Economist and data analyst · Neiva, Huila, Colombia

jhonstevenalvarezruiz.vercel.app  ·  LinkedIn  ·  humanizar.tech  ·  Live demo  ·  Type system

Economist and data analyst. Together with Steven Vallejo Ortiz I build the system that turns scattered AI agents into a company that actually operates.

I work with three materials that academic convention keeps apart for no good reason: the critique of political economy, the architecture of multi-agent systems, and the logic that lets you say when a set of parts constitutes a whole and when it remains a heap. These are not three fields. They are one problem approached with three instruments.

humanizar.tech — development companies, built from agents

An agent is not a company. The claim is neither rhetorical nor quantitative: it is categorial. Adding agents no more produces a company than adding coins produces capital, or adding individuals produces a body politic. What constitutes a company is a determinate form of organization — division of labour, attribution of responsibility, governance of capacity, accounting for spend — and that form is contained in none of its parts.

The opposite error has a name. It is the hypostatizing of what exists only as a relation: treating a structure as though it were a thing. Gustavo Bueno dismantled it in Ensayo sobre las categorías de la economía política; Luis Carlos Martín Jiménez carries it through in El mito del capitalismo, where he shows that "capitalism" names a myth precisely because it performs that hypostasis. Today's AI market repeats the operation on its own ground: it calls a process an "agent" and expects an organization to emerge from multiplying it.

CAUCE V3 begins by denying that premise. It does not orchestrate agents that already exist — it produces the teams and imposes the form that makes them accountable. From two to N specialized teams, governed from a single control plane.

The method is the one Juan Íñigo Carrera demands of any critique that means to be serious: not to represent the process from outside, but to reproduce its movement. A control plane does not describe the organization — it exercises it, which is why it can be audited.

And the architecture is Hegelian in a precise, non-decorative sense. The company is not presupposed as given: it starts from the minimal determination — a request in a channel — and is left to develop through its mediations into the concrete result. This is the requirement Stephen Houlgate defends in The Opening of Hegel's Logic: a presuppositionless beginning, whose categories are not imported from outside but generated in the development itself.

Every figure above is counted over the versioned files of the five production repositories. Source against test is split by path and by extension: anything under test/, tests/ or pruebas/, or whose filename contains test, counts as test. Nothing is estimated.

Security is not a layer added once the system already works. It is the constraint that decides what the system may do, and therefore comes first.

No credential enters a repository. Every project ships in two modes: the public version, without secrets, and the real configuration, which never leaves the machine. Where it matters, a scanner runs before every commit and aborts on a credential pattern or a local path. Continuous integration rejects the push if anything slips through.

Deleting a secret from a file does not remove it. It stays in history. The only valid repair is to rewrite history and rotate the credential. The second is not optional; the first without the second is theatre.

What I claim is tested. If a system says it works offline, a test cuts the network and verifies it. If it says it will not corrupt its state, a simulation of thousands of cycles checks the invariant. A claim with no test behind it is a promise, and promises do not deploy.

I work economic theory from the tradition that refuses to take equilibrium as its starting point, because equilibrium is not a fact: it is an assumption that decides in advance what the model will be able to see.

From the School of Oviedo I take the categorial apparatus: political economy is not ordered by magnitudes but by categories, and confusing the plane of magnitudes with the plane of categories produces operative myths — "capitalism", "the market", "technology" — that function as the subjects of sentences in which there is no subject at all.

From Juan Íñigo Carrera I take the demand on method: critique does not consist in setting one model against another, but in reproducing in thought the real movement of the form under critique. Whoever only represents, describes. Whoever reproduces, explains.

Ordered by technical difficulty measured against the code, not by age or size.

«Die Eule der Minerva beginnt erst mit der einbrechenden Dämmerung ihren Flug.»
_{Hegel, preface to the Elements of the Philosophy of Right}

_{Jhon Steven Alvarez Ruiz · Neiva, Huila, Colombia · jhonstevenalvarezruiz.vercel.app}