How I Would Learn To Code In 2026

Hey, let me show you how I would learn to code in 2026 as a complete beginner. Before we talk about anything else, I want to go through a couple statements that I see going viral all the time, and they are completely false, right? First of all, coding is not solved, okay? I don't know why so many AI hypers seem to believe that, but that is just not the case. Coding is not solved, not by a long shot. There is still a lot of work that goes into building a good app that these models are just not capable of doing right now and potentially probably not for a very long time. Which leads me to my next point, which is that learning is not obsolete. You can't just prompt your way to a good app, all right? Not unless you have the prerequisite knowledge of how to build a good app in the first place. If you don't know how to have good software architecture, if you don't know how to structure your code base in the right way, to identify performance bottlenecks to solve and patch security vulnerabilities and so on and so forth, you're not going to be able to prompt your way into building a performant, fast, secure app. So you still need to keep learning. And AI has actually made it harder to learn in some ways. I know that people use it as a guided coach or a tutor or something, and that's really helpful. I've been using it myself for those kinds of purposes, and it has, you know, some utility. But in many ways, at least in terms of coding, it's made it harder to learn things because it just abstracts you away from your mistakes, right? If you're writing the code yourself, then you're going to be making a lot of mistakes and you're going to learn from those mistakes because you're going to the consequences of those mistakes. If you then use AI to write the code for you and you just write a bunch of prompts, you've completely abstracted away the learning process. And so you're not going to learn pretty much anything if you're just using AI to write code. like if I pick two random people two completely junior programmers and I told both of them to I don't know create a mobile app of some sort and one of them is allowed to use AI and the other one is not allowed to use AI the person that is not using AI at all is going to learn a lot and the person who is using AI to build the entire app is probably not going to learn anything at all that's just how these things work you've abstracted away the learning process right but there are still some ways to learn how to code and learn to improve your code despite having AI write your code. I don't want this video to be something like, you know, if you want to learn then just stop using AI because that's just not realistic anymore, right? Everybody and their mom is using AI to write the code so if I give you some advice to not use AI-generated code that's just going to be bad advice on my part. But before we even talk about how to learn the code using AI in 2026, I want to just give you a couple of reasons why it's even important to keep learning. So this is a quick little illustration I made to just show you how much more effective seniors are with AI and how much more, how much less effective juniors can be with AI. So just for our purposes, right, let's just say that if you're a senior developer, you're, you know, this good at writing code right writing building apps And if you a junior you all the way down here right And this is without using AI right This is just handwritten code What happens if I then give a senior developer access to an AI coding agent? Well, they're going to skyrocket their productivity, right? They're going to become even better at their job because they already have the prerequisite knowledge of how to build good software so that they can just use AI to just leverage the hell out of their work and just produce so much more and so much better stuff, right? Because they already know exactly what they need to do. If I did the same thing with junior developers, their productivity would actually decrease. And the reason that happens is because they don't have the prerequisite knowledge, right? They can't actually prompt the AI to build good apps. If you look at a prompt that a senior developer gives to an AI, there's going to be so many more details in there, so many more, you know, implementation details in there, like, hey, use this data structure, use this pattern, implement it in this way, make sure to handle these constraints and these edge cases. The senior developer is going to know all of these things and how to handle them, right? The junior developer's prompt is probably going to be like, hey, go build this feature, thanks. It's a world of difference, right? So having access to these AI tools is a great blessing if you're a senior developer. It's a curse if you're a junior developer. And so you need to learn to code you need to keep improving your software engineering skills so that you can get to this level over here and not stay stuck or even degrade your skills down to this this level right now let's talk about the actual best way to learn how to code and this is the exact same answer that you would get even if you ask someone to learn to code like five ten years ago because this thing works what you need to do is you need to build a fuck ton of projects You need to build actually useful stuff in the real world, right? Not some tutorial cookie cutter app. AI can make those for you already. AI can probably make most of the projects you're going to make anyway. But this is still the best and the fastest way to learn. And the best, absolute best projects to build are the ones where you're solving some kind of problem. So pick your own problems to start with. Or if you know somebody who has a certain problem that can be solved with software, build an app to fix that problem. All right? Solve your own problems. solve your other people's problems and build projects, build apps, build software that can do that for you. And use AI to write all of the code for all these projects because let's be honest, you're not going to be writing all of this code yourself. That's just not the world we live in anymore. But constantly ask questions about the code to the AI. Constantly keep querying about the code that the AI is writing. A quick example of this is Phantom Editor, my automated video editor which i've talked about on this channel a lot this project has so many different areas that i have no clue about things like video and audio processing things like how ffmpeg works and how to write ffmpeg commands things like browser automation i had no idea how browser automation worked at all before this things like social media platform apis now this is probably the simplest one of all four of these because like apis are just these rest apis http apis and you you can probably just read the docs and understand them But like I had no clue how any of this stuff worked before i started building this project and i still managed to write i don know what how much is this yeah 19 000 lines of code over here all completely ai generated all completely well tested and everything just works and does exactly what i wanted to do i know exactly where everything is i know exactly how the code base is structured how everything is implemented how each tiny feature is implemented even though i i used ai to write all the code i still understand how everything works behind the scenes right and if you ask me about all of these different things now i can probably give you some really detailed answers on all four of them because i just understand them so much better now even though i used ai to build the entire project for me and it's because i asked a lot of questions and i asked about the you know areas that i'm confused about and i tried to learn as fast as possible and as best as possible about every single feature that I'm implementing in this code base. Now here's the thing, this doesn't need to be a complicated process at all, right? All you have to do is, let's say you sent a prompt to your AI to build some feature or write some code, right? What you should do next is you should just grab that section of code, paste it into your coding agent, and just ask it to explain that code to you, right? It's going to give you a long explanation, you're going to read all of that, you're going to understand some parts of it, and you're not going to understand some other parts of it. And you can then keep asking questions about those parts so that you develop a much deeper, much better understanding of the entire code base, right? Or at least of this one particular feature that you implemented with AI, right? And as you keep doing this, as you keep repeating this entire process, over time, what's going to happen is you're going to build familiarity with that code, and you'll understand it better. And when you repeat this entire process across your entire code base, you're going to build a lot of familiarity with your entire code base, no matter how many lines that is. Even if that's like 50, 100,000 lines of code, you're going to become familiar with it. You're going to understand and know where everything is and how each feature is implemented. Right. So really, like you asked me how I would learn the code in 2026. It's as simple as this. Solve problems, build projects. and as you keep doing this and as you keep asking AI and you keep patching the holes in your understanding by asking lots of questions you're going to learn the fundamental skills of software engineering right if you're building really good projects and you understand how they work under the hood you've pretty much learned software engineering right you've learned all the different skills over here like systems design architecture solving performance bottlenecks like just think about it. Let's say you build something like, I don't know, a 3D animation website or something, and it's really, really slow, right? You run it on your computer and it's really fucking slow. What are you going to do? You're going to go do some research on performance optimization, how to identify performance bottlenecks, how to improve 3D animation performance, and so on and so forth. And you're going to learn all of these things, right? And you're going to implement them in your own project, probably using AI, but you'll have such a a much better prompt to give to the ai because you have done your research on all of these things right and as you do all that you just learned how to solve performance bottlenecks right In fact one of the ways I did that is again in my Phantom Editor project I was using it to you know edit a lot of videos and it was just taking a lot of time because the video encoding process was really slow. And for a long time, I just didn't really care because, you know, all the processing happens in the background. I don't really need to think about it much. But one of these days, I just decided to, you know, go and optimize it, go and see what the heck is taking so long, right? So here's the entire process that I followed. I asked AI to analyze the entire file, analyze the entire feature set, and suggest performance improvements, right? Because again, I have no idea how video encoding and audio encoding works, right? I don't know those kinds of topics very well, right? I'm just a dumb TypeScript compiler, right? I'm just a dumb TypeScript web developer. So I don't know how those things work. So I asked the AI to analyze the entire thing and suggest performance improvements. And it gave me like four or five really good suggestions. I then asked a lot of questions. I mean, a lot of questions about every single optimization that it suggested, what it does, how it works, how it's going to be implemented in our code base, and so on and so forth, until I like deeply understood exactly everything that the AI was suggesting. And then I used AI to implement all of these performance suggestions. I tested the app again after that, looked at the new performance scores, and they were so much better than what we had before, right? So really, that's how you code, and that's how you fix performance bottlenecks and security issues and, you know, just build an entire app using AI. Finally, before I end this video, I'm just gonna give you a couple more pointers. Don't try to actually learn to code, right? Don't try to learn to code in the traditional sense, because that's just not gonna work anymore, because AI models are already writing code 1000 times faster than you, probably faster than that if they're using SpaceX's Colossus data centers. But yeah, the point is these models can write code way faster than you. There is no sense in trying to write code yourself or to learn coding syntax and all the programming language quirks and the semicolons and stuff. All of that is just completely abstracted away now. You probably shouldn't even try to learn all of those things anymore. What you should focus on instead is the skills involved in building good software, things like performance bottleneck identification and fixing, things like how to solve security vulnerabilities and common security issues in apps, things like software architecture, system design, and all the other things that we talked about before. And the thing is, you can learn all of these things even if you're using AI, because the AI's job is literally just to write the code, right? That's the easy part. The AI can write the code for you, but all the other skills involved in building good software, those are things you can still learn. And those are things you should still learn if you want to, you know, build a good app. So if you just focus on those things, you're going to be very, very successful in this industry. And all you have to do to learn all those things is build lots of projects, solve real world problems for yourself or for other people, because that's how you're going to learn how to build better apps. and just asking a lot of questions constantly. So yeah, hope that helps.