AI Developing Requires 1 Skill Most People Ignore
Other Videos📅 2025-11-05
Transcript
[00:00] Here is the most important skill you need to learn right now for 2026 when it comes to AI coding. I've been developing software and creating apps for 14 years. And just recently, in the last couple days, October 29th, 2025, one of the biggest breakthroughs in coding has incurred. I mean, this is the same level of breakthrough that we saw with Chad GBT3.5 in November of 2022. So, therefore, let's jump in. Welcome back, y'all. In today's video, we're going to be going over a massive breakthrough. Bye-bye box. This is 14 years of me
[00:31] doing coding and I'm going to run you through a little bit of a timeline and just for everyone that wants to know what the breakthrough is and what's happening is multi- aents and specifically the ability to run these in parallel. That doesn't make that much sense. Don't worry, let me explain more about that. But in order for me to best have you understand what's happening here, I need to give you context of just what coding used to be. And then yes, of course, I'm going to go over that specific skill associated with multi- aent parallel workflows that is entirely different than what AI coding used to
[01:01] be, what coding used to be. The entire industry essentially, I've been coding since I was 12. And I don't just bring this up because, oh cool, dude, this kid, this guy was coding since he was 12. He's so cool. No, no. I bring this up for you to understand that from 12 years old all the way to 23, you know, I was just coding. Coding apps on iOS store, coding web apps, coding AR apps. Coding for me is a passion. Coding for me is completely fun. I do it because I enjoy it. I could care less about making any money in coding. So from 12 to 23,
[01:33] there was a lot of frustration around coding. And specifically the frustration would be learning to code was extremely difficult. This has to do with the fact of a couple of things. The first thing is that we used to have to type every single line. Granted, we're not getting maybe code from an open- source repository. Essentially, if we're just going at it raw, we would have to type every single line. That takes time. So even if you know exactly what to type or you know exactly what you want, you still got to type it. The second frustrating thing with coding before AI
[02:05] and specifically 3.5 was troubleshooting when you ran into errors. Running into errors required you to go to platforms like Stack Overflow, Reddit, random blogs. Uh it was very very timeconuming. We're talking for errors to be solved in this context. Especially if it was a new tech stack or new piece of code you're working with, it could take days. No joke. Like genuinely days. There was situations where you ran into an error and you were completely new to it. It would take you two to three days to solve that air. We're talking multiple
[02:36] hours. We're talking about the original thing that I really only experienced pre-AII more than I do now, which was when I was coding and trying to solve those errors. It would take, you know, one or two days and then, you know, you go to sleep and then you'd be able to solve it in 10 minutes. Like your brain refreshed. This kind of eureka moment when it comes to coding is not as relevant as it used to be because of the fact that now with all this artificial intelligence if you know how to prompt it correctly you can solve errors extremely fast. We're talking under an hour used to be days. Another thing for
[03:06] you to understand preI when it came to coding is the education around coding was extremely limited. And people that would teach coding, I mean, you got to understand, y'all, I was 12 years old over here, 14, 15 years old, trying to watch YouTube videos on coding, they were extremely boring, dry, no one, they wanted to make a skill set that should be fun and creative, just boring. I didn't like that. And partly part of the reason why I'm where I'm at now, honestly, is because the fact that that 12-year-old version of me would have
[03:37] want to watch a video from the 26-year-old me that made coding fun, didn't have the BS developer jargon, just made it clear. So 12 to 23, 23 3.5 comes out, and this was a huge unlock. The reason this was a huge unlock, even though the code wasn't perfect, the code wasn't amazing, was if you knew how to code before GBT3.5, you would know how to essentially prompt it to handle a lot of the dead weight kind of code. A lot of the code that you knew exactly what to type, but you just didn't want to
[04:09] type it. That's a real thing. So 3.5 coming out at 23 was just one of the most massive unlocks in development in history for a long time. Now, I want to say one thing real quick. You know, 12 is 2012 for me. For all those OGs out there that was coding in the 90s and 2000s, shout out to y'all. Y'all had a different level of how difficult it was cuz y'all would have to read a book. [laughter] Like, for me, it was like just go to a form. I could post on the form. I don't like respect. Respect. All
[04:40] right. I know I know it was much harder back then, but I'm saying for someone like me at 12 years old, you know, I know for for y'all they were just like, "Oh, wow. You had a form. You had something you could post an air question on." My bad. My bad. Went back over here though, 3.5 23. You know what's kind of funny about this age as well is this is when I started the YouTube channel because at that moment in my life, I had realized from all the way back here that this is no joke. You got to understand when artificial intelligence first came out, specifically Chad GBT3.5 when it
[05:12] first started making movements, people thought this was a fad. 2022, November, December, January 1st of 2023, people thought this was going to go away. Well, obviously it hasn't. And now we're 3 years ahead. Which leads us to the next major breakthrough. I don't care about anything in between here. Yeah, we've obviously improved tremendously when it comes to context, the amount of data an AI model has. We've obviously tremendously improved when it comes to integrating thirdparty APIs through MCPs. But none of this really pushed the
[05:44] pedal forward like 3.5 did until until multi- aents running in parallel. This is one of the most powerful things I've ever seen. And the reason this is extremely powerful is that little idea we had back here which was can a solo engineer can an indie engineer really create out a large software company that is successful and usually the answer to that is no for obvious reasons. The reason is because in order for you to
[06:16] really create out real software you're going to need a team. And that is still true to an extent. You'd want other people on your team to help you build it. What I'm telling you right now is that these multi- aents just unlock the door that used to be a thought experiment. Now you have your team. The only thing limiting you between executing a real application now as a solo developer is the knowledge deficit of understanding how to associate labor to these agents. Eg. You're the CEO of your software company.
[06:47] Do you know actionably what to do next in your software for it to push the pedal forward? So, the skill you got to learn in 2026 is going to be how to be a senior engineer and specifically how to review code and associate labor in code. As a side note, anyone that's interested in seeing a quick demo of the multi- aents, check out my channel. It should be one of my more recent videos. Also, I'll try to leave it at the end here and possibly in the description. I think it's like a 9-minute video. I show you the power of this. Really simple prompts. You know, you got to understand, y'all, when I do those kind of videos showcasing a a very specific
[07:18] use case, especially when it's under 10 minutes, I got to show simple prompts for the outgo. If I get too complex of a video, uh, it turns off the algo. Essentially, the more complexity I usually leave to my bigger series that are like hours long. Some like what is what is going on behind this guy right now? Uh, let me explain the old way of running a software company. And when I say old way, I mean this is a very efficient way to run a software company, but the original junior engineers would be human be like you and me. Now they can be AI. One thing you need to understand about this topic and content
[07:50] around this topic and why people are still sitting in denial is because a lot of times software engineers they are quite literally just working on a task provided by them in the actual startup itself. What do I mean by that? How would you know that we're in a new era of coding if your entire role within that starter company was always being told what to do next? you weren't the one actually creating the maps, the agenda, and pointing in a direction. That's why there's a huge disconnect. And that's why in the last couple days,
[08:20] we are so early to this, like genuinely so early to this that I'm I'm so excited right now. I've been working extremely hard for the last 3 years. I'm going to work even harder because I just realized we're at an inception point in the market where the thought experiment is real. As a senior engineer, what I used to do with my internship classes at the software company I ran is I would associate labor. I know I know some of you are like, Corbin, I already watched your other video on this. Let me just give a TLDDR. All right, some people just found me for the first time or some
[08:51] people clicked on this video than my other videos. Point being, the old way would be a senior engineer would associate labor to each junior engineer. Each junior engineer then would create their own little branch here and their own little branch be like, you know, V1, whatever. Just call the branch something. And so they'll create their own little branch based off Maine. Uh if you don't know what I'm saying right now, I'm speaking alien language. Uh watch my 46-minute GitHub tutorial so you understand what GitHub is and branches and why it's relevant. This is Maine, you know, production. Uh the code that is read on the real live website
[09:22] link. Uh the junior engineers would create their PRs off of this and then when the code was ready, absolutely amazing code, all five branches would come back to the senior engineer. This is standard workflow in software companies. The juniors would do most of the heavy lifting when it comes to actually writing out the code and producing the code and then the senior engineer will vet the code and ensure the code works and it doesn't break the product and it looks good. Uh the typical workflow here would be a back and forth between each engineer until both confirmed or essentially the senior
[09:53] engineer says yes this is good to go and then would merge it. Really big companies have multi-layers of this where you could have juniors, senior and then it just the essentially the whole architecture just keeps going up. So really big companies might have multi-layers of this before it even touches zreal application. I teach this stuff so you understand in this new era of coding we've gotten to the point now and one thing that makes a startup extremely advantageous in this new era of coding is you don't have as much bureaucracy as you do in these larger companies. There's a lot of very very
[10:24] intelligent engineers that work for big companies and they are handcuffed behind their back because they they have so much bureaucracy. Uh their creativity is just shot at the legs because the senior engineers or the higherups don't care what they say just work a paycheck. this new era of coding is going to allow people like that that are very gifted to have the ability to be like you know what no I'm going to start my own company because the only thing that was stopping them before is in reality to start a company that's going to be
[10:55] successful you do need a team but now the AI models can be your team one thing I want to point out real quick because I wrote it here a sprint for all y'all wondering in terminology for software development is two weeks of labor And the idea is that these PRs that we would do together, uh, there'd be a twoe deadline and then when the twoe deadline is up, it's like how far did you get in your PR? Uh, typically the engineers will give you updates throughout the entire workflow. The whole point of a senior engineer as
[11:25] well is that you know the senior engineer if engineer 2 over here is like 80% done of the coding task and then they have an error run into a roadblock need help they can either talk to other junior engineers or alternatively talk to the senior engineer which will help them out. What I'm trying to say here and why I am making such a bold statement that 23 nothing very significant in my opinion has happened until now is this managing position here we couldn't do this before because the code wasn't good enough now it's good
[11:56] enough now the skill you need to learn is this role how do you associate labor corbin I have an amazing app idea I know exactly what I need to build well don't worry you can actually build it now you should watch my Google AI studio cursor series to learn how to do it from complete scratch. I'd treat it like you've never coded before because you don't need to know. You want to build an app. The skill you need to understand, and this is going to sound extremely dumb on the surface, but then once you actually understand what I'm saying, you're going to be like, "Oh, that makes a lot of sense." The skill you need to understand is the architecture of the
[12:27] app, how it's built, what does this mean, what's the back end, what's the front end, what's the data path, how you storing information, what's the security. This is more of the operational side. But the bigger thing that you are needing to learn that wasn't really that important pre this update or big movement is labor and specifically how you associate labor with the app itself. What's next? A very simple example of this could be you're the senior engineer and one of the
[12:59] pieces of labor you give engineer one is hey make the settings page make the user interface look like this. Another example of this would be like, "Hey, engineer 5, we need to work on the specific pipeline when it comes to the prompts we use when analyzing a user's PDF and providing value at the end of the PDF. This is how you associate labor, how you know what to do and what to build in an app. It's fundamentally more important now than you needing to write out the code." I could keep going down this rabbit hole, but I think you get the gist. TLDDR of everything I just
[13:31] said right now is why do a super complex math problem without a calculator? You might as well use a calculator. Oh, no. No, Corbin. Uh, you use a calculator to solve that math problem. This guy doesn't know math. Okay, don't use the calculator. Put yourself at a significant disadvantage. I'll use the calculator. But as you already know these style videos, I'll see you in the next. If you're watching this video right now, you are one of the most earliest of the earliest of the earliest adopters into this new phase of development. This stuff is so impressive that it's even better than GBT 3.5 when
[14:01] it first came out because of the fact that now we can take solo developers and create multi-million dollar companies if you actually lean in and learn the tech type of video. You had a form. You had something you could post an air question on. My bad. In my bed.