Growth and Marketing AI Agents Are Here To Stay (Avoid doing this)

Hey there, I'm Parker X. I led technology first startup that sold for 23 million bucks. After that, I tried my hand at building Airbnb for music. Don't recommend doing that. But what I did learn was how to use AI for coding, for marketing as well. How to leverage the absolute crap out of this. Did it every single day for the last three years. So, I know a thing or two about it. And on this channel, I share the journey to 100K in profit a month using AI. So, we go through a couple different things. There's questions and answers. There's this thing that's flying off the screen because I just set this up. We go through questions and answers. And if the best question people ask a good one, this this literally could be you. You could be the holder of the belt. Now, this is one of the best awards you could win in life, I think. But no, it basically I just unlocked badges as well because we hit 500 subs. So, that's really cool. But ask questions, you get answers to them. We rip through that. Then we rip through some news about things that are going on in AI. And my whoops going off. I don't know why I have a 9:45 a.m. We go through some news. We go through strategy of things that I'm going through. And then we go through some of the builds, things that I'm actually building. All that gets posted back into a community. Yeah, a school. You guessed it. You don't have to get it. There's plenty of free stuff on here, but I promise to deliver more value in there than any other school of the genre. That's what I pride myself in. So, let's jump in. Uh, do I want to add this cursor pro thing? Yeah, let's add this thing. Now, we can that Henrik says, "Thanks a lot, Parker." He wants turbo. So, people that are saying turbo, you'll notice a lot of people are saying uh turbo. It's because there's a problem that happens for a lot of newbie coders where they don't know how to set up what's called boilerplate. There should be a better name for that, but it's the setup of code that you just kind of have to have for everything, for every project. And that includes packages, that includes a lot of different things. So, I was running into this problem over and over and over where I spin up a lot of different projects, and I hate spending the time setting up the same thing over and over. I wanted to just have one central repository or place with all of it. And so someone in the YouTube audience reached out on Twitter and was like, "Hey, I want to help on map or on this Vibe thing. How do I help?" And so we put together the Vive Code Turbo Kit and added some docs to it yesterday to kind of better explain how it works. But it is a set of tools. So I'm going to be pushing this up later today. Has all the things that I really like in coding projects and it's free. We are going to continue to work on it and just use it and probably do a paid one eventually that's a little bit better. But for now, this one's free. And as a high level, it has biome, it has bun, it has the right tailwind version that you need for making sure LLMs are friendly with it. Then it has a bunch of different scripts that I use in my projects to move quicker. So things around the database, things around writing the types and it has a couple other things. The AI docs is really important. So all the popular LLM.txts. So we have Enthropic, Cloudflare, 11 Labs, Replexity, Resend, Stripe, the different Superbase ones, Subase, Swift, Tiptap, which is for notes. It's essentially a package that is notions note-taking stuff. Trigger.dev, which is for background jobs. It's an open source TypeScript project. Zapier. And then we have an MCP reference. This just has one in it, but I recommend that you guys use MCPS. They're really nice. And then a prompt library. And finally, uh there's going to be a set of what I think to be the best prompts that will eventually be an agentic workflow for new features. So if you're not familiar, I use code expanders for coding or text expander. So when I do F1, uh it doesn't do it in a codebase, but when I do it in here, F1, it'll pop up with the step one of six for creating new features. So that will be something maybe in the paid one, but I don't know, we're still refining it, so I'll probably put that version in here. And then other things just VS code extension. So making sure that you have the right things such as the Superbase extension, the Postgress tools that gives you type safe. Uh, it's basically an LSP or language server protocol. Language server something. I should know what that is. Gives you autocomplete for your database. And then biome is, if you've ever used linting or formatting, which you should. It's, you know, prettier or eslint. Biome is just a better one. It's more performant and it's really strict. So strictness in coding projects is great when you're using. That's what that is and we'll be talking about that a little bit more. So, we got a bunch of those comments. Where did my comments go? Uh, right channel. Seeing. Okay, here we go. Marlin. Marlin says, "Vibe marketing the new era." Shout out Marlin. Marlin just joined the community this morning, which is really cool. And as a part of everyone that's joining, you're going to get added to a directory so we can know what everyone's working on. If you're doing something that's private that you don't want to share, don't share it. But the point is when you have a group of people around you, you get a higher multiple on your knowledge because they are going to know something that you didn't know. I always think of the story of Mr. Beast aka Jimmy Donaldson and talking about how when he was trying to learn YouTube, he just talked about YouTube, worked on YouTube, and just literally it was his whole life. And he did that with a few of his friends. And that shared knowledge is just everything. So yeah, Vibe marketing is the new era. We're going to be talking about that. That's going to be the name of this video is something around Vibe Marketing because I'm entering into phase two of my content pipeline. Hot style says starter starter kit. I don't know what that means. Uh sir, you you've left something in here. You've you dropped an F here. Maybe learn to type sir. Okay. Uh there's no vibe coding if you're planning to develop a scalable and maintainable long-term project. I disagree. I think that vibe coding it just depends on your vernacular. It depends on your definition of it. I think looking at Andre Carpathy is the guy who coined it. When he talks to an LLM, he's saying stuff that we don't say. That's a big part of it. And I think that you have to have a ton of guardrails around it and you have to have good fundamentals. What I did is I read every one of his threads that he did. The first one that he put out was very technical. It's not like he's just saying, "Hey, make a tic-tac-toe app." You have to get good, kid. And not you specifically. That's just like, you know, understood you. But I I do agree where vibe coding, if you're just yapping and you have no underlying knowledge, you have no idea what you're doing, you couldn't code it without AI, then yeah, totally agree. And I ran into this with a client build where I tried to vibe code it without thinking. And it's a slippery slope. If you don't do the planning ahead of time, if you don't take the time to write out all the specs and everything, then yeah, it's going to be a bad time. So, Kunal Dubbie works says, "What did he use or what he did was use embeddings like cursor does, but local has access through literal vector search. Love the concept." Yeah, this was about talking about these node trees where a dev made made some visual representation of what your files look like in a knowledge graph is another way to put it if you use obsidian but it's all the nodes that are like the big dots that you see on the screen and it creates embeddings about everything so you can do what's called vector search vector search allows you to search things similarly across your codebase in vector space we think in 3D in 2D vector space is like a bazillion D. So you have a lot of different dimensions that allow you to find similarity across things. So when you want to make changes to your functions or your methods, it's a really good way of doing it. That's how rag works. Uh web 3v says make videos shorter and more concise. Bro, how do I say no? Uh no, this is not that channel. If you want add content, then jump over to my other channel, the main channel, which I'm gonna be rolling stuff out this week. But this is a daily upload. So I will partially I mean it is kind of true. I am trying to wrap this down to 30 minutes. So uh so Parker, I'm not Oh gosh. Uh when it comes to coding, I'm not very good. I just know zero about it, but I do see a feature in this, a future in this, you know, and my only way to code is vibe coding. Can I like send you what I'm trying to build privately, maybe on IG or something? Because I just don't want my idea in public. So maybe someone could take it, you know, running into these problems, blah blah blah. Okay. Uh, great comment. What I'd first say is you're not dumb, it's you're new, and you got to realize you're not going to be great at something if you're new to it. That'd be an anomaly. So when you want to learn something, you just want to speed up the iteration because everything that you want is on the other side of those iterations. You just have to get there faster by going faster, by putting in more hours. When I say that I've been coding for three years, I put a 2x multiple on it because I work doubles every day. So, you got to put in the time. And the whole vive coding thing is really it allowed a lot of people to on-ramp into this space, which is great, but then they immediately hit a wall as you are because you don't understand how the stuff works. You're just not even reading the code. You don't know what the code means. You don't know what a hook is. You didn't know that there were classes before hooks. You might not know what a class is. You've never rolled your own tech. You've never rolled your own off. You've never connected a database. You've never made your own database. You've never done a lot of these things. That should be a checklist of just basically here's the things you actually need to learn. I know a lot of people have tried to make content around that. This guy Mark is an indie hacker and he makes a bunch of money selling courses around this, but even then I think there's a lot of stuff you don't need. And Kunal has a good point. And if you're protective about your ideas as if it's the secret sauce instead of opening your arms wide and begging for implementation help you've lost before you even started. Don't stress about your ideas value. You'll be able to find more. Who knows your current idea might not even be tech to learn. Yeah, 100% agree. This is really really good point that I want to drive home. So whenever you're doing a new idea, first of all, the idea is worth nothing. Just remember that execution is everything. But I've learned this the easy way and the hard way where you have three big risks that you're trying to burn down when you're doing a new build. So you have the technical slash product risk. So is this thing technically feasible as he mentioned and that' be one and these are all equal and you need to burn down all three in order to better uh judge if the idea is going to work. You have market risk, which is I built something in the music space. I didn't take the time to burn down that market risk and understand, oh, the three companies, Sony, Universal, and Warner own the entire music industry globally. And in order to make any action in that space, you have to work with them. Example of this is Spotify. Spotify still doesn't make money to this day. They light money on fire because they've had to work with those three players in the space because the market risk is just and the last one is the founder risk. So are the people that are behind this project capable of delivering on it? So when I think of yours I and just think of this if you're like a an investor would you invest in yourself? someone that doesn't know how to code that is just starting and just starting to understand what the stuff that the codegen's doing, you know. So, I would look at this and then start to grade yourself on all of these. If you get checks on all of these, you spent the time, you know that it's technically feasible, you know that the market risk isn't there or there's a smaller one and there's the approach to how you're going to overcome that market risk and then your founders are capable, then sure, go after it. And another framework that I'd use for ideas is just the uh basically expected value is the impact that you can make times the probability that you can do said thing. So when I was doing a lot of SAS ideas, I was going for that uh10 billion outcome, right? And it's because I had put this huge preference on building a SAS and being the next Zuck. And you hear this all the time from people problem was that probability is really really low. So it's you know 0.01 or 1%. Whereas doing something that I'm doing now which is a multi-prong approach where I make content and there's a lot of ways to monetize that. That's a much higher percentage chance that you can get to $100,000 a month. So I'd put that as I'm just confident in myself but I think that's a 90% chance or 0.9. And so you just kind of rank these things. you go through this exercise. Same thing goes with AI services. And that can include just general dev work, but using the AI to get leverage on the work you're doing. I just know that's going to work. That's 100%. Because I have all these things. There's no market risk. There's no technical risk. There's no founder risk because I just do this. I've done it 10 years. And then as far as product building, right, and building automations before even AI existed. And then another one, you know, would be for me is this community And what makes me more excited about this is yeah, I can get to 100K a month off of the community. But there's also second and third order effects that you get out of doing something like this. The moat around community is hard to beat as long as you're a good person, as long as you do good work and put all effort into the open where it's like this is my intent behind it and this plays into the other ones. So, by doing this one, why I'm so interested in it is because it holds me accountable where I have to be on it. I can't not know my stuff because I what am I going to talk about, right? And so, it is a forcing function for me. It's like telling your friends that you're going to go to the gym and you don't go. Well, that's a social accountability thing. That's how I think about it. Uh, other questions? I saw one down here around MCPS. I'd love to see an MCP setup that connects cursor to CLLO 3D, a fashion design software that has an API. If I can vibe fashion design my garments, export the files as DXF, and send directly to manufacturer, that would be insane. That sounds really cool. I haven't used this, but this could be something that we look into. If they have an API, you can do pretty much anything. What could be interesting first though, and this plays into some other stuff that I've I'm going to talk about in a bit, but how you could just take the you could do the garment design outside of this to see what the outputs would make and then that would be something you could do in, you know, 10 minutes in chat.com versus going and building the thing. But what I would do is I'd look at the guy who made the one for Blender and I'd use that as context for how you build yours because you're just m all an MCP is doing is it's a collection of tools that allows you to connect the all. But I'm going to put this in a uh in a queue inside of the school just so I remember it. And I'm going to put it in here. tutorial requests uh from daily video view. I'm going to start putting in the requests that come in from the daily review videos into here as a running log. So, let's put that there. That seems fine. Cool. And I'll just pin that. So, eventually list of them And and I think that's all for Q&A. Let's get into some of the news. Oh, wait. Who's going to get the belt today? Let's see who had the best one. Who gets the belt? Because we're gonna have badges now. So, you're uh your comment will be highlighted. I I really do like this fashion one, but it's up in the air. I think I got to give it to Kunal Doobie Works. This guy, this is this is the belt winner. So congratulations Kunal Dubbie Works. You are the proud owner of the Comet Belt for today. This is a huge deal. Something you're going to want to tell your grandchildren about. So make sure you wear it with pride on this fine Monday, April 7th. Congratulations again and can't wait to see you wear it out. Please work. There we go. All right. And your award ceremony is on tomorrow's upload badge coming soon. Cool. So, let me change this. Kunal works great. What a what a day. What a day. Okay. Doovie Works just just a proud belt holder right there. Okay, next. So, on the new side, this is exciting. This guy Martin Casado, famous VC. Don't know what he did before that, but I think this is a pretty good way to put it. Starting to realize that it isn't about vibe coding a great game. Vibe coding is the game. I agree. It's figuring out the pieces of cluji workflows that will work today that won't work tomorrow because it'll be irrelevant. and then cobbling together the ones that are worth doing and then ignoring the ones that are just a waste of time because you see all the thread boys yapping about on so it comes back to still know the fundamentals, right? And on that point, here's one that I found interesting that I actually think may be worth checking out where this fell has been using MCPs as well as his own sort of meta framework where he adds tools and he adds a MCP that calls said tools and basically it keeps the running list of tasks that you need to do. Then it's in the CLI so you can see all these things. So, it starts with a PRD and then it has all the subtasks built out. And I like that it is in a CLI because while memory bank has been really helpful for features, I think it's still misses on some of them. And Gemini 2.5 should not be running in cursor. I've seen people saying that it's better, but I don't like to leave it up to the IDE. I don't like leaving it up to an additional system prompt. So, because this is open source, I can see what the system prompt is and then I can fork it and change change that. So, I'm going to be checking this one out today on my projects. Let's see. Next up, this is what I was talking about with the fashion designer. I made a bunch of these mocks in the native image gen with chat just by saying the name of the designer who made the cybert truck and saying hey I want to create an image remixing the attached photo to look something like Omid the chief designer Tesla would make don't change the fundamental shape of the design but rather the aesthetic so you could do something similar I think that that would be a good way of emulating different designs that are in the fashion space. So if you really like a certain designer, you could use that as inspiration and then have it update the piece and then you'd use that prompt as potentially a system prompt with a prompt variable where you put in the designer and that could be its own tool in MCP. So you could name it inspired by tool and that tool would go and call whatever it is that you need to attach to in in the tool itself. Again, when I say you're connecting tools, if we had an MTP that was operating Excaliraw, this whiteboarding software I'm using right now, you could have a tool that's called create shape or create rectangle, and then it's going to call this, which I've just selected, which is the rectangle, and you give it parameters such as the height, the width, the position on the screen, the color. Next up, I made a call. You should follow me on Twitter because I say stuff that I think is helpful, but I made a call on January 24th saying that the stock market's going to flip out next week over the Deepseek news and it was cool seeing the secretary uh Treasury yap about this. Mr. Secretary, thank you very much uh for joining us. Announcement of Deep Seek, right? So the so-called MAG 7, the tech stocks had been doing very well for about 18 months, led the market. Um, and I think that there's kind of a real dose of reality in what was going on in AI and I think US is going to remain the leader in AI. Yeah. So it's interesting. He basically says, look, as soon as this deepseek news came out, the market had a big old taste of medicine. So I yapped about that on Twitter and it was fun to see because as a developer you get to see around corners in a lot of areas of life. Sounds like a heavy sentence but it's just true in the stock market because the top stocks in the S&P 500 I think the top 10 of them make up 80% of the market capitalization of all the rest the 490. So there's a heavy skew towards technology, which means if you're deep in tech and you're in it and you're seeing what's going on and you follow Deep Seek 18 months prior to them launching this and then you see the white paper and you see that they claim it cost $6 million to build this. That's an advantage because Wall Street isn't technical. They're going to wake up, you know, six days later and be like, whoa, what's this thing? And that also plays into understanding the fundamentals of businesses because a lot of Wall Street won't get it. and you're paying some guy named Jackson who's 24 who's like zinned up on a bunch of aderall trying to figure this stuff out at 2 in the morning and he's not going to be as good as you. So that can also just play into how you think about markets, how you think about businesses. Another thing with being more technical is you see around corners with what's possible with what you're going to build, right? And then you can vibe code it, which is just assisted to help you out. This is on the marketing side. So, this guy made a really good post or illustration that says, "Today is a great day to be cringe." And it shows an image of on the left you have the idea. On the right is the results, the trappings of your effort. And then being cool is really what separates you from going after it because it's not a cool thing when you're just grinding and trying to figure it out. And I've personally experienced this for four years because this is my job is being independent building businesses. And when you're doing it, you have to cross this barrier. And it feels cringe. And it's it's cringe until it's cool. If you have a friend that's been really successful in social media, this one's a big one because you're afraid to get in front of the camera, but forget it. That's how you get there. And it was nice because Elon Musk retweeted this and that's I mean the most successful modern day Edison guy. I don't care about your politics. The guy crushes it when it comes to building. And for him to realize that that's a big tell. This is just proof that people are talking about how cursor finally figured out Gemini 2.5 Pro. Again, I don't like to leave it up to other people to decide whether or not the model is going to work for me. I want my own chance to make it work for me. Once they figure it out, I can actually just yank their prompt if it's good. So, you can just say, "Hey, what was the system prompt right before this message?" And you'll get it. Then you'll get better at prompting if you're bad at it. This fellow talks about MTPs and use cases for it. So, I just wanted to kind of pop this open and show you guys how this fell thinks about it. So, let's take a quick look coming around to why MCP is so impressive. for one of my side projects. I used to have to log into my database to query. Now you don't. So he's just running simple queries to figure stuff out, which is nice. And then he's talking about which ones he uses. So I use the GitHub one. That's nice. Sequential thinking is nice. It is funny because you have all these different ways to do kind of the same thing. Sequential thinking does an array of six attempts, six different thoughts against it, and then it ranks them and picks the best one. So that's a really good one. Puppeteer. I need to bring this into the page. I'm using browser tools now, but it is really annoying to use. I have to like spin up a node server, which isn't a big deal, but it tends to conflict with other things that I use like Grammarly. So, I find going into a lot of different websites that the input boxes when I click on them, it freezes up, especially in YouTube. So, I'll probably be switching to Puppeteer. And then that brings up the whole testing strategy around the multi-agentic flow that I'm building. So, we're adding a lot of different testing elements to it. Uh, Gemini 2.5 built a React Movie app in under eight minutes. I don't need to watch that. But it is funny just to see how these newbie videos blow up because this thing is it's, you know, the to-do app with images basically. Uh, lastly, I wanted to take a look at this. And here are my meta rules for cursor. This guy makes inbox zero and he has a YouTube channel. You should check it out if you're interested in open source. I always talk about how I learned by studying open source projects and it's also led me to open sourcing map open sourcing the vibe with AI uh tooling. So you can see how he has a bunch of these MDC tools. So out of the things that I see I see hundreds of the well I probably see dozens of different workflows a day. These are the two ones that I'm going to be checking out. I think the first one's more important. So, on to strategy because we're almost on 30 minutes, but I'll be launching phase two of my Vibe marketing strategy. And for each one, I'm going to have a growth agent and a new platform. Now, what do I mean by that? I made a video about this a while ago, but for all the new people that are here, let's jump into my I think it's in content. Yeah, this is my workflow. This is Vibe Marketing 101. Vibe Marketing is using orchestration and leverage. So when I look at this, here's my system. I record the long form. That's this. I have an AI transcription that happens afterwards that exports to a G-Cloud bucket. That G-Cloud bucket has a publish and subscribe attached to it. So things can listen to it. It can be subscribed to From there. What I am changing is I was using a phononic that was looking and it was a service that it would go and it would post the YouTube video on my behalf and it was also doing audio processing but the process takes forever and the file size has to be less than 25 gigs. I'm recording in 4K so that's a no for me. I'll be swapping this out to then get the uh pipeline edition for the subtitles as well as the summaries. So you get a summary that is short, a summary that is long, and the milestone chapters. So the things that you see along here. So I'll be adding something in there. That output, as I said, goes to YouTube as well as a process. I'll have something. I don't really love ClickUp, but it's just a DB with some statuses in it. But it would allow you to then know, okay, cool. We have all this stuff in here, and now we can go on to the next step, which is making shorts. So that's kind of where it's going is cool. You have the long form stuff and you reuse the long form to then create all the different shorts across all the different platforms, but none of it goes anywhere until I review it and then I can update the status myself. So I want to be looking at all these things. I don't want to hire anyone. I want to make sure that I dial it in, which means you get closer to the prompt. It might mean rolling my own video chopping service. I've seen people do this. I know how to do it, but it allows you to type and say, "This is the type of video." And then upload examples of the things that I like. Modifying the fonts for the fonts that I like so it doesn't look like your cheapo Walmart content that you see everywhere from a lot of these AI influencers. And then we'll go into text. So text will be across long form, medium form, and short form with option for carousel. So you need images as well. That'll be using foul.ai AI for image gen doing reverse prompting off of the content off of any sort of media. If there's articles that I talk about in this video, then we can go and do a scrape. We can do a fetch or a fire call to figure out what's going on in that article. See if there is an image in there and then use our prompt on top of it. Then by the end of it have every platform touched my website updated all just happening. But if I went and I just try to automate all of this at once, you'd end up with AI slop. That's what everyone's trying to do. That's why I hate on a lot of the content that's coming out because they say, "Oh my gosh, this is the best thing ever, and look at me. I made these automations." But you're generating slop at scale, content no one wants to watch. You're tarnishing your brand at the same time as wasting people's time. So, what I'm doing in this system is step two, which is I've done these ones manually for 28 days or 30 days starting tomorrow, and I'm getting the hang of it. Next, I'm going to add in three videos a week on the main channel. So, then I'll have 10 pieces of content going out every week. There's an additional three that are inside of the community. And all that long form is building up a library not only for people to watch, for me to be able to connect with all of you, but for training purposes so that I can make a really really good pipeline for myself. So, I'm excited about that. I've been talking about it. I'm happy I waited to to make sure that, you know, do the thing manually a bunch of times before you automate it. That's kind of the takeway here. Let's get back to the rest of this stuff. See the And then oh, sorry. And then on the growth agent, when I do add a new phase, so I'm adding the three videos a week. The growth agent for this will be likely the one that I talked about where it's going and it's looking at YouTube videos that don't or channels that don't have channel or uh video milestones on them. They don't have the markers. So, I always think of Joe Rogan and Theo Vaughn. Those videos get millions of views and they don't add markers. So, what I can do is I can set up a listener from my personal YouTube channel that goes yanks the video when we see a new one, transcribes it, runs the same pipeline, and then comes back and comments with it. And then it just has the milestone or the markers. And then at the bottom, it says, "If you want to learn how to make stuff like this, go check me out on my channel." Next, we need to pick up a domain uh or pick one. So, I own all of these and it'll be the thing that hosts everything to do with five with AI. So, it'll host the templates like the Turbo Starter Kit. It'll host a member directory. It'll host the free resources. It'll host a landing page or a squeeze page for people that are interested because I don't love the way that, you know, I just like custom customization. I love school in a lot of ways for what they provide, but I just want to have more stuff. So, this will include an SEO pipeline to own the keywords. I talk about how I'm trying to own Vibe Coding, own Vibe Marketing, own a lot of these things, these long tail tail newer keywords. So, if you guys have an idea of which one I should pick, drop the name of it below. That would be really helpful. I'm thinking it's going to be I vibewithai.com. Shout out Basil for picking that one. But I have build your AI skills. I have Fix My Vibe Code app. I have the Vibe Code Fix. I have Get AI vibes. And I have other ones. So I'll be picking one of those and then putting up the template and the directory later, probably today, and maybe just use Cloudflare pages just to get it out. Next is on the Discord side. So Discord live chat. School doesn't have it, but we have the ability to extend it. So, what I'm doing is I'm just adding new channels for each project. There'll be a web hook that fires from the school members so I can know when new people join and I can go actually talk to them. That's a big part of it. So, I make personalized videos. I do a little research on them to figure out what they do. Try to understand where they're trying to go, where they're at, so I can meet them where they're at, and then I can help them. And again, that context helps me when someone else joins and they're building something similar. We just had someone join that's doing QuickBooks integrations for clients. And I have talked to another member, Aiden. He's like, "Yo, I'm doing these." So, I know if I connect those two, they'll save time, make more money. And then same thing with the Vibe Dev. So, Vibe Dev will be listening for changes on the codebase. So, it's when things change, it can listen to it and then it can update it. I'll be adding in either Code Rabbit or GitHub Copilot as the PR review. So, that'll be nice and we can have every sort of update around it in here. We'll probably have a road map channel as well. Same thing for map. Map's going to be open source, so all the updates will be in there. And yeah, I'm excited about that. See, on the builds, I mentioned this before, but I'm going to be changing the website so that there's information about the community, the directory of members, how long they've been in it, what they're working on, so people can know and kind of understand what they're joining, so they can see, are there people like that in there? Now we have product managers, developers, a lot of AI services people. There's like seven of us right now, but now is the time to kind of set the plan for how we're going to operate and how we get the most value out of it for the long term. I said deploy the latest version. Yes. Code rabbit. Yes. This is the co-pilot version of it. Then I need to deploy the client build later today. This was the 5K contract. I ended up pricing 5K for their Web Flow site and then 15 for custom. I'm going to end up doing the 15 one because Web Flow is just a piece of river trash. So, I'm going to deploy that today and then later this week I am going to build something for the school that I want, which is iPhones have proximity features built into their U1 chip. And when you put two iPhones together, they start rumbling. It's kind of crazy. It's like, whoa, is there an API for this? Of course there is. So when I was looking into it, I was just curious what the use cases are that are native and then how to tap into it. So you can import nearby interaction. You use nearby interaction session to start the session and then you can pair another device using the discovery token. So it uses peer-to-peer networking over Wi-Fi and Bluetooth. You can send files, you can do multiplayer games, you can do whiteboard collaboration apps. But yeah, I'm curious what a good use case was for this. So I said, "Think of 20 high frequency actions that people take during their day that we can use this API to help with." And I'm going to just make a Swiss Army knife of stuff for this. So I was originally writing map in Swift for the companion app and I know how to write Swift. So I need to go do that and it was before co-pilots were a thing. So I literally raw dogging it. But now I want to make something because I know a lot of people want to make native iOS apps. It's definitely better. it feels better. Even if you are an expo boy, I don't care. Feels better if it's native. So went through that and then it gave me a bunch of interesting things that I wanted to just try out and then I'll post that back. Try to do it in 90 minutes. I like to time time cap things because it pushes me to go faster. So I think that is all. If you're interested in any of these things and you have questions, make sure you drop comments. If you want to learn more about what we're doing in Vibe with AI, you can go to the link description. If not, I hope you guys have an awesome day. Make sure you subscribe to the channel so you can stay up to date. You can stay ahead on what's going on in the world of AI. And I'll see you tomorrow.