Learning to code has changed
Chapters9
The speaker notes how learning software development has evolved by 2025-2026 and reflects on their own learning since 2007, highlighting changed tools, attention spans, and the need for a modern learning workflow applicable beyond web development.
Traversy Media explains a modern, fundamentals-first learning workflow for 2025 and beyond, blending traditional courses with AI and real projects to stay current.
Summary
Brad Traversy reflects on how learning software development has shifted since the late 2000s. He acknowledges his early inspirations from YouTube and classic tutorials, but notes that the ecosystem now demands a different approach. His core message is to build a solid foundation of concepts, then leverage AI and up-to-date docs to fill in the gaps. The centerpiece of his workflow combines instructor-led courses or guided curricula with real-world projects, while treating AI as a learning assistant rather than a shortcut. Traversy also walks through practical tools and practices, from Code Rabbit for code reviews to Excalidraw for resource planning, and he emphasizes staying current with evolving frameworks and documentation. He offers concrete strategies for using AI responsibly—asking the right prompts, contextualizing error messages, and avoiding blind copy-pasting. Across the video, he stresses pacing, measurable goals, and learning in public to accelerate problem solving and collaboration. While he teases an upcoming web development guide and an interactive platform, the takeaway remains clear: focus on fundamentals, learn how to find answers, and build continuously to become independent developers.
Key Takeaways
- Fundamentals first: understand underlying concepts (e.g., SSR basics, React hooks) and use syntax as a byproduct of repetition, not a memorization crutch.
- Treat AI as a learning assistant: use it to explain concepts, debug with context, and update knowledge with current docs, not to produce finished code without understanding.
- Real-world projects after courses enforce problem solving and debugging skills that handholding in tutorials cannot provide.
- Use a guided learning path for structure (courses, books, or bootcamps) combined with AI to fill gaps and tailor explanations to your pace.
- Integrate code reviews into your workflow with tools like Code Rabbit to catch issues early and define team patterns through custom rules.
- Build and learn in public: share progress, discuss solutions, and teach others to reinforce your own understanding.
- Prioritize sustainable learning habits: pace yourself, set measurable goals, celebrate small wins, and avoid burnout.
Who Is This For?
Aspiring developers and current engineers who want a modern, sustainable learning strategy that balances traditional coursework with AI assistance, real projects, and community learning. Especially useful for those navigating the expanding JavaScript/React ecosystem and learning paths beyond web development.
Notable Quotes
"Hey guys. So, learning software development in 2025, 2026 and beyond is a lot different than it was even three or four years ago."
—Opening point: the landscape has changed and demands a new approach.
"So, if I were starting from scratch today, my approach would look pretty different."
—Traversy outlines a reworked learning strategy from the ground up.
"One of the most important things we do are code reviews. But it's also a super tedious task and prone to mistakes since you never really know how well someone's reviewing your code."
—Introduces the value and challenge of code reviews, segueing to Code Rabbit.
"Code Rabbit... is an AI powered code review tool that integrates into your development workflow by automatically reviewing your pull requests or you can even just use the CLI before you commit your code."
—Describe the sponsor and the tool's core benefit.
"AI can put recent documentation into its context. So you have the most up-to-date information."
—Emphasizes using AI to stay current with evolving docs.
Questions This Video Answers
- How has software development learning changed in 2025 and beyond?
- What is Code Rabbit and how can it improve code reviews?
- How should you combine traditional courses with AI to learn programming effectively?
- What’s the best way to use AI to learn React and modern web stacks without losing understanding?
- Why is learning in public beneficial for new developers?
Traversy MediaBrad TraversyCode RabbitAI in codinglearning workflowsoftware development 2025 2026ReactSSRNext.jsTailwind docs and AI context
Full Transcript
Hey guys. So, learning software development in 2025, 2026 and beyond is a lot different than it was even three or four years ago. And I started around 2007, which seems like a lifetime ago. And I learned most of of what I what I know right here on YouTube from channels like the New Boston courses like Killer PHP. I learned jQuery and JavaScript from Jeffrey Way on Touch Plus. Uh I looked up to these people so much that I wanted to become one of them. And although I still don't think I'm as good of a teacher as some of them, but I'm proud of what I've accomplished and that I was able to help so many people, but here's the thing.
The way that I learned is a little outdated. And the fact is that there's just so much more to learn. You know, the internet in general, the the tools that we use, the um the way we think about learning, our attention spans, so much has changed. So, if I were starting from scratch today, my approach would look pretty different. And it does look different. and I'm going to share with you my exact learning workflow and what I suggest people do when learning something new. Now, this is not specific to just web development. I I will have my webdev technology guide coming out in January or February.
Uh but this is more for just anyone whether it's mobile apps or systems developer, machine learning, whatever it might be. Uh, I'll talk a little bit about the general categories of what you'll learn, but my main goal is to to talk about how to learn these things. So, let's get into it. [music] So, when it comes to software development, one of the most important things we do are code reviews. But it's also a super tedious task and prone to mistakes since you never really know how well someone's reviewing your code. So, today's sponsor, Code Rabbit, makes code reviews so much easier.
It's an AI powered code review tool that integrates into your development workflow by automatically reviewing your poll requests or you can even just use the CLI before you commit your code. So, Code Rabbit doesn't replace the need for code reviews. It just makes them more efficient and catches things that humans will often miss. So, instead of bugging teammates with things that don't need their attention yet, you can have your code reviewed in minutes and you can catch bugs before asking a c-orker for a final review. So let's take a a quick look at this PR where code rabbit caught some bugs.
So it catches that we initiated our client instance to our apprite backend in two separate components when instead we should have created one and then used the import throughout the app. So it immediately caught this mistake and provided us with suggestions for changes that we should make. So one of the coolest features is the ability to set custom rules for you and your team. So for example, we have a rule set that describes how headings should be set in articles. And when a PR is made, Code Rabbit will catch this. Even though it's not technically a bug, it's something that a teammate could have easily missed.
So a Code Rabbit will also learn from your entire codebase and help enforce patterns and and rules that you and your team set. So it's actually pretty awesome. So if you want to give Code Rabbit a try, make sure you check out the link in the description below. All right, guys. So, I have this Excaladraw document and I'll have this available for download in the description if you're interested if you want it for as a resource. And like I said in the intro, this is mostly about how to learn, not so much what to learn, but I did add just a very highlevel um area at the top here where we can talk about just in general what you'll learn as a software developer.
And this doesn't have to be webdev. It could be anything. So, generally, you're going to start off with a programming language. If you're learning webdev, you're probably going to start off with HTML, CSS, not really programming languages, but they are, you know, a markup and a style language. And then JavaScript and all that. Or you might be learning Python, Go, Rust, PHP, Java, whatever it might be. It's going to be a language. And then this stuff here, you're pretty much going to learn at the same time. So some kind of software framework. In webdev, you have frontend and backend frameworks.
And then you know you have Django for Python, Laravel, PHP, you might be learning.NET maybe C. Um so yeah framework you're at the same time you're going to learn a database and an OM so a way to connect and interact with your database. You might learn Postgres or MongoDB and then version control with Git and GitHub. Uh and then generally you're going to learn design and architecture. So certain patterns like MVC uh rest component-based development and then some kind of deployment DevOps stuff. So like AWS versel maybe you learn docker and containers. So generally everything you learn is going to kind of fall under this umbrella.
And this might look like a simple list but there's a lot that has that goes into all this. All right. So once you figure out what you learn that's that's really what I want to talk about is how to learn that stuff. So before I get into the actual methods and and workflow, I just want to kind of talk about the difference between back when I learned around 2007208 and now because it's very different. So back then my goal when learning something was to to basically open up a text editor and be able to just go to town and and just create stuff, right?
And that was possible. you were able to to memorize not all but most of the syntax, the important syntax by heart so that you could just open up an editor and code without even looking at the docs. Um, and that was possible because there wasn't so much to learn back then. You had your B for webdev at least you had HTML, CSS, JavaScript, uh, jQuery, WordPress, PHP. Those were kind of the big things. Maybe Ruby on Rails. Those were the big technologies back then. Right? So, uh, and you didn't really have like a a front [snorts] end was very simple.
It was basically just HTML, CSS, and a little bit of JavaScript. Right? Now, you have all these frameworks and different tooling and uh, SSR and uh, single page applications. So, it was it was okay to memorize things like jQuery or at least all the the the main methods, right? You could code without docs. you might use them occasionally or Stack Overflow, but um you didn't need to have them, right? Because you had a small stack. Now things are way different. There's so much more to learn. So what I say is un understand the fundamentals and the underlying concepts.
That's what you want to focus on the most. And then the syntax will follow, right? After so much repetition, that syntax will start to embed in your brain. and uh syntax changes so much especially with frameworks and libraries that I don't think it's as important anymore you know and you can just look up anything with AI so it's more about understanding those fundamentals so for instance react hooks if you're learning you want to understand that or grasp SSR concepts these are just examples but whatever it is that you're learning and then you want to use AI and the documentation um and using those together is very powerful you can use put the document mentation in the context of the AI.
And I'll talk a lot more about how to use AI in a minute. And you just have a huge ecosystem. So if you think of React, you have React, React Router, Tanstack, uh Nex.js, Astro, Gatsby, React Native. There's just so much to it, right? So I would say your goal now should be knowing what to build, how to find the answers, and where to find those answers rather than just remembering all the syntax by heart. So that's just kind of my my take on back then versus now. So this is the actual um workflow that I use when I learn something.
So I use a a mix of of a main learning method kind of a traditional learning I guess then use AI as a supplement. So basically you'll pick a learning method like a video course, a book if you you know if you prefer books, YouTube tutorials, whatever it might be. It could it could even be uh if you're going to college or or boot camp, that could be your main uh your main learning, right? As long as it has some kind of linear path. It has to be guided. It should it should be curriculum based.
Um concepts are reinforced. You have instructor context, whether that's a real life instructor or someone like me through video courses. Someone that uh that explains why things work, not just what works. and someone that you kind of vibe with that you that you can understand easily. It doesn't have to be me. It could be anybody, right? And then one of the uh what I think is a really important aspect of these these traditional methods is realworld projects, right? You have an actual application, not just theory. So the mindset for for this is to it have it be your main teacher, right?
So this learning all these methods are your main teacher. Then you have AI as a supplemental learning method where you can get much more specific about, you know, the topic in general or specific examples. You can have it fill in the gaps from courses you might take because a lot of times courses don't cover every single thing of that technology. Um, it'll help you debug your errors. And you know, back in the day, not even just I mean a couple years ago, if we had errors that we didn't understand, we would have to find something similar on Stack Overflow and kind of go off that very limited examples.
But now we can paste in that exact error message. Um, but the thing is you don't want AI to just, you know, give you the answer and you just implement it and not understand what happened. You want to have it explain everything to you. Um, and then also a lot of courses go up out of date very quickly, right? And I know this just as well as anybody. I created my React course and a month later React router completely changed and I had to update it. So what AI can do even even though it has a cutoff date which could be, you know, months and months ago, it can put recent documentation into its context.
So you can get the React docs or the Tailwind docs and put that in the context of your AI so you have the most up-to-date information. So the mindset with this, you know, supplemental AI is to have it as your learning assistant. So you have your main teacher, you have a learning assistant. All right. Now, as far as resources, of course, for you know, video courses, you have Udemy, Pluralsight, kind of these big, you know, big websites, and then you have creator platforms like my own, Traversy Media, Lariccast. There's also interactive platforms like free code camp, boot.dev.
Um, and I'm not going to say too much, but we're actually in the process of creating our own interactive platform, which hopefully will be ready by the beginning of next year. I'll talk more about that in a few months. Uh, let's see. If you're using books, don't use audiobooks for coding. I mean, you can't really learn learn coding with audiobooks. And then for AI, you have browserbased um AI like chat GPT or Claude AI or Grock. And that's okay because you're learning, right? You don't need an agent to create files for you and write your code.
But this can still be more helpful in my opinion. So idees like cursor or CLIs like Claude Code or Gemini CLI. Um, now they may be tempting for you to basically have it just do all your work for you, but you have to kind of be disciplined and um, not let it do that. So, put it in plan mode. Cursor, for instance, has ask mode where it won't change your files. It'll just put it in the, you know, chat interface. And you want to have it not just give you the code, but explain what's going on, right?
So, you might post an example from your course and have it, you know, tell you certain things about it, how it works, some maybe the instructor didn't cover it well enough or whatever. Now, I don't suggest just opening AI and asking random questions. I mean, you can do that, but you're not going to get as good of an exper experience. So, I would suggest putting in a prompt like this. So, for instance, it says, "I'm taking a course." You put the course name, instructor name, uh, talk about what it teaches. I want you to act as my personal coding assistant and tutor throughout the course.
Do not do the work for me. That's very important to mention. Um, and don't just give me the answer. Your job is to help me to understand each concept. Explain code examples, says English, but of course, whatever your language is. Suggest small practice challenges, keep track of what I've learned, um, help me apply what I learn in small projects. And I'll talk more about projects in a second. And then it says, "As we go, I'll paste code or concepts from the course. I want you to help me debug, improve, or modernize them with the latest best practices and then start off by creating a simple study plan and tell me what tools to set up, um, what I should, you know, have before starting.
So, I think this is a really good prompt to start with and, uh, and just have it work with you throughout the course. Have it create documents, um, cheat sheets, things like that. So yeah, I mean that's how I would suggest using AI. Now projects are very important and I'm not just talking about course projects. Courses themselves are they're they handhold a lot, right? They take you through and and you basically it's kind of a watch and and code watch and copy kind of thing. But it's good because it's structured learning, right? Teaches you the main concepts and syntax, but it's not real world development.
real world development, you don't have that handholding, right? So, it's on you to uh to debug and to refactor and stuff like that. So, you want to after your course or it could be just after a section in the course, build a project based on what you've learned because that's going to have you run into real issues that you have to fix. Um because again, in the real world, when you get a job as a developer or you start building your own projects, whatever your plan is, you don't have that handholding. You do have AI to help you and you can and I suggest using AI to help not to give you the answer but to help and this gives you a much deeper understanding and most importantly helps you with problem solving.
That's the most important thing in in software development is the ability to solve problems and just doing this without this isn't going to help you with that. It's going to teach you these main concepts and syntax which is fundamental which you need but it's not going to help you with real world development uh at least not you know not all of it. So this gives you independence. So projects are really important. Then I just have some general tips. So you want to pace yourself. Set a certain amount of time each day. Don't burn yourself out.
That's really important. Burnout can lead to just a lack of interest at all in in coding. It can also lead to depression, anxiety. I had a really bad burnout phase. I've talked about it in videos. Um, specific measurable goals, right? So, don't just say, "Oh, I want to learn React this week. I want to learn React hooks or I want to learn JavaScript event listeners." Whatever that particular thing is. Learn by doing. This kind of goes back to what I just said. Don't only do course projects. Create your own stuff. Real world experience. You want to celebrate small wins.
This is not easy. So, if you learn, you know, how to create a GitHub repo and commit and all that, that's a that's a an accomplishment and you should celebrate that. Give yourself credit. Uh, teach someone else what you've learned. I think that this is valuable and not talked about enough. Um, you don't have to be a teacher or a content creator, but just saying it out loud and showing an example to someone, it helps you just as much as it helps them. Okay? It reiterates what you've learned. Also, you're helping someone else, which is always good.
And then I would say build and learn in public. So, create a Twitter or an ex account, Instagram. Um, post what you learn. You can discuss with others. It always helps you kind of get a better understanding of things when you talk about it with people through, you know, you can even join like a discord. And just having those chats about uh what you're working on can really can help you learn and just get a you know get familiar with talking to other developers because if you plan to work in the industry you're going to be doing a lot of that.
All right. So again you know this is just kind of a general overview of of how I I suggest learning um any kind of software development in 2025 26 and and beyond. So hopefully you learned something. And hopefully you'll you'll implement some of this. Again, this document will be in the description for download. And that's it.
More from Traversy Media
Get daily recaps from
Traversy Media
AI-powered summaries delivered to your inbox. Save hours every week while staying fully informed.









