Building the MVP with Guatemala's Universities

By Anushka Appala and Dr. Janio Rosales

When people hear that NaciluzIA is being built with students, they sometimes assume we are cutting a corner — trading professional rigor for cheap labor and academic idealism. It is the opposite. Building the first version of a national algorithmic-governance platform inside Guatemala's universities is one of the most deliberate decisions we have made, and it solves three problems at once: it produces auditable code in the open, it builds the public-sector talent the country will need for a decade, and it lets us prove the model without the sensitive systems or the procurement machinery that a government contract would demand.

Let us explain how the university track actually works, and why it is a feature rather than a shortcut.

The starting condition is the opportunity

Guatemala ranks 122 of 193 on the UN E-Government Development Index (UN E-Government Survey, 2024), up from 126 in 2022. That is not a flattering number, and we do not pretend otherwise. But it is the right frame for what we are doing. A country in the middle of that ranking is not a country without capacity; it is a country whose capacity has not yet been organized around the right architecture. The distance between 122nd and the front of the table is the exact room a project like this exists to close — and it is more honest to name the gap than to dress it up.

Closing it is not primarily a hardware problem or a budget problem. It is a talent-and-trust problem. And universities are where both are made.

Who builds what

The MVP is co-built by students and faculty at two of Guatemala's anchor institutions — Universidad del Valle de Guatemala (UVG) and Universidad de San Carlos de Guatemala (USAC) — the private and public poles of the country's engineering and computer-science talent. Students write and review the code. Faculty supervise the work and hold it to a publishable standard. Because NaciluzIA is designed as an open, auditable standard rather than a proprietary product, what they build does not disappear into a vendor's repository — it is published, inspectable, and reusable. Academic incentives and open-source incentives point in the same direction: work you can put your name on, that others can read.

Language is not an afterthought bolted on at the end; it is validated at the source. The Academia de Lenguas Mayas de Guatemala validates the platform's indigenous-language interactions, beginning with Q'eqchi' — the largest Mayan language, with roughly 1.3 million native speakers (2019 census / Ethnologue). This matters because a mistranslation in a government process is not a cosmetic flaw. It is the difference between a citizen understanding what a form asks and being turned away for answering the wrong question. The Academia's role is to make sure that when NaciluzIA speaks to a Q'eqchi' speaker, it speaks correctly, in a register a public institution can stand behind.

Alongside the two universities, INTECAP and structured university tracks build the vocational and technical layer — the people who will operate, maintain, and extend the system after the first pilot ends. A platform that only its original authors can run is a liability. One that a pipeline of trained Guatemalans can operate is infrastructure.

What the students get, and why it is honest

We are candid that this is an exchange, not charity in one direction. Students earn academic credit and genuine research experience — the chance to work on a real system with real constraints, of the kind that is hard to simulate in a classroom. They contribute to publishable work. They graduate having built something that matters to their own country rather than a throwaway exercise. For faculty, it is a research agenda with a live subject. This is how we intend the relationship to be understood: mutual, named, and documented.

The instrument that makes it clean is a research memorandum of understanding. The MOU sets the terms — what the students build, who supervises it, how the results are published, and how intellectual work is credited. It frames the collaboration as academic research, which is both true and useful: because the proof stage runs on simulated data and produces open, publishable results, it does not require a government procurement process to begin. That is not a loophole. Procurement exists to govern the purchase of goods and services by the state; a university research collaboration on simulated data is a different activity, and treating it as what it is lets the proof stage move at the speed of a seminar rather than a tender. When the work later touches real programs and real money, the procurement rigor comes with it — as it should.

The point of building it this way

There is a version of this project that hires a firm, signs a contract, and delivers a black box. It would be faster to describe and far worse to trust. By building the MVP with universities, we get code that can be read, language that has been validated by its own guardians, a generation of engineers and technicians who know how the system works from the inside, and a proof stage that begins without waiting on a procurement cycle.

None of that decides anything on its own. NaciluzIA's binding rule — AI recommends, policies decide, officials oversee, audits verify — holds here as everywhere else. But the university track is how we make sure the thing we build is auditable not just in principle but in practice, by the people who built it and the country they built it for.

Cada decisión, a la luz.

— Anushka Appala and Dr. Janio Rosales