Halden Agent
The launch page for Halden's autonomous research agent.
A short note from us on what actually shifted in the way we work — and on the parts that did not shift at all. First, though, some of the work:
Tolla Robotics
Brand refresh and site rebuild, with a 3D asset kit and a design system.
Unstructured
A playful brand site that turns knotty data into something with a face.
Rowan Field
A platform site for lidar surveying, typeset end to end.
Ember Labs
Identity, motion and a documentation system for a research lab.
Born in Copenhagen,
where craft is
a discipline.
Where we come from
In 1924, a few streets from our studio, the Kunsthåndværkerskolen put cabinetmakers and architects under one roof. Chairs, type, posters, whole buildings — all of it held to the same standard.
Their idea was stubborn to the point of rudeness: the things people touch every day deserve as much care as the things that hang in a gallery. So they kept the makers close, and refused to let anything half-finished leave the workshop.
A century on we are around the corner, working much the same way. One team, one set of eyes, and nothing left to run on its own.
AI is good at
everything but the
final tenth.
What changed
The tools are genuinely good now, and everybody has the same ones. A first draft, ten variations, working code: what used to fill a week now fills an afternoon.
Which is why volume stopped being the hard part. Point a model at a blank page and it hands back the average of everything it has already seen. Competent enough, and forgotten by Thursday.
Brilliant at
Making a great deal of work quickly, and putting a rough first version of almost anything in front of you.
Hopeless at
Knowing which of those versions is any good, and having the nerve to throw the other nine away.
That final tenth is where the real work lives. It is still 90 percent of the effort, and it is still the part you have to do yourself.
Explore. Generate.
Refine. Ship.
The machine makes the options, we make the calls. In practice that works out to roughly 60 percent exploring, 20 building, 20 refining.
Explore
We can go further than we used to, and faster: interactions and shader work, not just static layouts.
Generate
It rarely lands first try, so we curate and keep tuning the prompt and the inputs until it does.
Refine
We keep what is working, fix what isn't, and take the rest of the way by hand.
Ship
Once something works, it stops being a one-off and becomes a system we can reuse.
The House Style
Protocol.
We make your brand machine-ready
Our first product. Every brand we work with ends up with one: a single source that holds the voice, the design and the rules, written so that a person and a model can both read it. It is what we generate from, and what we check everything against. The four parts below are how it works.
Read the full storyHouse Truth
The one place that holds the voice, the design and the rules we build everything from.
Skills
A job the system has been taught to do your way, ready whenever you ask for it.
Output
Finished work, drawn from House Truth and on-brand the moment it arrives.
Check
A score on every output, and whatever it learns goes back into the source.
Move your
organisation up the
levels of AI.
Passing on what we learn
This is the other half of what we do. We are a design and technology studio working with these tools every day, on our own craft and on our clients', and most of what we have learned is teachable. So we pass it on: pilots that prove a real use case, hands-on workshops, upskilling for your team, and the change management that makes any of it stick.
It starts with where you actually are. Adoption has levels, a little like self-driving cars: at one end somebody is pasting prompts into a chat window, at the other a system runs whole jobs on its own. Most organisations are idling in the assisted-parking phase, and we help you move up one level at a time, so the tools compound instead of gathering dust after the demo.
L1
Chats. People prompt by hand. Helpful, but nothing compounds.
L2
Templates. The good prompts get written down and shared around.
L3
Workflows. A task done the same good way every time, with a human reviewing.
L4
Systems. The workflow reads your own context and stops asking for it.
L5
Agents. Whole jobs run end to end, while people stay on the calls that matter.
We know where the line is, and we will happily talk you out of the AI idea that is going to embarrass you in six months.
The tools are everywhere.
The judgment isn't.
For our team, this has turned into a new layer of expression. The craft is not new, but it used to need an engineer in the loop. A designer can now build real interactions, motion and shader work directly, which lets us take a client's project somewhere more particular than a static screen ever could.
Atlas Week Athens
A complete event site in 1.5 weeks with 1.5 designers. Proof that taste is what makes the tools worth anything.
Softest possible
We asked how soft we could make a headline. The answer: very.
Three heads
Pushing a real-time frame library somewhere it was never meant to go.
One reasonable hat
Turning a studio in-joke into a working configurator over one weekend.
Get in touch
We are a remote studio spread across five time zones, so the odds are decent that one of us is awake and somewhere near you. Part workshop, mostly Copenhagen at heart, and happiest with a hard problem.
The nearest Nordhausers are Ida, Matthias and Petter in Copenhagen, about 480 km away.
hello@nordhaus.studio · Gothersgade 41, 1123 København.