest contracts, right? which ones have high failure accounts, you know, because that is an indication that, you know, maybe that is a front end that's broken or something is wrong with the smart contract. So, if there's a lot of failures, it's it's worth taking a look.

59. 23:58

    So, this is like a dab health monitor. And this third one, this is gas wasted, right? So, you know, even if there's a failed transaction, um I believe on year that that still burns gas. Um correct me if I'm wrong, but that's just my understanding. And so this basically shows, you know, which contracts are costing users money for transactions that still failed. So it's like a gas wasted query, right? So basically to

60. 24:23

    just sum it up, you know, you have this database now with a table filled with these failures that you created from a subset of the near execution outcome data. So pretty much, you know, you're just funneling in specifically what you need and then you're querying exactly what you want. So like as an example here this is 10 rows you know rather than a million or something like that.

61. 24:49

    So that is in a sense how a pipeline is built. It's really really simple and straightforward. You can also do this in the gold sky CLI and in the CLI you know you would just it would look kind of more like this but it's really nice to just visualize what's happening and so that was kind of why I wanted to show you the flow editor. Now, next up, what I wanted to talk a little bit about is

62. 25:11

    Turbo. So, Turbo is essentially the V2 of Mir, and it's basically just better in almost every single way. It's got significantly higher performance, meaning it can now process a lot more records per second. So you know really what was the need for turbo pipelines and kind of what was the genesis of us even needing to create it is that mirror pipelines they could support Salana but it was just not really functioning as

63. 25:41

    quick. The sheer amount of data was just insane. The speed was just crazy. So mirror pipelines were just not really built for Salana. So essentially because of the insane speed and large amount of data, we rebuilt our pipeline engine from scratch and we used you know a different language. We basically made a new framework and so turbo pipelines are now fast enough to handle Salana data, historical data, real-time data but all of the data sets that are available from

64. 26:10

    MIR are basically one to one available for turbo pipelines now too. So that includes near data sets. So you guys are lucked out. Basically you get to use Turbo to have really really crazy speeds. So higher performance that's obviously a really big benefit. The second one is there's TypeScript transforms. So basically the way you can transform uh the data it's it's even more customizable now and flexible to to really get you exactly what you're looking for. And a very cool thing is it's called dynamic tables where

65. 26:44

    basically you can update the lookup and you can filter without actually doing a redeployment. So you can make changes and and pretty much do whatever you need to do in a much faster way with less headache. And just to kind of give you even more insights, you know, this is our our Turbo Pipelines page. I mean, there's a crazy amount of benefits. It's it's really crazy what what this is able

66. 27:07

    to accomplish. So take a look at the at the page. It's on goldsyu.com products pipelines and you'll see just all the different ways that um pipelines can basically just streamline your experience. Um and data sets like near are optimized for this. So definitely definitely do recommend giving it a shot. And as far as creating the pipeline, it's super simple, right? Just like um you know just like with the standard pipeline you do new pipeline same flow

67. 27:38

    you hit turbo only the difference here is it's just going to be a YAML file. So just to give you an example of another one example that I made yesterday of of a turbo pipeline with near. This is essentially the YAML file that I used and it's almost basically the exact same as the example in the docs here. The only difference is here I hooked it up to um like Postgress versus here I did a black hole sync and a black hole sync it's basically not going anywhere right

68. 28:09

    so you can just test it out and just kind of get an idea of you know what what exactly is happening if it's working um but every time you do deploy a turbo pipeline you're going to get this page over here which is similar to the previous v1 mirror pipeline page where it shows you the sources it shows you the sync and it shows you the ammo file of your configuration.

69. 28:31

    You've got your logs as well just to give you some insights of what's going on over there. But the difference in the actual data is turbo pipelines because it's a different language and it's just a different framework. The metrics are actually available on graphana. So every turbo pipeline will get its own little dashboard like this which is super cool and I don't even know what half of this stuff means but I think it's important to to those devs that do. So definitely

70. 28:56

    take a look, deploy a turbo pipeline, see what's out here. And you know, one thing you'll notice I think when you deploy a mirror pipeline versus a turbo pipeline is whereas a mirror pipeline, you know, could be a few minutes to deploy, turbo is basically seconds, it's just significantly faster. It's got way more options for customizability and it's just all around, you know, essentially the superior product. So that was the turbo demo and that is basically it for today. Just kind of

71. 29:26

    wanted to give you guys you know a super quick rundown little bit of history of what is indexing you know what are subgraphs now you know it's basically how indexing all started out and then here here we are today we've got you know these real time data streaming pipelines mirror's the v1 turbo is the v2 give both a try out the flow editor try out the yaml file version with turbo and if you have any questions you know

72. 29:51

    definitely reach out to our support team they are absolute gurus and super fast to respond on. But of course, feel free to reach out to me, too. I'm probably not going to be as technical as the tech services team is, but I can definitely answer any highle questions and would also love to just talk just hear about what you're building out. It's always really interesting for me to learn about the use cases people are building out,

73. 30:14

    the headaches that they have as they're trying to figure out, you know, how to jump over obstacles. So feel free to reach out to me, send me a DM, you know, happy to schedule a call, whether it's to help you out or to just chat. But yeah, basically all in all, indexing is a is a really cool way to streamline your your process and you know, basically abstract away the headaches of maintenance

74. 30:37

    and you know, building indexing from scratch. you know, you just use Gold Sky, let us focus on that and you just build a good user experience. Focus on the branding and the marketing and getting people to use your product. Um, you know, you don't need to worry about the indexing. We'll worry about that for you. So, yeah, that's basically what I wanted to cover today. And, uh, I'm going to pass it back to you, Elliot.

75. 31:02

    Yeah, there we go. That was great. I That What a cool demo that you just did. I really liked uh you you well first of all round of applause you master class right there but um I really liked what you said at the beginning you how about how like it's indexing although it can kind of seem daunting I think it and it has maybe seemed daunting especially in this industry for a while because of how things have gone but um it's actually a

76. 31:29

    lot lower effort than RPC you mentioned that um you're not going to hit your rate limits you don't have to worry about like managing the RPC requests and the load balancing of it all and how it uh and kind of like that data pipeline on your side. Um and then that was so cool that you you guys have so you have the the guey that you can like move things around. You can actually see how the pipeline the data moves through the pipeline. Is it correct that the YAML dictates what that pipeline looks like?

77. 32:00

    Yeah, pretty much that pipeline builder is essentially creating a YAML file in the background. And so that's just like the visualization and and in the YAML file, there's like the three components of source, transform, sync, and those blocks are just basically those components just visualized.

78. 32:18

    Cool. All right. Hey, I got a couple questions for you. Um, one, uh, I was wondering, is there an API key to use? I think I saw that there was a Cly and you have like a login. Yeah. Yeah. Yeah. So, when you sign up uh for a Gold Sky account in your settings, you generate an API key and then you go to the CLI. It's super easy to install the Gold Sky CLI. I'm pretty sure it's just like a single line. Um you basically install the CLI, you uh

79. 32:44

    sign in with your API key, and then your project is connected and then you can, you know, fully build in the CLI or you can build in the in the UI on the web app. Either way. Gotcha. We do have one question. Is there an MCP for this uh at all? Do you know if there is one?

80. 33:01

    Uh, is that basically like an agent you talk to that builds for you? Uh, essentially it's it's it's a multicontext protocol. It it's Yes. In a way, it's like a way for agents to talk and communicate. Yeah. Yeah. So, I don't know how much I'm allowed to talk about it, but let's just say we're trying to make it as easy as possible. And we have AI in mind. We have agents in mind. We're looking at, you know, maybe you can just talk to

81. 33:31

    your pipeline, you know, maybe you don't even have to know what a YAML file is, you know. So, right now, not necessarily, but it's in the works and it's something we are very tapped into. Our whole team, we're constantly playing with AI. I mean, our our CEO is like vibe coding internal tooling for us, you know? I mean, it's it's crazy. We're very plugged into AI, so it is something that is top of mind, no doubt.

82. 33:56

    Awesome. Um, we got a couple people. They're excited to use it. Somebody's building a financial product uh about the financial. Think it's going to be very useful. They actually have a question for you. What would you recommend? store my data in my own DB or outsource it to Gold Sky.

83. 34:18

    Yeah, I mean basically when you're going to be using a mirror pipeline either way you're going to have to have some kind of database. So whether you connect your own for example Postgress, ClickHouse, S3 that could be your own database or you can just create a hosted Postgress.

84. 34:35

    So that's basically a database that we host on our end for you, but you still are owning the data because it's your database. But essentially, it is hosted by Gold Sky. So you have options for that. It really just depends what's easier for your flow. Um, I think at the end of the day, I would probably suggest you have your own database and you own your own data. um especially if you're going to be working with any kind of sensitive material or proprietary data

85. 35:01

    like you're just typically better off having your own database, you know. No, that I totally agree and actually that's that was one of the things that I really liked about this presentation was, you know, we don't got to worry about the GraphQL of of subgraphs of pass, you know, you can just stream this data directly into where you got to go.

86. 35:21

    Well, cool. Andrew, this was this was great. Thank you for coming on. Uh, I think any last words? You got your uh your socials there. Everybody give give Andrew a follow. Follow Gold Sky on there. Try out the product. Yeah. And uh check us out on X, too.

87. 35:38

    We're Gold Sky.io. And um yeah, we're always posting new updates on there. And basically, you know, we're we're plugged in to the ecosystem. You know, we want to hear about what you're building. We want to meet you. So, if you have any questions, feel free to reach out to me.

88. 35:52

    Ping us on X, email us, DM me on on Telegram, and uh yeah, we're just here to support you guys and and want to want to be there if there's anything we can do. Awesome. That's great. Well, thank you for coming on uh today, Andrew. I'm going to take you down now. Sounds good. Well, thanks for having me on, Elliot.

89. 36:11

    See you guys. See you later. All right, that was awesome. That was great. I I really like that workshop. I'm I'm excited to try it out for myself, honestly. But just to uh round this all out, so that was the third workshop in our innovation sandbox series. We got a couple more coming up.

90. 36:30

    Um just want to make sure I think the next one is going to be building with PingPay on Friday at what is that 15 UTC. So, I just want to remind everybody since we are uh we got about two weeks left, want to remind everybody of the tracks of the innovation sandbox. It's going until February 16th. You got about 12 more days. Three tracks uh across the private web and private life building consumer apps or web 3 widgets where privacy is the default across finance

91. 37:03

    assets, identity and user data. I think that this presentation that we just saw with Gold Sky and how you can stream data directly into your database, I think that becomes very relevant here in this track. AI that works for you is the other track. So, it's leveraging near's AI stack, the private cloud, private chat, shade agents, agentic payments, all this to build autonomous userowned AI projects.

92. 37:31

    And then the last track is open society from finance to the real world. So solutions that integrate stable coins, real world assets, tokenized assets, near intense or products that turn real world facts into reliable onchain outcomes. I want you to think about oracles um leverage near intense. It's a really easy SDK that you can use.

93. 37:52

    There's even uh we have some skills that are out there that you can install right now and you can find all this um or here very quick just rund down. You'll find that uh there's so there's prizes 3500 per track. There's also going to be a $2500 uh bonus for only on near. So that will be the best use of near's technology leveraging the uh near's tech stack to the best of its ability because there's all these really cool things about it.

94. 38:21

    yield, resume, chain signatures, the account model, all these type of things. And then of course there'll be two cross track bonuses of $1,000 each. Um you can find uh additional ecosystem bounties with Nova, who we just had a workshop on Monday. Hot pay has one out there, and then we have Ping, who will be doing a workshop on Friday.

95. 38:47

    You can find all of these resources, including the resources that Gold Sky shared with us today at this QR code. We're doing virtual workshops. We'll get some office hours up there. We've got uh if you follow this link, you can join a telegram and access all of developer support channels. You have plenty of developer de uh developer relations at your disposal. And then this is all

96. 39:11

    accessible via this notion hub if you need near AI credits. So this is specifically, you know, you're building for probably AI that works for you track or to private cloud or to uh private web track. If you want some near AI credits, you can scan this QR code. You can get $5 worth.

97. 39:31

    And then that's it. Uh thank you so much for tuning in and watching. You can scan this QR code. It'll bring you to uh the notion hub and then you can go ahead and get started. Join the telegram chats, form your teams, ask questions.

98. 39:46

    uh get the help that you need and let's start shipping. Thank you very much everybody for coming in. Uh and I'll catch you next time.