[00:00:00]

[00:00:44] This episode is brought to you by smile. io, the loyalty platform trusted by thousands of the fastest growing Shopify plus businesses to increase repeat purchases. Reward VIPs and build lasting customer relationships. Smile helps businesses become more profitable while discounting their products less.

[00:01:06] You got to be careful with the discounts. You don't want to push too hard on the discount pedal. If you're ready to turn transactional sales into lifetime customers, start building a more profitable business with smile. io today.

[00:01:25]

[00:01:25] Kalen: Okay dude, so what's the sitch with your hedge purchase? Where, where at? What are we looking at? Quick update. Percentage wise Status, I, uh, found

[00:01:34] David: my passport and it is valid until 2031. So we're good there. Nice. And I think I'm just gonna use my IT buffer budget to buy plane tickets.

[00:01:50] David: Nice. I think we're good. I think we're good

[00:01:53] Kalen: sounding like we're getting close to a lock.

[00:01:57] David: I saw you talking about some stuff going on.

[00:01:59] Kalen: Here is the event, pre-event situation. Okay. Okay. Last year, gadget and mantle put on a pre-event called Predis.

[00:02:10] Kalen: Perfect name by the way.

[00:02:11] David: Nice. Always gotta play off of the name of the conference. Yeah,

[00:02:14] Kalen: just like pre imagine it was perfect. And then, it was the day before and then they had this great little after party and so I was like, okay, I'm getting in the day before, so I could find a way to squeeze in there, but I'm, not getting until like 5:00 PM but so turns out they're at their pre-party or their pre-event is two days.

[00:02:32] Kalen: This year they changed the name to Tectonic.

[00:02:36] David: Whoa.

[00:02:37] Kalen: Okay. Whatever. Not no judgment

[00:02:39] David: sounds seismic.

[00:02:40] Kalen: Yeah, we're not gonna, we're not gonna get too hot and bothered about that, but it's a two day and then there's no after party on, uh, the 28th. I think they have something on the 27th. Maybe some kind of a nighttime situation.

[00:02:54] Kalen: Um, but anyways, , I just found out that Zapier and judge me are doing a little shindig on the 28th, so I'll shoot you that later. I just signed up. So I've got my, I've got my, March, May 28th night all worked out.

[00:03:15] David: Heck yeah. Okay, so conference is 29 and 30 Thursday and Friday, and there's some stuff happening on Wednesday, so I should get in.

[00:03:24] David: At some point on Wednesday.

[00:03:26] Kalen: Okay. Which day is Wednesday? Is that the 28th? So it's hap so the, the pre-event is 27th and 28th. Tuesday and Wednesday. Dang. That's a lot. Yeah, I know. It's a lot. It's a, it's a bit much,

[00:03:39] David: but I guess if you're cool, you'll be there.

[00:03:42] Kalen: Yeah. I mean, I'm bummed that I'm not gonna make it to that.

[00:03:45] Kalen: 'cause last year that, that event was real cool, I actually tried. I was like, you know what, let me see if I can reschedule. I've never rescheduled a flight ticket in my life. I was like, oh really? No. Who does that kind of thing.

[00:03:57] David: I do

[00:03:58] Kalen: that. I do

[00:03:59] David: that a lot. Okay.

[00:04:02] Kalen: I made several rookie mistakes.

[00:04:04] Kalen: I didn't use the promo code for Air Canada 'cause I was just being lazy. And then I got the cheapest rates. Which who doesn't get get you reschedule. Huh? And you can't reschedule. I was like, I, I was assuming they might let me pay a hundred bucks or something like that. Reschedule. But they just don't.

[00:04:22] Kalen: Yeah. Tell I'm like, why, like, isn't it in both of our best interest if I could pay a co, you know, a few bucks,

[00:04:29] David: you know, especially, well they added a new SKU so that you would, you would, you have a cheaper option, but you can't change anything at all about it. 'cause you used to be able to reschedule it no matter what.

[00:04:39] Kalen: Why not? I mean, yeah. So anyway, I couldn't reschedule it, so I'm kind of screwed.

[00:04:46] David: That's casual. I think, uh, three days will still be pretty awesome.

[00:04:50] Kalen: Yeah, dude. I mean, should be chill and then I'm gonna be there. I'm actually be there until Sunday. I just gave myself, I actually gave myself a little breathing room on the tail end, which hopefully some people will be around, probably, probably know, but probably be sitting in a bar in Toronto by myself.

[00:05:06] David: I'm sure you'll get some business prospects working.

[00:05:09] Kalen: Yeah, yeah. Hopefully

[00:05:10] David: meet with Toby, get some coffee with Toby and Harley,

[00:05:14] Kalen: yeah. Maybe hang out with Tobes a bit,

[00:05:16] David: you know?

[00:05:17] Kalen: Yeah. That's the

[00:05:17] David: plan.

[00:05:18] Kalen: , I noticed, he posted a tweet. He said that, , he's typing more into AI models now than email or Slack and, , and that I was like.

[00:05:30] Kalen: That is pretty crazy. Like yeah, I'm behind Think if you think about, I was trying to think if I do, I mean, I guess if I'm in Cursor I'm doing it a good amount. If I'm just talking about using like chat GPT or something, I'm not using it nearly that much, but ,

[00:05:46] David: yeah, I'm not either, but I bet Toby's got his own like whole system set up where he types it into some box and then it ships it off into like five different things and they all do something different.

[00:05:57] David: You know, he's probably got his own custom thing.

[00:06:00] Kalen: Yeah, he's probably got a set up. He had tweeted something about, like, he had this idea for something he wanted, it was basically a project idea. He was like, Hey, if somebody wants to do this, I'd be interested. And I was like, I gotta figure this out.

[00:06:12] Kalen: So it was like laser. It was so, yeah, it was something like, there was this open source voice to text thing that he saw and he was saying, if somebody can convert this to like, web assembly or, so, I don't know, something. I reread the tweet like 19 times, but I never, I didn't,

[00:06:33] David: I was like, you gotta pace that bad boy into Chad GPT and let it I know ca fire.

[00:06:38] David: I was like,

[00:06:39] Kalen: dude, if I could pull this off, , this would be my moment to get on his radar. Mm-hmm. But, yeah, didn't happen.

[00:06:45] David: I have to go see what it was. Maybe someone's figured it out. Yeah, probably. Somebody's figured it out by now.

[00:06:50] Kalen: . I mean it's a such a huge shift.

[00:06:53] Kalen: It's a pretty dramatic shift when you like,

[00:06:56] David: think

[00:06:56] Kalen: about it. Yeah.

[00:06:57] David: In that context, I still feel like I'm behind on cursor. I gotta try and I have an idea for something. , ' I'm not in VS code that often these days, which is unfortunate, but I would love to do this. Unbelievable. , we talked about the NetSuite two thing that's happening where they're replacing their suite analytics endpoint.

[00:07:15] David: Right? And it's completely shifting all of the schema. So like transactions are now called transaction and trans date is now like creation date or something. So there's like a small change to every single thing. I'm wondering if I can open my project, my DBT project and then also include the, with the DBT project.

[00:07:39] David: DBT is like. Imagine you have a transactions table that's coming out of NetSuite. Yeah. DBT is basically a bunch of joins that sit on top of each other to make it actually usable data. So like you'll have a DBT file that will run a SQL query against the raw transactions lines and it figures out if a certain transaction is new or repeat.

[00:08:05] David: So it does that by like ranking all of these customers orders. If there's one above one, then it's a repeat. Oh, okay. With the intention of , okay, we have these raw tables, DBT looks at all those creates new tables that like actually have useful information in them. Oh, okay. Okay. Gotcha. But I have like, there's almost 200 of these files and I have to go through all of them and update the schema.

[00:08:30] David: I'm wondering if I can just throw cursor at that bad boy and be like, Hey, cursor, here's a CSV that lists out all of the before and afters. Go to my code base and update it for me. I don't know if that's a thing, but I, I feel like that's where my cursor knowledge will come in. That'd

[00:08:50] Kalen: be worth, that'd be worth taking a stab at it.

[00:08:51] Kalen: I've seen it be, be pretty good at those, like those kind of like fuzzy find and replace type. Type deals. Okay. It's like if you're doing it on one file, I've seen it be, I'd say it'd be probably do a pretty decent job and then if you try to get it to do multiple files, it probably might do a pretty good job, but I wouldn't be surprised if it maybe hit a little snag on 200 files.

[00:09:15] Kalen: Okay. Um, each of the files are pretty

[00:09:17] David: small.

[00:09:18] Kalen: Oh, okay. But yeah, I'll have

[00:09:19] David: to give it a shot.

[00:09:20] Kalen: Yeah, give it a shot. Dude that, because that could, I mean it could just work.

[00:09:25] David: Do you usually give it like a, like here's an external resource that you can reference? Or do you just put the resources in the code base that you need it to,

[00:09:34] Kalen: to know about?

[00:09:34] Kalen: Sometimes I, I mean, sometimes I'll link to a documentation link or something like that. Usually by default, I probably don't. I just say, Hey, can you do this dude? Like it figured out the other day how to do a Shopify meta object. Search on field or so, or, , something similar to that.

[00:09:54] Kalen: And I didn't paste in any, documentation or anything. I was like, dude, that was, how did it get that so

[00:09:59] David: quick? Heck yeah. , has it done like, , JavaScript to RUS conversion or anything for you?

[00:10:04] Kalen: Yeah, I did that. I did. So, okay. So I did the JavaScript to Rust Conversion. And the way it got weird, like the first thing I did is I tried to, one shot it, I said, Hey, convert this JavaScript thing to Rust.

[00:10:20] Kalen: And I didn't like, understand types that well, so I just kept pasting errors in repeatedly, like an idiot. And it was sort of fixing them kind of a little bit. And then it finally got to where it was working and then I did some unit tests and I got it to pass all the unit tests. So I was like, cool.

[00:10:39] Kalen: I'm good. And then, and the unit tests were run, they run sort of like locally. They're not running, , the actual like web assembly stuff. Yeah. So then I went to deploy it and I hit some weird error. It was like the thing, the function isn't exporting the right run thing or so, so then I couldn't figure that out.

[00:11:02] Kalen: I got stuck. I started over with a basic rust app that, like the, outta the box Rust app. Mm-hmm. And I got that working. I deployed it, I made sure it had the right function, deployed, da da da da. Met, made sure it actually ran in the environment, logged stuff out. Okay, cool. And then I started adding onto it.

[00:11:22] Kalen: From that working state. And also by then I kind of understood the types a little better and the way that the GraphQL schema works versus like rust types and stuff like that. And then also those weird camel case stuff that it does. And so then, , I was able to finally get it working. , so I mean it definitely, I would've never had a prayer of doing the rust myself.

[00:11:47] Kalen: Like I just look at it and my eyes glaze over. Yeah. So I dunno anything about rust, I gotta go look at

[00:11:53] David: it.

[00:11:54] Kalen: And here's the kicker, dude. At the end of all of it, I am almost certain that all of the savings of rust are. Like the same as if you just don't parse, , JSO meta fields in JavaScript. Dang it. Which you don't have to anymore because you can just get the parsed, , JSON value for JSO meta fields now, I think that came out in like January or something.

[00:12:23] Kalen: Oh, like directly from the API or

[00:12:24] David: something like that? Yeah.

[00:12:25] Kalen: Instead of going value under the meta field, you go JSON value and it pulls it in pre parsed. Oh, okay. And so then when I ran 'em side by side, they were like, pretty much, I went from like 30 line items up to like 60, 70, and I think Russ was like 75 and like JavaScript was like 69 or something like that.

[00:12:45] Kalen: So it was a pretty much Is this number of items

[00:12:48] David: in the cart? Number of before it breaks

[00:12:50] Kalen: Yeah. Number of line items. Yeah. So. I was like, the risk of deploying this language that I don't understand at all. You know, I was like, I don't think it's worth it, you know? But so many people say that they see huge improvements, JavaScript to Rust.

[00:13:05] Kalen: So there's, there's probably more that I don't quite understand.

[00:13:09] David: I guess it depends on the complexity of the logic outside of parsing JSON.

[00:13:14] Kalen: Yeah, probably because, because honestly in our case, I don't think it's very complicated at all outside of mm-hmm. Outside of that. So yeah. You know, that's how it is sometimes is you kind of like, if you start prompting in a certain way, you can kind of get in a weird rut.

[00:13:29] David: Yeah. You're down a path and you gotta like back out.

[00:13:32] Kalen: Yeah. Dude, and this is so funny. Okay, so I was thinking about how like, with the models, there's that dynamic that if it's sometimes you're better off starting over because even if you say. No, don't do that. Or no, don't, you know, sometimes it can understand that but by and large it, like it doesn't do a great job of like forgetting stuff that you kind of want it to forget and you're better off just starting fresh.

[00:14:01] David: Yeah.

[00:14:02] Kalen: And I was thinking, I was like, that's so weird. Like. That doesn't make any sense. Like people aren't like that. Why is it like that? And then I was thinking about how, like I have this one client and we were talking about building some flows and stuff, and they said, you know, I said like, well, I don't know how many hours it takes.

[00:14:21] Kalen: Like I'm not gonna go off and spend 30 hours on this flow, without telling you. , and they're like, well, like I was thinking it would take more like a half hour for each flow. And then in my head I was like, oh man, that's really low. I was like, so then, so then anyway, so I started doing some work on it and I felt this like pressure that I had to, you know, so I even told them, I was like, Hey, listen, those, that's a bit low for me.

[00:14:48] Kalen: Maybe this isn't a fit. They were like, no, no, no. Honestly, it's fine. I was just kind of just saying that off the top. It, they basically told me to forget what they said. And then, but you couldn't, but I couldn't. I couldn't. I, I literally couldn't. And then I think I said something again, I was like, Hey, I got this flow done and I actually did get it done in this amount of time, but honestly I was like, I'm kind of rushing and I would've preferred to add a unit test here, but I didn't because of the da, da, da, da, da.

[00:15:15] Kalen: And they said it again. They said, don't worry about it. And still, it's still, it's stuck in my head that, that's their expectation. If I don't bang these out quick, then, you know what I mean? That really cracked me up.

[00:15:31] David: It's less of a like, you know, it's their expect, I mean, maybe it is, but also like you just probably internalized that as a challenge in that moment.

[00:15:39] Kalen: Yes.

[00:15:40] David: And you can't let it go.

[00:15:41] Kalen: Yeah. Yeah. I can't let, yeah. And then I was thinking like with models, you could start a new prompt and you're good to go, but with human interactions and stuff, there isn't really. A way to do that. Like, like

[00:15:55] David: to unsay something.

[00:15:56] Kalen: Yeah. You're just screwed. Yeah. I mean, you can kind of eventually get there and work through it, you know, like if, they say, Hey, no, no, no, I really, really, really want to, blah, blah, blah, blah, blah.

[00:16:08] Kalen: But it's like, it's so weird how I was like, people aren't like that. I was like, they are like, it's probably just the nature of like neural networks that when something, you know, like something puts a, a footprint on the network, yeah. Like, it's just hard, you know? So weird.

[00:16:28] David: It's super interesting.

[00:16:29] David: I'm like, yeah. All, all that stuff. And like us learning about how it works is all like so new.

[00:16:37] Kalen: Yeah. Still.

[00:16:38] David: And it's, fun.

[00:16:39] Kalen: Yeah.

[00:16:40] David: Yeah. I gotta fire up Cursor and see what I hate about it. I tried it once on like a 1000 line JavaScript file and it just destroyed it, so I gotta Right. Give it a try on this different project.

[00:16:53] David: Yeah, dude.

[00:16:53] Kalen: Take another crack at it. The NetSuite side of things, I'm kind of pumped because, I have this whole, plan to, you know, build an app to do the NetSuite integrations, and kind of like, cause I'm brought into all these issues with iPASS.

[00:17:09] Kalen: Mm-hmm. And I deployed a little piece of functionality live, uh, to, uh, yesterday, to sync some stuff in. , so it just, heck yeah, it just, yeah. And it's, so I just, actually, I used Cursor to do the soap XML 'cause like, who the hell wants? I was like, oh, I was gonna ask, how are you getting the data out of NetSuite?

[00:17:27] Kalen: It's the soap, huh? Yeah. I was like, , I was like, let me just try to get this thing working. And then I got it working. I was like, oh my gosh. , and so that's been, it's been doing a decent job with it. Like when it hits an error message, it's able to generally figure it out. . And I need to eventually get, this current project is only using soap, but this other project I have in mind is using the rest API and, , I'd like to get that figured out.

[00:17:53] Kalen: I took a stab at Yeah, at the rest, a p authentication, but I hit some snags. I gotta probably take another shot at that at some point.

[00:18:01] David: You know what could be interesting is, , if you're able to get in and like deploy a, a script within NetSuite, right? , is on some action in NetSuite or via workflow.

[00:18:16] David: You could have a script that just pings an API like pings a web hook with data that you need to something, right? And like kind of build your own data extraction. I don't know, like we, we do that simply in someplace, but I, I don't know what your rest API data needs are.

[00:18:36] Kalen: Yeah, that's not a bad idea.

[00:18:38] Kalen: Actually. , I haven't done that yet, but I was thinking about it and I just haven't gone in there and looked at it. That might be simpler in some ways. I do like to be able to pull though, because like mm-hmm. Especially one thing that drives me crazy that's hard to do in a lot of the I passes is like, I want to just test a specific record.

[00:19:01] Kalen: Like there's something about this one customer that I want to test and I have to either wait for it to come through on the update schedule mm-hmm. Uh, in order to test it or something like that. So I just want to have like some helper commands I can use to just like say, here's a specific thing, pull it and do whatever you're gonna do on it.

[00:19:24] Kalen: So certain things like that I'll wanna pull, but Gotcha. , definitely might be some use cases for, uh. Some, some on something on the sweet script side.

[00:19:33] David: Yeah. I think the way that Sligo does it is they have some sort of like in-between script that they can call to from outside. Like maybe there's a way to send a rest call into NetSuite that triggers a sweet script that then gives you back data based on the payload.

[00:19:53] David: Maybe that's a

[00:19:54] Kalen: thing, right? Yeah. There is like a rest, API that you can like hit with, uh, SQL queries. Oh, okay. That's pretty cool. I did that on this on the first project I worked on. I mean that's a pretty decent little API 'cause you can just run this, basically see your arbitrary SQL queries.

[00:20:16] Kalen: But it gets a little tricky if you want to get, like, the nice thing about SOAP API is you want to get a customer object and it's got everything on it. Mm-hmm. All the sub fields, all the custom field, versus having to query and join in a bunch of stuff. And then you've got multiple records and you got merge them in, you know.

[00:20:36] Kalen: Yeah.

[00:20:36] David: Deal. That's a little bit, I guess it depends, like, if you're going for one single thing, then soap is fine, but if you need like a thousand customers, that probably kills the internet. If you have that many, that much XML coming down the pipe.

[00:20:51] Kalen: Yeah it's pretty brutal, but yeah, I'm kind of excited because I got a couple bites from some people that might be interested.

[00:20:58] Kalen: And actually , the CEO at Gadget is really interested in kind of going after the IPAs market because in the same way that they have like these nice connectors for Shopify and you can just get in there, start writing code, and you've got your data bottles in there. Mm-hmm. Basically, he wants to do the same thing with.

[00:21:19] Kalen: ERP connectors and, 'cause he gets all this feedback from other agencies and stuff that they have all the same issues with IPAs and they're also super expensive. And so, okay. So he actually was telling me that they might be down to like sponsor me if there's a merchant that is, wants to migrate, he might be down to like sponsor me, do the migration for free so that it's like they don't even have to pay for it.

[00:21:48] Kalen: And then their monthly fee will be lower. , hopefully, probably be significantly lower on gadget. And then in the process, what he wants to do is take that and build it into their native gadget functionality so that, in the future they could have these really, you know, nice native connectors.

[00:22:08] Kalen: Heck yeah, that sounds like a gig. So, yeah. So that was pretty sweet. So, that was cool. , so hopefully, something will come through on that front, but I'm feeling like , this might be a thing for me. It's weird, it's a weird category, but

[00:22:21] David: somehow got into being a NetSuite expert.

[00:22:25] David: But that would be an interesting project. Then you could like focus on. Like how could I make this reusable? Exactly, exactly. Do some funky stuff.

[00:22:35] Kalen: Yeah. 'cause I'm constantly jumping from one project to another. I'm doing CS s tweaks over here, and then I'm doing, you know, and like, to be able to have one thing to focus on where like you're re you're leveraging that thing for different people, so that you can really focus on it.

[00:22:54] Kalen: And then o , other people can use it and maybe get a little bit of that internet recurring revenue money. Would be pretty cool. But, so yeah, we'll see. We'll see if that goes anywhere.

[00:23:06] David: Nice. How's the flow work going? You still working on flows or is it like a lot of gadget stuff these days?

[00:23:13] Kalen: I keep been getting some little flows. I got a couple, oh yeah, I did a little, uh, flow integration. There's this app called, um, admin buttons and it lets you add, different buttons onto , in the Shopify admin, like let's say on an order page using the, uh, admin extensions.

[00:23:33] Kalen: Oh, cool. So then it lets you define custom buttons. Like there's one that lets you update order, attribute custom attributes on an order, which you can't do. , or you can't create new custom attributes on an order or on a draft order. There's a button and they let you define arbitrary buttons, and then you can tie the button into a flow, or you can tie it into, there's like some other triggers you can tie it into.

[00:24:00] Kalen: So he has a bunch of templates for different buttons, like reship an order, or like reship certain line items from an order or like just a bunch of little things like that, that he's tying the flows. So I was working on a couple of those. , that's kind of

[00:24:15] David: cool. Was there one you said that lets you update, line item properties on a draft order?

[00:24:21] Kalen: The one I did does custom attributes on a draft order. So not the line item properties, but I think Gotcha. I think that was on their, that I think that was on his list too, was the line item properties.

[00:24:34] David: Yeah. I've, found that they are not well support, like it's hard to find apps that, that do that, which is weird to me.

[00:24:41] David: Exactly. Exactly. We had infinite options and it was like integral to our retail, situation because that's how, if a retail store needs to place an order for shipping from the warehouse. Yeah. , and the customer, once it monogrammed, like we had the, we had infinite options in there so they could click into the Right, the infinite options tile put on the monogramming size and letters and color.

[00:25:06] David: Right. And then it just went through like they normally do from the front end. But there was a, deprecation from Shopify on. I forget exactly what the API was for point of sale, but they, they removed it and we, so I had to do a bunch of quick research like yesterday I think, and find another app that

[00:25:25] Kalen: Oh, lets you do that.

[00:25:26] Kalen: Shoot. Wait, so what did they deprecate?

[00:25:30] David: Yeah. It was something to do with custom interfaces and point of sale. 'cause like the way it works, anytime you do something in point of sale, you have to click on one of those. Um, one of the tiles.

[00:25:41] David: Tiles, yeah. Yeah. , point of sale, app cart. App extensions. Ah, so something to do with the cart. But we had to switch over to something that has like a real, point of sale extension. What is the new extension called? POS UI extensions. So instead of a cart extension, it's a UI extension.

[00:26:06] Kalen: Oh, okay. Oh, so they're not up on the latest?

[00:26:09] Kalen: Uh,

[00:26:10] David: yeah. The app just like decided not to do it, I guess. It's like, alright, guess we're going for another one. And anytime you find something like that, it's always point of sale is something that's tacked onto like the, a lot of these are like, you can do this in the storefront and also it's in the point of sale.

[00:26:27] David: Oh yeah. It's a total afterthought a lot of times. Yeah.

[00:26:33] Kalen: That's interesting. Yeah. I really, that was another category I really wanted to get into was point of sale stuff and I just did a, just did like one thing there, but I just didn't end up getting enough it work to keep chugging on it, but that's, yeah.

[00:26:51] Kalen: It's understandable. I'm, there's not,

[00:26:53] David: yeah. Like point of sale is not nearly as big as. Online store.

[00:26:58] Kalen: Yeah. But I'm really jealous of people that are working on apps there because it's like, it's like a cool space to be working. It's like really early. And it's like they keep talking about how huge those are gonna be.

[00:27:11] Kalen: And they're in this kind of awkward stage now, like you're saying. So it, it's like easy to overlook them. But like, fast forward a year from now, there's gonna be some real killer apps that you're gonna be like, damn, why didn't I build that?

[00:27:24] David: You know, hopefully point of sale is really cool.

[00:27:28] David: It's really nice. Like compared to other point of sale. Yeah. , platforms. Shopify's point of sale is just so good. Nice. No waste.

[00:27:42] Kalen: Dude, I gotta give a man. , I gotta give a shout out to rj, the editor, because, , shout out on, on the episode 10. I recently listened back to, I remember being like. Feeling like a little bit off.

[00:27:57] Kalen: Like there was sort of like some extra awkward pauses. I remember. I was just, which one was that? I just remember it was number 10. It was, I guess we must have recorded it now, like four weeks ago, or three weeks ago. But I just remember in my head going, okay, I need to listen to number 10 because it's gonna be a bit of a mess.

[00:28:16] Kalen: And I listened to it and he tightened it all up. Dude, it sounded great. Nice.

[00:28:22] David: Get it rj Yeah. That episode is titled Bundle of Shenanigans. Yeah. Yeah.

[00:28:30] Kalen: Also, I told him to use the oh three mini high model for like titles and transcripts and stuff. And it, I think we said that in the episode.

[00:28:41] Kalen: I don't know, but it does a really good job of finding like titles. Most of the time we say like, show title at some point. So I think it might key off of that, but it's nice. That's perfect because it's just like, yeah, it's getting to the point where it's like pretty auto. It also did, does a good job of doing clips.

[00:28:58] Kalen: I'm gonna have 'em do the clips and, , so it's, ,

[00:29:01] David: the clips always seem so well produced. Like when you post 'em on Twitter, it like looks oh nice, looks nice with the waveforms behind it. And it's always an interesting part. Nice. Nice. Yeah, those are fun. I,

[00:29:13] Kalen: I was a little behind on posting 'em and want to get that cruising along.

[00:29:18] Kalen: , also shout out to smile.io dude, because they've been great. And it's, it's really nice to have, you know, to be able to afford the editor so that like Yeah. You don't have to be too, I think by now, if I was doing it all, I'd be getting to the point where I'd be like, all right, you know, we gotta do this once every other month.

[00:29:35] Kalen: Yeah. And like he told me, he is like, yeah, I just really want support it, you know, and you never know, like if they're gonna like, you know, nag you to death with like, add a popup banner to the website and, , like put three interstitials in. But they, you know, they like gave an ad read and then they're just like, chill, you know?

[00:29:55] Kalen: So it's, that's cool. I'm really, grateful for that. Yeah.

[00:29:58] David: Thanks.

[00:29:58] Kalen: Smile. Bill's the man. All right. I feel like I'm feeling cheesy. I'm doing all these shout outs, but, um,

[00:30:05] David: shout outs are important. People need to feel appreciated. Okay. And it's not something we should be ashamed of, right? '

[00:30:16] Kalen: cause

[00:30:16] David: we're nice guys.

[00:30:17] David: That's right.

[00:30:18] Kalen: Because we're good

[00:30:19] David: days and you just gotta live your life and you gotta let other people know that they are having a positive impact on you. Yeah. Go team.

[00:30:26] Kalen: All the shout outs. I was looking at the Algolia app. Do you guys use Algolia for C? Oh, no. You guys use the name search?

[00:30:32] Kalen: No.

[00:30:33] David: Yeah, I've used it in the past, but not, not at Kuta.

[00:30:36] Kalen: I was looking at it and I saw, that they have some flow integrations I thought was cool. They have like a, they have like a trigger for top searches and a trigger for, , top searches with no results, you know, so you can get notified of people if people are searching for something and not getting any results.

[00:30:58] Kalen: I was like, that's cool. That's that. I love seeing a flow. I love a flow integration.

[00:31:02] David: So does that mean that on some cadence they do something? On their end. That's like, okay, there, there have been 10 searches for this thing and there are no results. Now we send a web hook. Okay.

[00:31:14] Kalen: Yeah, they, yeah, they send a web hook into flow.

[00:31:16] Kalen: So then, you know, if you have the app installed, you can just set up a flow and use that trigger. You don't have to manually handle the deal with a web hook or anything like that. You just set up the flow trigger and then you can send yourself an email or do whatever. Gotcha. Oh, that's kind of

[00:31:32] David: cool.

[00:31:34] Kalen: Kind of slick.

[00:31:35] David: Yeah. I wonder what else you could do with that besides send an email, like, try and fold it into an existing search term group or something.

[00:31:45] Kalen: Yeah. Are there, man, there's, I wish I knew. I feel like the search, I don't understand very well. Is there API access into the search?

[00:31:55] David: Oh, I don't know.

[00:31:56] David: 'cause it is a separate app, so you have to go in and like. Change stuff within the app interface. Right. But maybe they, so usually with that, like there's not Right. An API interface to it. But that would be interesting. Yeah, I would because then yeah, like you, you could automate fixing zero result search terms.

[00:32:15] David: Yeah. That'd be cool.

[00:32:16] Kalen: Yeah. I don't think I've seen it. I feel like there might not be, but somebody recently asked, me about a mesa flow to , do some tie in some search terms, like popular search terms or something like that. And I was like, I didn't know if it was possible. I had a feeling probably what wasn't, but that would be super cool if there was a search.

[00:32:36] Kalen: Yeah.

[00:32:36] David: P that would be an interesting place to inject AI into flow, like based on the search term that is not getting any results and all the search terms that we currently have, fold this one into one. That makes sense. Yeah. There's been some cool stuff coming out with like. AI integrated flow things.

[00:32:58] David: It's getting crazy out there. Like you can, um, like we're looking at, this AI customer experience platform and you can do stuff , it has a chat bot and you can go in and look at the conversation and click into responses and see why the AI decided to make that response based on all of the information that you've given it.

[00:33:19] David: Oh, wow. I think that's where the interesting, like AI inside of flows that can just handle edge cases, so you don't have to be absolutely perfect and covering everything. Like you just got a thing that has context and can make decisions.

[00:33:37] Kalen: That's so cool. So is there like a percentage of customer support things that it's handling that it's like totally handling well?

[00:33:47] Kalen: And then

[00:33:47] David: no, we're, we're still talking to them. I, hopefully I can work on this in the next couple months and, be done like by the end of April. But we did a chat bot and, , it didn't work well. It was like the free version of Zendesk's chat bot. Right. And, what ended up happening is it was a different kind of customer that was asking questions of the chatbot.

[00:34:07] David: So we just, like, the intention was to help our CX team not have so many things to handle, but we ended up getting different, like people would email at the same rate. And we just had other customers who were now talking to the chatbot and they had questions about like, products and like, how do I make this decision?

[00:34:25] David: Whereas the email questions still came through that were like, where's my order? I need a refund. So it ended up causing more work for our customer experience team. So we shut it off last year. Right. So this will be like, try number two. Gotcha. Hopefully with some real AI in place and can also handle emails.

[00:34:44] David: Nice.

[00:34:46] Kalen: Yeah, I mean, I like when I hear the word ai, customer support, I just hate it, just because like, you know what I mean? Just 'cause you've been on the other end of it so many times that it's been tried for so long. Yeah. It's been tried for so long and like it almost makes it harder. if there is one out there that's actually pretty good, I'll write it off within the first three and a half seconds and not even realize that, okay, this actually is good.

[00:35:15] Kalen: I should actually be using this. Mm-hmm. But I'm so used to like writing off those AI customer support things that I probably won't even give it like a solid six second consideration. Yeah. Yeah. It's

[00:35:26] David: definitely an uphill battle. But if you think about like, I. Everybody's using chat interfaces for AI right now.

[00:35:33] David: It just makes sense. Like it makes so much sense if I could just talk to the AI of a company and I needed to return my order and it handled that for me and I didn't have to wait for someone. Yeah. And also like that person might make mistakes with my stuff. Yeah. That just seems like it makes sense.

[00:35:50] David: Yeah. Yeah. But we're definitely not there yet.

[00:35:53] Kalen: Yeah, dude. No, a hundred percent. , actually, gadget has an AI tool, within their web app, and I hadn't, \ they did a post showing like a one shot AI prompt creation of like Shopify, like a Shopify app. And that, \ Mohammad had tweeted and got a bunch of traction.

[00:36:10] Kalen: I thought it looked pretty cool, but I was like, well, I don't, I doubt you'll be able to like, create a lot of useful apps with like, like a one shot. Mm-hmm. But I finally went in and used it it kind of like a little bit of a replacement for customer support in the sense that like, certain things I've gotten stuck on before, like figuring out how to tie just getting a basic working thing where I've got a button on the front end in remix that talks to the database, does a thing on the database, and I have a backend place where I can start doing stuff.

[00:36:41] Kalen: Mm-hmm. Like I've spun my wheels on that a lot in the past and going through their docs and their docs are like, ah, they're like kind of hit and miss and there's a discord for chat, which in theory should be good because everybody, you can search, you can find things that have already been answered, blah, blah, blah.

[00:36:59] Kalen: Right. But that's always a hassle. But anyways, I used it to like create some of that boilerplate and it did a pretty solid job of it. So. Nice. That was cool. And then I was able to kind of take it, whereas like Cursor it's two gadget specific, like Cursor wouldn't have been able to help me at all with that.

[00:37:18] Kalen: Oh, I see. So the fact that,

[00:37:19] David: so they must have given it some like specific training or like the resources that are available to it are very specific.

[00:37:25] Kalen: Yeah. So that was, that was helpful. That was kind of like a little bit of a customer support replacement ish.

[00:37:32] David: That's cool. , yeah, when it works well like that, like, matrix, I, the app also has a really good support AI in there, right.

[00:37:39] David: In their, uh, slack. And they've been at that for a while. So really, I think it just depends on like, knowing what you're doing with this thing and making sure that it's good. Right.

[00:37:50] Kalen: Yeah. It seems like so much of it is in. How you're applying the model to your problem space and mm-hmm. How much effort you're putting into, getting the data in the right for , whatever they gotta do.

[00:38:03] Kalen: Like, , oh, and you sent that AI thing, the, browser man edition. Manus. , I took a quick peek at it and I was like, 'cause I tried operator, which if anybody should have the browser agent thing figured out, should be them. But it wasn't great. , you'd think, but it, you know, and I saw the website and I mean, it looked like a good website.

[00:38:23] Kalen: It, like, I could imagine it being really good. And I don't know if, if, uh, yeah,

[00:38:29] David: I watched their release video for a little bit. Oh, nice. And there's some crazy stuff in there. Like it, uh, you can upload A PDF and it opens the PDF and its own computer. Ah, so it's not just browser. It's like it can write code and then run it.

[00:38:44] David: And then, , CHATT can run Python code, but this thing apparently has its own entire system. So it is just like very general in terms of what it can do.

[00:38:56] Kalen: That's cool. Okay. Now I know what you mean by when you said it has its own computer. That's crazy. That's crazy that, yeah, like if they, pull it off, that would be so interesting because, you know, it's like Cursor is this small team and they're beating Microsoft, with copilot, like they're destroying them.

[00:39:19] David: Yeah.

[00:39:19] Kalen: And Microsoft they're the giant company with the billions that is creating, you know, helping to create these foundational models. 'cause you know, a lot of people are like, oh, well, you know, AI is just gonna make these, , few companies, rich.

[00:39:31] Kalen: But it really doesn't seem like that. It really seems like small teams are able to, , use these foundational models in ways to, build applications on top of 'em that they're able to actually, do better than the bigger companies, which is a really good sign.

[00:39:50] David: Yeah. And the only reason that that is possible is because of open source.

[00:39:56] David: It would not work. Yeah. To like, if Cursor was a small team and they had to make their own model Yeah. Or they had to like, start from scratch building an IDE, it just would not have happened. Right. So it's super interesting that like all, all of this, . Progress happens because of people's willingness to make things open source, which just makes you feel good.

[00:40:18] Kalen: It's true. 'cause it's like, open source has always been powerful, but it's kind of like powerful, but it's like always the second class. Citizen. Citizen a lot of times. Yeah. But like AI is sort of putting it on steroids a little bit.

[00:40:33] David: Yeah. It's coming back.

[00:40:35] Kalen: Yeah. Yeah.

[00:40:36] Kalen: It's almost like it's coming back, , because you, you, you could layer something on top of something open source, and have a really good AI application on top of it. Which is kind of nuts, man. Like

[00:40:51] David: Yeah. I, I feel like , that's where we are with this ai. Like everyone's working on base models and they're getting better.

[00:40:57] David: Cool. But the, like, the thing that makes it into the internet is things all fitting together.

[00:41:04] Kalen: Yeah.

[00:41:05] David: And you, if you've got all these things that are good and open source, it's easier for someone to like, have the aha moment to like, oh, these two things should work together. And baa bing baa boom. Yeah. Put 'em together.

[00:41:19] David: Yeah. Something new happens. Yeah, yeah. Yeah. I gotta check out that browser one. That'd be so cool to be able to just have your browser just do stuff. Dude, you gotta tell me a little bit about this duck hunt thing you were doing. What? Duck hunt. What was that? So,

[00:41:37] Kalen: so this dude on Twitter ball and Phil shout out to ball and Phil, he tweeted, like an idea, like a screenshot of like a thing where you get a discount code and there's like a little mini game you gotta play.

[00:41:49] Kalen: Where I think it was like that game where you gotta like, bang three, spin the wheel

[00:41:53] David: or something.

[00:41:54] Kalen: Well spin the wheel is like a pretty common discount thing where mm-hmm. , you get a certain discount by spinning a wheel. Like I see that all over the place. I kind of hate it.

[00:42:04] Kalen: I kind of hate all these things, but they seem to work, you know. But he was saying like, like, it was like a whack-a-mole type of a game or something where if you whack three moles you get a discount. And I was like, I think that idea of like mini games, which the spinning wheel is kind of like. A version of a mini game that seems to work and everybody seems to use it, but like I was thinking, yeah, like mini games connected to discounts or something seems like it should work.

[00:42:36] Kalen: So then I just used Cursor to create like Duck Hunt. I don't know why that was the first game I thought about,

[00:42:41] David: but it How long did it take you to get through the, uh, I'm sorry, but I can't remake Nintendo games for you, or did it just do it right away?

[00:42:48] Kalen: No, it just did it right away. Like it, you know, it's.

[00:42:51] Kalen: There's gotta be like, there's gotta be some open source code that it, or some code that it was able to find that did this exact, I mean, I'm sure people have like, done this like a billion times. Mm-hmm. So it grabbed it and then, , it pretty much worked fine the first time. So then I was like, so I actually have this one client that sells, like, hunting gear, and I pitched it to the, I shot them an email.

[00:43:16] Kalen: I was like, I think they're gonna go for this. 'cause they're pretty cool. Like they're funny and they're cool. And I haven't heard back at all on that email. And then, and then after that I was like, dude, I was like, this is so easy. We can create all sorts of, of little mini game, like little themed mini games and then like, you can brand them and whatever.

[00:43:37] Kalen: So then the next day I was like, all right, let me create one of those deer hunting 3D simulator games. Oh yeah. Because I was looking at this site, carnivore snacks that sells like beef Turkey or something. Yeah. And, so I started doing it. It was horrible. It, like, it sucked.

[00:43:56] Kalen: And there's this library called three Js that's pretty good for 3D stuff. Yeah. But so the, okay, so the thing with games is that you gotta have the artwork assets and without that it just puts in like a square or a rectangle and like, so then I was like, that

[00:44:16] David: beautiful 12 point buck poking out from behind a tree.

[00:44:19] Kalen: Yeah. Like it doesn't have, because I'm pretty sure because nobody open sources that artwork, like all those fucking game people, they put all their artwork, , behind licensing and stuff. So think so rude. I think that's why it wasn't able to really do it. , and then next I was like, all right, let me create like a basic fighting game where like a guy comes at you and you punch him, um, goes like, they have Joe Rogan as a highlight on their website.

[00:44:53] Kalen: And I was like, I could do a Joe Rogan fighting thing. And , it was horrible. It was just like squares. And then I was like, make it a character and it was a goofy, smiley stick figure character. And like I couldn't figure out how to get it to pull in like artwork. So then I was like, okay, yeah, there's like only really simple, like very simple games, I think are easy to do right now.

[00:45:19] Kalen: Like at least the way I was doing it. So it was really my just like shapes

[00:45:22] David: of stuff. I wonder if you could get, , like I wonder if Dolly can make Sprite sheets.

[00:45:29] Kalen: It probably can. It probably can. That's probably the way to do it, is figure out how to generate it. It probably takes a little bit of work to understand how, what those Sprite formats are supposed to be, and there's gotta be a way to generate those.

[00:45:42] Kalen: But like, that's a perfect example of where I just wasn't doing it in a smart enough way to, to feed in.

[00:45:50] David: Yeah. Just give it to me cursor. Just do, everything. Yeah. I have the idea. You do the work. Yeah. I was like, why can't it

[00:46:01] Kalen: , just do it. Just, just like make a deer, like just make a beautiful look.

[00:46:06] Kalen: Deer, come on. Yeah. Like, why can't you do this? You know? It's like, it's so much less. Complicated to draw a deer than to like, write all of this code. Yeah. But the only reason to write all that code is 'cause it has all this sample data, yeah, that's a perfect, I bet Adobe

[00:46:24] David: AI could make it Sprite.

[00:46:26] Kalen: Oh

[00:46:26] David: yeah. No, I'm sure there's gotta be Photoshop to do that, but you gotta pay them like a million dollars a year.

[00:46:32] Kalen: Oh, is their AI thing really expensive?

[00:46:35] David: , just Photoshop is expensive.

[00:46:38] Kalen: Oh, okay. I thought Photoshop was just like a monthly, like 20 bucks or something.

[00:46:45] David: Oh, maybe for a certain version I feel like I pay a lot for, you're on the enterprise plan, you have like 20 people with licenses and it's just like, I, I just see the end number.

[00:46:58] David: Oh man, this is so much.

[00:47:00] Kalen: Dude. You probably have a good support. You're probably the only person with like a good support contact at, at Adobe Photoshop too. Just like, just like at Shopify you have this great simple contact. I saw this dude post recently. He was like, Shopify support sucks.

[00:47:19] David: And I was like, oh, I do feel like it's getting worse.

[00:47:21] David: Like when they, every now they dump you into the help center automatically and have you talk to the ai Oh really? No matter what. So I feel like I've started opening fewer tickets and maybe that was their goal. Oh,

[00:47:37] Kalen: interesting. Interesting. The

[00:47:39] David: turntables have turned.

[00:47:42] Kalen: Yeah. That's so funny. I mean, you don't want to be too negative, but at the same time, like when I saw that, I was like, you know, sometimes you guys just say it how it is and like Shopify is a company that, you know, a lot of us deeply admire, but that support, a lot of times it's just.

[00:47:59] Kalen: It sucks. I mean, there's just no other way

[00:48:02] David: to say it. You know, maybe they're training the AI or something. Yeah. And it'll be better for all of us because we've dealt with this pain. But I know it's all I sense right now is the pain. So

[00:48:13] Kalen: I think also it's probably one of those things, just like I thought the deer should be easy to do.

[00:48:19] Kalen: Like I feel like because Shopify is a great company and they have a lot of smart people and they have great software that you should be able to just do support. Great. You know, like you should just be able to do that. Why can't you just do that? How hard is that? Because I'm only thinking about one individual.

[00:48:35] Kalen: Like I'm only thinking, well I could do a better job of support, for one person if I only had to do it for 20 minutes a day. Mm-hmm. And I didn't, you know, I didn't have to do it for a long time. But doing it at scale, like is probably just extremely. , difficult. Totally different problem and, and expensive so that's probably just what it is.

[00:48:59] David: It is what it is. Yeah. There's a lot of, , like expectation these days also that you can have some sort of chat interface and just get the data you want. Like I, I'm working on this data stuff because of the NetSuite thing, and it feels like there's always been this desire to have self-service analytics and reporting.

[00:49:21] David: Like, okay, all of our data is in this place now. I just want to know, like, the only thing I want to know is how many new versus repeat we had this week versus last week versus last year. And also make sure that it accounts for. Like our fiscal calendar, which ends in January and sometimes there's right years that have 53 weeks.

[00:49:42] David: So like Right. It is, when you get into it, there's a lot more difficulty than what you think about from the outside and like Right. Until you get there. It's like, why can't you just tell me what I need to know?

[00:49:54] Kalen: Right, right. Although I feel like those types of natural language queries, if there's not something doing them now, I feel like that's gonna happen pretty soon and it's gonna be glorious because, I mean, if I had to do what you just said, I, I am sitting there for a half hour just figuring out how to start thinking about building this thing.

[00:50:15] Kalen: Yeah. And then it's gonna take me time to go in and, build it. But that seems like there's many edge

[00:50:20] David: cases. But it feels like AI should be able to help with some of those edge cases as long as it has like the knowledge of the way that you want to do things. It's just that it hasn't all been tied together yet.

[00:50:33] Kalen: Like I'm almost certain that if I were to ask for that in JavaScripting cursor, it would nail it. Like it would nail it on the first time. But when you're talking about, having a data tool that you can talk to that, I don't know how it's working under the hood or how the data's stored, like you'd need.

[00:50:56] Kalen: You need a model that knew the data store and you just talk to it and like, I don't know. , but that's gonna be so helpful. You can do that to

[00:51:04] David: update its memory over time. Like what does demand mean for your company? Like, do you include shipping? Do you include tax?

[00:51:11] David: Mm-hmm. And just like build up all of that context over time and

[00:51:15] Kalen: mm-hmm.

[00:51:15] David: Then when someone asks for something, they don't have to go make sure you're using demand. That accounts for the fact that there was an extra week last month. It just mm-hmm. Knows all of that. Mm-hmm.

[00:51:26] Kalen: Right, right. Has like a set of rules that it knows.

[00:51:30] Kalen: I have a question for you. I have a client that has a loyalty program where if somebody spends, you know, a thousand dollars, they're in the bronze tier, 2000 silver, dah, dah, dah, and I'm doing that in Mesa. I'm just grabbing their orders, summing up the order totals and putting it towards that.

[00:51:52] Kalen: But then they have another CRM system, which counts up order, , counts up totals, and we're trying to reconcile them because their numbers are different than mine. And, we, what they do is they only count the product, line item basically. In their sort of customer value calculation. they don't count taxes, they don't count shipping.

[00:52:20] Kalen: And I was like, I was thinking about it, I was like, yeah, that's actually better. 'cause , it gets at kind of, you know, the actual amount, like the more, like the net amount you're getting as a business. Mm-hmm. But then I was like, wait a minute. but the customer, if they spent a hundred bucks and that included tax and shipping and everything else, they're gonna expect that that a hundred bucks went towards the numbers they're seeing towards the loyalty tiers.

[00:52:51] Kalen: So Yeah, I agree. You have to kind of have to go that route.

[00:52:55] David: Yeah. I think that's the way you go. But then you also include discounts. So like maybe the CRM is not including less discounts. Right. But I guess the way that you could think about it is if you've got. A discount that, , makes shipping free at $95.

[00:53:10] David: Do you expect tax to be included in that or not? Usually discount apps let you choose. But it's kind of a business decision.

[00:53:18] Kalen: Mm-hmm. And

[00:53:18] David: then the other thing that's weird is if someone decided to return something right. Does that net out? Or like in Shopify, if you're doing that calculation in real time, it probably nets out, but maybe it doesn't net out in the CRM, so there's always gonna be something that makes those numbers different.

[00:53:35] Kalen: Yeah. There's some weird edge cases and, stuff like that, , that we gotta figure out, but I think it, I think the returns are mostly being calculated the same way. , and then it was just the, the taxes and shipping stuff. So I think they probably just are doing it wrong with

[00:53:53] David: The's.

[00:53:53] David: Why? For our loyalty program, we do it based on. Like not, not amount of spend one because like buy more. Like that's not our, like our thing is fewer better, so we don't want you to like look at a number that you spent over time. Right. And instead it's more based on orders. But the convenient thing is then I don't have to worry about those numbers.

[00:54:15] David: It's just like, okay, you placed an order and then the flow that gives them loyalty weights like each, each flow waits for 30 days and then after 30 days if it hasn't been returned, then you get the loyalty points. So it's a little bit like having that separation helps with that. Right, right, right, right.

[00:54:36] David: Yeah, that makes sense.

[00:54:38] Kalen: And this is the loyalty thing you just recently rolled out? Yeah.

[00:54:43] David: Line. Yeah. Yeah. Nice. , gotta figure out how it's doing. I feel like we rolled it out and then we were like, okay. , it's working, I guess. Yeah. So we gotta get some, we're hiring like a, a new, um, like customer retention, marketing something.

[00:55:00] David: They start next week, so I'm excited about that.

[00:55:02] Kalen: Oh, nice. Run some numbers on it. Run some numbers. Run the numbers.