Why I'm building an AI operating system for government
Local government is the highest-leverage place to apply AI, and almost nobody is building for it seriously. That's the short answer to why I started NovaGovAI: the institutions that decide where you can build, how land is used, and how fast a citizen complaint gets resolved are running on tools a decade behind the private sector. AI can close that gap — if someone builds it for how government actually works.
The failure that's usually predictable
The idea started with a motorcycle crash in Cebu. Riding toward Moalboal at night — no street lights, no signage for a sharp turn. I was fine, but the lesson stuck: infrastructure and planning failures hurt real people, and most of them are predictable. Somewhere there was a plan, or a missing one, that could have flagged that turn. The information usually exists. What's missing is a system that turns it into a decision before the harm happens instead of a report afterward.
Why local government, specifically?
National governments get the attention, but local government units do the work that touches daily life: zoning, land use, permits, disaster response, citizen services. In the Philippines, an LGU planning office is often responsible for a comprehensive land use plan (CLUP) — mandated under the Local Government Code (RA 7160) — with a fraction of the tooling a mid-size company would consider normal. The result isn't incompetence. It's a tooling gap. When the software is spreadsheets and PDFs, the work moves at the speed of spreadsheets and PDFs.
That's a good place to apply AI, because the bottleneck is exactly what modern models are good at: reading unstructured documents, cross-referencing rules, surfacing what a human should look at next. Not replacing the planner — giving the planner an AI-native layer over their own data.
What "an operating system for government" means
I use "operating system" deliberately. NovaGovAI isn't one dashboard. It's the layer underneath: a model of an LGU's planning data, zoning, and citizen input that individual tools sit on top of — CLUP compliance, report triage, land-use analysis. The same underlying model powers the official-facing side and the citizen-facing side. On the citizen side, that's Nova Citizen, a super app for everyday government interactions, so a report a resident files flows into the same system a planner works in.
Building it this way — one ontology, many surfaces — is the same pattern we reuse across Zentarai Labs, including Nova Solutions on the private-sector side. Government is where the pattern earns its keep, because the coordination problem is the whole job.
Where this actually is
Honestly: early. Zentarai Labs is pre-pilot. NovaGovAI is built and demonstrated on modeled data, and the current work is hardening the platform and preparing the first LGU pilots. I'd rather say that plainly than dress it up — the interesting part isn't a claim of scale I don't have yet, it's the bet that government is the right place to point this work. I think it is, and I'm building from the Philippines because that's where the need is sharpest and where I can build closest to the people who'd use it.
Building in this space, or running an LGU that wants to see it? NovaGovAI →