The system that operates.
Your data and processes ordered and documented, and an agent connected to your real systems — orders, ERP, email — that carries out tasks with your team, rather than answering isolated questions.
Positioning · 2026
We build the system, leave it under your control, and teach you to take the baton: you go from using artificial intelligence to directing it — shaping it, correcting it, and making it work for you. Everything it does stays in plain sight, in a repository — an ordered archive, dated and with history — you can open and verify, without depending on anyone's word for it.
§ 01 · The problem
Today almost every company “has” artificial intelligence, and almost none puts it to work. A tool is bought, connected, and months later it is used to summarise an email or draft a text — a fraction of what it promised. The rest of the business carries on exactly as before.
It is not a lack of will. Artificial intelligence can only operate a business that is described in a way it can read: its data, its processes, who does what, which systems it accesses. Almost no company is set up this way — and that is why the AI neither operates, nor can be seen, nor can be directed.
A tool is hired and connected.
The AI does something, but whoever leads cannot see what.
The business is not described in a way a machine can read — and, for the same reason, it neither operates, nor can be seen, nor can be directed.
§ 02 · What you receive
What we deliver is not a presentation or an audit: it is a department of your company genuinely running on artificial intelligence, on a foundation a machine uses and a person understands. It works like a living repository — with its history and version control — not like asking a chatbot the odd question. The commitment that delivers it has a name and a deadline: the 90-day Start.
Your data and processes ordered and documented, and an agent connected to your real systems — orders, ERP, email — that carries out tasks with your team, rather than answering isolated questions.
Everything the artificial intelligence does is written down, dated, and versioned in a repository you can open. Not “trust that it works”: open it and see — what it did, when, with which data, and under which permission.
We do not hand over the system and leave: we build it with your people, and whoever will operate it learns, on the real case, the why of every decision — how it is structured, how it is corrected, how it grows without breaking. You keep the views to direct it: what it does, what it is forbidden to do, and where to correct it. And when Europe asks, the answer already exists — it comes from that same foundation, not from a separate formality.
§ 03 · The scene
Picture the department you chose to start with. Over the weekend, the system has classified what came in, prepared the drafts that were due, and set aside — untouched — what it is not entitled to decide. On Monday at nine, the person in charge does not ask anyone what happened: they open the repository and read it — what it did, with which data, what it left waiting for a person.
And when something is not to their liking, they do not open a ticket with a vendor: they change the rule, in writing, and the system obeys the new rule from that moment on. That is directing an artificial intelligence — not hoping it gets it right, but shaping it; not believing what you are told, but seeing it.
§ 04 · The idea
That foundation has a property that is not obvious. These are not two projects. When your business is described with the precision a machine needs to operate it, it is also described with the clarity you need to understand and direct it — and, along the way, with the trail that answers for you if anyone asks.
“Operable” means something concrete: your processes, your data and your permissions described in a structured way — readable by a machine — so that an artificial intelligence system can act on them with traceability: it knows which data it may use, what it may and may not do, and leaves a record of every step. That is what allows, for example, an AI to classify documentation, prepare a regulatory review, or carry out an operational step without you losing sight of what it did and why.
You should not have to take anyone's word for what your artificial intelligence does — not even ours. What we build is, precisely, your ability to verify it.
The business described with the precision a machine demands.
Data, processes, permissions: structured, legible, with traceability of every step.
Operable by the AI.
The artificial intelligence can read it and work on it.
Visible and yours.
You understand it, correct it and direct it — and can prove it to whoever asks.
§ 05 · How we work
We do not transform an entire company at once; no one serious does. We enter through a department that already wants it — the one with appetite and something to show, not the most reluctant: we order its data and processes, build or adjust the artificial intelligence that operates that part, and let you see how it does it. Once it works and is understood, the next department comes on its own, because it has already seen the result alongside.
It does not matter where you start. If you already have artificial intelligence that operates nothing, or that operates without you seeing what it does, we step in and leave it operable and visible. If you start from scratch, we build it for you. In both cases it is born with the same thing: legal certainty, results and control from day one, not added when it is already too late.
And we do not leave a black box behind. The department is not handed over finished: it is built with your people, and that building is the training. While we set up the system, whoever will hold the baton — chosen by appetite and by having already used AI, not at random — learns the why of every decision: how information is structured so the system can read it, what it needs in front of it to decide well, which working environment gives an edge, and where the human control goes. Solid foundations learned on your own case — a driving licence, not a course.
That part matters because a system like this does not improve just by being used: it improves when someone who understands it corrects and extends it. Artificial intelligence brings the good practices; the particularities of your business are taught to it by your people. That is why we train advanced users, not data scientists: people who decide with the system understanding how it works, not trusting whatever comes out. You stop depending on us as much as on being told.
Entry department
willing · operableNext department
extension…
as it worksNext
continuous extensionOperable company
horizon§ 06 · What sets us apart
We put your business in a state where artificial intelligence can genuinely work in it — and we build the one that operates on top. Without that foundation, any tool stays a demo.
What makes the AI work is what lets you see what it does, what risks it runs, and prove it if a regulator asks. European compliance — EU AI Act, data protection, the AI management standard ISO/IEC 42001 — is not a separate formality: it comes from the same architecture. And it gives you something more: no longer depending on being told.
What we deliver is read and used by an artificial intelligence system, and opened and understood by a person — it is not a folder of documents to file away. It is the same method this project is built with: the proof that it works is that we use it on ourselves.
§ 07 · The proof
The proof is two proprietary systems, live, built and operated by us: one works with people in regulated healthcare; the other publishes data intelligence with every figure verifiable at its source. You can open both.
The first, Strahlkraft40+, is an artificial intelligence system in production in German regulated healthcare. It runs, it passes audit, and we built it. It operates under an AI management system documented in line with the international standard ISO/IEC 42001, with its data protection resolved in the system's own operation, not on a separate sheet of paper. It is proof, in production, that we build and operate real AI, governed and auditable.
documents composing the management system, conforming to ISO/IEC 42001.
controls evaluated in the statement of applicability.
Cohen's κ — dual AI with formal rubric, substantial reliability.
The second piece is Despegue, verified-data intelligence on Argentina's investment wave, born by design as the demonstration case of our method: artificial intelligence researches at scale; a person directs it and verifies every data point against its primary source before publication. It is live — you can open it now and follow any figure to its source.
Behind them is Ignacio Aredez: a machine learning engineer certified by Google and a Lead Implementer of the ISO/IEC 42001 standard — the uncommon combination of someone who can both build the system and govern it. All credentials are verifiable.
To the ability to build and govern is added the ability to put AI under pressure: competitive validation in HackAPrompt, the largest language-model red-teaming competition, backed by OpenAI, Scale AI and Hugging Face. Those who know how to break an artificial intelligence know how to secure it.
Machine learning engineer, Google certification.
ISO/IEC 42001 Lead Implementer.
Competitive validation in AI red-teaming (HackAPrompt).
§ 08 · Why us
The market is full of those who promise transformation with artificial intelligence and could not explain what it does inside. The difference is simple to check: we show the method and the implementation before asking you for anything.
If this resonates, the next step is not a demo or a form. It is a conversation to see whether we fit: you tell us where you want artificial intelligence to start working and what you have documented of that process today, and we tell you frankly how and where we can help. We work with those who fit — and knowing it from the start is worth as much as the fit itself.
The question is no longer whether your company will use artificial intelligence. It is whether you will direct it — or keep believing what you are told.