Blog

Beelink SER8, Ubuntu server, Mosh and Tmux

For a really long time I was using the same setup for development.
Basically a macbook and Sublime text. Then using DDEV for Craft CMS sites, and Laravel Herd for Kirby or general PHP/static sites.

But things started to change when I dove into Omarchy late summer 2025.
I really started to play with the idea of having more Linux in my life, and be a bit less locked to MacOS.

Omarchy was a great adventure for me into Linux land, as I got knee deep into all kinds of configs and terminal challenges. I was close so close to feeling like this is the way that I ordered a Beelink SER8 mini-pc.

Long story short the Omarchy phase got me to a point where I felt I was now handling two competing computers on many areas. I still used the mac daily and the omarchy was kind of the dedicated dev-machine, but it being stationary ment it was sort of unpractical when moving around.
I was also experiencing all kinds of frustrating small issues, especially with wifi drivers. At one point I realized what a time-thief the whole project was, and that I couldn't any longer justify all this time spent on it.

So that left me in a situation where I just went back to the macbook again for everything, and the SER8 didn't have a purpose.

Then what sparked a new idea was hearing Scott Tolinsky talking about Tailscale: https://www.youtube.com/watch?v=G0sEM9ijkTE
That seemed really cool, and felt like something to try out. Only for me I'd use the Beelink and Ubuntu Server instead of a Mac Mini, and Claude Code instead of Opencode.

And man do i like this setup! Now I have the Beelink standing right next to the router, and it runs DDEV and Caddy (no Herd for Linux) with the added benefit of it being long running. So I basically have all sites running always here, and via cron scripts it pulls live content down daily so I always have a fresh copy to develop on.

Added benefits include running Plex on it, so I easily can access videos on the Apple TV.

But the last hurdle I had was long running Claude contexts. I basically dev via Zed to remote folder, and kept a terminal open for claude stuff. But when the macbook went to sleep it broke the ssh and thereby also the claude convo. So to the rescue came Mosh and Tmux.
Now with Mosh (ssh replacement more resilient to sleep/network breaks etc) and Tmux (persistant terminal sessions) I truly have long running ssh sessions regardless of network changes or the macbook going to sleep etc. and it's just such an enjoyable setup.

So, this post is an appreciation post to all the mentioned open source projects along the way!

Also really enjoy the Beelink! Fwiw its a Beelink SER8 AMD Ryzen 7 8745HS 64G DDR5 5600MHz + 2TB Crucial SSD for $669, which I really feel is a good price for what you get. Just remember to tune the fan to make it silent.

reMarkable editable pdf

One of the main issues I've had so far, is that I sometimes want to add a file from my mac to the remarkable, and still be able to interact with the graphical elements of it on the reMarkable. Simple example: Adding a pdf page with a circle as a pdf to the reMarkable only makes it a passive background layer by default.

To the rescue is Drawj2d: https://sourceforge.net/p/drawj2d/

What I needed to install to get it up and running was:

• An updated version of Java (brew install --cask temurin)
• Downloading and extracting Drawj2d, then moving the folder to /Applications
• Creating a wrapper script so drawj2d works as a terminal command:

printf '#!/bin/zsh\nexec java -jar /Applications/Drawj2d/drawj2d.jar "$@"\n' > ~/drawj2d_wrapper
sudo cp ~/drawj2d_wrapper /usr/local/bin/drawj2d
sudo chmod +x /usr/local/bin/drawj2d
rm ~/drawj2d_wrapper

After this I can run a command on the pdf file as follows:

echo image pageA4.pdf 1  0 0  0.7 | drawj2d -Trmn      ;# reMarkable 2
echo image pageA4.pdf 1  0 0  0.8 | drawj2d -Trmdoc    ;# reMarkable Paper Pro

The generated .rmdoc file can be drag and dropped into the reMarkable mac app, and now its editable on the device.

Would love to see this being possible in a more native way (like being able to export .rmdoc files from the reMarkable app directly).

reMarkable really is Remarkable

It took me a long time to make the jump, but I finally did, and my god is the reMarkable tablet a joyful product.

My long time colleague and friend have used the devices since they first came out, and have always been raving about them, but I always felt I have enough devices and my needs were met on the note-taking front with pen and paper. However over the last months I've been having issues with organizing all my handwritten notes and notebooks and it came to a point where I re-evaluated the reMarkable again.

Not only is it a pure joy to take notes in, but the whole way or organising it all felt very natural from the get-go. No fuss, no complicated tagging systems to deal with. Just simple folders and files and a clear and thought through way of navigating between it all.

The apps for Mac and my iPhone are also great. Especially useful is it to just scan documents (or notes I've had flying around on paper) directly into the tablet, and continue the document there.

The kids love it too, and I find it uncomparable to other digital devices they use in terms of "screen-time". They draw, colorize, write and most importantly create instead of consume with it.

As I'm quite into chess lately, I also made a little tool for generating chess puzzle pdf's for it, so that I can avoid the glaring OLED/LCD/whatever-screens and enjoy the calmness of e-ink. Also enjoying the absolute abundance of notifications. Link to the tool: remarkablechess.com

Oh, and did I mention it's a Norwegian company? As a norwegian myself it's pretty cool to see it have some success among giants like Amazon and Apple.

Check it out at remarkable.com. The one I got was the Paper Pro. Can highly recommend it.

Wabi-Sabi: On Embracing Imperfection

"In traditional Japanese aesthetics, wabi-sabi centers on the acceptance of transience and imperfection." – Wikipedia

It feels especially relevant now in AI-slop days, but I guess that knife could cut both ways.

What I take out of it is that the imperfections of being human, on having opinions and a unique taste, are immeasurably valuable even though machines arguably can deliver more "perfect" solutions.

Of course one could also make the counter argument that machines and their slop is itself the "imperfections", and that those should be embraced. This doesn't sit as well with me, but who am I to judge.

"Wabi-sabi nurtures all that is authentic by acknowledging three simple realities: nothing lasts, nothing is finished, and nothing is perfect."

Kintsugi: Waste not, want not

I also find the Japanese concept of Kintsugi interesting. Kintsugi is a Japanese art of repairing broken pottery, but where the end goal is not to hide the marks from the repair, but rather to embrace them and rather make it an actual visible embellishment.

Sometimes I feel like my kids see me as a cheapskate when I try fixing their broken stuff. Especially if it's fairly low-cost items from the start. Why bother, right? I wonder if kids in Japan naturally have more respect and understanding for this, or if they are also hit by the wave of consumption-first attitudes of the newer generations. As much as I find it important to teach my kids the value of fixing stuff, I am also aware of how mean kids can be to each other and that the concept of Kintsugi is not exactly ingrained in my kids' culture.

The Optimism Gap

Openclaw is still dominating the tech-news, and the recent acquisition from OpenAI raises a lot of opinions and reactions.

It's not that I don't have my own issues with that change, but I find it more fascinating to follow the whole situation happening in front of our eyes, where Peter fairly obviously feels alienated from Europeans and more attracted to the optimism in America.

Watching his interview with Zeit im Bild, and listening to his talk with Lex Fridman makes it painfully visible that most of the reactions he gets from the European side is criticism, whereas reactions in the US are generally positive.

Sure, there are a lot of aspects that need improvements. Sure, AI in general has a lot of negative aspects that need to be discussed. But there needs to be a certain level of optimism and willingness to look for solutions rather than problems. For opportunities more than risks. For more fun and creativity than limitations and regulations.

I'm not saying we should blindly use anything and everything that comes along as fancy new tools. But I am saying that the mindset and differences in general attitudes and optimism seem to have driven Peter out of Europe, and that is a sad realization.

The acquisition itself might have been inevitable, but the reaction to it says something uncomfortable about how Europe treats its own tech talent.

If we want the next OpenClaw to start here, we need to meet innovation with curiosity before criticism. If we want it to stay, we need to back it up with real support.

All that said, I find it immensely impressive what he has achieved with the Openclaw project.

Overwhelmed by opportunities

Lately I've been getting into generative coding. Stuff like p5.js and just svg's in general. Super inspired by people like George Francis and Daniel Shiffman.

But on completely different arenas I'm also really digging Suno for musical exploration. I've started to play with the endless amounts of half baked voicenotes full of ideas I've had through the years. It's really amazing to be able to give them more life with Suno instead of them just rotting in the voice notes. Very inspiring.

The last half of 2025 I was also really into Midjourney as a creative outlet.

Claude Code has become a huge part of my everyday life while coding the last 2-3 months. Claude makes things that I would've never dared to try before become realistic and solvable, which is amazing.

Each of these tools feels like a new door opening. Everything all at once.

But my problem was never lack of inspiration or ideas.

My weakness is focus and execution.

I feel simultaneously ever more inspired in all different creative directions, but also so scattered that I'm more stressed out than before.

I'm standing at a train station with a free ticket in any direction, but they're all interesting! And I'm standing there with massive decision paralysis.

I've always struggled with procrastination, but now more than ever.

Early mid life crisis anno 2026.

Coloki – A simple color palette generator

I built a simple color palette tool inspired by Flexoki by Steph Ango

It's available at coloki.vasser.no

I kept reaching for the Flexoki playground but wanted something simpler.

In it you basically just define a primary color, and it will generate a few darker, and some lighter colors for it.
It started with a very simple, almost fade-like solution, but with that you lose a lot of the saturation.

So there are some adjustments for retaining the saturation in the brighter colors.
But on the darker end, its rather opposite. Here it makes sense to desaturate a little bit.

The extreme ends (brightest and darkest) can't really be treated the same as the rest, so they have slight adjustments making them less affected by the sliders than the more mid-rangy colors.

Anyway. It all kind of makes sense for me, and feels useful, so I just put it out there in case anyone else finds it useful too.

My first Kirby template / starter-kit

This web site is now available as a Kirby template / starter kit.

Files can be downloaded either here or directly at Github.

Why release it as a template?

Learn to ship

Finishing something that isn't a client project has always been a hurdle for me. Since it's not paid work, it tends to end up last in the priority queue.

Learn to make fundamental decisions and stick to them.

Interestingly when packaging something for public release, I reevaluate a lot of decisions I make automatically.
For my own site, I could easily just use Composer and add in fancier SEO-tools etc.

But by deciding to make it a template, I ended up reading closer at Kirby Docs and their focus on lowering the barrier and accepting multiple ways of installing plugins.

I find that a very worthy pursuit, and made it a clear goal to not have any build tools in use. Also not to rely on plugins that likely will become paid in the future (like the SEO plugin, which is currently free, but will become a paid plugin).

Who is it for?

Me. As far as I know, I'm the only one using this template. Maybe someone else will in the future, or maybe not. But it's not any goal of mine to have x amount of installs. I make this for myself, and do it as a learning journey.

Here I draw inspiration from Rick Rubin, who believes creators should focus on making things that resonates with their own truth, not trying to please an audience.

What is Kirby?

Kirby is a really cool CMS. It's a bit overkill for the needs of this blog, which could have easily just been a SSG or so, but I still chose it because I also use it on other sites I make, and have a little bit of an allergy towards build tools.

Even though I use a VPS and Laravel Forge and easily could have a more advanced server setup, I really love that Kirby can essentially just be dropped onto shared hosting and work. And not just work, but work flawlessly both front and back-end. That is not the case of alternatives like Craft CMS and Statamic, although I also really like these systems.

Why still make things yourself?

Now that AI is getting so good, why are you even trying to do the work yourself?

Everywhere I look, tools promise to replace effort:

Fire your photographer
Create apps and websites by chatting with AI
Never start from scratch

And honestly, the results are impressive.

Sarcasm aside, it feels slightly surreal to watch the design and tech landscape lately.

There’s absolutely no shortage of impressive things popping up: graphics, music, full websites, even entire brands created in minutes. I’ve had my own revelations using tools like Claude Code. I don’t see AI as a villain. In many ways, it's helped me work faster and think differently.

But still… something feels kind of meh.

I think it's because we can tell when something was made with little effort, even if the result looks good. And for some reason, that's off-putting.

It's like we have a built-in sense for the relationship between effort and outcome. When they don't match, the work feels empty.
I have the same issue on stuff I build myself. Things I would have been proud of a few years ago now feel less impressive, simply because I got help from AI along the way. Maybe it's just self-defeating and nonsensical, but I can't shake the feeling nonetheless.

I see a similar thing happening at the Christmas markets here in Germany. A few years ago you'd find small handmade items, and you instinctively knew the person behind the counter made them. Fast forward to Christmas 2025: the stalls look the same, but we all know the items are mass-produced and shipped from China. They look handcrafted, but they don’t feel handcrafted.

Which makes me wonder:
Is the era of pure craftsmanship coming to an end?

As a web designer, I now compete with anyone who can open Lovable and prompt a website into existence. Sure, I can argue that my work is more thoughtful, more scalable, more maintainable… but the reality is that many clients don't care as much as I'd like them to.

And yet I can't shake the feeling that there will be a counter-reaction to all of this.
A moment where craft becomes valuable again precisely because everything is instant and effortless.

Right now, learning to code from scratch is probably less tempting than learning to prompt. But in a world where everything is mass-produced in seconds, I genuinely believe that the desire for real understanding, real skill, and real craftsmanship won't disappear. It might even become more valuable.

But right now, I really feel for everyone still doing their craft by hand.

It’s harder than ever, but maybe still worth caring about?

Fizzy from 37Signals out in the open

This article is not about Fizzy as a product, but rather about the decisions 37Signals make regarding how to launch their product.

First off, they have been working very openly on this project for a long time now. The founder, Jason Fried, have made multiple videos along the way showing WIP stages. I've found them very interesting to follow. Jason Fried has done this for many years also on other products, but it feels like he made it even more a focus while developing Fizzy.

Anyways, today they launched the product, and a few things stand out:

• It has a very generous free tier, and no card required.
• It is open source, so you can optionally run it yourself.
• They released the full codebase including all commits

The last point is actually the most interesting to me.

I mean, its almost 8000 commits, and it starts with "New Rails app" in june 2024. The whole process from there shows a lot of interesting decisions along the way. Combined with the process videos from Jason it really shows how 37Signals work.

That's pretty cool!