GASPEROHLAB
We design and ship games, applications and AI models.
An independent lab that takes hard problems from prototype to production — engineering software people actually use, not slide decks.
What we build
Four disciplines, one team
Every project starts as a question. These are the forms the answers take.
Games
Playable systems — from tight arcade loops to strange, systemic worlds. Mechanics first, polish always.
Applications
Web and mobile products built with intent. Fast, accessible and genuinely pleasant to use.
AI Models
Custom models and intelligent systems — training, fine-tuning and wiring intelligence into real tools.
Systems & Tools
Command-line tools, automations and engines. The quiet software that makes everything else possible.
Selected work
What we've shipped
Products in production today — with more taking shape in the lab. Open any project for the full story.
In production
Already running real businesses.
Our work isn't a portfolio of concepts. The Marapone suite — private, owned AI for construction and logistics — is built here and deployed on customers' own hardware today.
“We wanted AI our clients could actually own — running in their own building, with the source in their hands and no meter ticking. The lab shipped exactly that, in production.”
The numbers
Proof, not promises.
What the work is actually doing in production — measured on the customers' own machines, where it runs.
Documents & drawings audited on-prem
Faster tender & invoice turnaround
Runs on hardware the client already owns
Bytes of client data that leave the building
You own it
Software you own — not software you rent.
Most AI vendors want your data in their cloud and a meter running against you forever. We build the opposite. The way we ship Marapone's private AI is the way we like to work generally: it's yours, it runs where you say, and it keeps working when the invoice is paid.
The source is yours
You get the code, not a login. Nothing we build is a black box you can't open or a dependency you can't replace.
It runs on your metal
Deployed on hardware you already own. Your data never leaves the building, so there's nothing to leak and nothing to trust us with.
No meter, ever
You buy it once. No per-seat pricing, no per-token bill, no subscription quietly ticking against you for the life of the product.
How we work
From a question to production.
No two projects look the same, but they all move through the same four gates. Most ideas don't make it past the second — and that is exactly the point.
- 01
Ask
We start with a question worth answering, not a spec. If we can't say why it matters, we don't start.
- 02
Prototype
The smallest real version we can build, put in front of the actual problem — early, while walking away is still cheap.
- 03
Pressure-test
We break it on purpose. What survives contact with real use moves forward; what doesn't gets cut, cleanly.
- 04
Ship & own
What earns it goes to production — engineered, accessible, and still ours to maintain long after launch.
Ways to work with us
Three ways in.
Most engagements start small and earn their way forward. Pick the one that fits where you are — or just tell us the problem and we'll say which makes sense.
Prototype sprint
Weeks, not months. We build the smallest real version of your idea and put it in front of the problem — so you learn whether it's worth going further before committing to a roadmap.
- A working prototype
- An honest read on feasibility
- A clear go / no-go
Build to production
The prototype survives contact with reality, so we take it the rest of the way — engineered, accessible and shipped. You own the code and the decisions behind it.
- Production-grade build
- Full source, no black boxes
- Ongoing ownership
Private AI deployment
Domain-tuned models running entirely on your own infrastructure. Nothing leaves the building and there's no cloud meter — the same way we build and ship Marapone.
- On-prem, offline-capable
- Trained on your domain
- Bought once, owned forever
The ethos
A lab, not a factory.
GASPEROHLAB exists to explore. We move between games, apps, AI and systems freely, letting each discipline sharpen the others. The goal isn't volume — it's finding the ideas that deserve to become real, and building them properly.
A factory optimises for throughput. A lab optimises for learning. We stay small on purpose, keep the loop between idea and working software as short as we can, and let the work that survives that pressure be the work we put our name on.
- 0
- Products in production
- 0
- Disciplines under one roof
- 2025
- Founded
- 0%
- Independently owned
Curiosity first
Every project begins with a question, not a spec. We chase the interesting problem and see where it leads.
Build to learn
Prototypes over pitch decks. The fastest way to understand an idea is to make it real and press on it.
Ship what matters
Not everything survives the lab — and that's the point. We keep the work that earns its place in production.
Depth over breadth
We'd rather build one thing properly than ten things halfway. Focus is a feature, and saying no is part of the craft.
Own the whole stack
From the model to the metal, we build and hold our own tools. No black boxes we can't open, no dependencies we can't replace.
From the lab
What we're learning, out loud.
Devlogs, engineering notes and the reasoning behind what makes it out of the lab — and what doesn't.
Before you ask
How working with the lab works.
The short version of the questions we get most. Anything not here, just ask us directly.
Hard, interesting problems across games, applications, AI models and the systems that hold them together. If it can be prototyped, pressure-tested and shipped — and we can say why it matters — it's a fit. If it's a slide deck looking for a rubber stamp, it isn't.
Start a project
Have something worth building?
A game, an app, an AI model, or a problem that doesn't have a name yet — tell us about it and we'll build the prototype.