Adam Freisinger

Building a Home for My Stray Ideas

2026-03-21T00:00:00Z

I notice things. A weird-sounding word in a book I'm reading. An urge to understand how drone flutes work. An odd graffiti that keeps popping up in my neighbourhood. These aren't plans or tasks. They're fragments. And the part of my thinking that I really enjoy happens between them, in the sometimes accidental, organic connections that I make as I ponder about them.

I have been taking notes for as long as I can remember. Yearly bullet journals, Evernote, Obsidian, Bear, Apple Notes, and I even have a Johnny Decimal index. For the structured parts of life, these work. But for creative fragments, reflections, stray observations, nothing held. Every system either asked me to organise a lot (my brain ran faster than my organisational patience, so the spark would be gone) or let me dump things in and never surface them again.

About six weeks ago, three tools I was already using fit together in a way I didn't plan. Wispr Flow for voice dictation. Claude Code for building things through conversation. And an Obsidian vault with Smart Connections, a plugin that vector-embeds notes and exposes them to AI tools through MCP. I could dictate a thought, and Claude would find how it connected to things I'd written weeks ago. Capture and connection happened in the same place for the first time. That felt like something.

From my desk to my pocket

I started using the combination and building on it at the same time. Voice in, structured fragment out, a link to an older note I'd forgotten about. That loop was exciting. But it only worked at my desk. Obsidian had to be running, Claude Code had to be open, and I had to be physically there, while I kept having ideas away from it.

So I started building a solution I could reach from anywhere. This was not a weekend hack. It took about two weeks of focused work, dozens of architectural decisions, a lot of iteration and rethinking. But I could use it from day one, even while it was rough, and that kept the momentum going. I call it ContemPlace.

ContemPlace captures idea fragments in my own voice through a Telegram bot. I send a message, sometimes a photo, and the system structures it: title, body, tags, links to related notes. Everything gets vector-embedded for semantic search. My raw words are always preserved alongside the structured version.

Overnight, an automated gardener runs. It compares every fragment against every other by embedding similarity, creates connections I didn't explicitly make, and groups notes into thematic clusters using Louvain community detection. The next time I or any AI tool explores the knowledge base, the graph is richer than what I put in. The system speaks MCP, so my notes travel with me across tools. They aren't trapped in any vendor's memory.

Böhönc

Here is what it actually feels like when it works.

I'm browsing the internet and I spot the word "böhönc." Hungarian for ancient trees that have overgrown their neighbours. I love that. I take a screenshot, open Telegram, and send the photo with a one-line note to my capture bot:

very old trees that have overgrown their neighbours have a name in Hungarian: "böhönc", think like ents of Middle Earth

A few seconds later, the bot replies.

Capture in action

The title is a claim extracted from my words: Böhönc: Hungarian word for ancient trees that have overgrown their neighbours. Four tags. And a link. The system embedded my raw text, searched for semantically similar notes, and the capture agent decided a connection existed to a note about a book that uses unusual, evocative words. I didn't ask for that. My photo is stored. My exact words are preserved. I close Telegram and I don't think about organisation or administration at all.

That night at 2am, the automated gardener wakes up. The capture agent had already found one connection at the moment I sent the message, but it only saw the top few candidates and only compared against what existed at that point. The gardener is different. It looks at everything at once, compares every note to every other note, and works in bulk. It creates similarity links between fragments that are close in meaning. Then it does something the capture agent can't: it runs Louvain community detection on the full similarity graph, grouping notes into thematic neighbourhoods based on how closely they relate across the entire corpus.

My böhönc fragment ends up clustered with the note about unusual words. I didn't organise this. I didn't tag them the same way on purpose. The semantic proximity was enough. And the cluster reveals something I hadn't consciously noticed: I didn't just like the tree. I liked the word. The thread running through these fragments is my attraction to rare, evocative vocabulary. I can see this on my dashboard, a web interface I built for browsing clusters visually.

The gardener clustered them overnight

Days later, I'm in a Claude session. Claude has no memory of me, but it has an MCP connection to my ContemPlace, so it can query my entire knowledge base as if it had already built an extensive memory of me. I've just seen a forest I liked. On impulse, I type: "I saw a great looking forest, and that made me think about trees in general. Search my notes for trees, check for anything related, see what comes up!"

One prompt, and the agent traverses the knowledge base

The böhönc capture comes back. But so do notes I'd nearly forgotten: someone making prints from cross-sections of felled trees, an idea about making a drone flute by routing, wooden board games. The agent organises them into "direct tree connections" and "wood/material connections" without being told to. I hadn't thought of these things together. The instrument idea was from weeks ago, the board games from a different context entirely. But the embedding space puts them in the same neighbourhood, and the agent traverses it in one call.

I finish my coffee. My morning walk, which was going to be a routine stroll, now has a destination. Yesterday I noticed a huge tree by the river that had split in two. And something clicks. Could I make a print from its cross-section? Could a piece of it become material for an instrument or a board game I will be making soon, maybe even with a hidden nod to a böhönc, as another layer of connections?

The system didn't send me there. It didn't recommend a walk. It reminded me why I'd want to go, by connecting a word I liked to a material I work with to a craft I practice. A fragment I captured days ago, without thinking about it, became a thread that pulled me into the physical world.

My own böhönc by the river

Why I built it

The short answer is that I enjoy building. Building is how I think through problems and how I understand things. I could probably have assembled something close enough from existing products. Maybe Obsidian on a VPS would have worked. A few days ago, Anthropic announced ways to connect a Telegram bot to a remote Claude Code session. If that had existed three weeks earlier, I might not have started.

But building it gave me things that a ready-made product wouldn't. I know exactly what happens to my data. I know why the capture agent is forbidden from adding conclusions I didn't express. I know how reassuring it is that my raw input is always preserved, so I can recompute everything from first principles if I find a better way later.

A delightful detail from development: the development process could directly use ContemPlace for mining product ideas about itself, a kind of recursive dogfooding I enjoyed on a very meta level.

The repo has been public from day zero, and every design decision and the core philosophy are documented extensively.

A door left open

I'm proud that I built this, and I enjoy using it daily. I don't know what it will become.

I'm at the point in this project where my excitement is levelling off and my mind starts noticing new things outside its domain. Fewer new features from here. More real use, more observing how it shapes my thinking over a longer stretch.

If this made you curious, or if you recognise the itch from the opening paragraphs, the repository is public. You can have your own instance running in about half an hour, for virtually nothing. If it inspires you to build your own version, or to think differently about how you capture your ideas, or if it just sends you down an interesting train of thought, I would enjoy hearing about it.

The repository, public from day zero

Böhönc. An ancient tree that outgrew everything around it. I found the word on a screen, sent it to a bot, and a week later it sent me on a walk. That's the whole thing.

ContemPlace — Under the Hood

Cloudflare Workers (TypeScript) for compute. Supabase (Postgres with pgvector) for storage. OpenRouter for LLM calls. Cloudflare Pages for the dashboard. Four Workers handle the Telegram webhook, the MCP server, the nightly gardener, and the dashboard API. Everything except the LLM calls runs on free tiers. Running cost with daily use: about two to three dollars a month.

The MCP server exposes tools for: semantic search, single-note retrieval, related-note traversal, recent captures, cluster listing, capture, remove, and on-demand gardening. Any MCP-compatible tool can connect and query the full knowledge base. The repository has detailed documentation on every layer, from the capture pipeline to the gardening algorithms to the setup guide for deploying your own instance.

Hungarian Blue Trail Album

2026-02-27T00:00:00Z

I made a custom photo album about the Hungarian Blue Trail, as a gift for friends I walked it with. Handmade gifts always bring me a special kind of excitement: the freedom to explore and play, while still designing for someone else and moving toward a real deadline. The best part is that the giftee can stay on my mind the whole time.

From idea to held object

One day this was a note and a rough sketch in my notebook: a wooden cover, the trail cut into a map. The next day I could hold that idea in my hand. That transition still feels like wonder every time.

The map came from a script I created. It takes a trail in GPX format and merges it with Natural Earth map data to produce vector files for fabrication machines. I checked the result by running test plots on my pen plotter and adjusted from there: how to represent water bodies, what to do with orphaned river and border segments, how to cut the trail while leaving landmarks readable. I still enjoy the fast iterative loop of turning data into something visual with code.

Pen plotter in action

The plotted map

When the composition felt right, I borrowed a laser from a friend (thank you, balassa_wood!) and tried engraving for the first time. After a few videos about lasers and LightBurn, I started running material tests and worked toward the final cover step by step. The amount of shared knowledge available online still surprises me every time. I feel like I'm never alone when trying to learn something.

Laser engraving the cover

The engraved cover

The binding was already familiar territory, more meditative than exploratory. I folded the pages, stitched the spine with waxed thread, and trimmed the edges. Watching it come together was so satisfying.

The bookbinding setup

Coptic stitching

Photographing the finished album felt like closing the loop. An idea that started as a sketch, moved through code, machines, and my hands, and became something I could present with a happy heart.

The finished album

If you want to recreate something similar:

The fabrication-ready vector files are available as digital downloads. You can contribute by pledging any amount that feels right, the files are linked regardless in the description.

Preview of the vector layers

If you are more into tinkering, the gpx2fab script I created takes a trail in .gpx format and merges it with available map data of a specified country. I tested it on a few through-trail and country combinations I know and it holds up, but note that it is not a polished product yet. I can see myself returning to this later, as visualising data and digital fabrication can still make me very excited.
The covers are 4mm birch plywood, cut and engraved on a 20W diode laser operated using LightBurn. Cut-through was 350 mm/min at 100% power, engrave was 11,000 mm/min at 60%.
If you want to try your hand at Coptic binding:
- here is a longer walkthrough that is very easy to follow by Christine Guenard, and
- a beautiful take on Coptic stitch that is just mesmerising to watch by Ido Agassi.

Let me know, if you create something similar, I would really like to see it!

Internet Radios and a Standalone Concept

2026-02-05T00:00:00Z

Exploring New Music

As the next step in discovering — and re-discovering — my relationship with music, I noticed that from time to time, I want to listen to a fairly wide range of genres with open curiosity. Music with potential.

Where Spotify, YouTube Music, and most recommendation engines fail me is focus. Or rather: too much of it. If I start a Rival Consoles radio, I will mostly get instrumental, low-tempo electronica that gets fairly boring soon. If I try to trick the system by combining my various tastes into a single playlist, the result is usually more annoying than joyful.

User-curated playlists can be better, but they are hit-and-miss. They also require a lot of crate-digging energy before you strike gold. Sometimes that is fun. Often, it is not.

Theme Time Radio Hour

A few weeks ago, I rediscovered something that reminded me what I love about music discovery: Theme Time Radio Hour. In this show, Bob Dylan picks a theme and builds an episode around it using songs, short stories, poetry, and jokes. There are about 100 episodes. It is joyful and heartfelt, especially when contrasted with the joyless, heartless Top10 pop rotations of current FM radios I know. It's not just the music. It is curation with care.

I highly recommend it. Even more so because there used to be a wonderful website listing all the played music, with generous references and links. Thankfully, it has been rescued from the enshittification of the internet by the Web Archive.

Theme Time Radio Hour - Coffee episode

Radio Paradise

That ethos stuck with me. And it nudged me back toward internet radio. A thing I used to cherish (even had my own station alongside my DC++ server :D), but have long forgotten.

One of my current favourites is Radio Paradise – Main Mix, where Porcupine Tree, GoGo Penguin, Khruangbin, and The Doors can coexist peacefully. That alone makes my day. I wholeheartedly recommend this channel if you enjoy discovering music without genre fences.

Radio Paradise Main Mix

Eter

While Radio Paradise's interface is far better than most stations' I tried, I still felt a bit of friction. I did not want it living as a pinned browser tab. Since, unlike their many ad-ridden counterparts, they openly publish their stream URLs, it was, on one hand, surprisingly easy to start listening: just toss it in VLC, and it works. On the other hand, it was confusingly non-trivial to get just what I wanted.

My requirements turned out to be:

almost frictionless access to the stream,
a way to see what is playing right now (album art is a bonus),
a way to note music that catches my attention with as little effort as possible

I will spare you the list of apps and half-solutions I tried. The best solution I found so far that works equally well on my Mac, iPad, and iPhone is hands down Eter. Thank you, dear developer, Krystian Kozerawski, especially for having a one-time-purchase option and not forcing another subscription on me.

Eter living quietly in my toolbar

At the moment, Eter lives quietly in my toolbar. Whenever I am in a "let's explore some music" mood, I turn it on. The variety is great. And it is surprisingly fun to try guessing the band before checking the info.

A Standalone Device

This whole internet-radio rabbit hole made me wonder: What if this experience had a physical form?

A standalone device. In its purest version, it would do exactly one thing. You turn it on, it plays music. It shows what is currently playing. You adjust the volume. Maybe you can keep a small text list of songs that caught your attention. Only physical, single-purpose buttons. No touchscreen. No notifications. No other apps.

It could also solve a very practical problem: how difficult it is for my ageing mother to navigate the ever-changing internet, while still wanting to listen to better shows than what current FM stations offer.

I tried finding existing products, but everything I saw was either convoluted (defeating the idea of dedicatedness), or breathtakingly overpriced.

So I decided to try the DIY route.

ESP32 dev board ready for soldering

After some research, I picked an ESP32 development board. Built-in Wi-Fi, no extra OS layer, and a good chance of achieving that "instant-on" radio feeling.

To test the concept, I used a small OLED display, a mono amplifier, and a tiny speaker. I wired everything up on a breadboard and hacked together a prototype in a few hours.

The prototype wired up

list of components I used in this build:
- ESP32-S3-WROOM-1-N16R8-M development board
- D096-12864-I2C-GV-WH OLED display (0.96", 128×64, SSD1306, I2C)
- MAX98357-M I2S Class-D amplifier module
- 3525-SPKR-4R-3W speaker (4Ω, 3W)

And then it played Radio Paradise!

Bowie is playing (open for sound)

Does it play flawlessly? No. The first three seconds after startup are a garbled mess until some buffers stabilise (I assume). Is the music enjoyable from that tiny speaker? Not really. It distorts and clips, and it gets worse over time, probably because I am feeding a mono signal into something that expects stereo, without proper attenuation. Does it at least have volume control? Not yet. I already connected a rotary knob, but I accidentally used its GPIO pins for the amplifier. That will need fixing. But! About three seconds after getting power, without any further interaction, it connects to my Wi-Fi, starts streaming Radio Paradise, and displays the artist name and song title.

I had an idea. And now that idea is playing a Bowie song on my breadboard. I am happy. And I am proud.

Coding with an Agent

For this experiment, I also wanted to test current paradigm shifts in coding, so I used Amp to create the code. I guided it with text, channelling its vast powers toward something useful for me.

I have mixed feelings about coding with an agent. On one hand, it is incredibly powerful to "just build" something complex using descriptive language. It gives me what I always loved about coding: going from idea to working system quickly. And it hides many of the parts I never enjoyed, like coaxing libraries to behave or guessing undocumented parameters.

On the other hand, it definitely weakens the "I made this!" feeling that usually deepens my emotional connection to the things I build. I still recommend trying it yourself. Especially if you allow yourself to let go of control for a while. If you are curious, the code as it stood at the first "wow, the concept works" moment is available on GitHub.

A handful of commits got it working

Enough for Today

Today, the radio sits on my desk exactly like this: unfinished, slightly flawed, and quietly doing its one job. I turned it on when I wanted music in the morning, and I turned it off when I’d had enough. Maybe I’ll do so again tomorrow.

Maybe that’s the real test. Not whether it becomes a product with a roadmap, but whether it finds a place in my daily life in a way that feels good.

Today, it definitely did — and that feels like enough.

Instruments That Currently Inspire Me

2026-01-29T00:00:00Z

I've been looking for an instrument I can fall in love with for years now. Not just admire or find fascinating (fortunately, I have several of those), but truly love. An instrument that becomes a meditation device, a companion to noodle on, a way to express my current state intuitively with as little friction as possible.

The Journey So Far

I learned to play the flute as a kid. Later, I was a bassist in a few small bands. A couple of years ago, I started exploring electronic music-making devices, and the rabbit holes went deep: quirky standalone grooveboxes, live-coded music, live performance samplers. Each new discovery taught me something, but none quite clicked in the way I was searching for.

What I miss is the "simplicity" of getting to know the ins and outs of a limited instrument. One that needs no patch initialization. One that isn’t an infinite canvas, but offers an opinionated framework, a space where I can get better and better at one thing through practice.

I like more than one music-making device of mine. I admire them, they fascinate me, each in their own way. But I really miss that one special instrument that I love.

What I'm Drawn To

I'm inspired by the world of organic-electronic standalone instruments. Devices that combine organic or tactile elements (wooden enclosures, acoustic components, natural materials) with unconventional interaction methods (magnetic fields, touch surfaces, physical gestures).

I know how deeply aesthetics affect me, and how all the small sensory things add up. Touching mostly wooden and metallic surfaces calms me. Touching plastic repels and drains me. These aren't trivial details, they shape the entire experience of playing for me.

The Dream Instrument

For a while now, I’ve been thinking about creating such an instrument myself.

Maybe it won't sound astonishing at first. Maybe I won't succeed on the first try. But this idea inspires me and keeps pulling me back, again and again.

I want to create something that feels genuinely organic — a standalone hybrid of acoustic and electronic, beautiful enough to create awe simply by being seen. Something that invites playful experimentation through touch and gesture rather than screens and menus, and that needs no formal musical knowledge to coax hauntingly beautiful sounds from. Ideally, it wouldn't even need a cable or a line out to sing.

Building Toward It

As a kid, I built a lot of weird percussion instruments. Then, as I developed some woodworking skills over the past years, I built a proper cajón. A few months back, I learned soldering and built the ingenious Touch 2 kit by SynthUX. Now I'm working on my C++ skills to program the Daisy Seed, connecting various sensors to it, and designing custom PCBs.

Current state of my Daisy Seed experiments

I hope that, one day, these paths will merge into something beautiful. Something that can be played. Something I can play.

Current Strongest Inspirations

I've been keeping an eye on the music-making world for a while, learning about instruments and creations, and my list of fascinating gear is near-infinite at this point. But these three speak to me more loudly than anything else right now.

Mulatar by Lockruf-music / Rubi Mo

Image courtesy of Rubi Mo / Lockruf-music

A "modular" string and percussion instrument. You can decide whether you prefer a sitar-like or slide-guitar-like string setup, and pick the types of percussion heads. What fascinates me most is the clever dual output: you can pull the volume knobs in or out, routing the audio signal to one of the outputs, so you can change the effect chains for the pickups and internal mics.

The previous, triangular version was built by bending wood. The current bamboo version (as far as I understand) uses a single piece of bamboo as the body. I can imagine myself trying a much simpler experiment based on this idea.

rubimo.net

TERRA by Soma Laboratories

Image courtesy of Soma Laboratories

A gorgeous instrument with touch interaction, a limited set of sounds, and movement-based modulation. The layout of the notes, the transposing keys, and the modulation macros is unconventional, yet somehow makes total sense. The "menu" is a triangle of LEDs that reminds me of learning a new language. I can feel the pull of the TERRA every time I look at it.

My current Daisy Seed-based project uses an MPR121 touch sensor, and I previously dabbled in gyroscope-driven sound design. Out of the three, this feels the closest to something I could imagine building a “lite” version of soon.

somasynths.com

Aurelia by Terraphones

Image courtesy of Terraphones

Shaped like a giant ocarina, made of wood, using silent mechanical Cherry keys (why hadn’t I thought of using mechanical keys for my instrument before?), with a built-in power source and speaker. I'm curious how my experience playing the flute (and the EWI Solo) could be translated into playing the Aurelia. I would very much like to spend time with it.

terraphones.com

These three instruments embody what I'm searching for: the marriage of organic materials and thoughtful electronics, the priority of touch and gesture over menus and presets, the confidence to be a specific thing rather than anything you want. They're not trying to be infinite. They're trying to be unapologetically themselves. And they succeed beautifully.

I hope that one day I'll build something that belongs in this family.

Photo of the Year

2026-01-02T00:00:00Z

A few years ago, I had one of my photographs printed in a larger format, on proper fine art paper, and hung it on the wall of my living room. I treated it as if it were part of a small, private gallery: carefully placed, given space, allowed to exist as more than just an image on a screen.

That was also the year I decided not to buy a Christmas tree. Instead, I made one from sticks and rope. It could be assembled, disassembled, and reused year after year, without cutting down another tree for a few days of decoration.

When I was looking for a good place to hang one of the 2D ornaments, the hanger of the photo turned out to be the best spot. So I took the picture off the wall, fully intending to put it back once the holidays were over.

But the holidays ended, the year turned, and something unexpected happened.

With the new year came a kind of quiet excitement. Instead of rushing to put the same photograph back, I found myself thinking about what it would be like to live with a different image for the year ahead. One that felt right for now. One that could set a tone, rather than just fill a space. And just like that, a small ritual was born.

Since then, every year, when the Christmas ornaments are packed away, I look through my own photographs. I wait for one to speak to me. When it does, I have it printed, and it becomes one of the focal points of my living room—and, indirectly, of my daily life.

The spot matters. The image is visible not only when passing through the room, but also from the dining table and from my bed. I start my days with it. I end my days with it. For a whole year, we look at each other countless times.

In a way, this is my own, very personal “photo of the year” award. One I give quietly, without juries or votes or announcements. And in return, the photograph gets to shape my days. Sometimes subtly, sometimes more directly, simply by being there.

This year, I chose a submerging bird to accompany me.

Photo of the Year 2026

Notes on the object itself

The print is made as a giclée on 100% cotton, acid-free, gently textured Hahnemühle Photo Rag Baryta paper. It’s mounted on aluminium dibond, without any framing. A hanger and spacers are glued to the back, so the shadow cast by the print itself becomes an ever-changing frame, depending on the light and the time of day. When choosing the print's size, I usually aim for a resolution of around 240 PPI.

Re-building My Music Library

2025-12-02T00:00:00Z

For a long time, music was something I didn’t just listen to. I lived with it.

I knew the exact moment when a scratch on one of our LPs would line up with the music. I could hear it approaching like a tiny, familiar glitch in the timeline. I made mixtapes on cassette, waiting with absurd patience for a song to come on the radio so I could catch it without the host talking over the intro. I knew precisely when to flip the tape in my walkman so I could replay my favourite song without rewinding and draining the batteries. Later, with my discman, I could predict whether a bump on the bus would be within the tolerance of the prebuffered stream or if it would choke. I had a growing CD collection, then a growing mp3 collection. I scrobbled everything compulsively to last.fm. On long car rides I’d get goosebumps from the tracks that were playing.

Somewhere along the way, that relationship got diluted.

Streaming and the quiet erosion of attention

I’m not anti-technology. I love good tools. But the shift to streaming did something subtle to my connection with music. With “everything” available at all times, it became very easy to listen to “something”. I stopped tending to a library and started just consuming a feed. Music moved from being a carefully chosen companion to becoming background radiation. Playlists appeared out of nowhere. Algorithms tried to guess what I wanted. The friction almost completely disappeared. In theory, that should be great. In practice, it made my relationship to music feel less personal. I rarely listened to an album from start to finish. I didn’t know which track was number 3 anymore, or where the emotional high point of an album lived. I wasn’t building anything; I was just tapping “next”. And I missed that older version of myself: the one who would happily spend an entire evening fine-tuning a mixtape for someone, or sit on a bed with a stack of CDs, deciding which seven could come on a trip.

What I want back

I’d like to reintroduce a slower kind of discovery.

I want to stumble onto new music that feels like a golden nugget. Something that makes me move or smile or go quiet for a bit. Something I want to listen to on repeat for days or weeks, then forget, then rediscover later and get hit by the whole package again: the sound plus the memories plus the person I was when I first heard it. I don’t need a perfectly complete or objectively good music collection. I want a personal one again. A library that looks like a strange, accurate map of my life at different points in time.

A glimpse to my current collection

LPs, rituals and limits

A few years ago, I tried to fix this by going back to vinyl. LPs did help in some ways. Putting a record on is automatically a ritual: choosing it, taking it out of the sleeve, placing the needle, committing to listening to a whole side. I still enjoy this, and I’d like to keep that as a special thing. But LPs are expensive, take up physical space, and for me they are tied to being at home and having time and energy for the ritual. That combination makes it less accessible than I'd like to. Also, discovery through LPs alone is too limited for me. I’m not going to casually buy five random LPs just to see if one of them sticks. So LPs stayed in the “beautiful, occasional ritual” category, not the “everyday music companion” one. Whenever I attend a live concert, I try to get my hands on an LP to have a memory capsule of the experience.

Bandcamp and ripping the old shelves

My next iteration was Bandcamp plus local downloads. Bandcamp is a good compromise: it supports artists directly, I can actually own the music, and I can download high-quality files. Discovery is manual and scattered, but at least I’m picking things with intention. I’m paying attention to album art, labels, tags, and recommendations from people whose taste I find similar. Lately I’ve started going backwards in time: I’m ripping my old CDs. Not all of them. Just a few selected pieces at a time – the albums that were once important enough to buy and keep, even when money and shelf space were both limited. I’m using iTunes (or Music.app, or whatever the name is this year) and ripping them to lossless, which turns out to be pleasantly straightforward. There’s something nice about the slowness of this process. Taking a CD out of its case after years, noticing the tiny scratches and fingerprints, remembering the weird local shop where I bought it. Dropping it into the drive. Hearing it spin. Watching the track titles appear, or not appear, and then having to fix them (for tagging I use MusicBrainz Picard, which is somewhat less fiddly than it sounds,and my current choice for music playback at home is Swinsian).

A tiny player with real buttons

The other key piece was getting a dedicated music player again.

Karnivool comes to Budapest in 2026 🥳

I bought a Snowsky Echo Mini and bumped its storage with a 256 GB microSD card. Fifteen years after my last iPod shuffle retired, I have a small object in my hand that is just for playing music. No notifications, no messages, no browsers. Just files and buttons. I can press play in my pocket. I can skip. I can pause. I don’t have to unlock anything or look at a glowing rectangle. It feels oddly freeing and very nostalgic at the same time. I copy over the albums, the few carefully chosen things. It’s not “everything”. It’s my current small universe.

Yes, it takes more effort

Is this more work than opening Spotify and tapping on a personalised playlist? Absolutely. Ripping CDs, organising folders, tagging files with Picard one by one, hunting down missing covers. All of that takes time. Finding new music that actually catches my attention takes multiple tries. But that album that did catch my attention, now means something again. Because I touched it. I decided to include it. I know which folder it lives in, which track feels like the core, and where I first listened to it. The friction makes me care again. Do I feel more connected to the music I listen to now? Hell yes!

This isn’t a manifesto against streaming. I’ll probably still use it for certain things: quick checks, background playlists, sharing links. But next to that, I want to slowly re-grow a small, dense, personal music library that I handle with my own hands. Something I can carry in my pocket, not because it contains all the music in the world, but because it contains just enough of the right music for me, right now.

Ginkgo Notebook

2025-10-20T00:00:00Z

This year was heavy with woodworking, and lately, lighter crafts started calling me again.

Carving can be so satisfying

I carved a ginkgo leaf from lino—it turned out surprisingly decent for a first try. Then I bound a small A6 notebook: cardboard covers, Rives textured paper, simple pamphlet binding with waxed thread.

Fresh from the workbench

Without a paper guillotine, I trimmed the edges by hand and rounded each corner one by one. I stamped the ginkgo on the cover with black paint, and apart from a single rogue fingerprint, it came out just as I’d hoped.

It’s simple, fits in my hand perfectly, and makes me excited to see what I’ll fill it with.

Ginkgo Notebook

A Logic Game from Leftover Wood

2025-09-20T00:00:00Z

A few days ago I came across a logic game I had never seen before. I’m always curious about games, and this one struck me as both simple and ingenious. The rules: each piece has four dimensions (height, shape, colour, hollowness), and the goal is to get four in a row sharing any single dimension. The twist is that you don’t place your own piece, rather you choose one for your opponent (since then, I've learnt that the name of the game is Quarto!).

Out of curiosity, I asked a friend to introduce me to resin printing, and recreating this game from my Shapr3D plans was a perfect first project. I quickly realised I still dislike the feel of plastic, so I decided to build a wooden version instead.

Resin printed prototype

This project gave me several firsts. I had never created a rod before, and now I managed one using a roundover bit, pushing a slat against a simple router fence, shaping some leftover aniegre. For the game board, I found a piece of cherry that was just the perfect size, and carved it by hand with a V-bit. Later, I improvised a small container from birch offcuts, on which the game board could double as a lid. That was my first time using a guide bushing with the router instead of a templating bit with a bearing. Next time, I’ll probably take out most of the material with a forstner bit first.

Satisfying shaping of a rod

Starting to take shape

I considered various hinge mechanisms for the lid, but magnets have always fascinated me. I embedded small neodymium magnets, carefully aligning all four pairs, so that the lid can snap onto the base from any direction. It works beautifully, though I’ll remember not to try fitting the magnets next time before sanding and finishing again. Once inside, they refused to come out by any non-destructive way.

My professional "zen" press, holding the magnets in place until the PU glue cures

Now that the piece is finished, I find it lovely to look at and satisfying to interact with. It turned out polished in some ways, while still carrying the imperfections that make it unmistakably handmade. I enjoy the interplay of three different woods and the two pigments, the contrast of rounded and rectangular features, and the satisfying snap of the magnets. It makes me smile every time I pick it up.

I love the interplay of patterns, shapes and colours

Piano Keys in Oak: A Coat Rack That Refused to Be Ordinary

2025-09-03T00:00:00Z

What began as “just a coat rack” for a new fitness room quickly turned into something far more personal. The set included a modern oak shoe rack and a dining-style oak tabletop, but this piece became the wildcard: equal parts design experiment and personal challenge.

I loved making this one

At the lumberyard, I stumbled upon an oak slab with a striking live edge. It wasn’t part of the original plan, but the moment I saw it, I knew I had to keep that raw, organic contour. It meant extra work and custom solutions, but it also promised a chance to make something truly mine.

A live edge that demands attention

The hinge system was its own adventure. Before committing to oak, I started with real CAD but later I turned to another flavour of CAD that stands for Cardboard Aided Design. Prototyping the folding motion with cardboard, then refining it in plywood, gave me the confidence to move on to solid wood.

Another type of CAD - Cardboard Aided Design 😅

It works!

Only after the plywood template felt right did I commit to oak. The mechanism itself is simple in principle but demanding in execution: each hook folds out like the keys of a piano, disappearing flush when not in use.

Ordnung muss sein! :D

CAD version 1.0

Looking good so far

Getting there meant meticulous, almost meditative cycles of assembly, reassembly, sanding, finishing, and testing again. Every adjustment was a respectful discussion with the material: avoiding cracks, respecting the grain, and finding harmony between precision and the organic irregularity.

Starting to take shape

By the time I lifted the finished piece above my head for the hero photo, I realised it had become more than a commission. It was a synthesis of many things I care about: careful craft, playful engineering, and, above all, the meeting of function and aesthetics.

I really am proud of this one

Now it hangs as a statement piece and a quiet reminder: even in everyday objects, beauty and utility can coexist—sometimes all it takes is patience, persistence, and a live edge you refuse to cut away.

Crafting My Own Fregoli: A Ceiling-Mounted Drying Rack from Wood

2025-03-09T00:00:00Z

My drying rack was always practical—but oh, how it irritated me. Its plastic wheels never cooperated, and every threshold became a battle of rattling metal and collapsing legs. One day, as I struggled with it yet again, I thought, "There has to be something better that takes up less floor space."

Enter the "fregoli"—a traditional Hungarian ceiling-mounted drying rack from my childhood. But commercial options disappointed me—either ugly plastic contraptions or overly complicated motorised units priced like sports cars. And the memories also weren’t great: nylon cords frayed and tangled, the constant threat of falling, and a user experience so frustratingly poor it felt like punishment every time. Surprisingly, this very memory sparked my inspiration. Could I create something practical, beautiful, and enjoyable to use?

Although the initial CAD sketches looked promising, transferring the plans onto the actual wood felt strangely off. Trusting my instincts, I kept the geometric logic but adjusted the spacing parameters on the fly. I'm discovering more and more how important it is for me to plan thoughtfully, yet still leave room for improvisation and intuitive choices.

I enjoyed geometry classes as a teenager 🤷‍♂️

With many curves, arches, and rounded corners, I initially cut the first piece freehand using a jigsaw and sanded it meticulously until it felt just right. I then used this piece as a template for the rest, routing them precisely with a templating bit. Realising how effortlessly I could reuse the leftover freehand-cut edges for subsequent pieces was incredibly satisfying—a tangible indicator that my jigsawing and routing skills have genuinely leveled up through steady practice. This step felt particularly important, marking a clear increase in my confidence thanks to continuous learning and practice.

Template in action

At the lumberyard, I stumbled upon heat-treated oak dowels: a deep, rich colour, and a comforting scent intriguingly reminiscent of something I could not place my finger on yet. Paired with the leftover birch plywood, I envisioned the striking contrast between the dark rods and the plywood’s layered edges.

Beautiful contrasts

Sanding the rods was something I dreaded initially, but building a simple jig around an inverted orbital sander turned a tedious task into a surprisingly efficient one. Hours into the process, I realised why the smell seemed so oddly familiar: heat-treated oak first smells like something freshly baked, then immediately like some very luxurious whiskey. That sensory discovery alone made the tedious sanding feel worthwhile.

Oiling is always satisfying

My favourite design detail: a sliding centrepiece. I wanted stability without sacrificing the flexibility to dry larger towels or sheets. By slightly widening the centre holes, I achieved a snug yet adjustable fit for the oak rods.

Trying to avoid some sanding

As always, there were unexpected challenges. Double-sided tape wasn’t strong enough to hold the pieces together for template routing, forcing me to improvise with screws that doubled as very handy centre markers. And when one oak rod snapped due to a hidden knot, the spare piece I instinctively bought saved the day—even presenting me with an opportunity to test whether PVA glue will bond to heat-treated oak.

All glued up

Mounting the rack was a saga in itself. After hours drilling into the stubborn ceiling, dodging iron rebars, and getting mortar in my ears, the pulley system finally held securely. It was exhausting, and something I'm not really looking forward to doing again.

There, I fixed it.

Looking at the completed fregoli and feeling the texture of the wood fills me with quiet satisfaction. Sure, there are a few details I'd approach differently next time, but right now, I'm resisting the urge to immediately chase perfection. I'm enjoying the imperfect beauty of what I’ve created, content in knowing that it already brings me joy.

Testing mounting options

Lifting the rack for the first time was not the moment satisfaction hit me, but the next day, when I stepped into the room and saw how it transformed the whole atmosphere. The contrasting woods, the arching crosspieces, and the black lift strings create an intriguing shape resembling the ribcage of some graceful sea creature, casting intricate, ever-changing shadows across the room. The hardoils need a few weeks of curing to completely seal the woods from moisture, and now I find myself looking forward to laundry days, simply for the joy of interacting with something I've made with care.

In a nutshell – it's mounted!

Creating this drying rack was particularly meaningful because it marked my first entirely original design—not a remix, nor inspired by something I'd seen elsewhere. Having the freedom to draw directly onto the wood, cut freehand, and make intuitive changes on the fly was liberating. It feels like a personal milestone: a confirmation that I can create something with confidence, from concept to finished object, that will live up to my (sometimes unrelenting) standards.

Some friends asked if I’d consider making more to sell. But part of me knows the joy lies in the creation itself—the planning, the woodworking, the problem-solving—not in mass production. Sure, I might enjoy creating another one, perhaps refining the process, working more efficiently. But by the third repetition, I suspect I'd already be redesigning it for CNC manufacturing and finding ways to simplify construction for outsourcing. Honestly, I'd rather just share the design files openly online, making it accessible for others to replicate and improve freely. Maybe this makes me a terrible businessman—but I’m very much at peace with that idea. Perhaps that’s a good lesson learned: the things I love most lose their charm when mass-produced.

So here it hangs—useful, elegant, mine. A reminder that surrounding myself intentionally with beautiful things that reflect back the qualities I want to nurture in myself is a worthy aspiration.