Understand a country by cross-referencing its open data, with AI agents

6/23/20265 min read
Understand a country by cross-referencing its open data, with AI agents

Direct answer: a law is an idea. The data tell you what it actually became — who applied it, which rulings bent it, which choices and which results it produced. The idea behind Open·Parlamento goes beyond the law: an agentic system that searches and automatically cross-references every public database — statutes, cadastre, statistics, procurement, archives — to help people understand the reality we live in by analysing the data, not the opinions.

TL;DR

  • The law is an idea; the data tell you what it became. You start from the statute and follow laws, rulings, choices and real results.
  • An agentic system cross-references every public source and always cites the original.
  • Five layers: sources → knowledge → reasoning (agents) → use → new infrastructure.
  • Domains: not just law — territory/cadastre, statistics/economy, history/archives, businesses.
  • With guardrails: provenance always cited; you cross-reference public roles and data, you don’t profile private citizens.

A statute is an idea; the data tell you what it became

The starting point isn’t the past in itself, it’s this: exploring a statute as an idea. A law is born to achieve something. Then it enters reality and changes: other laws amend it, rulings interpret it, administrative choices apply it (or not), and in the end there are real results — money spent, territories transformed, rights exercised or denied.

Understanding that path — idea → laws/rulings/choices → real results — is the only way to understand the reality we live in with data, not with opinions. And yes, the past is in there too: how it brought us here. But that’s one facet of the chain, not the goal.

The five-layer model

All of this holds only if it’s organised. I think of it like this:

Layer What it is What it’s for
1. Sources Every public DB via MCP connectors Reach the data where it lives, citing the source
2. Knowledge The knowledge graph: statutes, relations, entities, territory, time Make relations first-class
3. Reasoning Agents that cross-reference sources and find patterns The «lenses»: law ↔ data ↔ choices
4. Use Study, map, cross-reference, narrate, administer The universal tool
5. Infrastructure A self-managed, public, AI-native open-data layer Move beyond legacy portals

The agents (layer 3) are the juiciest part: they make the right calls to the right source — a question about healthcare queries the health data, one about a contract goes to ANAC, one about a recovery-fund project to OpenPNRR — and they put together the pieces that, on their own, don’t speak.

Which data, beyond the law

The law is the first vertical. But the same method works for the whole universe of public open data:

Domain Example sources What it enables
Statutes & case law Normattiva, EUR-Lex, Constitutional Court, Court of Cassation how statutes evolve, inconsistencies
Territory & cadastre OpenGIS, cadastre, municipal boundaries, toponymy the history and geography of places
Statistics & economy ISTAT, Eurostat, budgets, procurement/ANAC, recovery fund cross-referencing data and choices, public money
History & archives State Archives, historical gazettes, time series how we got here
Businesses & interests business register, state holdings, offices held conflicts of interest on public data

Understanding the past (and what brought us here)

There’s a reason the time spine runs through every layer: without time you don’t understand the present. How a statute changed over twenty years, how boundaries and municipalities shifted, which chain of decisions produced the rule we’re clashing with today. Cross-referencing the law with territory, statistics and archives means reconstructing not an opinion about how things are, but their documented history.

The tensions, declared

A tool like this is powerful, and it has to be said: also delicate. That’s why the guardrails come first, not after.

  • Provenance at scale. Every cross-referenced statement cites the source. «Real sources, not invented» becomes harder and more important as the sources grow.
  • Privacy. You cross-reference public roles and data (who holds an office, how they voted, where the funds go), you don’t profile private citizens. The line between accountability and surveillance is exactly here.
  • Licences. Third-party data is used in compliance with its licences, not «hoovered up» casually.
  • Governance. A «public and self-managed» infrastructure needs a governance model, otherwise it stays a slogan.

FAQ

What does «understanding reality with data, not opinions» mean?

It means that instead of starting from a thesis and looking for confirmations, you start from public sources — laws, rulings, budgets, statistics — and let the data, cross-referenced and cited, tell you what actually happened. The opinion comes after, and on verifiable grounds.

Is it only for the law?

No. The law is the first vertical, but the system is designed to cross-reference any public database: territory and cadastre, statistics and economy, historical archives, businesses. It’s a universal tool for reading public affairs.

Isn’t it dangerous to cross-reference all this data?

It can be, and that’s why the guardrails are explicit: you work on public roles and data, not on profiles of private citizens; you always cite the sources; you respect the licences. The goal is transparency and accountability, not surveillance.


Understand reality with data, not with opinions. See how Open·Parlamento works, why the graph is only the door and how you surface inconsistencies and gaps, or let’s talk.

Open DataAICivic TechKnowledge Graph

Scritto da Giulio Garofalo