[00:00:00] David: It was plotting out all of the, for every channel, so like direct, organic, paid shopping, paid search, blah, blah, blah.

[00:00:07] the conversion rate versus the sessions last year versus this year, uhhuh. And so there were just a bunch of bubbles and then I linked them together with a, an arrow and it's kind of a.

[00:00:18] Crazy amount of things on one chart, dude. So I understand.

[00:00:21] Kalen: That sounds freaking awesome. I don't know what the hell they're talking about.

[00:00:32] 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.

[00:00:54] 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.

[00:01:13] I gotta get rid of this pen. Right now

[00:01:15] Kalen Jordan: Wait, wait, wait. What happened this week?

[00:01:18] David: I've been fidgeting, like with a pen. I've been clicking it

[00:01:21] Kalen Jordan: Right.

[00:01:22] David: and

[00:01:23] Kalen Jordan: You're like, I gotta,

[00:01:24] David: rid of this pen right

[00:01:25] Kalen Jordan: I gotta clean up my Nice, nice. So you're take, you're taking the production quality up to a whole nother level

[00:01:33] David: We're getting there

[00:01:35] Kalen Jordan: to a hoe

[00:01:36] David: chair

[00:01:36] Kalen Jordan: never level. Have you seen that? Uh, I think it's a key and peel skit to a hoe never level.

[00:01:46] David: No, I don't think I have

[00:01:47] Kalen Jordan: Ugh.

[00:01:47] David: one, but I, I

[00:01:48] Kalen Jordan: How many key and peel skits have you seen?

[00:01:53] David: I feel like I see them mostly on like Twitter or something.

[00:01:56] Kalen Jordan: This is like a KPI for like, you know, as like as a friend, it's like what's our overlap on Key and peel skits?

[00:02:05] David: I am

[00:02:05] Kalen Jordan: There's

[00:02:06] David: peel horror fan, I think,

[00:02:08] Kalen Jordan: really.

[00:02:09] David: though.

[00:02:10] Kalen Jordan: Wait, I always forget which one's key and which one's peel. The tall, skinny one is

[00:02:15] David: Key,

[00:02:16] Kalen Jordan: or, and then the peel. Yeah. Yeah. Yeah. I,

[00:02:18] David: Peele. Yeah,

[00:02:19] Kalen Jordan: I don't think I watched that, but, it was like really good. Right.

[00:02:24] David: there's been a few of 'em I

[00:02:25] Kalen Jordan: There's been a couple

[00:02:27] David: Nope. If you haven't seen Nope. Don't look anything up about it and just watch it without the kids.

[00:02:33] Kalen Jordan: Nice. I I'm not like super duper into horror movies. Like, I mean, uh, like,

[00:02:43] David: just

[00:02:43] Kalen Jordan: yeah.

[00:02:44] David: remind yourself that you need to be scared, you know,

[00:02:46] Kalen Jordan: Right, right.

[00:02:48] David: just in, in general,

[00:02:49] Kalen Jordan: I use, AI podcasts for that, to fulfill that need in my life. I was like.

[00:02:56] David: to all these things. I'm not thinking about.

[00:02:58] Kalen Jordan: I was literally driving a pickleball. It was like a Friday, it was like, , this was like last week. It was like, end of the week, gonna go have fun, start the weekend off with, pickleball.

[00:03:10] And I'm like, what should I listen to? Like, yeah, let me listen to this AI podcast where they talk about how AI is gonna kill us for like three hours straight in. Like, and this, the guy actually wrote a couple books on it. He was like super smart, and he explained it in such a like simple and concise way that it was just like, oh yeah, no, that's definitely gonna happen.

[00:03:34] That a hundred percent makes sense. He wasn't like alarmist, he was just sort of like very matter of fact about everything. And I was like, damn.

[00:03:42] David: It's kind of scary when someone like with conviction about something scary is also really good at explaining things.

[00:03:50] Kalen Jordan: Yeah.

[00:03:51] David: that's the superpower is to bring other people along and it's like, oh

[00:03:55] Kalen Jordan: Yeah. Especially when they do it a lot like shorter and more concise than other explanations you've heard.

[00:04:04] David: Mm-hmm.

[00:04:05] Kalen Jordan: they just say it in like eight words and then stop talking and then you're just like, oh shit.

[00:04:13] David: Two sentence horror or something like that?

[00:04:15] Kalen Jordan: No, that sounds amazing.

[00:04:18] David: It's pretty interesting.

[00:04:20] Kalen Jordan: Give me a taste. Give me a little, like Do you remember any of them? How could you not remember a single thing from a two sentence horror Reddit? That sounds like the type of thing you'd remember offhand.

[00:04:33] David: I don't remember anything anymore.

[00:04:35] Kalen Jordan: I know I say I know. It's like I say that and I'm sitting here like not remembering. I don't remember anything. It's ridiculous.

[00:04:45] David: Some of these I have to edit, I think, and not

[00:04:51] Kalen Jordan: Rated TV ma

[00:04:54] David: Yeah, let's do, uh, top all time.

[00:04:57] The boy woke up in a strange hospital room and disappointed that his family was nowhere to be seen And then, so like they always start with the title

[00:05:06] Kalen Jordan: Right,

[00:05:06] David: then the next sentence is the,

[00:05:08] Kalen Jordan: right.

[00:05:09] David: matters worse, his hands seemed to have wrinkled overnight.

[00:05:13] Kalen Jordan: That's brutal. Nice.

[00:05:15] David: just a little,

[00:05:17] Kalen Jordan: Yeah.

[00:05:18] David: that

[00:05:18] Kalen Jordan: So

[00:05:19] David: wasn't that great?

[00:05:20] Kalen Jordan: I was like,

[00:05:21] David: own

[00:05:21] Kalen Jordan: I was, I was expecting something to blow my mind. I was like, okay. But, oh, and then I know we were, we were talking about the MCP storefront thing. We could try to do a little screen share if you want. I mean, it's, I'm not sure. Yeah, I don't,

[00:05:36] David: that before.

[00:05:37] Kalen Jordan: yeah.

[00:05:37] I don't know. I don't know if it'll work. I think Riverside has a way to do it, , share screen, which might work, but, there we go. All right. Now I need to greatly enhance my font size. Let me,

[00:05:55] David: Yeah, I can do that too.

[00:05:56] Kalen Jordan: let me jump over to my other, pewter.

[00:06:03] All right. So.

[00:06:04] David: start from scratch.

[00:06:07] Kalen Jordan: And bearing in mind that our vast, uh, audience is primarily listening via audio will, will need,

[00:06:15] David: So,

[00:06:16] Kalen Jordan: to narrate.

[00:06:16] David: I started up Claude, also setting up the storefront. MCP for Claude Code is super easy. I.

[00:06:23] Kalen Jordan: Okay.

[00:06:23] David: It's a quick, maybe we can tweet about it at some point, but you just put in like the storefront name.my shopify.com/api/cp

[00:06:33] Kalen Jordan: Right.

[00:06:34] David: and cloud code knows what to do with that.

[00:06:36] Kalen Jordan: Okay.

[00:06:36] David: So then, uh, I have the CP set up, and when you're in here, you can do, , MCP just to see where you're connected

[00:06:43] Kalen Jordan: Right,

[00:06:44] David: are available.

[00:06:44] So everyone's storefront has all these tools available. Like

[00:06:47] Kalen Jordan: right.

[00:06:48] David: setup that I've done for this, but I, since it's set up, I can just say something like, what backpacks does Quiana have? So it's gonna use the tool to do, search Shopify catalog. the interesting thing here is that when it comes back. Like the query that Claude sent was backpack, and then the context is users looking for a backpack available. And if you go in, you can look at the actual response that Shopify sent, and it's a bunch of, it's a bunch of JSON, but then the context part is at the bottom here. is also part of something that Shopify sends back, which is just a prompt. So it's instructions and it says, use markdown to render product tiles as links to their respective product pages using the URL property for product data

[00:07:35] Kalen Jordan: Right,

[00:07:36] David: blah, blah, blah, blah, blah. It's a bunch of stuff the LLM to understand.

[00:07:39] Kalen Jordan: right.

[00:07:40] David: And then it takes what it gave me back from that response and tells me like, okay, here's the leather backpack. 16 inch, it's $498 and it's available in black. then I can say, add it to the cart and let's check out. So like super simple use case you could continue going through and like asking stuff about things.

[00:07:59] There's also a, a tool for,, understanding the store's like privacy policy and shipping policies.

[00:08:05] Kalen Jordan: Right.

[00:08:06] David: imagine like all of the baseline stuff that you would need to understand as a customer

[00:08:10] Kalen Jordan: Right.

[00:08:11] David: here.

[00:08:12] Kalen Jordan: So,

[00:08:13] David: it's asking me which size I want.

[00:08:15] Kalen Jordan: right. That's cool. So like you're demoing this using the cloud code terminal, , and I don't understand how it's adding to the cart. Oh, it's giving you a cart link.

[00:08:25] David: yep.

[00:08:26] Kalen Jordan: Wh where,

[00:08:27] David: takes me right to check out.

[00:08:29] Kalen Jordan: okay. Is it using like the API behind the scenes to create that session or?

[00:08:36] David: it's Storefront, so there's a storefront, API available for like adding, like create a new cart, add something to it, and then now I have the cart token. So that's how it can update the

[00:08:45] Kalen Jordan: Okay, so there's no front end. Yeah,

[00:08:48] David: yeah, you can tell it's like very simple, customers aren't gonna be using this,

[00:08:53] Kalen Jordan: That's what I'm trying to understand is like like I've seen a couple demos of like a thing that does something on the front end of the site doing some things, but like, I'm trying to understand like how can you actually use this kind of a thing for

[00:09:10] David: So there's a few things. One, the reason why I wanted to look at this is because I wanted to know what product data, it's it exposing, because those are potentially things that I can optimize and then. Two, like I wanted to show you because of the step function stuff that, Toby was talking about with

[00:09:27] Kalen Jordan: Right.

[00:09:28] David: remote. so imagine like this is the text level. So right now everything's operating on text and the LLM operates on text. Okay, cool.

[00:09:36] Kalen Jordan: Right.

[00:09:37] David: with, more context, Shopify could say something like, display these product results for across using, these, like what are those things called that they have, that they're releasing like

[00:09:47] Kalen Jordan: you're not talking about the AI blocks, right?

[00:09:50] David: the remote dom support?

[00:09:52] Kalen Jordan: Remote dom. Yeah. Yeah, that's what they tweeted about was remote dom support in MCP ui.

[00:09:57] David: Yeah. So like this is a text level. Now the next level is to actually give the user something that is interactable and like a real time

[00:10:06] Kalen Jordan: Uh, okay. Uhhuh,

[00:10:08] David: And that's when I think customers will come in because at that point, like. to set up an MCP, I have to do like weird terminal stuff. Even. Even now, it's like

[00:10:17] Kalen Jordan: right.

[00:10:18] David: but for a normal user, they're not gonna do that.

[00:10:20] So then what you're looking for is like Claude and OpenAI and everyone to like make that super simple. actually aggregate some of these cps and like, I'm in shopping mode, so let's see which brands have cps and like be able to give the user a better experience. this is all like, feels like low level technology.

[00:10:41] Kalen Jordan: Yeah, it feels low level and I and May and I'm trying to, better understand what the higher level stuff is gonna look like. But is it like you could build mean, like, I think the demos I've seen is they build a chat widget and the chat widget is talking to the MCP, the storefront MCP, to like basically kind of do what you just did in a chat widget on the front end.

[00:11:04] And so then with the remote dom stuff, it's like also. It's like gonna have access to the dom so it could like insert, you could like, I don't know, it could just modify the website or like your app would be modifying the website, but it would know what HTML elements to deal with.

[00:11:29] David: a way to, seed the MCP with like as kuana, these are our brand guidelines. Anytime someone's interacting with MCP, through an LLM, the LLM should utilize these kinds of things. Like our product tiles look like this. That's what I imagine this

[00:11:45] Kalen Jordan: Um, okay.

[00:11:47] David: because then you could render it anywhere and it wouldn't have to be an I frame.

[00:11:50] It would just be like chat, BT is rendering stuff. Or it could be a third party app that makes their own decisions on how to render those things.

[00:11:59] Kalen Jordan: Okay. Dude, I think I'm just not smart enough to extrapolate the higher, higher, um, level stuff that's gonna be.

[00:12:09] David: either. It's, it's gonna be interesting, like I, I don't know, I don't know how long it takes for this to be an actual thing that matters for customers who are shopping versus the, like, everyone's starting to see referers from chat GBT and stuff like that, but

[00:12:23] Kalen Jordan: Right,

[00:12:24] David: a link that, was generated by the LLM.

[00:12:26] Kalen Jordan: right, right,

[00:12:27] David: what if that shopping takes place somewhere else,

[00:12:30] Kalen Jordan: right,

[00:12:30] David: shopping on your products is somewhere else and you don't get any click data about that. You just like all of a sudden see orders coming in that, that's interesting.

[00:12:38] Kalen Jordan: right. And when you were saying earlier that you can see, you wanna see what data comes back 'cause you can optimize it. Is that the use case you had in mind, like optimizing data for like when people using chat TBT for referral traffic?

[00:12:56] David: Maybe, or just in general, I know that this is something that LLMs will have access to, and so I probably need to care about it. And, and so like, what are the fields that come back and do they

[00:13:07] Kalen Jordan: Right.

[00:13:07] David: One thing that kind of sucks in this scenario is you have a lot of things like meta fields that, on a product that adjust the way your site works, that's not gonna apply here.

[00:13:18] Like we have a, a meta field for hiding a variant,

[00:13:21] Kalen Jordan: Right,

[00:13:22] David: that's not gonna work unless Shopify builds in support for bringing back those. And then we also get to explain like, this is what a hidden meta field is.

[00:13:31] Kalen Jordan: right, right, right, right, right, right, right. That's right. I totally forgot about that because Yeah. But then, is that already a problem for you on like the shop app? I can't remember if you guys use the shop app.

[00:13:45] David: Yeah, probably, but we

[00:13:46] Kalen Jordan: Yeah.

[00:13:46] David: people using the shop app,

[00:13:48] Kalen Jordan: Right. It's like now this.

[00:13:51] David: are like, don't have any inventory. But same problem we have with feeds in general, I feel like

[00:13:58] Kalen Jordan: right, right, right, right, right. So it's basically, it's something everybody's sort of aware of, but it's one of those things where like probably volume is low enough on the shop app for certain use cases that people are just like, ah, I'll deal with that later. But then as Chad GPG referral traffic is spiking, people are probably wanting to pay more attention to that.

[00:14:18] Um, quick, sort of semi pivot topic wise, when you talked about, cleaning up the data, that product data and making sure it looks good, this falls into this whole PIM category I've been thinking about a bit

[00:14:32] David: Oh, nice

[00:14:33] Kalen Jordan: Right. Which is, I was thinking like, there's so much messy data, messy product data, and I see that in my projects and stuff.

[00:14:42] There's gotta be a way to like automate the process of like imp cleaning up data, improving it, I don't know, finding everything from just the length of a field to running some prompts against them to see if they're like, I don't know, whatever.

[00:14:57] David: Mm-hmm.

[00:14:58] Kalen Jordan: is that the kind of thing you would do with, like, you'd want to do with your product data?

[00:15:05] David: Yeah. One thing I can think of is we have this, this section on the PDP called,, dimensions and Care,

[00:15:11] Kalen Jordan: Mm-hmm.

[00:15:12] David: right now some, a copywriter goes in and, the product team puts in the dimensions and then also something about how to care for it. So that's actually a product team

[00:15:21] Kalen Jordan: Right,

[00:15:22] David: but it's it's copy that's written and imported as generated.

[00:15:26] Kalen Jordan: right,

[00:15:27] David: that I would like to have, like today because of something I need is I, I would like all of those product dimensions just as the length, width and height, split out.

[00:15:36] Kalen Jordan: right,

[00:15:37] David: One use case where I would take an, I would say, AI, pim, look at this field and pull out the that information, put it into column for me in A CSV or something like that, and then it just let it chug through your products.

[00:15:51] That

[00:15:51] Kalen Jordan: right, , and I wonder if Sidekick probably can't do bulk exports, but it probably will be able to pretty soon, man. I mean, it's because I. It actually, it could probably just, well, I don't know, but you, it could hit a bulk, , API endpoint. That'd probably be a good way to do that. And then,

[00:16:12] David: Do you

[00:16:12] Kalen Jordan: I don't know,

[00:16:14] David: sidekick access to certain admin APIs or like let you configure it to be able to do that?

[00:16:19] Kalen Jordan: do, I don't know. I saw somebody say somewhere that they were working on like a really user-friendly, like way to confirm before, like making changes. So I'm guessing that's their approach to like, being careful with like rights, , potentially destructive rights. Um,

[00:16:41] David: thing they called, like the first thing they called out was edits to the. To the theme. Right? So

[00:16:46] Kalen Jordan: oh, okay.

[00:16:48] David: maybe

[00:16:48] Kalen Jordan: Right.

[00:16:49] David: first, but that would be interesting to like utilize the bulk export or bulk query API to do some

[00:16:56] Kalen Jordan: Yeah. Yeah.

[00:16:57] David: on things. That would be great.

[00:16:59] Kalen Jordan: Yeah. Yeah. 'cause uh, it's funny because, I'm working on this search, it's like a product data sync app. So there's a bunch of data sync stuff. And then, I just pinged you about, uh, I was talking to Jay about his app,, that does like ai, marketplace connectors, like sending your data to like target marketplace and then some AI to like clean data up and stuff like that.

[00:17:24] And, he was, saying that he was trying to figure out if he's in a PIM category, if he should build out a PIM or, things like that or just plug into pim. So it's like, I've been thinking about this whole thing from like a bunch of, and then I, I remember you saying you'd love a Shopify native pim.

[00:17:40] I remember Jake also had said that, and I don't know exactly what his use case is, but it's just, and then I have a, there's actually this client I'm gonna talk to next week about a potential project of like pulling data into Airtable, , so they can modify it more easily, including like, translations in markets and stuff like that.

[00:18:01] I don't know, man. , It's an interesting one.

[00:18:05] David: It feels like a lot of the time when I look at pmms, one of the big things that they kind of like surface in their marketing is, helping with the process of enriching or hydrating the data

[00:18:17] Kalen Jordan: Right,

[00:18:18] David: To be site ready.

[00:18:19] Kalen Jordan: right,

[00:18:20] David: We just kind of do that in a Google sheet.

[00:18:22] Kalen Jordan: right, right,

[00:18:23] David: like that,

[00:18:24] Kalen Jordan: right.

[00:18:24] David: not that, that's not like a really interesting use case for me in terms of a pim.

[00:18:29] But if, if a PIM was like, we understand all the data that you have and, , like use AI to, to slice it differently or into different, , like how should I structure my product data taxonomy based on the data that I have? That would be interesting. I.

[00:18:45] Kalen Jordan: Mm. Interesting.

[00:18:50] David: Going back to my example of the dimensions in care, that's two different things.

[00:18:54] Kalen Jordan: Right.

[00:18:54] David: if, there was like some like assistant in within a pim, I imagine what it could do is me split that up into two different fields that would actually be more helpful for like long-term data availability or whatever.

[00:19:07] Kalen Jordan: Right? Yeah. Like I feel like the question of how to structure the taxonomy and stuff, you maybe could feed like, your taxonomy or maybe feed like a snippet of your product data, feed into like oh three and ask it if there's any changes you should make. And it prob might come back with a decent answer.

[00:19:29] But then on the implementation side, saying, okay, I want to do this. And then have it run through them, that feels like something you could build into an app pretty nicely. \, Also, the question side, you could build into an app too. It just feels like whatever the question was, you'd want to feed in the right, like probably some very specific data before you ask the question.

[00:19:53] David: so that you make sure that, yeah, I mean, I, I guess the other thing that would be interesting for some sort of AI assistant here is that it, it has to have way to aggregate what it is that all of your data is like a, a schema specifically for your data. Like , what did you title things, what is the data in it and how would you describe it? then the LLM would have more context for answering your questions.

[00:20:19] Kalen Jordan: Right,

[00:20:19] David: 'cause what, like what you were saying is like if you, would have to specify certain attributes. So that you focus the, the ai,

[00:20:27] Kalen Jordan: right.

[00:20:28] David: it would be nice if you didn't have to focus it and it just kind of knew a bit more about what's going on.

[00:20:34] Kalen Jordan: Well, yeah, I mean, I feel like that's sidekick, like sidekick and I wonder if Sidekick would know, would have a good answer on, , your schema? I'm not sure.

[00:20:45] David: seeing if it can access meta fields. Oops.

[00:20:49] Kalen Jordan: Yeah. Oh, it can. Okay dude. So use case sidekick, use case of the week, dude. I actually posted about this and a bunch of people came out of the woodwork and they were like, this is awesome. But basically, when you're doing something with a meta field, , or set of meta fields, you're writing some code or whatever, and you're looking for a product that has that meta field populated so you can like see an example value or like find a product to test with type of thing.

[00:21:18] And then you know how sometimes it takes forever. You're like hunting around for a product that has some random meta field populated.

[00:21:28] David: Yeah. Especially if it's not like available as one of the search filters on products or something.

[00:21:33] Kalen Jordan: Right. Oh, right, right, right, right. And so anyways, you can ask sidekick, give me an example product that has this meta field populated and it grabs one for you. , Which is a pretty straightforward, like it's pretty straightforward GraphQL query. , And I only tested it on like my small test store, so it'd be interesting to give it a better use case to make sure it's actually doing like a proper search.

[00:21:55] But that was a nice little finesse one. See, yeah, I can definitely access, I can definitely access minefield,

[00:22:02] David: meta fields.

[00:22:03] Kalen Jordan: Sweet moment of truth.

[00:22:05] David: those.

[00:22:09] Kalen Jordan: That would be cool. I.

[00:22:10] David: launch date. Oh, it has to fix the query. It might be,

[00:22:15] Kalen Jordan: Oh no. It'd be funny if when I did it, it just found the first product and then looked for the meta field value or something. It just got lucky. Yeah, it,

[00:22:26] David: kind of weird. I'll have to try it again later.

[00:22:28] Kalen Jordan: yeah.

[00:22:29] David: any release dates that are in 2025, which is not true.

[00:22:33] Kalen Jordan: Oh, okay. Take it back. Sidekick use case failed.

[00:22:39] But yeah, on the PI thing, I feel like, there's all these things related to like data cleanup that, you know, there's gotta be a, a ton of different things you could do. I mean, even things like cleaning up your images, maybe you want to crop them or, , crop them differently or, I dunno.

[00:22:58] David: Oh, that would be cool.

[00:22:59] Kalen Jordan: yeah.

[00:23:00] David: I want all of the images that feature a model to be in the second spot.

[00:23:05] Kalen Jordan: Right. Okay. Per, okay. These are good use cases. I'm gonna write these suckers down.

[00:23:12] David: Business ideas for Caitlin.

[00:23:14] Kalen Jordan: Yeah, I literally gotta write that down. Keep going, keep going, keep going. Hook me up with the use cases. But I feel like there's a ton of those and, um,

[00:23:28] David: Yeah. You

[00:23:29] Kalen Jordan: you know,

[00:23:30] David: could just keep coming up with them. I bet.

[00:23:31] Kalen Jordan: I mean, you're basically just are gonna like pipe some queries, in, and then pipe some mutations in with a prompt in between, and then hope that it actually works.

[00:23:45] like

[00:23:45] David: Like a, that's like an innate N thing. Have you seen N eight N

[00:23:49] Kalen Jordan: I see people talking about it constantly. I haven't, I haven't played with it.

[00:23:54] David: it or not. Like

[00:23:55] Kalen Jordan: I think so.

[00:23:56] David: or,

[00:23:57] Kalen Jordan: I think it's N eight n. That's how I say it inside of my head. Yeah.

[00:24:01] David: too.

[00:24:02] Kalen Jordan: have you messed around a little bit.

[00:24:04] David: I haven't, I, I feel

[00:24:05] Kalen Jordan: Oh, gotcha.

[00:24:06] David: because everyone

[00:24:07] Kalen Jordan: I know, I know, I know, I know. I feel the same way. I,

[00:24:10] David: go-to flow that has some level of AI involvement

[00:24:14] Kalen Jordan: right, right. Yeah. I gotta check that out. Yeah, and, and like with the MCP thing, it's like, it's all I hear about now, like, I feel like everybody that I've shared some idea with you know, something that to me, I think of as like an app idea. Like let's say what I was just talking about with the PIM and literally like the last four people in a row I've talked to have been like, have you thought about making an MCP?

[00:24:39] Or they'll be like, are you doing an MCP? And I can't tell if like, it's become this buzzword that people repeat without knowing exactly what they're talking about or if I just don't like see it yet, how like everything is gonna be an MCP or something like that,

[00:24:58] David: like when you first mentioned that I, it made me think about like, how would you monetize it, but the way you monetize it is through it requiring off.

[00:25:07] Kalen Jordan: Right.

[00:25:07] David: be interesting.

[00:25:09] Kalen Jordan: and that's what's cool. I mean, that's what's kind of cool is how you can, monetize it like that. But I haven't, seen any paid cps. I mean, I would think if there were good ones out there, people would be starting to pay for them. So, I mean, if it's such a, kind of no brainer and all you're doing is stitching some data together, why aren't there any paid ones out yet?

[00:25:32] You know, uh, or maybe,

[00:25:34] David: nerds who are messing with it. Like

[00:25:36] Kalen Jordan: yeah.

[00:25:36] David: one's, no one's actually going to, unless they're like some business who, like Shopify, I guess would be, , a company that you could market a paid NCP to because they are so gung-ho on like everyone using, LLMs to help them

[00:25:51] Kalen Jordan: Yeah. Like

[00:25:52] David: it doesn't feel like

[00:25:53] Kalen Jordan: I.

[00:25:53] David: real, like, , widely addressable thing.

[00:25:57] Kalen Jordan: Right. And by the way, I will say I was starting to worry about Shopify not being like AI native enough. Like in the sense, I think I had mentioned this, but I felt like things were moving slowly with ai, like the first, I don't know how long sidekicks been out for, but like, things were going slowly and like Shopify has so much momentum and infrastructure around the rest of their product.

[00:26:20] In my head, I'm thinking they're not gonna pivot fast enough to ai. They're, you know, they're just, they're moving way too slowly with it. And it feels like I, I'm starting to feel like, they are on it with AI because Toby's whole leaked memo and about using AI tools, and then I've heard about it from multiple places.

[00:26:38] Like I've heard, well mostly tweets from Shopify executives I guess, but they're saying that like a lot of their non-technical people are using ai. I talked to the sales sales guy I know at, uh, additions for that's, not working at Shopify, but he's building AI tools for his shop, his sales team or whatever.

[00:26:58] And, it feels like they are full steam ahead with ai and Sidekick is starting to get good, like,

[00:27:05] David: Mm-hmm.

[00:27:06] Kalen Jordan: you know, sidekick started to do some cool stuff, so.

[00:27:10] David: It was really interesting. I saw, uh, like I caught part of a, keynote at Addition Stev

[00:27:15] Kalen Jordan: Mm-hmm.

[00:27:16] David: with, , Glen Coates,

[00:27:17] Kalen Jordan: Mm-hmm.

[00:27:18] David: was talking about this scenario they found themselves in, 'cause I feel like they felt the same way about themselves and they had to say something about

[00:27:26] Kalen Jordan: Mm-hmm.

[00:27:26] David: we talked about, it felt like Shopify was slow on ai.

[00:27:29] Kalen Jordan: Yeah. Yeah. Yeah.

[00:27:30] David: uh, like he said that there was a team that he was working with internally who had spent something like three months trying to build this like specialized, image generation thing.

[00:27:42] Kalen Jordan: Mm-hmm.

[00:27:42] David: And. , As soon as they were ready to provide a demo, like an internal working demo,

[00:27:48] Kalen Jordan: Mm-hmm.

[00:27:48] David: Google came out with this like amazing AI image generation

[00:27:53] Kalen Jordan: Mm-hmm.

[00:27:53] David: And like Glenn, as the product lead or whatever, was like, okay, our thing is pretty good, but this thing that Google just came out with is way better.

[00:28:02] Kalen Jordan: Right.

[00:28:02] David: like, I know we just spent a long time working on this, but it's just not as good as what the state of the art is now. And so it feels like they probably struggled a bit with that and , are starting to make I, I guess, better

[00:28:14] Kalen Jordan: Yeah.

[00:28:15] David: on or something.

[00:28:16] Kalen Jordan: It's interesting because I feel like probably the tendency for companies is to be like, okay, AI's a thing. We're gonna get our smartest people on a project to build a thing. And then I think by default , they tend to sort of gravitate towards things that sort of compete with the models, the frontier models or like they build something which is in the implementation path for the frontier models.

[00:28:45] David: Mm-hmm.

[00:28:45] Kalen Jordan: because like they're trying to do something as challenging as possible and they're getting their smartest people. And then, and I think also the, the guy, the, the s ski o founder guy who started, oh, it's called Icon, it was this AI thing that generates ads and stuff like that, and everybody was super hyped about it.

[00:29:03] And then a similar thing happened with, , GPT images or something like that. And then what everybody now, all the Twitter posts I see are like, oh, it's not good, blah, blah, blah, blah, blah. And so, and then I feel like when it works is when they kind of democratize it and they just like. Push it out to everybody, not just the smartest people, but you know, the, the dumbs, guy or the dumb or like the dumb me who like, and I feel like then the general, like public finds use cases where it's almost like less is more.

[00:29:34] Like they're not trying to create an image model, you know, they're just doing a relatively straightforward wrapper around the models, which I think tends to be like the better path even though it feels like it's kind of too easy or something like that. And you know what I mean?

[00:29:50] David: if it's wrapping, then, it can benefit from the underlying model being improved.

[00:29:54] Kalen Jordan: Yeah. Yeah, exactly. And it's interesting how they said that. Like, I, I saw 'em tweet that like they're getting, they're seeing the most usage outta their like sales and operations teams, which is kind of nuts. Yeah. Which is like, and I'm sure like they have pretty smart salespeople on it.

[00:30:12] Like, you know, I'm sure they hire like, you know, talented and people in every discipline. So, you know, they're probably like cracking away at stuff,

[00:30:22] David: Yeah, just building like ad hoc reports or charts or whatever it

[00:30:26] Kalen Jordan: I,

[00:30:26] David: They want to actually like the, those people probably have certain things that they want to get at but can't

[00:30:31] Kalen Jordan: yeah.

[00:30:31] David: and like the data team or whatever,

[00:30:33] Kalen Jordan: Right.

[00:30:34] David: have a bot help them through it, which

[00:30:36] Kalen Jordan: A.

[00:30:36] David: sense.

[00:30:37] Kalen Jordan: And this, what's so cool about Toes is that like if it weren't for what he did with the memo and then publishing the memo, it might've continued in that direction where it's like, okay, let's have a certain team work on ai. It's like, it's a thing for a team to work on versus it needs to infiltrate every fiber of our company type of thing.

[00:30:59] And I feel like the companies that are ultimately gonna like, get hurt by not moving fast enough with AI are the ones that take it that direction, and then they wake up four years later and they're like, we haven't done anything. But like, I feel like Toby just maybe was like, dude, we need to, he like hit, he communicated the urgency and then it feels like it shifted like the, you know what I mean?

[00:31:23] David: just for technical people to use, like it's, is capable enough now that everyone should be

[00:31:30] Kalen Jordan: Yeah.

[00:31:31] David: some efficiency through using it.

[00:31:33] Kalen Jordan: Yeah.

[00:31:33] David: it's super like it's, it's awesome that they prioritized releasing that functionality to merchants because every merchant can like do some crazy analysis of bot traffic just

[00:31:45] Kalen Jordan: Yeah,

[00:31:45] David: and pasting this thing into a window versus needing a data team.

[00:31:49] Kalen Jordan: yeah, yeah. And, I mean, the way he framed it too was like, and it's a touchy subject because you don't wanna say, Hey, you're gonna lose your job. And he didn't say that, but what he said was, you are all expected to be using AI as a baseline. And I think that, , that does a couple things, like the people who are not looking into it, whatever, are gonna be more likely to look into it.

[00:32:10] And then also there's a kind of a way in which sometimes AI feels like cheating. Even if you're using it for something, you might not say that, because it feels like if you say that, it's gonna sound like. You cheated a little bit. Even if the end result is actually, you know what I'm talking about, like sometimes

[00:32:27] David: summaries. Like I, I was just working on a summary of a retrospective we did from holiday last year. I

[00:32:33] Kalen Jordan: right?

[00:32:34] David: it took me like 30 seconds and I'm like, I feel like this isn't good enough because it

[00:32:38] Kalen Jordan: E exactly. I, but it's like,

[00:32:42] David: shorten this more or just

[00:32:44] Kalen Jordan: or

[00:32:44] David: it or something

[00:32:45] Kalen Jordan: right. But it's like, yeah, you take the 30 seconds to do it and then of course you're gonna take some time to like review it, make sure it's actually good, but then if it is, it's like, okay, you spent two minutes instead of an hour. So I think it normalizes it as a, as a good thing where, and that's what it means to be like, AI first is like, people are gonna be like, assuming I'm using AI as much as humanly possible.

[00:33:09] So that it probably doesn't need to be this like guilty little, secret type of thing, you know? So.

[00:33:16] David: You can tell that what I generated is AI generated and that's fine.

[00:33:22] Kalen Jordan: Well, but I, but I feel like depending on the type of result you're looking for it, like, it depends, like in certain cases you shouldn't be able to tell, say I generated if, depending on the level of quality. But in other cases you want like a rough summary of some analytics. You might, maybe it's okay as long as like the number, like it's okay if the text looks a little autogenerated as long as the numbers are accurate.

[00:33:48] But if the numbers aren't accurate, then it's like, it's a bad, that's a bad usage, you know what I mean?

[00:33:54] David: That's what I feel like. I find myself doing that on Metrics day is like, okay, that looks like an interesting chart

[00:34:01] Kalen Jordan: Mm-hmm.

[00:34:02] David: I have never had a concept of before. But is it, true?

[00:34:06] Kalen Jordan: Right.

[00:34:07] David: like go through and check all of the numbers and everything, but it's

[00:34:11] Kalen Jordan: I check, yeah. Check everything by hand.

[00:34:14] David: from scratch.

[00:34:15] Kalen Jordan: Right? Totally. Yeah. And that's the whole thing is using the tools for what they are and not what they could be. And we're now curators of all this stuff and like you can do a really bad job of curating it or like a perfect job or somewhere in between. But that's like, that's the whole skill now.

[00:34:33] Did you say on metrics day?

[00:34:36] David: yeah. I, I spend a lot of the day on Monday looking at

[00:34:39] Kalen Jordan: Oh, cool.

[00:34:40] David: performance.

[00:34:42] Kalen Jordan: Oh, that's cool. So that's your metrics day. I,

[00:34:44] David: Yeah. I.

[00:34:45] Kalen Jordan: nice. Do you usually find some nuggets? Every, I feel like I need to have a metric stay now. It sounds like a cool thing.

[00:34:54] David: I try to like, it's one of those time boxing things where like every Monday, like I have the stuff that I need to generate. And then also I try and do like something that was a little bit

[00:35:02] Kalen Jordan: Is metrics day code word for like relaxing and taking it easy to, to like recover from the weekend. You call it metrics day. Just kidding.

[00:35:15] David: I, I actually like when I get into it, I'm just sitting there with Claude for like hours

[00:35:21] Kalen Jordan: Nice.

[00:35:21] David: to at stuff and

[00:35:23] Kalen Jordan: That's so cool.

[00:35:25] David: and it's really fun when it's putting together charts and stuff , and it like helps you visualize things.

[00:35:31] Kalen Jordan: That's so cool.

[00:35:31] David: made this, uh, this one chart. thought it was really cool. And then I, I showed my boss and she was like, this kind of sucks. So

[00:35:41] Kalen Jordan: What was the chart?

[00:35:44] David: it was plotting out all of the, for every channel, so like direct, organic, paid shopping, and paid search, blah, blah, blah. the conversion rate versus the sessions last year versus this

[00:35:56] Kalen Jordan: Uh huh.

[00:35:56] David: And so there were just a bunch of bubbles and then I linked them together with a, an arrow and it's kind of a crazy amount of things on one

[00:36:04] Kalen Jordan: Dude,

[00:36:05] David: understand,

[00:36:06] Kalen Jordan: sounds freaking awesome. I don't know what the hell they're talking about.

[00:36:11] David: oh, I'm still gonna make this chart every week, but just use it for myself.

[00:36:14] Kalen Jordan: Hell yeah, dude. So it shows one bubble to another bubble is showing last year to this year within that metric or within that channel. Dude, that's freaking cool. Were there any, any interesting, , deltas

[00:36:31] David: Yeah, there's a bunch of deltas, but it's, interesting to see like, okay, this

[00:36:35] Kalen Jordan: I,

[00:36:36] David: was down by this amount, but these two other channels are up by kind of the same amount just based on the length of the line. So

[00:36:42] Kalen Jordan: oh,

[00:36:42] David: kind of cool

[00:36:43] Kalen Jordan: that's cool. Yeah.

[00:36:45] David: And then like where they show up in the plot, like up higher is higher conversion rate and over to the right is more sessions, so you

[00:36:53] Kalen Jordan: Oh, sweet. It's fucking rad, dude. Don't let them tell.

[00:37:01] David: feedback on that one.

[00:37:02] Kalen Jordan: Yeah, dude, whatever. Whatever. I, he,

[00:37:06] David: that chart.

[00:37:07] Kalen Jordan: yeah, dude, there's nothing worse than being hyped about something and then like, your boss or client or whatever is like, like, like when I made that freaking duck hunt game, dude, that thing was gold, man. My client sh they didn't even reply to me.

[00:37:23] They never even emailed me back, dude. I was like, and I didn't, I didn't ha I was like, I'm not gonna follow up. I'm not gonna send a follow up email when the subject line is duck hunt, dude. Like there's certain lines I just won't cross, you know?

[00:37:39] David: Did you still build 'em for the hours though? Just

[00:37:41] Kalen Jordan: No. No. 'cause I'm, no, I, no, I did. No, I didn't. I, I spent like an hour and a half just myself on like a Saturday or something like that.

[00:37:51] I probably should have billed, I probably should have billed them though. I still have some outstanding hours. I gotta, um,

[00:37:58] David: Chase him down.

[00:37:59] Kalen Jordan: yeah, so I, I had this, client asking about some work and I have this, my main agency client I've been doing more work with, and it's been going good, like it's been going really smooth, working on interesting stuff.

[00:38:13] So then I kind of turned down some work from this other client, which is a little nerve wracking because. I like having my risk distributed a little bit between different clients. Like no one person can tell you, you know what I mean, what to do, but it's just like working with certain clients is just kind of draining.

[00:38:33] And also depending on the type of work and the amount of back and forth and the amount of like general confusion, which tends to be pretty high among merchants. Like there's a significant amount of confusion, ,

[00:38:45] David: Do you charge based on, on those kind of like internal metrics that you have?

[00:38:49] Kalen Jordan: I mean, I do charge them more, like I'm charging less per hour and I'm also doing, a decent chunk of hours, but I'm still charging a bit less than what I would charge merchants. But I still, 'cause like I'm not.

[00:39:03] David: hours overall,

[00:39:04] Kalen Jordan: It is more hours overall. And also, like, I'm not primarily concerned with, maximizing my hourly earnings.

[00:39:11] What I really want to do is have the space and like mind space to like build an app at some point. And, and also to be working on things that are kind of like, could become products or, are like in a strategic vein kind of a thing.

[00:39:29] That's the other thing. But anyway, but I was just like, I'm a little, a little nervous about, you know.

[00:39:35] David: There's gotta be a lot of benefit to doing the, like freelance work and understanding more closely what the problems are that need to be solved.

[00:39:46] Kalen Jordan: Yeah, I mean it's nice to have like exposure to a broad mar like swath of the market, I guess, or to different tiers in the market on the one hand. But on the other hand,, sometimes they just have weird like problems that are just stem from them kind of doing things in a very weird way, and they're not necessarily, you know what I mean?

[00:40:06] Whereas like, you might be, and if you're working with an agency, you're working on some different, your different clients, so you're getting some variety, but it's almost like you want, you, it's almost like you want to see the problems more that are from, I don't know, I don't know how to say it.

[00:40:23] Like, It's just weird. It's like sometimes.

[00:40:27] David: be faced by a more like a wider merchant population.

[00:40:33] Kalen Jordan: Yeah, yeah. Wider, like more best practice type thing. But anyways, that's my little dilemma to deal with. Let's see. Next up, next up on the podcast. , Dude, so I've been wanting to search for tweets for what feels like my entire life, but I guess I've been on Twitter now for like 13 years, something like that.

[00:41:00] And freaking, I used Grok the other day. there was this , log system logging app that, , Jake had,, replied to me about like, I don't know, a week or two ago. Normally I, when that happens, I already know search is not gonna work. So I scroll through my thing. , Dude, I've scrolled back like two months before looking, you know what I mean?

[00:41:23] Looking for the thing. And so anyways, I asked it and it found, it, found the tweet. It was like, and it showed what it was doing. Like it showed, it was like querying tweets for this user ID with these keywords. So dude, it was magical. It was a magical moment. I was so excited.

[00:41:45] David: I wonder if internally grok has some, , like it's probably not MCP, but something like that. I'm sure it does.

[00:41:52] Kalen Jordan: Well, I think it's gotta just have a database con or probably an API connection to Twitter's API and then it just con constructs. And so I don't understand the MCP thing. Like you, you would, if you're building gr you would just have the inner, I don't know. I don't know how it talks to it, but I feel like you'd want it to just direct sort of more directly talk to the database.

[00:42:14] But no, maybe you're right. Okay. Maybe you're right. Maybe they do have an MCP for that.

[00:42:19] David: I guess it depends on if, is is there more context outside of your request that's important to that question? not. maybe they just have a

[00:42:28] Kalen Jordan: Right?

[00:42:29] David: like a query API that it can use. But then why is that working better than the function?

[00:42:37] Kalen Jordan: I guess the search function, if I, I can't remember if you can search for tweets from a person or not, but I'm pretty, I remember. Okay, so then I guess if you did that and then you searched on keywords, it probably would work. I guess it's just making,

[00:42:57] David: it faster or

[00:42:58] Kalen Jordan: yeah.

[00:43:00] David: like

[00:43:00] Kalen Jordan: Yeah.

[00:43:01] David: bigger amount of responses and then it can process the responses faster. Maybe that's it too,

[00:43:05] Kalen Jordan: I don't know. And I'm, I'm pretty sure it's gotta have more flexible access to the underlying API than that advanced search page. 'cause that advanced, and I've tried using the advanced search many a time and like not being able, for one reason or another, not being able to find the thing that I was looking for, or the results were really bad.

[00:43:25] It was weird. So.

[00:43:28] David: search of any search I've used is, is Max Finder search. I don't remember if we've ever talked

[00:43:34] Kalen Jordan: nightmare. I think we have,

[00:43:35] David: in the exact wording, like the exact file name. Still nothing.

[00:43:43] Kalen Jordan: it's a nightmare, but, and like Twitter is literally, it's like the primary place I learn about useful things for my job, so there's a lot of times where I want to search for something, you know? So Pretty sweet.

[00:44:01] David: I do like bookmarking on Twitter. I've been using that a little bit.

[00:44:04] Kalen Jordan: Yeah, yeah, yeah. I'll bookmark either funny videos to show my family or like things for the podcast pretty, pretty much. Um.

[00:44:14] David: Also interesting that Grok just kind of like Grock is the only that is kind of like living in the public

[00:44:22] Kalen Jordan: Mm-hmm.

[00:44:22] David: Like no, no one else is, is being as consumed, I feel like, by a wider set of people

[00:44:28] Kalen Jordan: Mm-hmm.

[00:44:28] David: just in the tweet responses everywhere.

[00:44:30] Kalen Jordan: Right. Yeah. You're probably right. And people just tag rock and stuff like that all the time and yeah, it's, it's a, it's a pretty interesting What's your take on the, on, did you see the news about, , ROC four and the benchmarks and stuff?

[00:44:46] David: I saw that and I thought it was kind of interesting that it was like days, like one day or two days after Grock started getting real racist with everyone. Did you see that

[00:44:55] Kalen Jordan: No.

[00:44:57] David: it was bad, like they had to shut him down for like the whole evening? 'cause he was saying some real bad stuff.

[00:45:04] Kalen Jordan: We all get a little racist sometimes. You got,

[00:45:08] David: I guess if you're an ai, it's like way worse.

[00:45:11] Kalen Jordan: that's way worse. Yeah. That's funny, but Well, and it's like, I don't know. I wanna believe that, , it's the best. But like, I don't know, I do feel like Elon exaggerates sometimes. And um,

[00:45:25] David: Oh, definitely. I.

[00:45:26] Kalen Jordan: and like, I mean, I still think he's probably the greatest entrepreneur of all time, but, um, he does like, there's some weird things he's exaggerated.

[00:45:34] And then also the thing I don't understand with these benchmarks is like, it feels like you would just train the model specifically for the benchmark. I don't know how randomized the benchmarks are or how the hell they work, but, and I've heard people say that, that once a benchmark comes out, I'm pretty sure I've heard them say that you can train.

[00:45:56] There are ways to train for it, but,

[00:45:59] David: Then the question is, how good is it for like actual use cases, like

[00:46:04] Kalen Jordan: yeah.

[00:46:04] David: people are, how many of your users are gonna be scientists trying to get editing help on there? Like 5,000 line white paper that's talking about new science.

[00:46:14] Kalen Jordan: Right,

[00:46:15] David: I guess that's the frontier, but like the, from a product perspective, need to like using it also.

[00:46:22] Kalen Jordan: right, right. Yeah, it's, it'll be interesting to see. It's, it is interesting. I think his approach on it is like first principles, so it's like trying to like put all the first principles of physics in there. I, I don't know. I don't quite get it, but,

[00:46:38] David: it feels like rock is the one that's like more heavily engineering focused,

[00:46:42] Kalen Jordan: right.

[00:46:42] David: and then anthropic, I, I feel like is inter, like, sciencey,

[00:46:48] Kalen Jordan: Mm-hmm.

[00:46:49] David: Like poking at things a little bit more interestingly, just based on the things that they release.

[00:46:54] Kalen Jordan: Mm-hmm.

[00:46:54] David: kind of fun to watch them all compete and come out with different things.

[00:46:58] Kalen Jordan: Yeah, it is cool. And You wouldn't have imagined the ways in which they differentiate, like early on it felt like, well, they're can all, all gonna kind of be the same. And you wouldn't imagine people would use multiple of them. And now you're seeing like everybody's using multiple of the models in d in different ways, at different times.

[00:47:15] David: I wish I was better at using agents. Like I, I feel like I should just have agents running

[00:47:19] Kalen Jordan: dude,

[00:47:20] David: but I don't know what,

[00:47:21] Kalen Jordan: I know, I know, I know. This is the thing, dude, I, and I see these tweets, how I got an agent to, and I've stopped, I've stopped watching the videos 'cause I'm like, I don't know man. I don't, I don't buy it. , But at the same time,

[00:47:35] David: interactions, like one-on-one with an LLM, like it, it takes a lot of

[00:47:39] Kalen Jordan: yeah.

[00:47:39] David: So how, how does an agent.

[00:47:41] Kalen Jordan: I don't get it, but I see people saying they're using it and it changed their whole business overnight and stuff. And I know that things are changing fast and I also know that I haven't really been adopting anything new recently. Like I just have my boring two to three x productivity increase with cursor sonnet four, you know,

[00:48:00] David: Hmm,

[00:48:01] Kalen Jordan: and I'm just like cranking along and I haven't done anything significantly new with like, my kind of AI workflows in like a month or two.

[00:48:11] And I'm like, I'm so far behind right now. Like, it's funny.

[00:48:18] David: I used, uh, Claude Code to like, help me make some updates to the privacy policy

[00:48:24] Kalen Jordan: Nice.

[00:48:25] David: downloading some, so it wasn't code at all. And when it, when I did Quadin it, it was like, see there's like no code here. It's just all regular files, and it was fine. It worked.

[00:48:35] Kalen Jordan: Nice. That's cool. I saw this tweet, this chick, like coded her own music. , And

[00:48:43] David: I saw that too.

[00:48:44] Kalen Jordan: she was,

[00:48:46] David: was like tweeting about that,

[00:48:47] Kalen Jordan: yeah, that was so cool. And it, like, she was showing the code and I was like, dude, because like we're all starting to think like we're all starting to be able to make more stuff than we could make before.

[00:48:57] Like, because, because of the AI coding tools. Like I could probably make an iPhone app now. Before, I never would've even tried to make an iPhone app, but I'm like,

[00:49:06] David: right?

[00:49:06] Kalen Jordan: you know, but everybody's just building their own little fun little personal software and I'm like, dude, I should totally like. Do that like, 'cause I listen to electronic music all the time when I work, and it'd be so cool, like you could have it respond to your mouse or you could have, I don't know, keyboard shortcuts, whatever, like, you could slightly speed up the tempo or like, I don't know, man, there's all sorts of cool stuff you could do,

[00:49:32] David: Yeah, that editor that she was using was really interesting 'cause it, it also like highlights the line of code that's being executed to make each sound.

[00:49:40] Kalen Jordan: right?

[00:49:41] David: through and like click in and change this parameter and it updates immediately. Like she had five different functions, defined for each part of the song.

[00:49:50] So it's like

[00:49:51] Kalen Jordan: Yeah. I had no idea. I had no idea what that editor was. I didn't know if that was like just a visualization or what the hell was going on.

[00:50:01] David: It's some sort of open source project that like you can just go and, and do.

[00:50:06] Kalen Jordan: Oh, okay. Nice.

[00:50:08] David: I don't remember what it was.

[00:50:09] Kalen Jordan: Nice. I gotta check that out.

[00:50:11] David: yeah, , you can go do it and then like write some code and see what it sounds like.

[00:50:15] Kalen Jordan: Nice. Nice. , Yeah, no, that's pretty cool. I gotta check that out. You got anything on your list? I Don't wanna monopolize the, i all the topics if you have anything on the old list,

[00:50:29] David: My list mainly was the MCP thing. 'cause I, I

[00:50:32] Kalen Jordan: right?

[00:50:33] David: time like actually setting it up week. And I'm, interested in kind of understanding how stuff will keep progressing. there's also a, a dev topic specifically for LLMs and there's only a few, topics there so far. So it'll be interesting to see how that fills up over time.

[00:50:50] Kalen Jordan: Yeah, yeah.

[00:50:51] David: Like, will we get any access to being able to adjust those things

[00:50:57] Kalen Jordan: When you say topics, like, what do you mean topics where?

[00:51:01] David: On the dev forms,

[00:51:03] Kalen Jordan: Oh, gotcha, gotcha, gotcha, gotcha. Yeah.

[00:51:06] David: a category for. LM and MCPI think is what it is.

[00:51:10] Kalen Jordan: Oh, sweet. Okay. I, I gotta check that out. , Yeah, no, the,

[00:51:14] David: one person in there who posted about the MCP, responding with details about the specific inventory that's available for a product. So

[00:51:23] Kalen Jordan: oh, nice.

[00:51:24] David: it's early 'cause they're like, oh that shouldn't be happening. And so then they're making adjustments based on those posts.

[00:51:31] Kalen Jordan: Oh, you're not supposed to leak inventory levels, right?

[00:51:36] David: Yeah. Like I, I typically, I, I feel like on the storefront, it's like, you don't want that to be easily scrapable, I

[00:51:43] Kalen Jordan: Your competitors.

[00:51:45] David: of these. Yeah.

[00:51:47] Kalen Jordan: Right, right, right. Yeah. No, that's cool. , And the storefront MCP stuff is, is super cool. I was, was having a hard time. Wrapping my head around that, I gotta dig into their MCP. I'm experiencing a little bit of vendor lock-in pain at the moment without

[00:52:06] David: oh.

[00:52:06] Kalen Jordan: naming any specific vendors.

[00:52:08] It's just, sometimes you're excited about something and then you kinda become, , dependent on it for your project and then you're like, oh man you just run into more issues. Um.

[00:52:25] David: pricing

[00:52:26] Kalen Jordan: Limitations, you know, limitations or like little bugs and you kinda gotta wait for someone else to, deal with it.

[00:52:34] Like, whereas, you know, if you're on something like CloudFlare Workers or cloud, there's probably very few actual bugs you're gonna run into. Like it's, and you're kind of just on your own. Like, if you don't know how to do something, you're not gonna, you'd have to pay someone to help.

[00:52:49] David: harder,

[00:52:50] Kalen Jordan: Yeah. Yeah. But, the only limitations you're gonna run into are like the core architecture, like, infrastructure limitations type stuff. Yeah, a little frustrating. Have You experienced that before? Vendor lock in pane.

[00:53:06] David: Experience it like every time we select a vendor.

[00:53:12] Kalen Jordan: I've never not experienced it.

[00:53:15] David: Yeah. Well,

[00:53:17] Kalen Jordan: life is a disaster.

[00:53:19] David: and then you're like, yeah, you're, with them for at least a year most of the time. And so you just have to kind of like really try and get their support. And if it's not happening, then build on top of it. But yeah, it, it

[00:53:32] Kalen Jordan: right,

[00:53:33] David: like there's something weird like that.

[00:53:34] And it comes up sometime near when you've fully explored things. Like

[00:53:40] Kalen Jordan: right,

[00:53:40] David: explore everything before you sign a contract most of the time.

[00:53:43] Kalen Jordan: right, right. Right.

[00:53:45] David: that part sucks, but you, you try and limit it as much as possible and it happens.

[00:53:49] Kalen Jordan: Yeah, it's like in theory you fully explore and fully vet and everything's fully verified, but in practice, like

[00:53:57] David: Mm-hmm.

[00:53:58] Kalen Jordan: just always something that you know is gonna come up. I don't know, I guess everything, nothing's perfect, so there's always gonna be some frustration, but

[00:54:09] David: Yeah.

[00:54:10] Kalen Jordan: I mean,

[00:54:11] David: are slightly different for one reason or another, but sometimes it happens and it's like, how is anyone using this?

[00:54:18] Kalen Jordan: but yeah, like, I think the difference is when, like with Shopify, there's tons of frustrations. Support is horrible. I mean, there's lots of frustrations, but I wouldn't say I, I feel like I have vendor lockin, like I want to leave, you know?

[00:54:36] David: Yeah.

[00:54:37] Kalen Jordan: I think when you get to that po Yeah.

[00:54:40] Like when you get to that point emotionally with a piece of software where you're like, I wanna leave, but it's just not practical because of X, Y, and Z. That's the feeling I'm talking about.

[00:54:52] David: Yeah. with Shopify, the limitations are where your work is at, right?

[00:54:56] Kalen Jordan: Right.

[00:54:57] David: value on top of what the base

[00:54:59] Kalen Jordan: Right,

[00:55:00] David: is,

[00:55:00] Kalen Jordan: right, right,

[00:55:02] David: but in some cases it's like this thing has to be this way. had something that was like gonna be done in three days, like some big chugging project to update some data.

[00:55:12] Kalen Jordan: Mm-hmm.

[00:55:13] David: I Had one piece of stack that was like, Nope, that's not gonna happen. And so it went from update taking like three days to taking like a full month. And there's just nothing I can do about that. I just have to with it.

[00:55:28] Kalen Jordan: That's brutal. That's crazy. That's a pretty long process. Did you end up, , that thing we're talking about where you were gonna, do all those mass updates in Google Cloud to like,

[00:55:41] David: that's the project.

[00:55:42] Kalen Jordan: oh, okay. Okay.

[00:55:43] David: It's finally

[00:55:44] Kalen Jordan: That's what I was thinking. That's what I was thinking.

[00:55:47] David: Yeah, A month later.

[00:55:48] Kalen Jordan: What was the, without getting into specifics if you don't want to, but what was the bottleneck with a vendor that was, I mean, it would've been in Google Cloud that couldn't go faster,

[00:55:58] David: it wasn't Google Cloud. It wasn't Shopify. 'cause we have like infinite API

[00:56:03] Kalen Jordan: right.

[00:56:03] David: there's no way I would ever, it was, the part that like syncs the orders from Shopify to NetSuite.

[00:56:10] Kalen Jordan: Oh, okay. Gotcha. I.

[00:56:11] David: It counted all of those updates as order updates and slow. like the first day I was chugging through stuff and then like the next day, no new orders were in NetSuite because. We were still processing all the old ones that I had just adjusted. And so

[00:56:29] Kalen Jordan: Gotcha.

[00:56:30] David: from that, so it, we went from like, it was like thousands of times more order volume than what we normally see. And so,

[00:56:42] Kalen Jordan: Wait, so I thought you were updating data directly in Google Cloud that had originated from Shopify SA.

[00:56:54] David: the, um, I forget what it's called, but it's a fulfillment order move. So I wanted to change the location on the fulfillment orders.

[00:57:01] Kalen Jordan: Gotcha. Okay. Oh, that's right, that's right, that's right. This, okay. Got it. So then all of those updates triggered the other software to send the updates to NetSuite.

[00:57:12] David: The worst part about it is it within the flow, we filtered those out, so there was no update to NetSuite, but it still had to process the fact that there was an update to an order that we wanted to ignore.

[00:57:25] Kalen Jordan: Gotcha. That's, that's awesome. Hey, man, if you need a NetSuite guy, actually, uh.

[00:57:32] David: heard of someone working on some like admin blocks. That's pretty sweet.

[00:57:36] Kalen Jordan: Yeah. So, in certain cases the sync doesn't send an order over because one piece of data or another is missing or something like that. And so they wanted a way if they're looking at an order to manually resend it. So it's like, oh, cool, I can do an admin block for that. So it just, it just gives them a little, and then I was like, well, okay.

[00:57:57] Then I also went ahead and pulled in the, like a link to the, so you can easily click a link to get to NetSuite, stuff like that,

[00:58:04] David: okay.

[00:58:05] Kalen Jordan: but

[00:58:05] David: felt like that was the main functionality and I was like, yeah, I want that,

[00:58:09] Kalen Jordan: Right, right, right.

[00:58:10] David: the link.

[00:58:10] Kalen Jordan: Yeah. It's funny because I was primarily thinking about the resend functionality, and then as I was building it, I was like, oh, I should put a link in here. And then I was like, and then I felt the same way. Like the link is kind of like the coolest part, even though it's like super easy.

[00:58:26] David: yeah. Well it's interesting, like you can't get the link if it's not there. So could, I guess it

[00:58:31] Kalen Jordan: Yeah.

[00:58:32] David: updates to be a button that you can click

[00:58:33] Kalen Jordan: It, yeah. Ex.

[00:58:35] David: cool.

[00:58:35] Kalen Jordan: Yeah, exactly. Yeah, exactly. So yeah, that was good times. Good times. But, uh,

[00:58:42] David: Someday I will be unlocked from this vendor and we'll find the new

[00:58:47] Kalen Jordan: we'll build some s build something sweet. Heck yeah. Dude. Wait, what's the cp

[00:58:55] David: KKCP, Kalyn context protocol. I don't know.

[00:58:58] Kalen Jordan: Oh, context protocol, right? , Dude, , okay, so we're about at our time. Oh my gosh. I wanted to, , quick pop quiz. Quick pop quiz for you,, commit hashes in, in get,

[00:59:11] David: Mm-hmm.

[00:59:12] Kalen Jordan: it's always a combination of letters and numbers, right? I.

[00:59:16] David: Yeah.

[00:59:17] Kalen Jordan: That's what I thought too. But I randomly had a commit hash that was only numbers.

[00:59:24] And I was like, I was like, I'm pretty sure I've never seen this in like 15 years or whatever the hell it's been. But then I'm like, well, yeah, it's probably just, it can randomly be all,

[00:59:37] David: Or is it Base 64?

[00:59:39] Kalen Jordan: it's,

[00:59:40] David: I don't know.

[00:59:42] Kalen Jordan: I always forget, I always forget, but

[00:59:44] David: If

[00:59:44] Kalen Jordan: kind of cracked me up.

[00:59:46] David: yeah. That's interesting. That's like super rare. You should, frame that.

[00:59:50] Kalen Jordan: I should buy some, I should buy some Bitcoin right now. Right.

[00:59:55] David: Yeah, exactly.

[00:59:56] Kalen Jordan: I hear it's gone up. I still haven't Googled the price, by the way.

[01:00:02] David: Yeah. I don't know anything about Bitcoin. Either

[01:00:04] Kalen Jordan: I still haven't Googled it 'cause, uh. It's funny, it's like I literally, I'm trying to emotionally prepare for the fact of how depressed I'm gonna be. Like the higher it goes, like literally, it's gonna be so bad like five years from now or something.

[01:00:21] David: we will

[01:00:21] Kalen Jordan: um,

[01:00:22] David: not.

[01:00:22] Kalen Jordan: yeah. When it was going up, I kept thinking to myself like, the more this goes up, like people are gonna get targeted to get robbed and stuff like that, you know, like you start,

[01:00:33] David: Yeah.

[01:00:34] Kalen Jordan: you know what I mean?

[01:00:35] You start thinking about that

[01:00:36] David: have to be a, like, some sort of like better, what is it called?

[01:00:41] Kalen Jordan: like protocol security protocol. Yeah,

[01:00:49] exactly. Exactly. Well, that's the thing is I really believed in the whole like, self custody thing and it's like, it's pretty cool that it's like. The, it's sort of like the first time in history that you can, you can do that,

[01:01:03] David: Yeah.

[01:01:04] Kalen Jordan: in the way that you can, like it creates a new, because like there's ways you can do it to where even if you're physically coerced, it wouldn't be possible to be physically coerced to give your Bitcoin if you had like a two outta three signature distributed across different cities.

[01:01:21] You can memorize your key and then cross borders with the money in an, in a way they couldn't possibly get it unless they coerced you, they could get it. But, it, you know, like I really,

[01:01:32] David: How many people are gonna actually do that?

[01:01:34] Kalen Jordan: yeah, I really

[01:01:35] David: many.

[01:01:36] Kalen Jordan: Right, right, right. Yeah, I mean, I think it's, I think it's pretty cool, but, um.

[01:01:43] David: It is, it's really cool. It's interesting technology. that's another thing that I'm like, I don't, I don't know how this is all gonna shake out, but fun to watch.

[01:01:50] Kalen Jordan: Yeah. Yeah. But I think everybody has something they did stupidly with, like, there's a lot of people who almost bought it when it was like 50 cents and they carry that trauma with them, like to this day, you know, so.

[01:02:04] David: I have some of that trauma. I had, a friend who, a long time ago was like mining Dogecoin for some reason, and he gave me a bunch.

[01:02:12] Kalen Jordan: Right.

[01:02:13] David: it didn't seem like a bunch back then. It was like 5 cents.

[01:02:16] Kalen Jordan: Yeah.

[01:02:17] David: if I still had that, it would be nice. But

[01:02:20] Kalen Jordan: Nice,

[01:02:21] David: gone forever.

[01:02:22] Kalen Jordan: nice.

[01:02:23] David: hard drives like multiple times.

[01:02:25] Kalen Jordan: You hired a private investigator.

[01:02:27] David: Yeah. Maybe I should

[01:02:30] Kalen Jordan: Nice.

[01:02:30] David: I?

[01:02:31] Kalen Jordan: That's crazy. I. Yeah, that's, I guess that's just, yeah, and I never had any of that 'cause I never really thought about Bitcoin until, I guess whenever it was, a few years back when I got some. Um, so I never felt like that pain. But now I'll get to feel that for the rest of my life.

[01:02:48] I, so that should be cool. Okay. Last thing, dude. 'cause this drove me nuts. I was going through my tweets and I got emotional, just remembering it was how they, some people posted the built with chart. They were like, like marketing people. Well, one of them is actually a founder of a SaaS. I think it's like an analytics SaaS called Ferma or something like that.

[01:03:09] And then I just kept seeing people post about it. But the built with chart goes down for the first time ever for Shopify. And they thought. Like that stores were closing. And then they posted that and then everybody was repeating that. And I was like, in my head, I was like, there's no way store.

[01:03:29] There's no way the number of active stores has gone down. I just, I refuse to believe it. So, like I looked on the, , store leads site, which tracks Shopify stores specifically, and that one was showing that stores have been increasing. So then it's like, okay, well the built with chart might mean that analytics have dropped or that traffic levels have dropped, which could have taken some of them off of the top 10 k, top a hundred K, top one K lists.

[01:03:58] That's possible. But also, like if you look at, like, a lot of things are down, like Cloudflare's down, Facebook's down to on built with like, I think it's a CloudFlare bot detection thing or something like that, that is skewing their numbers. But, um.

[01:04:13] David: is built with using CloudFlare data or something.

[01:04:16] Kalen Jordan: I mean, it tracks CloudFlare just like anything else. But if, which there's no world in which CloudFlare usage has gone down. Like that's why I was interested in what CloudFlare would show

[01:04:27] David: That was

[01:04:27] Kalen Jordan: and

[01:04:28] David: that you posted, was that like, this is actually across CloudFlare and Facebook too,

[01:04:32] Kalen Jordan: Yeah.

[01:04:33] David: it that's causing this downtrend?

[01:04:35] Kalen Jordan: Yeah.

[01:04:36] David: also did you see that, Amazon, I saw something about Amazon's, prime day was down by like 40% or something crazy on the

[01:04:44] Kalen Jordan: Really. Holy shit. That's crazy. I,

[01:04:48] David: there is some level of like, maybe not stores closing, but customers purchase behavior changing these last couple months.

[01:04:57] Kalen Jordan: I could see that. I could see, . And I think that's probably why they're seeing the numbers in that light and interpreting it. And then like, people were like, is our industry collapsing? I think that like the tariffs, all that kinda stuff, like may people are spending less or whatever. And I know some stores have closed, but like, I don't think net they're down.

[01:05:21] David: I'm glad you're a champion for Shopify stores opening.

[01:05:27] Kalen Jordan: well it just didn't make sense to me as I'm like, it's like Shopify's like a freight train. But I mean, yeah, at some point it, it could happen if the economy takes more of a hit.

[01:05:39] David: that means for Facebook though. Like if, if the signal is that shoppers are not shopping as much, then why does it impact Facebook traffic also?

[01:05:47] Kalen Jordan: Yeah. That's why I don't, think traffic levels are down let's say, even if shopping levels are down, I don't think traffic levels are down. People are probably still clicking around on stuff. They're just like, oh, I can't buy it right now.

[01:06:01] I think the measurement is just off in that, on that thing. But dude, like,

[01:06:06] David: like next month.

[01:06:08] Kalen Jordan: yeah.

[01:06:09] David: monthly chart.

[01:06:11] Kalen Jordan: Right, right. Yeah, yeah, yeah. , But yeah, this one dude, , you know, be, have you heard of Beard Brand? The,

[01:06:16] David: Mm-hmm.

[01:06:17] Kalen Jordan: yeah. There's that guy Eric, that's the founder on Twitter, the Beard brand founder. He's like the fame famous.

[01:06:23] He's been on like TV commercials and stuff. Have yous ha.

[01:06:28] David: I read through that, that article about their problems that they were going through.

[01:06:32] Kalen Jordan: Did I? Yeah, that's cra I didn't read it, but I, I, I tweeted him about it and he mentioned that Target dropped them and like stiffed them for 200 grand and then some other things happened or whatever, and they were like, lost a million dollars or like a million dollars down in losses. I'm like, holy shit.

[01:06:49] Like that would stress me out. Yeah. But I don't even know if that was that related to the tariffs. Well, probably was somewhere related to the tariffs.

[01:07:01] David: Yeah, it sounds like there were a bunch of problems. They were dealing with one of them potentially being shipping and, logistics or something like that.

[01:07:09] Kalen Jordan: Oh, okay. Gotcha. Yeah, I don't know. I mean I'm, oh, I did hear about. I think I saw someone posted that, like this merchant, like closed down their store because of tariffs and they like asked them for a job or something, or somebody did some cold outreach to a merchant to promote their product or something.

[01:07:27] And then the person's like, actually I'm, going outta business 'cause of the tariffs and let me know if you, if you have a job or something like that.

[01:07:33] David: Oh

[01:07:34] Kalen Jordan: which sucks. Yeah. , So I'm sure it's happening.

[01:07:39] David: love e-commerce. I want it to still be like unique and not like just a few big ones that are doing it, you know?

[01:07:46] Kalen Jordan: Yeah, no, and I mean that's the thing, despite all this stuff, the fact there's still more stores opened every month, and those are paying stores. Well, I don't know exactly how they measure that, but if they didn't pay, they'd disappear after 90 days or something. So.

[01:08:03] David: Gotcha. That's good.

[01:08:04] Kalen Jordan: Yeah. They're not just like,

[01:08:06] David: I,

[01:08:06] Kalen Jordan: like the Magento days when we were counting stores and nobody had any idea if it was even like an active store or not.

[01:08:14] Bunch of demo stores posted at random URLs,

[01:08:20] David: they're all Shopify go stores that are just still live for some reason.

[01:08:24] Kalen Jordan: Shopify go. , Nice. . Good stuff. Good stuff.

[01:08:30] David: Good stuff.

[01:08:32]