Deep Intelligence Across the Dutch State

The Challenge

Dutch government-relations work runs on the official record: what parliament debated, what a minister promised, where a bill sits in the pipeline, how a court read the law, what the budget actually allocates, and which municipality moved first on a contested issue. Every piece of it is public. Almost none of it is usable together. The parliamentary record, the statute book, the legislative calendar, the official gazettes, the case-law database, the national statistics office, the budget data and the municipal publications each live in their own system, with their own search interface, their own identifiers and their own quirks, some of which quietly bury the most relevant document three pages deep.

The questions that actually matter to a consultant live between those systems. Did the minister deliver on the commitment she made in that debate? What happened after the motion passed: anything, or nothing? Has this dossier quietly crossed to the Senate? What did the Council of State warn about, and did the government listen? Which EU directive is forcing this Dutch law, and what is the transposition deadline? No search box on any single portal can answer them, so firms answer them with analyst hours, and the picture is only ever as complete as the time available to assemble it.

Stromy built the engine that answers them directly.

Our Approach

The whole official record, one surface. The engine unifies more than ten families of official Dutch and EU sources: the full parliamentary record of the Tweede Kamer (dossiers, debates, motions, per-faction votes, committees, members and thousands of tracked ministerial commitments) alongside the consolidated statute book, the forward-looking legislative calendar, government communications, the national gazettes, case law, the statistics office’s catalogue of roughly five thousand data tables, ministry budgets down to individual budget articles, and the regulations of municipalities, provinces and water boards. It reaches across to Brussels too: EU directives are linked to the specific Dutch laws that implement them, deadlines included. Over sixty purpose-built query tools sit on top, each engineered around the real-world quirks of its source, including the ones where the official search is the problem.

Intelligence that is composed, not searched. Real questions span sources, so the engine answers them as single operations. It reconstructs a dossier’s full timeline across documents, debates, votes and publications. It follows a motion from sponsorship through the vote to the ministerial commitment it produced and the follow-up documents that honoured it, or didn’t. It watches a petition until the government’s formal response arrives, because that response is often where the state’s framing of a contested topic gets locked in. It detects when a bill crosses from the House to the Senate, and pulls the Council of State’s advisory opinion and the Court of Audit’s findings into the same view, so the oversight layer is never an afterthought. And when the record cannot prove a connection, the engine says so rather than guessing: no keyword-matched maybes presented as facts. For firms whose entire product is judgement, that discipline is the point.

It reads the documents, not just the metadata. Most tools stop at titles and dates. This one retrieves and parses the actual text (debate transcripts, explanatory memoranda, government letters, court rulings, law articles), navigating a document’s structure, pulling targeted sections from files that run to hundreds of pages, and extracting every reference it finds: the laws cited, the related dossiers, the attachments worth opening next. Analysis comes back as quotes with sources, not summaries of things nobody read.

An analytical layer shaped by public-affairs practice. Fifteen ready-made workflows cover the recurring work: topic monitoring, dossier tracking, actor briefs, committee watch, legislative horizon scans, ministry-narrative analysis, case-law watch, quantitative grounding against official statistics. Three complementary lenses answer the strategic questions: what official activity exists around a topic, who is moving it (with full coalition history, so “was that party in government when it voted this way?” is a one-line question), and how a company, technology or issue is being framed across the official record. On top of that sits a tension engine with seven detectors that surface what the record contradicts: passed motions whose own deadlines have lapsed, dossiers silent for years while ministers claim momentum, commitments overdue, spending that doesn’t match the stated ambition. That is the material a consultant takes into a roundtable.

The Outcome

A government-relations team working on the engine stops aggregating and starts advising. An issue is followed as one continuous thread, from a municipal gazette to a committee room to a Senate reading to the budget line that funds it, instead of six separate tabs. A dossier briefing that took an afternoon compiles in minutes, with every claim carrying its source. And because the whole surface is MCP-native, teams interrogate it in plain language inside their own AI workspace and capture every good question as a repeatable workflow, so the intelligence operation compounds with use.

The engine is in production inside Stromy’s own intelligence work today, and we are opening access to a limited number of partner firms in the Dutch public-affairs market. The same model (official sources unified, cross-source intelligence engineered on top, delivered natively into an AI workspace) extends to any jurisdiction where the record is public but the picture is not.

---

Technology used in this project: MCP Data Connectors · AI Orchestration · Skills & AI Workspaces