This AI Agent Trend Will Die Hard, Avoid It (do this instead)

The amount of grift in the AI agent space is ridiculous. Had to dress up. You know, this is a very serious video, so I had to dress up very serious. No. Uh, hey, I'm Parker X and I led tech for a startup that sold for 23 million bucks. After that, I tried my hand at building Airbnb for music. Don't do that. You need a big market. You need to identify what works, what doesn't in a market. But yeah, on this channel I make daily uploads where I cover the stuff that I'm doing, what I'm learning as I scale an AI services code to 100K a month in profit. We use as much AI leverage as possible without generating slop. That's huge. So we quickly go through some questions that people ask and I go through those and then I go through a couple news things just real fast with predictions on them. So it's evergreen. It's not just like a sloppy AI newsletter and then we go through the strategy that I'm executing on and some of the builds. So, let's rip it. It's so funny seeing me in a suit, but you know, this is high quality. I'm like Lex Freriedman, but I look different. So, first questions and answers. Uh, is there any way to get in contact with you sharing some ideas about YT strategy? my dog Marlin. Dude, how you going to strategize if you can't figure out how to get someone's email? Come on, bro. Uh, there's three levels of nerdom that you can go through here. I can click on your name. This would be like the Vzero noobs version. Uh, I can see if there's an email in there. I think there's not one for you. So, you can do that. I'm pretty sure if you wanted to go a little further, you could just type in my name, which is it's hard to find. It's the name of my channel. and then you'd click there and then you'd find a bunch of stuff in there. Um, I guess another way if you wanted to go and get like my passport number, you could run awesome otint oint OS int which is open source intelligence and you could run one of these and you might find my passport number probably not. But those nerds out there if you wanted to learn it's kind of crazy what you can do with these tools. So yeah, that's that's how I'd do it. I would just start basic. But if you don't feel like doing any of that, it's me at parkerx.com. Cool. Yo, I love your content. I'm trying to build something. Do you have any tips for getting people to invest in your ideas and when the product is finished, how to launch it successfully? Yeah, there's a lot in there. Uh, I did the first company. I was like a 20-year-old, 20some, maybe 19, and I joined my almost brother-in-law who were no longer at that level. my sister uh and him broke broke apart. But when I joined him, we did it bootstrapped. We didn't raise money. When I did it the second time with Airbnb for music, I raised 150K. And what I would say is make something that you want. I'm not going to just parrot what Y Cominator says, which is like make what people want. That's like obvious, right? But there's steps to that. What I would do first is, it's going to sound silly, but as an exercise, go rip the Y Combinator interview questions, like the form, the application. And then as an exercise, go make like a intro deck, which is like eight slides. And what that'll do is it forces a lot of things. This is trash, by the way. This was made in June of last year when I was still trying to raise. But what it does, it makes you think through all the like reasons why this thing needs to exist. Um, and then in the slides, you actually have screenshots of the thing, right? So that would require you to actually make it. So yeah, I'd start with a problem that I personally have and then I would go solve it because it will make you not have to talk to customers as much when you're making the first version of it. And then once you've scratched your own itch and figure out how to market it, talk about it on the internet. I think building public's this funny thing where like a lot of people are like, "Oh, building public's for losers, indie hackers." Guess what? Y Combinator is building public. Look at what Sam Albin saying. He's like, "I'm back on my YC game building." So yeah, having like a little deck, knowing why you're doing it. Um, at least if you're trying to raise money, that's going to be specific to that because then you'll figure out all this information of like how much am I going to charge them? How's that going to work? How I'm going to distribute this thing? This is what I'm doing right now, right? It's funny. And then like why are you the guy to do it? So, there's a standard one to making these. I would use a teaser deck is seven slides, intro decks 10, proper decks like 15, but then you have an appendex with a bunch of stuff. And then when it comes to launching it, well, talk about it on the internet. Get people interested in it while you're building it. That's the best way. That's what I'm doing now. And loving your content. It's criminal how low your views are while a dozen video creators can make videos on same N templates and get hundreds of thousands of views. Hey, Medwis Mariner. Great name. Yeah, most NANM is garbage. I gotta I gotta get on my my my high horse here. This same thing happened in drop shipping. That was a huge trend. And what happens is a space gets hot. It gets trendy. People trend hop. They jump into it. They're good at marketing. They're good at making content. They don't actually do the thing that they're talking about. 99% of the creators on YouTube right now making make and n stuff, they haven't made a dime in the real world of using these things. They couldn't have because you just can tell like the video intro opens with them taking a voice of some stupid movie character and being like I can control my uh calendar and it'll like ring your doorbell and uh it's like kind of like woof if you're familiar with the office where you get a fax, an email, a text, all this stuff and you're just like dude this is the stupidest thing ever. But I feel bad for most people because they're just like, "I'll buy your slop." It's crazy to me. I should be nicer, but it just I don't know. It it there's a lot of um edutainment that actually isn't edu. It's just entertainment. So yeah, that's why I made this flywheel where it's like everything that I must do must play into the next thing. So the videos that I'm making now are things that I already am forced to do. Because I make the video, I have to know what's going on. Because I make the video, I have to do the research. And because I have the school, I have to produce really valuable things. And then I use said valuable things to do client work to advance my own SAS. So I'm like marketing and building public. Hard to keep up daily. Still need to focus on this. Insane value. Thanks, bro. You rock. Can we have a video of your story? how you build your startup. Uh what's your life story that might be a little helpful to us. Yeah, I'll give you the uh two-minute version of it. So then I'll make a proper video just explaining it. So it was my sister's boyfriend startup. His name is Jason. Company was called Delivery Dudes. I always say at the beginning of the videos, I led tech at a startup that sold for $23 million. That's because I had gone to school. I had gone to a college, right? And I made it into about the first year and every time that I go for a semester I'd come back and I'd see him like delivering food for restaurants. I was like, "Wow, this is cool." And I was throwing parties and I was charging people to come to the parties. I was making money that way. Then I decided, well, I can recruit these people to come and work at this company. They need money. Let's launch lunch delivery. So I was the first lunch delivery guy. And then you figure out, oh, we probably need like technology beyond a whiteboard to figure out who's going to pick up the orders. Okay, let's build some tech. How do you build tech? Parker, can you help build tech? There's a guy named Austin who built the first version of it. He was awesome. I learned a bunch from him of how to just get stuff out in the world. And then it kind of evolves like after you're making 100k a month, then you have a different system, then 200, then 250, then 500, then a million. You know, that's just kind of how it works. At our peak, I think it was 73 million. I really want to get like a screenshot of that so I can verify it. But that's what we'd say in the meetings and P&L. You got I guess you just kind of have to trust that. So I do trust it, I guess. But yeah, however many times 12 goes into 73 was what we peaked at a month. So yeah, doing six million a month in gross food sales, you have different stuff than you do in the early days. So it just, you know, the faster you can learn the real skill, the best skill you can learn how to do in anything is learning how to learn really fast. Get to competency fast. Leverage AI to do that. Yeah. And then I'll make a proper video kind of explaining everything because done a lot of stuff. I've had a YouTube channel that got kicked off by the Department of Homeland Security because I was teaching people how to make pyrochnics. It was called Park of the Pyro when I was 13. I got addicted to Ableton Live and I produced a bunch of music. I started scripting and running my own servers like dedicated servers for Counterstrike. Had a community there. After that produced music, made over 40 tracks. Spun in a club in Miami. realized I didn't want to do that because my career would be living in clubs. After that, got into design, interned under the guy who started Liquid Death, which is a water company. That showed me that it's incredible what you can do with the power of ideas. I was like, whoa, this guy just turned water into a billion dollar company. I could do anything. Dropped out of school, joined my sister's boyfriend named Jason, did delivery dudes, started with gorilla marketing, literally flying stuff, and then designing the flyers and then designing the car wraps. trained under a guy who's a total gangster creative then realized I was into tech. Started designing apps, started managing those apps and then by the end of it was doing like just front-end styly stuff. Um but then also like leading all the engineering and then like you get a really cursory really like broad understanding of how it all works. Cool. So that actually was most of it. Now, some of the other news. Let's see. Uh, I guess I missed all the stuff after delivery, dudes, but I need to get on to the next and I'll make a proper video just for you. So, thank you for the comment and please leave more of them because it's fun. They launched Clotted for Education. Okay, what does that even mean? Let's see what Mr. Thread Boy has to say about it because we don't really care about his take, care about what it is. Okay. So, what makes it different than normal project? Okay, cool. So, yeah, they basically just took a prompt and then built one button around it. Talk about innovation. I'll dig into it, but like when I look at that, I just think cool. You did what uh Notebook LM did a year ago. Now, let's talk about it. The name of this video is Beware of the Amount of Shyers. James, my boy. Like, there's something going on in the space where everyone's getting fast attention by saying, "Here's how I'm making all this money using this thing that takes no effort at all. I just flicked my finger and it printed money." No, dude. If you were doing that, you wouldn't share. Like, that's why I was so hesitant about even making a paid offer for the AI stuff within the platform called School. I was so hesitant about I was like, "How do I stand out?" Well, it's because I'm actually making money with client work and I'm also launching a SAS. And my idea instead of raising venture capital was to productize that, put 70 pieces of content that will be evergreen frameworks and agree to myself that I'm going to make content in there around the builds that I'm doing. So, it's like I'm already doing the thing and putting it out and selling it as a skew. And then I can monetize that to then force myself to be better at making content, being at the edge, become the AI guy. Well, that's kind of cringe. But yeah, I feel like the strategy for a lot of people is like new trend, go print, and they haven't made real stuff before. They haven't like worked in technology. Um, so if you like go look at most of them, like I jumped in this guy's telegram. That's how far I was down the funnel. and he's like, "I'm gonna like teach you how to blow up." And then his first thing is like, "I'm blowing up on TikTok." And you check his thing, he's got like five followers. You're like, "Okay." And he's talking about how he's making 500 posts a day. You're just you're making stuff that nobody wants to hear or see at scale. What's the point? And then he talks about his conversion rates. 01%. That's awful, dude. make like a quarter as many posts better. H uh, okay, other news. There's a video editing thing that I'm gonna reverse engineer using my reverse engineering prompt. It lets you, this lets you get uh, closer to the metal. Now, that's just a cool phrase for saying you want to be able to own the prompt. A lot of the stuff in the video gen space or video chop space goes something like this. You have your long form LF and then it goes into opus clips or descript and it's for like the lady who cooks in her kitchen who's non-technical and wants shorts because she heard shorts are good and that's good. That's a good cohort. It's going to service a lot of people. But I'm not a grandma cooking and I don't want to have cringe content. They make cringe content at scale. They are good at identifying stuff in the video that will likely be good. They transcribe it. They probably prompt it to be like, "What's going to work? What's piffy? I don't know." But it ends up like looking like an AI made it. And that is instakkill on brand. Like for me, the difference is how do you make it feel dope? Like how do you make it feel like I made it? That's it. That's why I'm not like releasing any of the uh like workflows I have out into the wild till they're really good. Uh I'll be adding like three videos a week on the main next week, but that's 10 across a week. That'll probably be painful for a week and then I'm going to go into this, which is the OSS prompt. Uh basically, you can prompt and own the part of the video editing before it goes to become short. So, it's actually not open source. So, I'm going to use my reverse engineering setup to go and figure out how to make it. So, I'm excited about that. Uh, this I don't even want to talk about it. It doesn't look interesting. Okay. So, on the reverse engineering thing, just so you guys know, this isn't a school. Sorry. I I I'm actually gatekeeping this because I don't even know if I'm gonna keep it in school because I don't know but this just I worked really hard on this so I don't want to give it away but it's a reverse engineering thing. Basically like step one is you do research and again all you're doing with aentic flows is replace manual steps and check outputs. You do that in a loop. This part you check the outputs until it's good and then you automate it. I always refer to Elon's five-steper, which is not on my background, but he basically says you want to remove parts that you don't need and you don't automate until the end. There's probably a lot of dumb requirements and then speed things up and then automate. So, in this case, I'd do research. I'd grab Wapalyzer. I'd look at any of the stuff I can find on the open internet about the thing that I'm trying to build. Then, I run the reverse engineering prompt. that has like a stack in it that I'm going to use. And then you run F1, F2, F3, which is a product manager, which was me in the past, which is so funny. It's like this is this is literally my whole job. And you take the output of the reverse engineering things, feed it into the technical product manager. You go from technical product manager into expert solutions architect which is like the guy you used to pay a quarter mill to come up with good ideas and work with your technical PM. And so each one ladders off the next within ton of context and research and then PRD goes to step-by-step atomic road map and then that would go into the implementation. Right? So that's going to be doing that and it should go quickly and then I'll use that in the stack. Next, uh, I had a really good call with my buddy Dane. He's the one that we did the Airbnb MU for music thing. And by the way, the tech was sick. Like literally every single thing in Airbnb, like even to date with their stack, like we built, which is crazy. And when I say we, I was just PMing and designing. He was actually building it. So kudos to him. It was built in Flutter. He's like the Flutter G. You can hate on it if you want, but I don't care. We did it. and uh he just was talking all about like his test flow and so I wanted to bring that up because that's what will be at the tail end of this flow or it's like if I'm not reverse engineering things then I just go straight to F1, F2, F3 and you can screenshot and grab the tops of those but I'm gatekeeping. Sorry. And then I'll add testing to it. And I'm not sure I'm going to test I'm going to experiment with adding testing within F2 and then also afterwards and then also in the CI/CD because it's like when you crack that code that is mega alpha and that's not something I'll even share in the school honestly. Like it's it that's the thing is like so many people they like proclaim that they've found this amazing alpha but like they wouldn't share it if it was really printing that much money because that is the advantage. So I share a lot of the stuff that when I find something that's really going to crush and this is advice to you like that is your alpha. So uh other things this is strategy. Okay, so I'm going to stand out this agent's Python point uh proof of concept today. So, shout out to Hari, one of the viewers of the channel. He was like, "Yo, this map thing sounds cool." And actually, another guy from like California that's like building an iPhone app that like does something with a calendar. He reached out too. So, it's funny like people are interested in map, but the whole idea is like I want to make it feel like a video game. What makes video games addicting is the psychology and the hooks that they use. So, how do you make the game of life feel like a video game? I've tried it every different way. It's like I put streaks on the board when I was learning to code. I'd have little blocks that I'd get similar to like GitHub contribution. That's okay. There's psychology around like urgency there, but like how do you take the XP bar, the life bar, the hearts, the flashes, the pulsations, like all these UX elements that drive behavior in a video game to get them to go do the quest, to get them to get the gold. That's what I'm thinking about with map and then driven by multi-aggentic platform yada yada jing jang like there's a bunch of cool stuff but like no customer on the planet cares about that. I just want to know for any business will this make me money, save me money or save me time. Um there's some like things that just always sell like sex always sells. Am I going to look good? Am I going to get laid? Am I going to make more money? Like if you have any of those then you're winning. Um, so anyways, I'm going to be building like really really basic PC just to get it hooked up and then some of the stuff I'm playing around with just like the video game ideas is like okay at the end of the day it is kind of just like a calendar as a primitive underneath the hood like how do you make it so that you can tell like if you didn't speak the English language if you if you weren't blind but like if you were deaf if you didn't speak English like how could you determine what time of day it is and what's coming up and like this kind of idea is where I'm going where it's like it doesn't look the way I want it to look. But you could basically have this runner that's moving along and then as events come up throughout the day that you did have scheduled then you can see those. This is for the person who hates calendars. I hate calendars. I've used them forever but like and I know the value of them. But I want to make something that like feels again like a video game. And this is simulated at a 30 second rate but it would be a uh you know full day. And then this should just be part of it. And then we'll have all the other primitives that people are used to using in here. So yeah, that brings up this topic of data compression. Uh I can't find the article. The idea is that I don't think chat is the final form factor. I think it's nice, but we've had the same exact thing in like a lot of different cases where it's like right now basically chat bots, it's awesome. are smarter, but this has been around since 2015. So, I think what's going to end up happening is instead of it being a chatbot that can do tool calls within the thing, it's recognizing what you would likely do next and then offering you the ability to just like nod your head or be like, "Yeah." And then it like does the thing. So, as it learns you, as it gets more context based on your behavior in the app, it's taking product analytics, so like what things does the person click? What are their workflows? It's tracking those, learning those, training against them, and then suggesting the next thing you should do, and then eventually just doing it. So, that's really what it is, but it requires all the tools under the hood. So, like the calendar will have like schedule assistant and like all these little point solution-y AI things, but like no one cares about what's underneath it. It's just like, does the thing solve my problem or not? Uh, the golden test. This is about tests. There's us this morning. I dressed up for you guys. So, Oh, yeah. And he's building like this cool app that goes in golf carts, tells you where you're at, it's built, and uh he built the whole thing. So, that's cool. Very, very cool. I spoke about this a little bit, but I I again like I've just been like staring at these things. I'm like, how do you like get people addicted to the game of life? So when I look at this, there's levels, there's quests, there's like enemies that you have to take down. So I'm almost like framing like is a project an enemy you have to take it down and then how would XP play into it? What would be interesting and I just thought of this so it's incredibly underdeveloped in terms of thought but if you could give some sort of incentive on the cost of the product for getting XP it had to be done right where it's like you balance the profitability and opportunity of the business with the success of the user. I'm just like that's the business cap side. It's like, oh, what's your LTV to CAC? You want a really good ratio there of like you're printing off each customer and it doesn't cost a lot. CAC gets figured out when your shit's good. So, it's like just make the thing awesome. So, that' be figured out, but it will take a long time to like figure out how to make it good. But in the age of AI, like abundance is obviously going to be a thing. Software going to be commoditized. So, you can't be like a cookie cutter Shad CN app, right? Like, you got to be different. You got to be cool. So, I'm gonna be building like World of Warcraft for your goals. It won't look like this obviously, but maybe even something like to the idea of like a character. It's like I always go to this idea of like you're here like if you're at if you ever been to like a shopping mall like that's like a funny idea but like you're going to somewhere. So there's like you are here you want to get here and on like a shopping mall map you'd see it and it's like the list of all the like stores or whatever. Then you have some sort of map that like explains how to get places. So I don't know. I'm trying to just think, you know, of something like that exists in the world that is a heruristic that people are already used to. Like the use of heruristics is really important. I'm trying to draw a pin. It's like you're here and you need to get here and then here's the steps. Um like a heruristic in in ecom would be like you wouldn't put the shopping cart at the bottom left. Like everyone under the sun knows that it goes at the top right because that's what Amazon does. It's with Shopify, like everyone just knows that. Um, same thing with email. Like there's heruristics around email. So, what heristics exist in the app that you're building, the thing you're trying to build that you could take from other industries because that's what makes it interesting. That's why Liquid Death worked. They didn't copy any of the heristics, any of the design, any of the things that were going on in water. They looked at like when we named the product I that was like part of a session there and like Liquid Death and Noble were the top two names and it came out of a session of looking at proathletes and being like why did Red Bull win? Why are they winning? It's like oh because they don't sell an energy drink. They sell an emotion that you're going to be able to take anything on. you're going to be pushing the edge of what's possible in human performance. So, in this case, it's really like how do you give them a feeling of accomplishment on their day-to-day stuff? Like I really I'd be looking at this and I'd be like, "Okay, like maybe as an exercise, if I have the things of life that you need to do this, it's like what things does everybody do every day?" It's like trying to move forward, right? Like there's a couple daily drivers and this is how I scaffolded the first version of the app, which is like you have your goals You have your tasks that are the daily things you do to move towards those goals and there's patterns around that. Like are these actual next actions that are atomic steps or does it say like get jacked? No, you want to say oh I need to go to the gym three days a week. I need to do this program D. You have your goals. You have your tasks. You probably have notes. Probably have some sort of health information. So health dash So these are things that people all over the world use at least like three of them, right? And so when I look at those, it's like, okay, well that's in IRL. What's in video game world? Well, you have like XP, you have uh quests, you have skills, you have your level, which is XP, you have a heads up display uh of a map of like where you're navigating. You have NPCs. You have what else? Inventory. I I dumped a lot of hours into Guild Wars one, not WoW, but I'm using WoW because it was more addictive. And then you have like this general like kind of look like looks kind of interesting to the art. Art's a big one. That's more on the design side than the primitives of the app. Then you have like this open world like open world chat which is interesting because you can see other people can talk to them. What else do we see in here? Enemies. So we're running up on 30 minutes which is how long I I take to do these videos now. But this is kind of where I'm going with this where I've built this before. I had all these apps. I've built them, unbuilt them, rebuilt them, built them sideways, upside down, whatever. Even if I put all these agents under there to like help you, like I'm not going to out cursor, the guy who makes cursor notes, right? Even if he bundled it, that's still commoditized. But if you have this brand that's built around this sort of thing, I mean, I just go back to the liquid death. It's like water is the biggest commodity on the world, but they can't get copied because they were so different. And so, yeah, I think I think if I looked at this, it's like, well, which ones match? It's like, okay, if I did like, well, goals tied to maybe your level and maybe your quests and then I guess tasks kind of ladder up to quests as well and XP and then maybe your notes. The calendar and the map are pretty relevant. So, it's like trying to figure out how do you make calendar kind of look like and feel like a map. Like what I had had before was that thing I was just showing where it's like if you were on the calendar, you know, you kind of have this motion that's going on and then as things pop up that you have to do that are scheduled, they're already there. But then as things that you do in real time, like you do a workout, like it just kind of shows up and then it looks different and it has all the contextual information. Before the workout, it's going to look different. You need to know what the heck you're doing. That'd be your workout assistant. It's like, "Yo, time to get jacked. Here's what you're going to do." Um, then after it's like, "Yo, you got jacked, but it wasn't, you know, your heart rate wasn't as good as the last one." Stuff like that. Um, but yeah, this is kind of what I'm thinking. Okay, then on the Vibe TV side, I built the first version of this. So, the thing that I do at the top of this show is I do a little bit of a news section, but what I am doing now is like, okay, I add something to a Twitter bookmark. This is the next step. And then the Twitter bookmark turns into a site that has all the things and then I can check them. So this is V1. V 0 is done. V 0 is just RSS to summaries, but it looks like crap. So, but it's done. Love to know what you guys think about this. This is actually the call to action. Um, yeah. Please let me know what you guys think about this. like what what video game am I not thinking of that would help with this and then like the video obviously you know I have to say that and comment below too if you have a good unit testing strategy like what am I missing there I'm not sure to be totally honest I'm supposed to say something in the video that's like a prompt template or like a get the freebie below, but I can't think of anything. So, see you guys.