From Pilot to Law: The Honest Path to a Legal Mandate
By Anushka Appala and Dr. Janio Rosales
It is tempting, when you believe in an idea, to describe its destination as though it were near. We would rather be honest. A national platform for algorithmic governance does not begin with a law; it earns one, rung by rung, over years, by proving at each step that it deserves the next. Anyone who tells a government that transformation arrives fast is selling something. What we are proposing is slower, and far more durable: a ladder from a small proof to a regional reference, where every rung is earned before the next is attempted.
Here is the whole climb, honestly drawn.
___________
| 5. REGIONAL |
| REFERENCE |
___________|_____________|
| 4. NATIONAL |
| LAW |
___________|_____________|
| 3. EXECUTIVE |
| MANDATE |
___________|______________|
| 2. PILOT & |
| COMMITMENT |
___________|______________|
| 1. PROOF |
| (3-mo MVP) |
_____|____________|________________________________________________
One program, One live Acuerdo Ley de A model other
one town, program, Gubernativo: Gobernanza: nations adopt;
simulated data; measured an executive an open the open standard
weeks to minutes leakage order to standard crosses borders.
recovered adopt the passed by
open standard Congress
Rung 1 — Proof: the three-month MVP
The climb begins with the smallest thing that can be true. Our first step is a three-month MVP: one MIDES social program, one municipality, Spanish and Q'eqchi', running entirely on simulated data so no sensitive systems are touched. It proves exactly one thing, and proves it concretely — the time to complete a social-program application drops from weeks to minutes, while 100% of decisions remain in human hands. Nothing binding runs on real citizens. The point is not scale; it is a demonstrable, auditable proof that the architecture works and the principle holds.
Rung 2 — Pilot and commitment
A proof on simulated data earns the right to a live pilot, not a national rollout. Here the platform runs against a real program with real oversight, and the measure changes from "does it work" to "what does it recover" — welfare leakage measured, service times measured, every decision still signed by a named official and recorded in the ledger. A successful pilot produces something no slide deck can: evidence a minister can stand behind. That evidence is what earns an institutional commitment to go further.
Rung 3 — Executive mandate (Acuerdo Gubernativo)
With a pilot that recovered real money and served real citizens, an executive can act. An Acuerdo Gubernativo — a government order within the executive's own authority — can direct participating institutions to adopt the standard. This is faster than legislation and reversible by the next administration, which is precisely why it is a rung and not the summit. It proves the model can operate at the level of the state, under real accountability, while the case for permanence is built.
Rung 4 — National law (Ley de Gobernanza)
Only a law makes the model outlast a political cycle. A Ley de Gobernanza — a law of algorithmic governance passed by Congress — is what turns a working system into a durable institution, binding future administrations to the same principle: AI recommends, policies decide, officials oversee, audits verify. This is a multi-year destination. It requires a track record, cross-party trust, and a text careful enough to survive a legislature. It is not promised by the pilot; it is made possible by it.
Rung 5 — Regional reference
The final rung is not something we build; it is something others choose. If Guatemala proves that public-sector AI can be fast, transparent, multilingual, and fully accountable, the model becomes a reference other nations adopt. The context makes the opening plain: only seven Latin American countries have a national AI strategy (OECD-OPSI, Strategic and Responsible Use of AI in the Public Sector of LAC), even as the OECD.AI Policy Observatory tracks more than 900 AI-policy initiatives across over 80 jurisdictions (OECD.AI, 2026). The frameworks are multiplying; deployable, legally grounded implementations are scarce. A working reference is worth more than another framework.
The one design choice that makes the ladder climbable
Everything above depends on a single decision about how the mandate is written. It must be written as an open standard, with NaciluzIA as the reference implementation — not as "every entity must install one vendor's product."
The distinction is not ideological; it is what makes the law fundable, fair, and passable. A mandate that names a single supplier is a mandate a competitor will challenge, a legislature will distrust, and a development bank cannot in good conscience fund. A mandate that defines an open, auditable, published standard — which any qualified party can implement, and which NaciluzIA happens to implement first — is one a bank can finance, a court can uphold, and a Congress can pass without handing a monopoly to anyone. The nonprofit stewards the standard; a for-profit arm delivers against it; the public interest owns the specification. That structure is what lets the mandate survive scrutiny at every rung above it.
We are, in other words, describing a multi-year destination reached only by proof — a small MVP that earns a pilot, a pilot that earns a mandate, a mandate that earns a law, and a law that, if it holds, earns imitation. Each rung is contingent on the one below it actually working. That is the honest version of the path, and it is the only version worth walking.
Cada decisión, a la luz.
— Anushka Appala and Dr. Janio Rosales