Libraries & tech stack I'm excited about (for 2026)

00:00:00: Obviously, it's all about AI these days

00:00:04: for me, I use AI all the time. I write a lot of code with

00:00:08: AI, I use tools like Cursor and so on, but in this episode, the focus is

00:00:12: not on AI. Instead, I quite often get the question, which

00:00:15: technologies should I learn in 2026 or which technologies are

00:00:19: you, so I, using in 2026? And that's what I wanna

00:00:23: share here. I wanna share some technologies

00:00:27: frameworks I'm excited about in 2026.

00:00:30: And this list is not necessarily comprehensive or

00:00:34: complete. Instead it's just a list of technologies I

00:00:37: personally will be using this year and I'm very interested in

00:00:41: this year. And I've talked about it before, there is,

00:00:45: course, kind of the default stack if we're talking about web development,

00:00:49: which I am, which dominates everything, especially with

00:00:53: AI or due to AI. And that is, TypeScript,

00:00:56: React, Next.js, and Tailwind. That is

00:01:00: AI's favorite and you can see that

00:01:04: in the download charts of React for example, and it's the same for

00:01:08: Tailwind and the other technologies.

00:01:10: And obviously that is, therefore, what you should definitely learn or use,

00:01:14: especially if you wanna find a job.

00:01:16: It's not the only choice, of course, and I'm not saying Angular, Vue,

00:01:20: stuff is bad and it can be a good niche to, for example, dive

00:01:24: deeper into Vue, but that is AI's

00:01:27: favorite, that is the favorite stack of AI.

00:01:30: So naturally, I'm using some of these technologies also

00:01:34: so before AI. I'm using TypeScript for all my projects

00:01:38: because TypeScript is amazing and yeah, it can sometimes be annoying

00:01:42: to define more complex types, but for one, you don't need to do

00:01:46: that that often, to be honest. And two, AI can

00:01:50: help with that. AI is pretty good at writing TypeScript code

00:01:54: advantage of using TypeScript before AI was already that it

00:01:57: can help you catch certain errors and can lead to

00:02:02: higher quality code in the end, but with AI especially, the

00:02:05: advantage of using TypeScript is that it's one important building block

00:02:09: can give AI useful feedback on whether the code it wrote is

00:02:13: valid or not. Combined with unit tests for example,

00:02:17: or with giving AI browser access with the Playwright

00:02:21: MCP or the built in browser in Cursor and so on.

00:02:24: So TypeScript is something I always used, I always liked way

00:02:28: before AI but now with AI it's even more useful, so it's part of my

00:02:32: tech stack too. The same is true for React.

00:02:34: I also love Angular, I love Vue. I have courses on both and if

00:02:38: you've followed me, you know I love all these frameworks

00:02:42: has only one favorite framework but over the last years, it's been

00:02:46: mostly React that I used for all my projects and especially with

00:02:50: AI and with AI knowing so much about React, though I will

00:02:54: say it can produce quite garbage React code too with

00:02:58: bunch of use of fact and so on, but with all that, React

00:03:02: is still part of my tech stack.

00:03:05: Next.js has been my main React

00:03:09: meta framework for a very long time and it's still great.

00:03:12: I wanna be very clear here. It's still great, but that's the first

00:03:16: new thing maybe, I really like TanStack

00:03:20: Start. I've built projects with it, like this Build My Graphic

00:03:23: project I built as a side project last year and then I released it.

00:03:27: It's a project that simplifies the process of generating

00:03:31: infographics with AI like this one here, which

00:03:34: shows the finished overview of my favorite tech

00:03:38: stack and I used TanStack Start for that and I'm also

00:03:42: using TanStack Start for other projects I'm working on.

00:03:46: I'm, for example, also using it for this canvas drawing app

00:03:49: you're seeing right here and it's amazing. I really like it.

00:03:53: I like that it's faster than Next.js during development, that it

00:03:57: does certain things differently. I like its syntax.

00:04:00: I like the simplicity compared to some of the Next.js stuff.

00:04:04: There are no use server, use client and so on directives and therefore,

00:04:08: TanStack Start is definitely one of the new things

00:04:12: I started using last year and I will continue using this

00:04:16: year. Now I typically use TanStack Start with

00:04:20: Bun for the backend code and in general I use Bun for installing

00:04:23: packages, for running my code, also the dev server because it's

00:04:28: just a faster Node.js. I mean simply using it for installing

00:04:31: packages with Bun install is faster than NPM and I really

00:04:35: like Bun. It is, of course, Node.js compatible for the most

00:04:39: part and it has a bunch of extra things built

00:04:43: in that can be super useful, like the built in S3

00:04:46: client which allows you to interact with the S3 object

00:04:50: store and compatible stores like R2 by Cloudflare to store

00:04:54: objects, images, whatever it is and that's built in,

00:04:58: package. It's super fast, it's faster than the official AWS SDK, for

00:05:02: example and yeah, for all these reasons I really like

00:05:06: Bun and therefore Bun is pretty much my default too in

00:05:10: all the projects I'm working on. It's of course not new and it's also not

00:05:14: new for me. I've used it last year too and even before that but

00:05:18: it is my default now so that's definitely also something I'm using a

00:05:22: lot.I wanna point out though that Node.js has come a

00:05:25: long way too. It's easy to dismiss Node.js as kind of

00:05:29: and old and Bun as being much better, and I do think

00:05:33: Bun is better in many regards. But I wrote an entire article, and you

00:05:37: find it linked below in case you wanna read it,

00:05:41: TypeScript support, native TypeScript support that are built

00:05:45: into Node.js that you can use today with the latest versions of Node.js

00:05:49: for which you don't need extra packages or Bun because Node can do

00:05:52: more than many of us think it can. Am I still

00:05:56: using Bun because I like its speed and so on? Yes, I am.

00:06:00: But if you're in a project where you need to use Node or you just prefer

00:06:03: it, it's worth noting that you can do much more nowadays with

00:06:07: than a couple of years ago. So modern Node.js is

00:06:11: definitely also something I would consider taking a look at again this year if you

00:06:15: haven't in a while. Speaking of modern things, I'm also super

00:06:19: excited about many modern browser APIs and CSS features

00:06:23: that are available nowadays because a lot has changed there over the

00:06:27: last, I guess, 18 months or so. There are a

00:06:31: lot of browser APIs and CSS features like, for example, the

00:06:34: popover API, but also many others that have reached

00:06:38: broad major browser support so you can use them in all those

00:06:42: major browsers today. You can use them in your applications

00:06:46: of JavaScript code or extra packages you would need to

00:06:50: otherwise by using these features.

00:06:53: Now, I created some videos on my other channel,

00:06:56: channel, where I dive into some of these things

00:07:00: APIs, and I definitely recommend taking a look

00:07:04: these videos and these features in general.

00:07:07: The browser has come a long way and there's no need to solve a lot

00:07:11: of problems with your own code or extra packages that can be solved

00:07:15: just by leveraging these new features.

00:07:17: And it's worth noting that AI by default might

00:07:20: not suggest or use these features because, of course, it was

00:07:24: trained on data that had all the old code but

00:07:28: you can teach it. You can share the documentation, copy and

00:07:32: paste it into your prompts or share links depending on

00:07:36: using, and then you can have AI write modern code

00:07:40: and that is absolutely something I would recommend

00:07:44: doing, that you teach AI all that

00:07:47: modern stuff. You really should do that because there's

00:07:51: so much you can do here. Now, speaking of AI, of course,

00:07:55: we can't entirely skip it in this video and I've said it

00:07:59: before in other episodes. I'm using AI a lot.

00:08:03: I write all my code with AI and AI writes a lot of

00:08:07: code for me. Now, not talking about vibe coding here.

00:08:10: I wanna be very clear and I know some people will still get it

00:08:13: I'm talking about controlling the AI, steering it, providing the right prompts,

00:08:17: right context, getting into the code yourself

00:08:21: to and really doing agentic engineering as it's called.

00:08:24: And therefore, of course, I work with tools like Cursor.

00:08:27: And one tool I'm very excited about this year is OpenCode, for example.

00:08:31: There has been a lot of hype about it recently.

00:08:33: I started using it back in December so I was cool before it got

00:08:37: cool. And this is essentially an alternative to

00:08:41: It's a CLI tool, a terminal-based coding tool, and

00:08:45: you can use it to do agentic engineering and it's very nice, has a nice interface

00:08:49: even though it's in a terminal, has a lot of features,

00:08:52: like agents and agent skills, and yeah, you

00:08:56: can really do great agentic engineering with

00:08:59: OpenCode. And therefore Cursor and OpenCode, for

00:09:03: example, are other technologies I'm using this year

00:09:07: is one I'm very hyped about. Now, there are two things most

00:09:10: applications need that you might be building,

00:09:15: a database. And for the database, my favorites are PostgreSQL or

00:09:18: SQLite. Both are amazing and both are of course not

00:09:22: new. I'm totally aware of that and I know there are more hyped

00:09:26: alternatives like Convex and others and I might look into

00:09:30: those too. I haven't really used them to be very

00:09:33: But I like these old school databases for very different

00:09:37: reasons. Postgres, of course, is very, very popular.

00:09:39: It's super performant. When using AWS, for example, you

00:09:43: can use managed services like RDS or Aurora to make

00:09:47: hosting it simple if you want to, but you can also self-host it with

00:09:50: Docker. Just need to make sure that you back up stuff and so on.

00:09:54: And what's great about Postgres, e- especially now in this

00:09:58: modern age is, of course, that it supports all those

00:10:01: different extensions like pgvector if you wanna have a

00:10:05: vector database for building a rag system.

00:10:07: And extensions have been there forever, that's not new, but now in the AI

00:10:11: age we got new stuff like vector databases and you can simply do

00:10:15: that with Postgres without using extra tools or paying for extra

00:10:19: services. And that's one of the reasons why I really

00:10:23: for many of my projects. But for some projects where I wanna keep

00:10:27: simpler, where I maybe don't need those extensions,

00:10:31: And in case you didn't know, SQLite is really just a file in your

00:10:34: project. The database is embedded in your code,

00:10:38: mean. So there is no extra database server you

00:10:41: Instead it's just an extra package you install or if you're

00:10:45: using Bun you don't even need to do that.

00:10:47: It has built-in SQLite support so you can just use Bun without

00:10:51: installing anything and you can start writing to the database

00:10:55: just a file in your project. You can back up the database by just

00:10:59: copying that file somewhere else and that makes it super simple to get

00:11:02: started. And this might sound like something that's

00:11:06: or suitable for major projects but you would be

00:11:10: wrong. SQLite can scale to-... tens and

00:11:13: hundreds of thousands of users and thousands of concurrent

00:11:17: requests, so it's super scalable and powerful.

00:11:21: Of course, at some point, you might need

00:11:24: more than it, but that point is pretty far away and most

00:11:28: applications never reach it, to be honest.

00:11:30: And then, this simplicity really is worth a lot.

00:11:33: And it's easy to think of it as a toy database but it has lots of

00:11:37: features built in. It's an amazing database

00:11:40: production projects, like Build My Graphic, for example.

00:11:44: It uses SQLite, and therefore, that's another technology I've been

00:11:48: using for the last years and I'll keep on using, and I'm,

00:11:52: still very excited about. And I (laughs) just wanna do some

00:11:56: here because many people think of it as a development only database or

00:12:00: anything like that, and it's absolutely not.

00:12:02: Now, for authentication, it clearly is BetterAuth for

00:12:06: me. Now, there are many ways of adding

00:12:09: and of course, in case you have noticed, I'm talking about the JavaScript ecosystem

00:12:13: here. But BetterAuth is a relatively new library.

00:12:17: I mean, it's over a year old already, but it's still not super old.

00:12:21: It's extremely well-maintained and extremely powerful.

00:12:24: It allows you to roll your own authentication without writing any code

00:12:28: related to it. You just use this package

00:12:32: the auth providers it supports, and it makes authentication

00:12:36: a breeze. It's the default for all my projects.

00:12:40: I'm using BetterAuth for all of them, and it's just amazing.

00:12:43: And I'll keep on using it. I'm still super excited about it,

00:12:47: recommend taking a look at BetterAuth if you haven't already.

00:12:50: So, let me quickly create that infographic I showed

00:12:54: you. And whilst that's working, let me be very

00:12:58: clear. There's, of course, way more interesting stuff happening, way

00:13:01: more packages that are worth exploring and that I'll be using.

00:13:05: For example, if I'm building mobile apps, I'll use React Native.

00:13:08: This is just a selection of the libraries or

00:13:12: technologies I'm most excited about that came to

00:13:16: my head now that I thought about it.

00:13:17: There is absolutely a possibility that I forgot something

00:13:21: or that something will, of course, emerge over the course of the year

00:13:25: already exists but I haven't really used it thus far, and of

00:13:28: course, I'm interested in hearing your thoughts

00:13:32: excited about, maybe outside of AI.

00:13:35: And obviously, AI dominates everything and will have an impact on everything,

00:13:38: but it's worth keeping in mind that we, as developers,

00:13:42: can and should still make choices about the technologies we use and

00:13:45: we let AI work on. And yeah, that is a

00:13:49: selection of the stuff I'm most excited

00:13:52: about.