A brand existed on paper and not yet in the world. We handled almost everything needed to get it there: the direction, the digital system, the conversational systems and part of the operational work.
For confidentiality reasons, the client’s identity, specific industry and selected visual elements are not disclosed. The process, decisions and systems described reflect work actually carried out on the project.
The client arrived with a clear idea and two concrete things: a name and a mark. The logo had already been designed by others, before us, and we did not touch it beyond rebuilding it as a vector where digital use required it.
Everything else did not exist. There was no system saying how that mark should behave outside its own outline: which colours, which typography, which tone, which rhythm. There was no place to meet it. There was no way of answering the people who would ask questions.
A mark without a system around it is not a brand: it is a drawing. The distance between those two things is exactly the work we were asked to do.
A project starting from nothing has a problem that rebrands do not: there is nothing to correct, so there is nothing pointing the way. Every decision is the first one, and each constrains the ones after it.
So the decisions had to come in a particular order — first what the mark should mean, then how it should look, then where it should live — and each layer had to carry the weight of the next. A visual system decided badly forces the site to be rebuilt; a site decided badly forces the way of selling to be rebuilt.
On a confidential project this is the section that matters: without the client’s name, the only thing that makes the account checkable is how precisely we state the boundary.
The logo is worth being explicit about, because it is the item most often credited to whoever did not make it: the logo was pre-existing and was not designed by us. We rebuilt it as a vector so it would hold up in digital use — a technical task, not a mark design project.
The first delivery was not an image: it was a document. What the brand promises, to whom, in what tone, and — above all — what it must never do. A brand system is recognised by the limits it accepts more than by the elements it owns.
From there came the colour system, deliberately reduced to a few families, the material language and the typographic direction. Few rules, written so that someone else can apply them without asking permission: that is the proof a system genuinely exists rather than living in the head of whoever made it.
A second document carried the same decisions onto digital ground — how it moves, how much it breathes, what happens as you scroll. Written, again, before opening the editor.
The site was designed and developed by us, from the structure of the pages down to the server side that feeds them. Not an adapted template: an architecture decided for this project.
The decision that matters most is not visible while browsing.
Every piece of copy, every price, every image is edited from a management panel, without touching the code and without calling us. It is the difference between handing over a site and handing over something that can be kept alive. A site that needs a developer for every comma is stale within six months, because nobody updates it.
The pages read from a single source, not from copy pasted into the code. Changing something changes it everywhere it appears — which is how inconsistency is avoided once a project grows.
Every function serving the pages has an automated test checking it. Not a formality: it is what allows something to be changed a year from now while knowing immediately whether something else broke.
The management panel sits behind authentication. We mention it because in the first version it did not: it was found during testing and closed. An honest project reports the findings too, not only the results.
The project includes a conversational assistant and a voice agent, built as systems separate from the site and meant to accompany whoever arrives without replacing contact with a person.
We describe them by function rather than by configuration: the configuration belongs to the client, and is not ours to publish.
The answers rest on a knowledge base built for the purpose, not on a scraped website. It is the difference between an assistant that can explain how something works and one that repeats the promotional material.
Before settling what it should do, we wrote what it must not do. Outside that perimeter the assistant says so and hands over, instead of improvising on the client’s behalf.
When a conversation shows real intent, the system gathers what is needed and passes it to a person. Nothing is forced: anyone who only wants to look around stays free to do so.
Whoever runs the business can see what was asked and what did not work. Without that, an assistant degrades quietly and nobody notices.
Alongside the chat, a voice agent was built for the project. Voice changes the problem: the timing is different, interruptions exist, and an assistant that works in chat can be unfit on the phone.
A brand that has to reach the world needs physical things, and physical things have suppliers. This part of the work produces nothing to show, and it is the part that took the most patience.
We researched and evaluated suppliers for components and for packaging, contacted them directly, and held the thread of the requests and the replies ourselves.
We mention it because it says what kind of involvement this was: we did not stop at the edge of what was convenient to call “digital”.
This project is why we talk about brands and systems in the same sentence: they are the same work seen from two sides. If you are bringing something into the world that today exists only as an idea, the first step is a conversation.
Book a Discovery Session