Develop Yourself

#296 - The 5 Worst Tech Trends of 2025 (And How to Win in 2026)

• Brian Jenney

Share episode

Grab this project to learn the basics of RAG and working with agents https://www.parsity.io/ai-with-rag

2025 was one of the most confusing years to be a software developer.

We were told AI would replace us.

Then we were told it would make us 10× more productive.

Neither actually happened — but the hype affected careers, hiring, education, and mental health in real ways.

In this episode, I break down the 5 worst tech trends of 2025, based on what I saw firsthand as a working engineer and bootcamp owner — layoffs that never stopped, juniors getting squeezed out, AI being used as a crutch, the collapse of coding bootcamps, and the biggest lie in tech: the AI productivity myth.

Then we flip the script.

I also cover the best trends that quietly emerged and how I’m personally planning to take advantage of them going into 2026 — from AI agents and coding tools that actually help, to why generalist engineers and system-level thinking are winning again.

If you’re:

  • A junior developer trying to break in
  • A senior wondering how AI really affects your career
  • Or just tired of LinkedIn hype and doom headlines

No magic prompts. No fake roadmaps. Just reality — and a practical way forward.

Send us a text

Shameless Plugs

Join Our Christmas Coding Launchpad below to land your first dev role in 2026 👉 www.parsity.io/christmaslaunchpad

🧑‍💻 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, work with LLMs and launch your career.

AI Bootcamp (NEW) - for software developers who want to be the expert on their team when it comes to integrating AI into web applications.

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. 2025 felt like the tech industry was kind of sobering up from this decade-long binge of overspending, overhiring, and a lot of hype and magical thinking. Some trends really did change how we're going to build software in the future, and I'm genuinely excited for them, like I'm sure you are as well. Others were pure industrial grade BS, the likes of which we have never really seen, driven by tons of speculation, FOMO, and billions, if not trillions, perhaps, of dollars going into questionable technologies. For me personally, this was one of the most confusing years as a developer, because at one point we were all told AI was going to replace us all. Then we were told it was going to make us 10x more productive. Now, no one actually saw these things happen, but the marketing was so strong that I'd be lying if I didn't say it did affect me in some way. I truly questioned whether this was a good career for me to continue going into, whether I should still encourage others to go into web development or software as a whole. And then I came full circle and realized okay, this hype machine is so strong, it's affecting me just like a lot of really new developers out there. And I feel really bad if you're a new developer that is coming into the field right now because this is a super confusing time. But now in hindsight and reflecting on this crazy year and all these socially awkward CEOs and they're overpromising, the patterns that emerged are pretty clear, in my opinion. And I want to start with the five worst tech trends before we get to the five best ones, and how I plan personally to take advantage of these and how you might as well. The number five worst tech trend was the layoffs that felt like they just never ever stopped. It feels kind of like we've been in perpetual layoff mode for years now. Now there was never really any stability in any career, even in tech. And I think that that myth has been completely shattered. It used to be that your parents worked the same job for maybe 30, 40, 50 years. Now it feels like if you're lucky if you work at the same company for five years. So the pace actually did slow, but it just didn't end. Tens of thousands of people lost their jobs. The cuts were marketed as strategic, but what they really seemed to tell people was hey, we're still scared, or hey, we overinvested in AI and you're gonna take the blame. You're gonna be the scapegoat for this round of layoffs. It's not you. We're just investing in AI right now. So we gotta make room for that investment, right? And this had a greater psychological effect too, because now the people that kept their jobs, they feel like they need to hold on even tighter. Teams got leaner, middle management kind of vanished. I was one of these people, I was a middle manager at a large company. I found myself laid off and I found it really difficult to find another middle manager position. Essentially, they wanted managers that could code. So I took an IC role where I'm actually a lot happier, but I do feel really bad for the middle managers and for the tech people that had more human-centric roles because those feel like they're just disappearing. Non-core roles disappeared. And the fear and instability that this has created, even if it's just the perception that things are less stable, has led a lot of people to be a lot more safe in how they approach the job market. It means people are staying in one position longer, they're not job hopping as much. And I've even heard people say, Don't negotiate in this market. Whatever you do, don't negotiate. Now that is awful advice. And I hope that trend reverses going into 2026. Be cautious, not dumb. Number four worst trend was using AI as a crutch instead of a tool. We saw this with vibe coding, we saw this with all the 10x productivity bros on LinkedIn and Twitter. This one really actually kind of scares me, and I think it should kind of scare you too. We're seeing developers now and non-developers who can scaffold impressive looking apps with AI and they can create a really beautiful interface or even a full-fledged mobile or web app and then deploy it and get users and money for it. But these same people can't debug a simple issue using the network tab, console logs, or know how to inspect the logs in their cloud service provider. This should scare us all. We are treating AI like a magic wand, and we're building a house on quicksand, or at least the people that are going this route are doing exactly this. And if you're a newer developer, using these tools, you should be very cautious. It's a seductive trap. You don't know how screwed you are until it's way too late. I've seen this in Parsity, the boot camp that I own. People that reach for AI too quickly and they offload their thinking onto an agent or some sort of large language model tool like Claude or Cursor. Then they sit down at their computer and realize they don't actually know how to code anything. They just know how to prompt. I've talked extensively why this is not a career path or why this is not even a way to get into the industry and have even a short career. It's not really a way to make anything beyond something kind of like a weekend project. And I think we need to get way away from this idea that you can just use AI to scaffold a billion-dollar project or a million-dollar company or even a small SaaS app that has users. I recently did a video on Code Rabbit's report that clearly outlines that AI generated code is almost two times more likely to have critical issues and errors. Please don't use this as a crutch or a way to build actual software. Or if you are, have some guardrails and actual software engineers that know what they're doing to keep things in check and balanced. Now, this is one of those awful trends which I don't really see going away. I think we're gonna have a lot more slop, a lot more bugs, and a lot more outages. And number three is one that I personally felt the coding boot camp collapse. Maybe you're really happy about this. Maybe like, yeah, screw you coding boot camps. And I'd be lying if I didn't say I kind of feel you, right? A brutal number of coding boot camps shut down in 2025. I own a coding boot camp in 2025. Don't hate me too much, right? Now, thousands of people paid money for these programs. They overpromised, they underdelivered, they said you can get six figures from spending three months learning HTML, CSS, and React, and maybe a little bit of Note and Express. Now, this was always a bit of a lie or a dubious kind of myth, but it held up a little bit before the pandemic when actually you could get hired with some of these really basic skills. And even back then, that wasn't really the case. Let's be honest here. But now that promise is undeniably false, right? The bootcamp industry as a whole kind of preyed on people that were looking for an easy way to get into a six-figure job and said anybody could learn to code. Anybody, right? You don't know how to use a computer, you can learn to code. I've worked at coding boot camps throughout my career while being a full-time software developer. I did it as a side gig and now I own one. I see a lot of value in them because people honestly do get hired from them. I went to one and I got hired from it. People need to learn how to build software. So I kind of feel sad that boot camps are closing and people are kind of cheering this on because I'm thinking, well, what happens to the adult that's like 30 and does want to learn to code? And they say, Hey, I'm trying the self-teaching method, which I've met hundreds of people who have tried this and they're just not getting it. Where do they go? Where do they get the mentorship? Where do they go to a trusted program where they can actually learn this stuff? And you can hate all you want and say, oh, these programs suck and they're just for losers or whatever. But at the same time, we need somewhere to teach people, and it can't just be all online. I mean, I think adults deserve more from this, and it's a little bit sad to see basically every large coding boot camp close down. Now, at the same time, part of me is a little bit glad to see this happen because I think some of these coding boot camps did not deliver on their promises and they didn't give people the experience or the education they deserved. We've never really encouraged people to get into any other industry besides tech with this kind of blind optimism. It cheapened the work we do, and I think it set up learners with the wrong expectations. So even though I'm sad to see these go, I do think there is maybe a silver lining or a light at the end of the tunnel or insert cliche there about something good happening after something bad happens. Number two trend that really sucks, and this one hurts me to say also the junior developer market just does suck. Let's just be honest here. The entry-level market got significantly harder in 2025. If you're a senior developer, you probably saw the exact opposite. Despite what the news is saying, senior developers were in high demand. I felt this. Maybe you felt this too. If you're at the senior level, it felt like this was kind of our year. I got more offers than I've ever got, I got more interview opportunities than I've ever got in my entire life. Pretty wild. I hope that doesn't come off as some sort of weird humble bragger. I don't know, maybe it does. But the benefits that perhaps you and I felt as senior developers weren't felt downstream by the juniors. Companies dramatically reduced junior hiring, they raised the requirements during interviews, they shifted towards senior-only roles. Companies basically want people that can come on board and ship with minimal onboarding. Now, I think this is also partly due to layoffs and restructuring and having less middle managers and less people in general. Teams want to do more with less. So they hire a lot of seniors. They hey hit the ground running, here's your laptop, and here's the first feature we're gonna get out this Friday. So, this old model, this traditional model, which is prevalent in almost all industries, of hey, you're gonna learn on the job. We're gonna hire for attitude, and we're gonna hire you for your potential ability to be good on this job. That doesn't really exist anymore, and that's not just in tech. That is a trend we're seeing across industries. So you can no longer just show up with a tutorial project, shallow framework knowledge, like I know React or View or Angular, and a portfolio as your experience. You really need proof now. Deployed apps, you need to have some sort of systems level thinking, you need to understand system design, I think, at a very basic level. The opportunity obviously still exists. Hiring did incrementally go up, and we'll get to that soon. But pretending otherwise, pretending that the market is the same and that your portfolio or a certificate is going to be the way you enter into tech in 2026 is a fantasy. So stop worrying about your portfolios, stop worrying about certificates unless you're in DevOps or something like that, and then focus on building something real. And we'll get to that because I do want to leave some hope for junior developers who are trying to enter into the market. I don't think it's as doom and gloom as people make it out to be. I do think you need a very different strategy. And the number one worst freaking trend, you and I all know this already. We hate this trend, the AI productivity myth, the biggest lie perhaps in tech. A controlled study found that open source developers were 19% slower. I've cited this study many, many times. Code Rabbit released a report at the end of the year saying AI generated code produces 1.7 times more bugs than humans. We've read all the posts from all the AI bros about the prompts and the 10x productivity and the dude that fired his team and has a bunch of sexy prompts. It's nonsense. It's complete nonsense. Teams with good engineering culture, strong talent will benefit from AI. They probably will see modest gains, or maybe not so modest gains. Maybe they'll see some really incredible gains. I think teams that are bad will get worse. AI will speed run them towards failure, just like it can speed run really top-performing teams towards advancement and the gold star. I've already done a video on this exact same subject, this pressure that we've all felt from the top down to not only use the AI tools, but to also unlock unheard of productivity gains. And then we were expected to do this, released a lot of shoddy software. I think this is the reason why we've seen an historic number of outages on the internet this year. And I really hope that developers take back the narrative and say, hey, listen, here's what's possible, here's what's not. We love the tools, we don't like the hype. Here's how we're actually using them, and here's what you can honestly expect to gain from using these tools. Here's how we should, how we shouldn't, the pitfalls, the gains, all the reality of using these tools. I do think we're heading towards that, so I am optimistic about that in the future. Okay, enough with the bad stuff. Let's get to the best trends in 2025 and how you can use this to your advantage going into 2026. Okay, the number four best trend, which I also said was one of the worst ones, is the bootcamp collapse. Now, here's the good part about this, and yes, it's on both lists. This collapse wiped out some of the worst offenders. It also wiped out some really good programs that I am sad to see go. But if you're a program that was selling certainty, like saying three months or a few weeks to$200,000, or some of these cybersecurity boot camps, which were honestly always super scammy to me, some of these ones, I'm glad to see that they are getting wiped out. I think that's good. I think this creates space for honest education, no fake timelines, no job guarantees, no pretending that this is easy. Here's the thing at a fundamental level, education still works and it's necessary. We need to educate people. And I think that's even more important now with all the AI tools on the rise because people can quote unquote learn from AI tools, which are prone to hallucinations, giving you really generic and buggy code. I don't think we want people learning how to code from AI tools when they're objectively worse than human programmers. We still need humans to impart some of this knowledge. And as companies want to raise the bar and only hire people that can hit the ground running, I think it's going to be even more important for people to get education that matters. Now, whether you get this in a college or at a boot camp or whatever the new wave of education will be, I'm not so concerned with that. But the idea that we just don't need education for people is ridiculous. So, yes, it is good to see boot camps kind of fade away in many ways. And I'm hopeful and excited about the new wave of education. And I do believe that the new wave is going to be much more human-centric, which is why at Parsity we're leaning further into personal development, further into human and one-on-one mentorship, because I do see this as a future of coding education and generally for education for adults into the future in 2026 and beyond. Now, number three, really good trend hiring quietly recovered. I'm kind of sick of all the how cooked am I poster? The market is completely cooked. Two years the market sucked really bad, right? It never came back to that pre-pandemic hiring spree, and it will never go back there. That was an unprecedented time. It's just not going to happen again unless we have another pandemic or something like that. And I don't think any of us want that at all. So that insanity of all that overhiring and stuff and giving you$300,000 because you knew how to make a React component, that's not going to come back. Hiring is up 30% from the low of 2023. Companies are hiring intentionally. For generalists, it seems like generalists are really winning in this game, especially for startups as they get smaller, more focused, more lean. More teams need a generalist that knows some AI. And if you're a software developer who's kind of thinking, well, how do I learn AI? What do I do right now? I have some resources in the show notes I think you should grab to teach you the basics of retrieval augmented generation and building agents. I think building agents using retrieval augmented generation, fine-tuning LLM ops, basically observability into large language models, learning how to integrate and use these in a modern full stack web app is going to be a skill that you're going to want to gain in 2026. That is probably the safest bet you can make to be in higher demand as a software developer going into 2026 and beyond. Speaking of AI, because how can we not in 2025? The number two best trend, in my opinion, were AI agents started delivering some real value. Now, the jury's still out a little bit here, but it felt like AI agents stopped being so research and demo oriented, and we actually started seeing companies ship real useful products. We're finally consolidating around some usable patterns for building agents as well: reason and act loops, tool using, multi-agent workflows, reflect and verification steps. I'm seeing more conversations about when to use agents versus workflows, the differences between them, the kinds of patterns that people are using, and the ones that people are actually shipping out into the world that are bringing some actual value. We're moving away from this mental model of agents as human coworkers and more as task doers, which I think is the exact right way to go. I'm making a bet that AI agents are gonna essentially become infrastructure for a lot of companies out there. It's gonna be no different than building an API or deploying something into the cloud. I think there's gonna be a lot of mess in between, but if you learn how to build agents, some of these patterns that are emerging and how to implement them and understand the pitfalls, the trade-offs, the risks, I think you're gonna be in a very good position going forward as more companies adopt boring, stable use cases for AI agents. And the number one best trend was AI coding tools became pretty good. I mean, the AI hype was BS, but the AI tools got pretty cool. Like we really benefited from the tools that came out this year. Can you believe it? Like this year, Claude, Cursor, Codex, they took over the industry by storm, and now they're part of basically all of our daily workflow. It felt like GitHub Copilot times 100 this year, right? The tools are cool. Let's not lie to ourselves here. I don't like the BS, the hype, but I do love the tools, and I think you probably do too, because we see the retention rate on these tools is incredibly high. I think somewhere like 80% retention rate on tools like Claude and Cursor. And the difference between January and December of this year is pretty massive. We got new and improved models, we have patterns that are emerging, we have planning and delegation agents, we have multi-agent workflows. Really, it's become a very, very fun time to write code, in my opinion. Now, to a lot of people, they might miss the old days of hand coding. You can still do that, obviously. But these tools have saved our delicate hands from so many keystrokes: boilerplate, experimental refactors, test scaffolding, documentation. Geez, thank God for these tools for documentation, actually, because that is one of my least favorite things to do. And I love it when I can just say, Hey, Claude, look over this set of files, look over this service, write a small doc about this so I can share it with the team. Tell a non-technical stakeholder what I just created and why the deployment process is hanging up or something like that. This has been a real big save for not just me, but I know for all of you as well. And also that repetitive code, like, hey, break up this React component into a bunch of smaller components here. Hey, make four API routes for some crud operations on this particular model. This kind of stuff is exactly what AI was built for. Still not perfect, but man, does save our delicate little hands, doesn't it? And it's the soul crushing and more boring parts of the job that I can't wait to see automated away more. And despite some of the pitfalls and very real concerns about AI generated code, these tools have massive potential and they've genuinely changed the way we write code. So, what do we do with this information? And here's the practical stuff that I plan on doing and that I am doing. Stop chasing hype. I am as guilty of this as anybody out there. I like shiny stuff, but I think we have to ignore the 10x productivity gains. We have to ignore anybody selling certainty or clarity in an increasingly unstable market. We have to stop looking to social media for hot takes and how to plan our careers and start looking to other developers around us and believing what we're seeing. We have to stop believing the gurus on LinkedIn and Twitter that are selling us stuff and trying to make us believe in a reality that just doesn't exist. I'm going a bit more old school and going further into books and documentation. One of my goals next year is to learn some Rust. And I'm starting with the documentation with the Rust Lang book, which is actually a really good book, if I'm being honest. I mean, it's not like night reading or something like that, but it's not bad at all to read to learn a programming language like Rust. So that is what I plan to do. And I'm spending a lot less time worrying about what all the bros on LinkedIn and Twitter are saying about X technology or the ridiculous claims they're making and having emotional reactions to that kind of stupidness. Now, if you're a newer developer, I think you need to build complex software. What does that mean? Build something that solves a problem for you or somebody you know, and then deploy it to the cloud and maybe even try to charge money for it. Now Charging money is optional. I think you need to have that mindset though when you're building software, if you're doing it to get a job, because people want to know hey, can you hit the ground running? What's the best way to prove you can hit the ground running? Build software for somebody else. No one has to pay for it. But if you build it with the intention to sell it, you're likely going to come across all the more interesting problems that happen when you try to do something that ambitious. I think a little hard work is going to go a long way in an age where everyone is looking for a shortcut. So being just a little bit diligent, taking a little more time in your craft is going to really set you apart in an age where everybody is just trying to vibe code or AI their way into the industry instead of actually putting in the work to get there. Next up, I'm renewing my focus on system design and building systems. Knowing front end alone isn't enough anymore. It hasn't really been enough for a long time now. So I am a full stack software developer, but I want to go deeper into system design. And I mean just doing some basic system design overview. I've done this for interviews, but I want to know how are large companies building systems with AI integrations? What are the pitfalls? How are they building things like RAG at scale? How are they building agents at scale? How are they working with cloud service providers in mitigating the risks of creating so much code with AI? But in general, I think if you just brush up a bit on system design, that's going to be really important heading into 2026. Is more companies are going to start expecting you to know system design, even at the more junior levels. And it's not that hard to pick up. You can just read Alex Zoo's books on system design, they're pretty good. System design one and two, not sponsored. I did buy the books, very good, highly recommend. And lastly, maybe just maybe be a little bit more optimistic about this next year. Last year was non-stop doom and gloom news. This year, I think despite all the bad things happening, there's some really good stuff going on in the industry right now. And if you're considering getting into the industry, I don't think it's as bad as people are making it out to be. But I don't know, maybe I'm way off base here. If you had a trend that you either loved or hated that you hope doesn't go into 2026 or you hope continues into 2026, I'd love to know. Leave a comment and let me know what I missed, what I'm wrong about, and what you suggest people do to get prepared for 2026 because I think we're all anxious about the coming new year. Thanks for listening, and I'll see you around later. 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.