All of the changes I've made to my tech stack in 2025

We're quickly coming up to the end of 2025 and this is the time of year where I start to reflect on everything that's happened throughout it. Uh specifically when it comes to some of the choices that I've made. This year in particular has been absolutely eventfilled, especially when it comes to tech and software development. And for myself, there's been a number of big changes on that front. One of these changes has been my preferred tech stack, which I actually did a video about earlier this year. In that video, I went over what my chosen tech stack was for the second half of 2024 and the earlier part of this year. [snorts] And whilst I did have a few things that I planned to change, it ended up being more than I anticipated. This is because both that video and my tech stack at the time were defined in a period I like to call BCC, aka before Claude Code. Since Claude Code found its way to being installed into my system, however, I'm not entirely too sure how that happened. my chosen tech stack began to change in a different direction. And whilst some of the changes have been somewhat AIdriven, a decent number of them have actually ended up being more quality of life improvements with a couple being quite surprising. Therefore, I thought it would be a fun idea to do a more reflection video talking about everything that's changed in my tech stack throughout 2025. The first big change is one that I was sort of expecting going into the year, uh, but ended up going in a different direction than I originally thought it would. This change is when it came to my chosen programming languages. In case you're unaware, around the earlier part of 2024, I decided to only use a single language for an entire year, which also fell into the first couple of months in 2025. That language was Go, which is still my chosen language when it comes to back-end web services and building command line applications. Originally, however, I thought I was going to start picking up Rust again in 2025. Uh, one because I just had that itch and two because I was excited to start learning and trying out some of the Rust front-end frameworks such as Leptos and U. As it turns out, however, because of clawed code entering the scene, this didn't end up being the case. And instead, I decided to go with another language, one that I haven't really used that much in the past, TypeScript, which honestly I've started to appreciate more than I ever thought I would. Now, whilst clawed code was a big reason as to why I decided to pick up TypeScript, one of the more minor ones was because the TypeScript compiler is now being written in Go, which as a Go developer made it feel kind of appealing. So, I decided to pick it up. Uh, but what am I using Typescript for? Well, as it's probably no surprise to anyone who works as a software developer, I'm using it mainly for front end. Although, I do want to try some back-end development with it, uh, using something such as Hono, which I keep hearing good things about. So, maybe that's something I'll try in 2026. And hopefully, I'll also pick up Rust once again. As for 2025, however, I've been exclusively using TypeScript on the front end, specifically when it comes to my chosen front-end framework of choice. Next, JS. I know you just hate to hear it, but for some reason, I can't get away. Now, I did play with both SolidJS and Solid Start a little bit earlier on in the year, but for some reason, I just keep coming back to React. In my case, I like to use Nex.js JS as a backend for front end which not only gives me the ability to use React quite easily uh but it also provides serverside rendering which is something I actually really like when it comes to web development. In fact, this is why I originally gravitated towards Go with Temple Alpine and a dash of HTMX as it allowed me to have serverside rendering in a context I was familiar with. Unfortunately, whilst I did love that front-end stack, it just wasn't the most productive. For starters, LLMs are ridiculously good at writing React, for better or worse. Although, interestingly, they're not so good when it comes to Next.js. Not unless you don't mind use clients being littered everywhere. Perhaps more importantly than LLM compatibility, however, is the total ecosystem, which again, for better or worse, React and Next.js are just second to none, except for maybe Go on the back end, which makes using both of these a really strong combination. As for my favorite packages from the NextJS/Ract ecosystem, these include the fantastic component library, Shadcen, of course, Frameotion, which if you've seen any of my videos, you know I love a good animation, and perhaps my favorite integration of all, one that's modified another area of my chosen tech stack, Orth or for me in 2025 has been a bit of a roller coaster to say the least. I originally started out by using my own home rolled OR solution. Uh but thanks to anxiety, I was spending a lot of time worrying about all of the possible edge cases rather than just focusing on the core of what I wanted to build. Therefore, because of this, I decided to make the executive decision to migrate over to another solution in order to save a bit of time. That solution ended up being Clark, which I also learned is kind of controversial in the YouTube space. Despite this, however, it ended up being a really good timesaver and was definitely a decision I was happy with. Of course, I knew that this was only going to be temporary, and eventually I would want to migrate over once my course platform was fully launched. So, once it was, I decided to try out a couple of other orth providers and ended up using one that was not only recommended to me in that video's comments, uh, but also ended up becoming my favorite orth solution ever. Better, which is an open source or package for TypeScript, one that happens to be extremely extensible, uh, thanks to its fantastic plug-in system. In fact, better or ended up being so good that at one point this year, I almost considered changing my backend language from Go to just full-on dual wielding TypeScript. Fortunately, I didn't have to make this change thanks to better or plug-in system, which allow me to use it with a Go backend through the use of JWTs. Uh, by the way, if you're interested, I have yet another video that goes into more detail on how to do this in the description down below. In addition to Better Orth's plug-in system, one of my other favorite things about it is the fact that it works with pretty much any database driver you might use, making it a really robust solution for your preferred database stack. Speaking of databases, my go-to is still Postgres, and I don't think that's going to change anytime soon. However, one thing that has changed is that I now use an OM, which is definitely a departure from where I used to be. Now to be fair, I don't use an OM that much. In fact, I only really use it when it comes to working with TypeScript on Next.js, uh, specifically with better orth when it comes to building backends with Go, I'm still a massive fan of using SQL C. As for which OM that I'm using on TypeScript, however, well, in this case, it's Drizzle, which for me feels like a really nice balance between an OM and just directly using SQL. As for actually deploying Postgres, well, in this case, I'm still using Neon, who full disclosure, I do have a working relationship with, uh, but one that I actually sought out because I both use the product and I really enjoy it. This is also something that I've been looking to change throughout 2025, making sure that I work exclusively with sponsors that I actually use. Speaking of which, whilst Neon isn't sponsoring this video, there is another brand that I really like who is one that's great for learning the two languages that I now use. Boot.dev. Boot.dev is an online platform that will teach you the skills to become a back-end web developer using either TypeScript or Go. However, unlike traditional e-learning platforms, boot.dev does this by trying to make the whole experience enjoyable rather than being just a grind. They achieve this by taking concepts from role- playinging video games. Meaning that whilst you're learning to code, you'll also be gaining experience points, leveling up, obtaining achievements, and competing for first place on the global leader board. If that sounds too good to be true, then fortunately, you don't just have to take my word for it, as all of the material on boot.dev is free to both watch and read. However, if you do want to make use of the interactive features, such as code challenges, AI assistance with Boots, the magical wizard bear, and all of the game-like behaviors that I talked about before, then you can get 25% off your first purchase by using my coupon code, dreams a code when you check out. That's 25% off either your first month when you choose a monthly subscription or your first year if you go with an annual one, which is honestly really good value. Not only that, but if you find that the platform just doesn't work for your style of learning, then there's no risk as boot.dev offer a no questions asked refund for the first 30 days. So to try everything that Bootdev has to offer, then make sure you use my link in the description down below and use my coupon code dreams of code when checking out to get that 25% off your first purchase. A big thank you to boot.dev for sponsoring this video. Okay, so perhaps one of the biggest stack changes that I've made this year has been one that's allowed me to buy back quite a bit of time. [snorts] This is a change to my go-to payment provider, which used to be Stripe, but is now Polar.sh. This was a change that came about due to my own incompetence when it came to filling out tax returns, specifically when it comes to VAT in both the EU and UK. I've done a whole video talking about this change in more detail, but the general idea is that I moved away from having to handle these sales taxes myself to instead just using an MO or merchant of record, which handles all of that for you for an additional fee. For myself, that fee is only an extra 1.1% and 10 cents extra per transaction. So, whilst it's not nothing, at the scale that I'm doing, it's really not that much. And given the amount of late fees that I kept having to pay due to my own affforementioned incompetence, I honestly wouldn't be surprised if I was breaking even. For me, migrating to Polar has been a really good decision as it solved a few key issues I had when it came to filling out VAT returns. That being said, however, I do have a small proof of concept that I'm currently working on to effectively solve filing of VAT returns for both the EU and UK without needing to use an M. So, if it ends up working out, then there's the potential for me to move back to Stripe. But that's not likely going to happen until some point in 2026. However, if it does, I'll be certain to let you know. In addition to payments, another proposed change that I mentioned in my earlier text video was migrating away from my transactional email provider, Mail Gun, which I've been using off and on now for almost a decade. Whilst Mail Gun is absolutely fine, the problem I had with it is it's kind of expensive, coming in at $15 a month. Whilst this doesn't sound like much, it was probably the most expensive part of my actual infrastructure stack, especially as it only allowed me to have one sending domain. Now, Mail Gun does have a free tier, which allows you to send up to a 100 emails a day. Uh, but I found I was constantly exceeding this, especially on days there was a lot of traffic. So, [snorts] early on in this year, I decided that I was going to migrate to another provider. And originally I thought this was going to be AWS SCES aka simple email service which is incredibly affordable when it comes to sending transactional emails. Surprisingly however whilst I did end up migrating my transactional email provider I didn't end up going with SCS. Instead I ended up migrating to another service one that is actually more expensive than Mail Gun, Resend, which now I pay $20 a month for. Not only is Resend more expensive, however, fundamentally it's just a wrapper around SCES, which begs the question, why did I migrate to Resend? Well, honestly, this was 90% to do with developer experience. Resend is just extremely nice to use, especially when compared to AWS. Not only is it really easy to get up and running with, uh, but it also has a lot of integrations to other services. However, [snorts] the additional 10% reason as to why I decided to use SCS is that they allow you to configure up to 10 sending domains for the $20 a month plan. Another associated reason as to why I'm using Resend is for React email, which if you haven't used yet is perhaps one of my favorite things that I've discovered this year. I have a video plan to talk more about it in the near future. Uh, but for the meantime, it basically allows you to style and configure an email using React, which is absolutely amazing compared to the horror that is styling an email traditionally, unless you're one of like the five people in this world that loves working with HTML tables and inline styles. Now, to be fair, you can use React email with other providers, uh, which is pretty cool. And for me, it's likely going to be needed in 2026 as the current issue that I have with Resend, other than the price, is the fact that I'm starting to exceed those 10 sending domains. So, unless I want to upgrade the $90 a month plan, which I absolutely do not, then I'm going to have to go back to my original plan to migrate to SCES. So, there's a strong chance that that's going to happen in 2026, unless of course Resend want to reach out and maybe we can strike up a good deal. Okay, so the last big change that occurred in my tech stack throughout 2025 has been when it comes to application deployment. This change was entirely driven by the rise of Agentic AI. As by using it, I realized that my setup of Docker Stack was missing a key feature that I needed when it came to the new meta of software development. This feature was preview deployments, which is where you're able to review the changes that a pull request will make before merging it into your application. Unfortunately, [clears throat] because of my Docker Stack setup, I didn't have this feature enabled, and it was something that I really needed in order to improve my workflow uh when it came to using tools like claw code. Therefore, I needed to find another option when it came to application deployment. Believe it or not, I'm actually a huge fan of running Kubernetes, which are pretty easy to get preview deployments up and running with, especially if you use something like Argo CD. However, whilst I do love Kubernetes, I will confess that it can be quite difficult for a lot of people to understand. And because I like to broaden my horizons and try new things, I decided to look elsewhere. So, initially I began using another service uh railway, which provides a more consumer-like version of running kubernetes uh which personally I kind of liked. Unfortunately, however, whilst railway is fun to use, it's not the cheapest thing in the world. And as my application usage was starting to scale, so too were my costs, especially when compared to just using a VPS. Therefore, I decided to stick with my trusty virtual private server, uh, but look for an option that provided preview deployments that wasn't going to be K3s and wasn't going to be Docker Stack. After trying a couple of different platforms, I ended up discovering Dock Ploy, which has pretty much met all of my needs uh when it comes to application deployments. Of course, it's not perfect, and I've had to build a couple of additional tools and pipelines to get it to where I want it to be, both with review apps, uh, but also with my secrets management platform, Doppler. However, now that I have all of that in place, I'm pretty happy with it. Although, I'm not too happy about the license that it's using. However, it's no worse than something like Railway or VEL. As for the rest of my infrastructure stack, I plan on doing a couple more videos about it in 2026, uh, talking about things that I use for proxying, uh, but also my secrets management solution, Doppler. Additionally, I've also got a couple of HomeLab videos coming up as well, with one talking about how I managed to keep my home lab always online and how it compares to just using a VPS in the first place. In any case, that covers some of the major changes that I've made to my techstack in 2025. And whilst there might be some small tweaks coming in the near future, for the most part, I'm pretty happy where it is. In fact, I'm so happy with it that I've actually been building something to make deploying applications that use my chosen tech stack as quick as possible. Zenstart. This is my new starter kit SAS application, which I know starter kits tend to have a bit of a negative connotation. Uh, but this one I've really been building to focus on solving a lot of my own problems when it comes to deploying a new project and having to deal with boiler plates. For myself, I really wanted a solution that at a click of a button, it would be able to create a project with everything I need configured such as orth admin, payments, UI components, SEO, etc., etc., using my favorite packages. In addition to this, however, I've also been getting it to work with some of my favorite integrations, including automatically creating a new database on Neon whenever I generate a new project, as well as also creating a new project with my secrets on Doppler with the database URL from Neon already populated. So far, I've really been enjoying using this kit, and it's allowed me to start new projects quicker than ever before. At the moment, it only has the starter template for Next.js, JS uh but I'm starting to add one in for my go back end as well uh making use of better orth and JWTs. If all this sounds like something of interest to you, then there's a link in the description down below to either check it out or sign up for a wait list and get an early bird discount uh depending on when you're watching this video. In any case, that's all from me and all of the changes that I've been making in 2025. As always, however, I want to give a big thank you to you for watching and I'll see you on the next one.

Some of them have been somewhat surprising... Click this link https://boot.dev/?promo=DREAMSOFCODE and use my code DREAMSOFCODE to get 25% off your first payment for boot.dev. Whilst some of these changes have been brought about by perhaps the biggest change to the way I code, a good number of them have been quality of life changes. A section of this video is sponsored by Boot.dev Mentioned Videos: - Original Tech Stack Video: https://youtu.be/XbN66h912NU - Better-Auth + Go: https://youtu.be/dNY4FKXwTsM - Stripe to Polar: https://youtu.be/zM9k0WI1FrM Join the waitlist for Zenstart at: https://zenst.art My Gear: - Camera: https://amzn.to/3E3ORuX - Microphone: https://amzn.to/40wHBPP - Audio Interface: https://amzn.to/4jwbd8o - Headphones: https://amzn.to/4gasmla - Keyboard: ZSA Voyager Join this channel to get access to perks: https://www.youtube.com/channel/UCWQaM7SpSECp9FELz-cHzuQ/join Join Discord: https://discord.com/invite/eMjRTvscyt Join Twitter: https://twitter.com/dreamsofcode_io