Dashboards, admin systems, APIs and the workflows between them. The test we apply is not whether it demonstrates well, but whether somebody who did not build it can operate it, extend it, and find out why it did something last Tuesday.
What people arrive with
It works, until two people edit it, until a formula is pasted over, until the one person who understands it is away. The replacement has to be at least as fast to use as the spreadsheet was, which is the requirement most internal tools miss.
The work lives in one system, the money in another, the customer in a third, and a person moves data between them. That person is the integration, and they are the part that does not scale.
No tests, no documentation, no staging, and a deploy that is a person on a Friday. The first deliverable in that situation is not a feature; it is the ability to change something safely.
What the work includes
Scope is where a project is decided, so this list is specific rather than reassuring.
Built around the decisions somebody actually makes, with every figure derived from its source rather than typed in. A number kept in step by hand will eventually be wrong, and will look exactly as confident as a correct one.
Typed boundaries, explicit error shapes, and the same request producing the same result. Where data moves between systems, the move is a job with an identity rather than a script somebody runs.
A record that can be in any combination of seven booleans has no state machine, and every bug in it is a combination nobody considered. Naming the states is usually the design work.
Not coverage for its own sake. The tests worth writing are the ones that fail when a business rule is broken, which means they double as the place the rules are written down.
Configuration through environment variables and nothing else, no secret in a repository, a documented way to run it locally, and a build that fails rather than shipping something half-checked.
How it runs
Not a process diagram written for this page. It is the production loop the studio already runs, which is why it has a measure step and a human gate in it.
Evidence
Everything shown as evidence on these pages is a NOVEMADE product - our idea, our money, our name on the listing. There is no client logo wall here, no testimonials and no case-study metrics, because there have been no clients to take them from. What there is instead is software you can open.
Design a tattoo, turn it into a working stencil, and see it on your own skin first.
Check it yourself
The fight is the game’s. These are level 1 of 1000, with the hit points, damage, counts and timings the shipped build reads out of its own campaign file; the shrine offers the two routes that level really offers, and the Bloomwell’s +24 kin and the Thorn Oath’s 1.35× attack are its own balance numbers. The Warden is boss family 1 of 50, with its real root and nova patterns and its real colour, and the grove is drawn in the palette region 1 declares. Every word is the game’s own string, in any of its 10 languages. What is rewritten is the combat loop itself: Godot’s web runtime is 10.3 MB before any content and this campaign is another 5.9 MB, so a real build would be roughly fifty times the weight of anything else on this site.
Goes further
Ten regions, a hundred difficulty tiers and fifty boss families, generated from a model — and the acceptance run that had to prove every route through it is winnable.
Where this stops
We publish no client names and no case-study numbers, because the work shown on this page is our own and there are none to publish. We also do not staff a seat on somebody else’s team by the month - this is project work with a defined end and a handover that includes the keys.
Also from the studio
Collaborate
Publishing, partnerships, investment, or a technical conversation about how we build. Tell us which one and we will skip the rest of the form.
