keebloom. makes mechanical keyboards and input devices. Each unit is handwired by me. Each enclosure is FDM printed by me. The brand exists because the keyboard market mostly defaults to two looks, gamer RGB or boring office, and I wanted something that could sit on a desk and be worth a conversation.
These are notes on how it gets made.
Why it stays handmade
The obvious move at some point is to design custom PCBs and move production to a board house. It would be cheaper per unit. It would scale. It would let me ship more boards faster.
I'm not doing it.
The moment keebloom. moves to custom PCBs, the project becomes about manufacturing optimization. Right now it's about making. Every joint is one I soldered. Every enclosure came off my printer. Every switch was hand-pressed into a hand-drilled plate. The labor is part of what you're buying, the same way it is with a hand-built guitar or a small-batch ceramic.
That sets a hard ceiling on volume. It also sets a price floor and a brand position. keebloom. is not competing with Keychron or GMMK. It's a different category, closer to small-batch audio gear or independent watchmaking. Limited by design.
The switches
Most of the choices in keebloom. are about feel. The switch decision was the longest one.
I tried Gateron Browns. Too soft, no personality. Holy Pandas. Too snappy, too aggressive on the bottom-out. Boba U4Ts came close but the tactile event happens too high in the travel.
I landed on Akko V3 Creamy Purple Pro. The tactile bump is pronounced enough to feel deliberate but soft enough to type on for hours. Quiet, which matters in a shared space. The bottom-out is dampened without going mushy.
The line I keep coming back to: the Akko is just “considered.” That's the word for it. Same as the rest of keebloom. The switch matches the brand because both are tuned for the same kind of attention.
For anyone who flies, the closest reference I have is activating a switch in an R44 cockpit. Defined break, satisfying engagement, no slop. The Akko Creamy Purples sit in that family.
Encoders, the boring choice
The rotary encoder on the numia macropad is a standard EC11 with a click function. Industry standard, works with QMK, available everywhere, costs almost nothing.
This is the only place in keebloom. where I went with the obvious option. Custom encoders exist, the high-end keyboard world has opinions about feel and detent count and torque curves. None of it was worth the time. The EC11 does the job. The interesting decisions in keebloom. are not the encoder.
Worth knowing what to optimize and what to leave alone.
FDM is hard
Every keebloom. enclosure is printed on a Bambu A1. Single-color FDM, 256x256 mm bed, no multi-material.
The bed size is the hard constraint. Anything larger than a 65% layout has to be split into pieces, which means seams. Seams on a product that lives at eye level on a desk are visible. So the design has to either embrace the seam (treat it as a deliberate joinery line) or hide it (route it under an edge, into a shadow line, behind a wrist rest). I do both, depending on the piece.
The other constant fight is overhangs. The aesthetic I want includes recessed details, undercuts, soft chamfers that don't sit flat to the build plate. FDM hates all of these. Every enclosure goes through several iterations of orientation and support strategy before it prints clean. Most of my failed prints are not failures of the slicer or the printer, they're failures of me asking the process to do something it doesn't want to do.
I'm getting better at designing for FDM rather than against it. The dahlia65 enclosure went through six revisions before the surface was acceptable. The numia is on revision three. The desk speaker hasn't passed a print test yet.
This is the part of keebloom. I'm worst at and most engaged with. Most of the design choices I'm proudest of came from a printer failure that forced me to rethink a detail.
Materials, briefly
The material palette is small and consistent across products.
Matte polymers for primary enclosures. Soft-touch, non-gloss, subtle texture. Warm in the hand.
Concrete for weighting elements where structure benefits from mass. Smooth, not industrial.
Fabric for acoustic surfaces (the desk speaker, eventually). Reduces visual noise, signals that the object is meant to be lived with rather than looked at.
Metal only where function demands it. Screws, internal frames. Never decorative.
The principle behind all of it: matte over gloss, soft over sharp, subtle texture over a perfect surface. Products that age well and don't demand attention.
What's live and what's not
Three products in development under the same design language.
Dahlia65 is the flagship. 65% layout, the most refined of the three, the one I'd ship first if I were taking orders today.
Numia is a modular macropad and numpad system. Smaller, simpler, but the modularity adds complexity to the housing design. Currently on revision three of the enclosure.
The desk speaker is the most ambitious and the furthest from done. Fabric grille, concrete base, FDM body. Acoustically not yet tuned. Visually closer than the print quality is.
Bloom keys, the QMK editor I built as internal tooling, is shipping separately as a paid desktop app. It started as a tool for me and turned into something other keyboard builders wanted.
What's next
More prints. The dahlia65 is closest to a unit I'd hand to a customer. The bottleneck is finish quality, not function. I'm slowly closing the gap between the renders and the printed object.
Once one product is at sellable quality, keebloom. opens for small-batch orders. No injection molding, no PCB house. Just me, my printer, my soldering iron, and a waitlist.
keebloom. is at keebloom.com. Questions or feedback: max@skogsdue.com.