Develop Yourself

#299 - Yes, You Must Learn to Code and Then Never Write Code Ever Again

• Brian Jenney

Share episode

AI is writing more code than ever, layoffs are everywhere, and it feels like software careers might be disappearing. In this episode, we explore why that narrative is misleading—and what’s actually changing in tech.

We break down why learning to code still matters, how to use AI without becoming dependent on it, and which core skills will help you stay relevant as the industry evolves. If you’re trying to make sense of the noise and figure out how to move forward in software development, this episode is here to help you navigate what’s next.

If you want to join our next cohort to learn AI through project based learning - sign up here: https://parsity.io/ai-dev

Send us a text

Shameless Plugs

🧑‍💻 Free 5 day email course to go from HTML to AI

🤔 Got a question you want answered on the pod? Drop it here

Apply for 1 of 12 spots at Parsity - Learn to build complex software with AI integrations.

AI Bootcamp - for software developers who want to be the expert on their team when it comes to AI

SPEAKER_00:

Welcome to the Develop Yourself podcast, where we teach you everything you need to land your first job as a software developer by learning to develop yourself, your skills, your network, and more. I'm Brian, your host. Hey, this is Brian, and I just want to thank you for being a listener to this show. This is a very exciting week for Parsity. We're kicking off the largest Dev30 class we've ever had. Really amazing. I'm super excited about it. And we just ended our second week of an AI cohort. We built a fairly complex data pipeline to upload a whole bunch of articles and LinkedIn posts into a vector database. And next week we're going into agents. This year at Parsity, we're going to do a lot more programs for people that are already past the learning to code stage and want to jump into things like AI or interview prep or learning how to use AI tools. And I really want to hear from you out there what you would like to see more of at Parsity. If you're interested in joining our next AI cohort, which will be happening in March or April, the ultimate goal is to make you the AI person in your company, the go-to person when it comes to integrating AI into web applications. You can go to parsity.io slash AI-dev to learn more. And please write me, hit me up, use the show notes, use the comments on Spotify. However, you want to get in touch with me to tell me what you would like to see. Anyways, without further ado, thank you so much and I hope you enjoy the show. If you're trying to become a programmer in 2026, it probably feels like you showed up to a party five minutes after it already ended. Everywhere you look, AI is writing code, boot camps are closing, tech layoffs are dominating every piece of the news around tech. Influencers say the jobs are dead. So it's like, what are you supposed to do? The average Reddit's advice is just give up. There's no hope at all. This is truly big-brained activity. And I'd be lying to you if I told you that things aren't harder than they used to be. But no, I don't believe this is the end of software careers. In fact, I think something a lot more interesting is happening if you look just below the surface. And if I'm being honest, I think that you need to learn to code and then get used to never ever touching code ever again. We're at a really strange moment in history where AI can now generate entire applications and thousands of lines of code in a flash. Code is really cheap. Software, though, is still expensive to create. The fact that it's so easy to create code, though, leads people to think, well, why should I even learn to code if AI can do it? And this is the consensus among a lot of CEOs, tech bros, marketing bros, product manager bros out there that are saying we don't need developers anymore. And I get it. It can be a really empowering thing to not have to ask a developer or know how to code to build an application and go from thought to something out in the cloud in minutes or days with very little tech background. Now, obviously, this has a lot of pitfalls, but it's actually possible. And this wasn't the case pre-AI. And you may be asking, well, then why even learn to code at all? Here's the thing, though your browser, your phone, servers, databases, APIs, they all require machine readable instructions. AI doesn't eliminate the code, it just changes who's writing the code. And if you don't understand it, this can have catastrophic consequences. We've seen a massive uptick in the number of bugs that are getting released. We've seen a massive uptick in the number of outages. We've seen some pretty embarrassing blunders from vibe coders who end up leaking all sorts of sensitive data on the internet. And this is only going to increase. Like guaranteed, the more people that put out slop, unsecure slop, and don't have any clue of how it's made is only going to make the internet a much less safe place for all of us out there. If you don't understand the code and you just blindly trust whatever the LLM outputs, then you've outsourced the only thing that really matters, which is your thinking. If the model is wrong or hallucinating or just offline, you're completely screwed. So, yes, you still must learn how to code and reason through trade-offs and a bit of system design and just understanding at a basic level, like how the internet or code or applications work, and understanding at more than a basic level how code works, and have some moral and more importantly, legal obligations that you'll have to face. But let's be honest about why this feels harder than ever to become a software developer or just to stay employed. I think a lot of us are feeling this feeling like, dang, is the end coming near. And I'm of the opinion that the internet is basically lying to you. We're all guilty of this, myself included. Everybody wants to predict the future. It's a really exciting time and things change up so quickly that it feels like we can see just what's on the horizon, but we really can't. We're humans. So every couple days or weeks or months, you'll hear another take like AI killed web dev, AI replaces programmers. Learn this prompting technique, or you'll be irrelevant. You won't be replaced by AI. You'll be replaced by someone who is using AI. Some of these cliches actually are truer than I used to want to believe, but a lot of them are just people getting hot takes and clicks online. The reality is, I don't know, they don't know nobody. Anyone who claims that they know what jobs will look like in six years, let alone six months, is either lying or just trying to sell you something. There are two things in this world you can worry about, right? Things you can control and things you can't control. The future of AI is not in our control, but learning and how we prepare ourselves is fully in our control. Now you can call me biased and you should because I own a coding boot camp, but this whole just self-teach everything or learn from AI, I think is really bad advice. Most people who successfully self-teach, I've noticed, fall into one of a few groups. One, they had a mentor or a family member that is involved in tech. They have a strong technical background, they might have a ton of free time, they're extremely disciplined. I had a business partner who was a former lawyer. He took, I think, three years to teach himself. I met another guy in his 40s, took him about three years to kind of self-teach and finally land a role. Now, these people were incredibly dedicated, they had a really solid plan, and they had more importantly, a lot of time. It took them over two years to do it. Essentially, they went through what a person would likely go through in a college program, but on their own. If you don't fit into those categories, I'll be honest with you, I think self-teaching is brutally hard. And AI actually makes it worse in many ways because you're probably not using it in the way you should be. Because it's way too easy to get in this habit of not knowing something, reaching out for AI, it generates the code. You're like, oh, that works. And you feel like you're making progress. So you copy code, you paste it, and you're doing this faster and faster and faster and making more and more perceived progress in quotes, right? But what you're really doing is just training yourself how to be incredibly dependent on AI. And then when you get into an interview or any situation where you can't use it, you never develop that ability to actually think on your own or work through hard problems. And so you're really just unemployable at that point. Or worse, you get into a job and you don't know what to do beyond just how to prompt something. So you'll find it really hard to move up because no one's going to pay you a lot of money to just learn how to prompt and then say, Hey, I don't know how to fix that thing. Uh, AI, can you do it for me? AI can be an incredibly powerful tool, however, if you're using it correctly when you're learning. It's really good at giving examples, explaining generic concepts, rephrasing ideas. The way I use it when I want to learn things like linear algebra or I'm having to learn Kubernetes or Terraform for work is to basically talk into Chat GPT or Anthropic or whatever you're using and explain how I understand something and then ask for feedback on my understanding. These aren't unique or novel things that I'm learning, they're just foreign to me. If you're learning how to code or trying to transfer from front end to back end, the kinds of questions you're going to ask aren't unique or novel. They've been solved a million times over. The trick is, though, is seeing how your thinking maps to the thinking of other people out there, all the training data it has. If you're talking about APIs or REST architecture, you can quickly say, hey, here's my understanding. Is this correct or is it not? And ChatGPT will give you an answer you can kind of trust. But if you ask it to just make you a learning plan, it'll just give you the most generic random plan out there. I've seen people do this, like, give me a learning plan to learn full stack JavaScript or something. It'll either give you way too much or too little or not much context. It doesn't know you, where you're starting from, it doesn't know your market, it doesn't know how you're going to learn, it doesn't know the projects that you're going to need to make. It doesn't give you the kind of feedback and instruction that you're going to ultimately need if you want to go from zero or one into hired as a software developer somewhere. If you want a real edge in 2026 and not just some sort of hype or some workaround or some shortcut or some junk like that, you're going to have to stop trying to future proof. Everybody is trying to guess the future skills. But in reality, you just need durable skills. There's some timeless things that are going to be more and more important. In fact, the fundamentals, counterintuitively, are going to be even more important going forward because when code becomes so cheap and easy to write, we have to worry about things like scalability, bugs. We see that AI generates bugs at a rate almost two times more than humans. So we are objectively better at creating code. The more code generated by AI, the more bugs, the more important it will be to debug multi-part systems that may live in the cloud, local infrastructure, and distribute it across multiple stacks. This is not something an AI can just do. This requires actual thinking and understanding of how software is generally built. The fact that code will ultimately be executed in machines is unlikely to change. Just because AI will write it doesn't mean it doesn't live somewhere and you're not responsible for it. If AI writes a thousand line long file that you don't understand that takes in people's credit card information, you deploy it and it breaks and you have no clue how to go about fixing it, then you're not only going to be potentially fired, but you could be in legal hot water and it would be ethically shady to do. So when I read LinkedIn posts or on Twitter about some idiot that made a$10,000 a month SaaS app over a weekend that they don't know how it works, I cringe a little bit because I'm thinking that's not a flex. That's basically telling your customers, hey, I don't really care about you. I don't know how this thing works. I don't care to know how it works. I just want to make money. I built this thing in a weekend and screw you if it breaks, and I have no clue what's going to happen if it does. But hey, I'm getting paid, so see ya. The next thing I think it's really important to learn, whether you're a very junior developer or a very senior developer, is just learning a bit about how large language models work under the hood. You don't need to be a machine learning research engineer or really understand most of the math behind large language models. But I do think it's important that you understand what tokens are, what training is, what attention and transformers are, and why hallucinations happen. You should be ultimately able to explain something like how words get turned into numbers and how those numbers are then used to generate the next token in a sequence and give you something that sounds plausible, right? If you can kind of explain that at a really, really high level, I think you're ahead of most people, software engineers included, that have no clue how these things really work, other than I prompt something, it spits back out code, me happy. And in addition to that, I think that learning how to integrate AI into web apps will ultimately be your secret weapon. Most boot camps don't teach this, except for Parsity. Most senior developers don't really have any experience yet either. So if you learn how to do something really simple, like call a large language model behind an API using Anthropics SDK or Vercell's SDK or whatever you're using out there, that's a good start. Then learn how do you handle things like embeddings and vector databases. This is how you build RAG systems, retrieval augmented generation. I've said this a million times and I'll continue saying it because it's one of the most practical and simple ways to get started with AI in a full stack web application. If you can do this and then learn a little bit about building agents and the different flavors of agents, like the router pattern, human in the loop, React pattern. If you can learn when and why to use these kinds of things and then maybe learn how to implement an agent using something like Langgraph or Versailles SDK with tool calling, that's a really simple way to get used to building agents and understand the pitfalls and different approaches and why you might use one over the other. You can become very valuable very fast at this particular time because the rules are still being written. There's not a lot of expertise in the field. And in the land of the blind, the one-eyed man is king. You could be that one-eyed man or woman. Next up is learn some basic system design. You don't need to be like an AWS architect or something like that, but I think at a really high level, you need to understand how services talk to each other over a network, what a distributed system is, and some of the pitfalls and nuances of working within a distributed system, how data can flow, whether with messages or cues or streaming, and some of the most common ways to achieve that, how systems can scale with things like caching or again using messages and cues, and then how these things break and how you can track these things and what kind of logs and services are out there that you can use to log and then debug errors that might happen in a distributed system. I think these skills are gonna matter a whole lot more now that raw coding just becomes kind of automated. That's still gonna be a thing we're gonna have to plan, plot, operate, strategize, and then actually deliver the code. But there's infrastructure, there's infrastructure, there's testing, there's deployment strategies. There's a lot out there beyond just writing the code on your machine. I'm not saying that's not difficult, that's just one piece of a much larger puzzle. And learning how the puzzle fits all together can really elevate you as a software developer, especially if you're more on a front-end role or full stack and haven't really gone into beyond what's on your machine. And lastly, just read some freaking books. Jesus Christ, man. The internet is tons of noise and people are incentivized, myself included, to have hot takes, to tell you things that might not be fully true, but get a lot of clicks and views. It's really exciting to hear about the company that replaced all their developers with AI. It's a lot less exciting if I showed you how I actually use Claude Code or something in an enterprise code base. People find that a lot less interesting. They find it a lot more interesting to hear about the guy that applied to a thousand jobs and couldn't get a single callback, or the team that replaced their entire development team with one sexy prompt, and now they fire their entire and now they fire their entire engineering department. But think about it, large language models have been trained on high quality writing and low quality writing as well. I mean, I have a lot of stuff out there that I'm sure they've read and put into their system and training as well. But a lot of really great books out there that you might want to read. Designing data intensive applications, clean code, clean code in JavaScript, maybe the Phoenix Project, which is a novel, which is actually a pretty good one about DevOps and just kind of how large enterprise systems can work and you can deliver with speed. This is how you build depth and also shut out a lot of the noise that's happening online from people that honestly don't usually have a lot of experience in the first place. Now, here's my prediction after I just told you that no one should make predictions. But I do think in the future, it seems incredibly clear that AI is going to write most of, if not all, the code. Humans are still going to design the systems, define the products, integrate the tools, make the decisions. And the people who are going to really kill it in software 3.0 are the people that think in systems and understand system design at a high level or a lower level, the people who can integrate AI smartly and use it to automate workflows or build products that actually drive revenue. And people that use AI to do their work without letting it replace them. So if you're using it to do things that you already know how to do faster, well, obviously that's a net positive. If you're using it to do things you normally wouldn't know how to do, and you're becoming so reliant or dependent on it that you find it either impossible or very, very difficult to work without it, then I think you're probably going to automate yourself out of a job. And if you're brand new to coding, then yes, get used to the idea that it's going to be harder than it was before. But the last generation said that before the previous generation, and they said that before the previous generation before them. It'll always get a little harder moving forward. It doesn't mean it's impossible. And honestly, the playing field in many ways has kind of been reset. And this is a very unique time where if you learn a set of core skills, you make some safe bets on what to learn, like AI integrations or working with RAG or agents and things like that. You could set yourself up to be in a very, very good position going into 2026 and further. So I hope you found that helpful. I hope you're not tripping out on all this doom and gloom news. If you do want to get your hands dirty with a project that is actually going to teach you the basics of retrieval augmented generation, get it in the show notes. It'll also be pinned in the comments. People love working on it. I'm going to have some more of these coming out. And let me know if there's something else that you would like to see a project, no videos, a project you'd like me to see me create, and I'll try to have it out to you. Hope that's helpful. See you around. That'll do it for today's episode of the Develop Yourself podcast. If you're serious about switching careers and becoming a software developer and building complex software and want to work directly with me and my team, go to parsity.io. And if you want more information, feel free to schedule a chat by just clicking the link in the show notes. See you next week.

Podcasts we love

Check out these other fine podcasts recommended by us, not an algorithm.