Resources / Podcast / The Factory Floor

Corey's AI marketing repo hit 15,000 GitHub stars

The Factory Floor · Episode 21 · March 27, 2026 · 38 min

Watch
Transcript50 sections · auto-generated, so expect the odd wrong word

Me, me, me, me. I'm good. All right, ladies and gentlemen, welcome to the factory floor. We are uh joined by, of course, >> ourselves, >> the two the two tried and true, well, it's me and then the listeners, of course, I am the factory floor and you guys are merely my guests every single week. >> Recurring guests. Yeah, that's right. Uh Corey Haynes and Zach Stevens. Uh how we doing today? Are we good? >> Good. Thanks for having us on your podcast. Appreciate it. >> Yeah, no problem. >> Thank you so much. Pleasure to be here. >> Yeah, you guys better perform. But uh because if this doesn't work out, then next week we're getting somebody else. Um, okay. So, this week I what I would like to do is um

ask Corey some questions because in case you haven't been on Twitter recently, uh there's a lot of things happening in the world of AI news flash and uh Corey was a little bit of like a a key part of it recently. Yeah. Um which he released uh uh marketing skills for Claude Code. And so what I'd like to do is just ask him about that. Like what what did he put out? What is it? Like what is the skill if anyone doesn't know if they're listening? And then what spawned from that? Cuz you've been on some podcasts. You've gone and spoken at some places. So give us a rundown like what happened a few weeks ago that sparked all this uh all this uh happenings. Yeah, I think it was like the last week of January and I was just starting to

play around with some skills in cloud code for coding and design. And then I was like, hm, I wonder if there's any for marketing and andor copywriting or how that would work. And so I made my own to kind of experiment with it. And I just like kept it open source and and public in GitHub. And uh and then I was like, oh, this actually seems to be working like really well. It's really good at copywriting. It's good at like a lot of the recurring types of tasks that we do at Conversion Factory. Um, and I was already trying to get my team, my the marketing team at Conversion Factory up to speed on like here are some workflows, here's some different types of things that we can and should be doing where we can, you know, connect to

the HFS MCP and do keyword research or we can um go look through uh a bunch of, you know, ideas and from our 140 SAS marketing ideas ebook and then match them, I guess, a client and all those types of things. Um, and then I was like, I think that this is probably something that could do well that other people are going to use if they're already using cloud code. And so I like kind of launched it, but then it really took off like honestly way more than I initially uh expected. And I think because every other skill out there, which I'll explain what a skill is in a second, every other skill out there already um was just focused on like super super technical things. Um but as we know, coding and programming

is getting really really easy. And so what people have appetite for now are all the things that are not related for coding and programming. It's for marketing. It's for design. It's for sales. It's for customer support. Um and and cloud code is this this tool, this this vehicle that allows you to do so much more than a chatbt. And so these skills kind of give cloud code, which is already a really powerful tool, even more tools and capabilities and basically steers it in the right direction. So, like a skill is a markdown file with a bunch of rules about what to do and what not to do. And so, I made a skill pack that was just a bunch of skills all together in an open source repo.

The way that you described skills to me was that they are essentially an SOP, like a standard operating procedure for a for an AI uh tool to follow, which is great because then what you're doing it's like essentially you're giving the the bot a framework for this is how I think about these scenarios and how I make the decisions. In the context of marketing, it could be something like lots of if this then that scenarios, right? Do you have a positioning doc? If no, create one. If you do have one, great. Does your homepage reflect it? Uh does the copy on the homepage reflect like the CTAs and the value propositions and all those other items. Does it have a glossery for the terms that we need to

use for this industry? So it's essentially taking all the things that you do and putting them into a very codified format for the bot to follow. >> Yeah. Yeah. And then giving the bot instructions about how to do it too. That's one that's one of the biggest differences is how most people are used to using AI is you you talk with chatbt. Hey, I need to do this thing. It tells you cool here's step one, step two, step three, and here's the thing that you paste into your email marketing tool. and then you have to go and do that and follow the directions that the AI gives you. Now with cloud code, it just does it for you and skills make that SOP and that process, that workflow much clearer. Um, >> and so now you're not like the middleman

anymore, uh, of, you know, being in between AI and your tool where the work is actually happening. Cloud code now is, um, the thing bouncing between you and the tool and just doing it on your behalf. And that unlocks a lot because AI is, >> you know, thousands of times faster than we are. And so when you can give it these guard rails, um, you can do a lot more than you ever thought was possible. >> Yeah. It seems like there's a based on what we've seen over the last few weeks with with your marketing skills thing like there's probably a big market for these like adjacent hard skills that aren't like directly related to coding that [clears throat] AI can use to make products and make services a lot better.

So I'm really curious to see how that develops. I do want to ask how did your so take me from you wrote this thing and then you tweeted about it to like then the the next echelon of things that happened which is you were on some podcasts and in some YouTube videos and speaking at uh different companies and conferences and things. How did that how did that take place? >> Yeah. So how GitHub works is when you have an open source project people can get a star which kind of like saves it for them but it's also just like a public you know count of endorsement essentially. And so um you know pretty quickly it hit like a thousand stars and then it started trending on the front page of GitHub and then it started circulating

around Twitter because it got all these likes and I think the original tweet got close to like a million impressions and so just a lot of people saw it and cloud code and skills had already just started to take off and so it was kind of already in the zeitgeist of like what tech Twitter was talking about at the moment and then my repo popped up and just started circulating everywhere. The other factor was that there was uh there's this company Verscell which one of their guys Andrew Woo he created a skills directory called skills.sh and by the way I just learned sh stands for shell and shell is like a uh a command language more or less within your terminal. So um it's called skills.sh SH and it's a directory, a marketplace, you

know, a big listing of all the skills out there. And then they made they kind of wrapped it around some uh like a package manager um in NPX. And so then my skills are all started trending on the skills directory and then they had done a lot of marketing around it and it it blew up on their end which just kind of helped perpetuate the momentum that was gaining with the marketing skills repo. And um that was crazy because then you know there's all these companies publishing skills for the first time. You know Versel's published skills, Anthropic um Microsoft and then it was like Cory Haynes 31. [laughter] And so I think for a lot of people that was the first time that they're like who is this guy and like what is this you

know who what are the marketing skills? There were so many. I think I published 22 initially. Like Verscell published one. They published Verscell React best practices I think it's called. >> I published 22. So it just kind of dominated the front page of the directory and everybody started seeing it. It's only one command away from installing and um you know and got a lot of a lot of hype that way. So then Verscell decided that they wanted to launch a couple of parties um around the US. So San Francisco, New York, and London. And they offered to fly me to attend SF and then to speak in New York, which I just did last week. Um to talk about how I made the skills, what they do. It was like a six-minute lightning

demo. But um and then people have asked me to go on podcast. Oh, show me this, show me how this works. I've been getting comments even on Twitter like people are like, "Isn't this just like another way to make a bunch of slop?" Yeah, that's fair. That's what like most people think about AI still. But I think that then they see, oh, here's how you vibe code, but vibe market and they're like, "Oh, that's actually very interesting." Um, >> so yeah, it's just captured people's attention. It sounds like it'd be the equivalent of, you know, putting a uh like a synth in the hands of a really capable musician. It's like, sure, it could it could make it could very much make slop, >> but in the hands of somebody who knows

what they're doing and is very creative, it's actually just going to make your life a lot easier. >> Yeah. Yeah. 100%. And that's my suspicion is I think that, you know, it's got over 10,000 stars now on GitHub. um I think like something maybe like 3,000 forks or something which are like people directly copying it into their own uh GitHub. I think maybe a couple of hundred have like used it end to end with like one skill. Probably only a handful have really used all the skills their full capabilities. So, it's still like uh you've got this powerful tool, but no one's really fully utilizing it yet. We'll see if that changes, but >> um >> yeah, it can be a slot machine if you're if you have sloppy taste. Um but if

you're an expert, then it's a really powerful tool. Yeah, it was funny when we were all together because in case anyone doesn't know, obviously we run an agency and so we're like adopting and learning all these AI tools and like adding to so we were all together like setting up Claude and doing some things for uh just the the three of us and Cory was like okay go to skills.sh SH and I was reading through the the list and it was like enthropic anthropic versel Corey Haynes 31 anthropic and I was I was like this is like a comical this is so funny and random. Um when does marketing skills 2.0 come out or is it like are you going to add to it as it goes or how does this work in terms of like the evolution of

these skills? Like is this something that you want to add to? Is it like there's new ones that will come out that you're like, "Well, this is a better version." Or how will that look as it evolves? >> Yeah, we're already on uh version 1.3, so we've had three updates, 1.1, 1.2, 1.3. I've been doing it like every week or twoish, and just trying to listen to feedback. There's people that have because it's open source. You can you can either add an issue which is like hey this thing is missing or broken or you can add a pull request which is I built this and now I want to merge it into the repo. So there's been quite a few people who are like I need this I need this or it'd be cool if it worked this way and did that thing. Um so it's

already evolved like a lot since the beginning where from the beginning there was just the 22 skills. Now I think there's 32. I also added a bunch of tools and integrations where basically cloud would have to write a lot of manual integrations with some of these tools like uh let's just say web flow for example where it would have to figure out okay let's go find web flow's API let's see how it works let's write the scripts to be able to like interact with web flow and do the things that we want it to maybe we want to like you know send a blog post into the CMS well now I just built all the integrations and CLI into the repo so that it didn't have to build it from scratch. It just says oh okay let's use this script for

web flow and that's how we know to you know get the blog posts in the right way. So there's like 52 I think tools and integrations and those can either be um MCPs CLI command line interface which is basically a a little wrapper in your terminal to be able to more easily do things and and it just tells Cloud how to interact with outside services. And yeah, I'm going to be adding some more soon. Probably the next stage what I'm going to be doing is um just like short commands. So how the skills get triggered is you have to kind of trigger them via a keyword. It would be like um you know cloud I'm working on SEO. What should I do to improve? And then it's like SEO do we have a skill for that? Oh

we do. Let's use the SEO audit. But now you can maybe just type slash SEO and then it'll initialize directly calling that skill and then be like, "Oh, let me start an audit for you. Let me suggest some improvements. Let me d I the other thing like I don't know where it's going to go. I have no idea where this is going to land. I don't even know." It was funny because a week after um publishing the repo, Verscell came out with a study in a blog post that I think flew under the radar a little bit, which was that um skills in their current form of being individual markdown files all separate. AI had like a 60% hit rate of correctly calling and invoking the skills at the right time. But if they compressed all

the files into one massive markdown file that was like digitally compressed so that it was a smaller file size and easier to read, it had 100% hit rate. >> So they were like, "Everybody compress your skills." Um, but there's no good way to do that. So there's probably going to be a future where marketing skills is just one giant file versus right now it's like >> a couple hundred files. Yeah. >> Um like altogether if you split them out, you know. >> So rather than having like oh this one's for copyrightiting, this one's for SEO, this one's for X, Y, and Z, those would all be compressed into one. It's just marketing. So anytime you're asking anything related to marketing to Claude, it'll just reference that and then find

that piece of the markdown. Is that what you're saying? >> Right. It'll just go to that section. If you say SEO, it'll go to the marketing skill and then it'll search for the SEO section and then follow that. It'll all be the exact same content. It's just >> Yeah. >> packaged differently. >> How easy is it for for somebody to get started with it? What should they count on there being the the most value within those skills? Like if you're going to pick one thing out of these skills, this is the goat. Maybe like the top three, you know, go over the top three things inside those skills that everybody should utilize or perhaps even better question like what's the order by which you establish them or

does the AI already know how to do that? Like it will walk you through the steps. >> So the skill repo will tell you like the order of operations and it can also give you recommendations. In fact, one of the things I mentioned in New York at my talk was one of the things I found to be the most useful when starting out with a skill is you don't want to have to read through the skills themselves. Like it's not really important for you to know and memorize the information, but if you just ask Claude to do that for you and say, "Hey, how do I use this skill?" It can do all the work of thinking for you and oh, you're supposed to do this first and then you do this and then you do this. Because sometimes you can wrongly

prescribe directions to Claude and you're like, >> "Hey Claude, um, start with the SEO skill and do this, this, and this." And when in reality, you're supposed to start with the product marketing context skill, which is how I have it set up where everything starts with product marketing context, which is your positioning, targeting, audience, uh information about your product, capabilities, things that it does, things that it doesn't do, so that you're not like hallucinating features. Uh that skill sets all that up for you. And then every other skill references the product marketing context skill so that >> it's loaded with the right info about you and your product and what you're working on.

>> And then there's other skills that reference each other. Like there's an SEO audit skill, but then if you wanted to get into AI SEO things, then it actually branches off into a different skill, you know, dedicated to AI SEO. Um, so probably where I would start is product marketing context. That's always like and and if you just use that one skill and then he just told Claude, don't use the other marketing skills, but write copy for my website, it's probably still going to get you better results because >> it's actually knows about your product instead of just trying to make it up or do like a generic web search. um which it can hallucinate all the time because it'll you know maybe it goes to uh some

review blog post from three years ago about your product and now all those things are out of date or absolute right it might not even go to your own website or it might go through your website but it doesn't go through all the pages just looks at the homepage and your homepage only has 30% of what your product can really do right so um product marketing context Um, I do think that the copywriting skill is like pretty good at writing decent copy. The programmatic SEO one is really cool because that's one of the most difficult things to do and to do, right? And I've even had a couple people do it. You guys know Carlos Carlos Harbono, shout out to him. >> Yeah. >> He used it to publish like 4,000 No, no,

no. Like 24,000 pages on his new startup. Um, and like they're all perfect. Like they're just they're just done. And he did it in a couple of days kind of going through it step by step and went through a couple of iterations, but that's one of those very niche things that's hard to execute, right? >> And then the SEO audit one is also really good because especially when you're spinning up a new website in Nex.js, for example, it's going to do things more built for an app, not a website. And so you do the SEO audit and then it's going to clean up and add all the things that you need in order to be up to spec for SEO. >> Very cool. That's especially useful if you have a a product that creates a lot

of like a huge amount of data. You know, like Swipe Well, right, is you know, >> super cool. Nice. Um, I was [clears throat] going to ask because we one of my favorite things is like there's a lot of agencies that are like a lot of agencies I think die because they're unwilling to adapt. Like they have a certain way that they've done things for a really long time and it worked for a set period of time and then as industries shift and as tools shift, they kind of refuse to adapt and die out. That's one of the things I really like about working with you two specifically is because both of you are like like the new thing and want to see if we can get the new thing to work with our business. And so that's like that's

kind of my first the context for my question which is it seems like there's a bunch of marketers out there who are like just scraping the surface of AI and being like look how awesome this is. Look at this snow I just like scraped off the top. It's all powdered and it's it's beautiful. Like they're like, "Oh, I can feed this into ChachiBT and it will tell me if the headings are messed up or I can I can write a blog with with ChachiBT." Like that is the surface that they're scraping and they have no idea that there's like all this valuable ore and stuff just under if they just digged a little bit deeper. And this to me seems like kind of the start of that. like you're we're digging in and we're

being like, "Okay, this is a a way to implement a very serious and deep workflow using AI to do a ton of things and it has the brain sitting right here. How long before the normal marketing people can handle this kind of like like normal market people who have no coding background or even no like development background even using no code tools. how long before they can do this kind of thing without needing like serious handholding. That's like kind of my biggest concern with the way that marketing is shifting with AI. >> I think until the time until it feels normal for everyday marketers is like this is your standard tool set is going to be like a few years probably. um >> they're just I mean we went to a

conference you know just a few weeks ago and I don't think I talked to a single person who had even who had even touched cloud code or really even knew what it was >> really yeah >> yeah like >> but okay but the flip side of that is that I don't think cloud code is like the final tool that we all should be using for marketing um the fact is that like using the terminal is not a very userfriendly experience and uh and it takes a lot of finessing your terminal cloud cloud code your setup integrating with other tools um and so like I think that that's going to be that's going to be the thing that we do really really well and like if you're cutting edge that's going to be the way that you do it but long term we're all

going to find ways to take cloud code and put it into the browser all set up for you. >> That's basically the point of Magister, >> I've never even touched a terminal before how to do this. And I made coding formarketers.com. And then everybody in the Discord is like, can I just use Claude and Cloud Co-work? I'm like, no, [laughter] the point the point is to use something that has all the tools outside of what you can do in the browser. So now we're setting up a way to take all the power of the terminal but put it in the browser. So there's going to be tools like Magister and others. There's even like Manis now um >> that have these built-in connectors and essentially like are literally running

cloud code under the hood. They're using the cloud code SDK which means that they are >> taking cloud code. >> Yeah. Manis is running off of the cloud code SDK even though it's powered by meta. >> Yeah. And so is OpenClaw. >> Um OpenClaw uses the Cloud Code SDK. >> I mean Yeah, but we I mean we all knew that. [laughter] >> Yeah. No, but but that's what Manis did was they basically forked OpenClaw and then they like it's like okay, did they do it? Did they really not? It's kind of a bunch of same difference, right? It doesn't really matter. Um, but yeah, detail-wise, like cloud code right now is the gold standard for agentic AI. And so, everybody is using cloud code under the hood. It's just a matter of the

interface and the connectors and how they're harnessing cloud code under the hood. >> Yeah, we we started I know you showed us you were like, hey, use the we should use this the the meat the blueberry. And I was like, this is a new thing there. There's other versions of this, right? There's, you know, like what is uh I was basically like researching Yeah. cursor and all the other ones. I was like, okay, what are the other options and basically AI was like >> they're all good and this is brand new, so just pick one cuz they all are trying to learn and do new things. So like Claude is the only thing that's kind of like right now it's kind of settled that this is what you should use for this specific purpose. But like even where

you use it is not set in stone. Like there's not a standard toolkit that makes the most sense >> for all the use cases which is like >> dude I cannot wait to figure like there's going to be in 3 months there's going to be a new tool that's like oh clawed inside this is the greatest thing since sliced bread it's like going to be so awesome. So I'm just waiting to see what that'll be. >> Yeah. Yeah. That's that's the difference right now too I think is the user experience. So cloud code in the terminal is not a great user experience, >> but that will not be true probably in 3 to 6 months like you said. Um we're we're already seeing that like with with Magister just a little sneak peek really quick is that we've gotten like our MVP

up and it's basically cloud code in the browser. Now you still have to know how to prompt it, what to do, you have to do some work to grab API keys, things like that, right? Um, now that we've solved the cloud code thing of like agentic AI workflows, now we're going back to the interface, the UX, the user experience um, and making that a friendlier, easier uh, easier to use. Yeah, we Okay, so we had another topic that we were going to talk about. I'm wondering if we save it for another day because we're we've already been talking I think for about 30 minutes on this. Yeah, unless you guys want to hit that. >> Save it. Okay, cool. Any other final thoughts on the Claude marketing skills uh adventure that you went on over the

last three weeks? What was the best food you had in New York? How about that? >> Um I had some really killer pizza and pasta, of course. It's my go-to. No, I think the the let me only other part about this that I think I would love to hear from you guys on is um you know building is no longer reserved for programmers and developers which has been a theme and it's a little feels even a little bit cliche to say because over the last year especially the agent has come a long way and there's things like cloud code that make make that very easy but now you guys are seeing we're not not just even for building apps we're using cloud code for marketing for design, for other things. Like now anybody and everybody can build

and especially like if what you were good at was writing code and now you don't need to write code. Well, what else were you good at? Now you kind of have to reinvent yourself a little bit to be honest. But if you were really good at design and now you don't need to code, well, you're still really, really good at design. In fact, now you're better at design because you can build more faster. This is where Zach owns us. [laughter] Zach runs the whole world cuz he's the only one who has like the true eye for what looks good. and he's >> now wielding a new >> like weapon that is more powerful than anything he's wielded >> and uh his his artistry can like go straight through it and work really really fast. So I I think the world is

changing and Zach is going to uh Zach is going to poop on all of us honestly. [laughter] >> Double you guys can hear me okay. I know my mic was acting kind of weird. Uh, just pull it up. >> I think it switched, but [snorts] >> it did switch to my MacBook, but I think we're okay. >> I can still We can still hear you. It's all good. >> It sounds fine. >> Not coming through a tin can or anything. Okay. >> No. >> Um, no. I think the paradigm the paradigm shifted really really fast. It was like within within a week, there were things that I've been saying for years like it's the AI is good, but I don't have the kind of control that I need to get designs where I want them. And that's always been the curse. It's

like I don't want to play with it until it is going to give me the results that I'm looking for because it's just wasting my time. it became a net like loss for me to take what it could do in Figma and Web Flow and then try and tell Claude to do it for me because we tried like we we gave this shots and like a year ago and the answer was no. Like this is not something we can pass up to clients. This is not something that we can ship internally. And all of a sudden, it was like out of the blue, Cloud Code was good enough to be the the hands and feet behind the design ideas that we have. And we're seeing this right now like with the I'm able to do things like WebGL and create 3D rotating icons from SVGs that I had made in

Figma. And I can because I have web flow experience and I know how to somewhat talk like developer speak and I can give it the proper frameworks for like even just knowing what webgl is. It's a good command for me to say like no I want you to make it like this. um or I need you to adjust these elements within something on a web page or understanding things like the proper structure and giving like putting our workflows to to AI things like this is how we experiment like you know we do a series of three versions then we refine that one version we like and then we jump into a style guide and build off of that. It's cut like there is no design or develop process anymore. It's just making all the things. [laughter] So, and then I

can literally communicate at the speed of thought. Like I'm talking to the machine and saying, "I need you to fix this." It's like me directing an army of very technically skilled designers and developers at the same time to build stuff. And it's amazing. [laughter] Zach on several occasions has said, "I'm having the most fun with work that I've ever had or I've had in several years at least." Which is uh which is fun. >> Well, and the and the cool thing is that the design stuff doesn't go away, right? Like especi the the cloud code to Figma thing, and I thought it was genius. I was like, "This is amazing. I can actually I can have it oneshot something and then I can take it in for granular refinement if I'm not

able to get the point across the same way or there's just some things that are harder to talk about and describe than it would be for me to jump into Figma and adjust it more more specifically. So it's amazing the I don't I feel like design and now and design has always been the idea of crafting with intention. It's the pinnacle because all of it comes down to design. It's not like the technical limitation isn't there really anymore. It's the experience of light. I was listening to a podcast. I listened to the full three hours of Peter Steinberger talking to Lex Friedman about OpenClot. The thing that came up over and over and over again was, "Well, yeah, you just get to focus more of now on the the

delight of the product." And like Peter talking about his German accent [laughter] and that word delight, delight, delight, delight. It came up over and over and over again. And now we I feel like we get to do those things that would have been like that would be really really cool, but man, that is going to take like a day just to get that really bespoke interaction that you're looking for. And now it's done in 15 minutes, even less. Yeah, it is. It's nuts. I I think you kind of hit the nail on the head with the um like the the lines are blurring so much harder between developers and designers and it's going to turn into like you have builders and maybe like strategists but even strategists I feel

like is going to end up being in the same builders category like I don't see a world too far away like if developers can adopt some of the same design visuals and use claw to to build beautiful websit ites and beautiful products. It doesn't seem that crazy that copyriters and marketers could do the same thing. Like it's not that far away. It's just kind of picking up some of the lingo like you were mentioning. But >> uh it's pretty nuts. >> Yeah. >> The the big thing I've been noticing for myself, right, is >> uh like I took Zach's design in Figma for my personal website and then I started building it in X.js with cloud code. And there are certain subjective uh things open to interpretation about

the design. You know, color gradients, animations, um the use of certain accents and elements across the site that I'm not as practiced or familiar in or even have the vocabulary about. Like I found myself in Cloud Code being like, okay, in my head I have this idea of like a shimmer, but then when I say shimmer, what it produces is not what I imagine. What am I imagining that's was another word for shimmer that's not shimmer? You know what I mean? So like the more you can expand your vocabulary, the better you're going to be at whatever you're trying to produce. Um, so I found myself being like, dang, I need to I just need to know more design lingo because I don't know all the I don't know how to describe what's in my

head and tell it to cloud code. >> Yeah, I I don't know. I think there's there's so much experimenting for us to do and a lot of like taking projects and being like, hey, this is how we're doing it now. Switching it up. Um, but yeah, I had the same thing where I was like, "Oh, I should start building my because again, I also have a we have all personal sites that Zach designed that we need to spin up." I was like, "Oh, let me start taking this." And I did kind of the same thing. And I was like, "Okay, I have a gradient. It's a radial gradient. This is this is the color, you know, the hex codes or whatever." And it gave me like instead of like a from the bottom corner up, it was like a center

gradient going out. It looked like a big like rainbow. And I was like, "Dude, this is not what I wanted." and just having to like be more specific and be more like, oh, I should have just given it direct access to the thing I was explaining rather than just pasting it in and being like, this is what you should do. Like, how do we make all those interconnections and all that? >> Yeah, that's it. It is still you get way quicker haptic feedback. This is why the Figma to cloud code thing was fantastic because I can move something where I want it and like get more specific or change an element like that. I had to do this with the buttons on conversion factories site because they were pissing me off. I was like you

just you're not getting me right now [laughter] blueberry and cloud code. So what I did was I took the button in and I wanted a very specific hover state and I said okay I'm going to make this hover state real quick. Nailed it. Two minutes and then send it back to Cloud Code and it says make the transitions between all of these elements really smooth and easy. Nailed it. Good to go. >> Yeah. >> Mhm. >> We could we could go back and forth. This is so like there's so much to talk about. We could just do this on a bunch of stuff for a very long time. So, let's wrap. Uh super cool stuff, Corey. Thanks for sharing all the awesome stuff about the skills and uh really excited to see how Claude and all that stuff is going

to change our our workflow with conversion factory. Uh but yeah, thanks for joining the factory floor. We will see you on the next episode. Go >> and there's one other thing to add to this which is what we're talking about right now like the line between design and development. It's not happening tomorrow. It's not happening next week. It's not happening next month. We are doing this right now. Like the design team at conversion factory is >> yeah pod code. We have the terminal up. We are skipping between like that process like it's no longer bifurcated between the dev team and the design team. The designers are in there making changes and it's freaking awesome. So the next time you see us, we are going

to have websites that have been coded in Nex.js by our design team. That's it. >> Signing off. Thanks for joining the the factory floor. We'll see you next [music] week or next time, who knows? Peace.

What it's about

Corey's open-source marketing skills for Claude Code, why they took off so fast, and what changes once AI moves past prompting into execution.