Curation is the product. The interface is just how you reach it.

Counting courts is not measuring competition.

RacqI knows not only who operates a venue, but what kind of operator runs it. Pricing, utilisation, opening hours and capital requirement differ completely between them — a school hall and a commercial centre two streets apart are not the same competition.

A directory sees two venues. We see two different businesses.

Where the operating basis is established, the calculation counts the venue as what it actually is. Where it is not, the answer flags the venue instead of assuming a category.

Stored, searchable by meaning and by exact term.

Every part of the base is human-curated: chosen deliberately, gathered over years, checked by people who know what the material is worth — and in part not publicly available at all. None of it is a scrape. Someone had to decide the rules.

Research library

Studies, standards and hand-read regulation — the fine print that decides a permit or a subsidy, which a crawler skips.

Selected by hand over years. Part of it is not publicly accessible.

Facility database

The venue, the operator behind it and what kind of operator they are — commercial, municipal, non-profit, club, school, academy — with sports, court counts, pricing, positioning, programming and format mix.

Official registers, statistical offices and supranational bodies — consolidated, de-duplicated, verified, then enriched venue by venue. Not purchased market data.

Benchmark modules

CapEx ratios, standards and density baselines distilled from decades of operating experience — available in no report or dataset.

Curated by practitioners. Specific know-how, in a quality and quantity available nowhere else.

The engine can be built. The evidence took forty years to assemble.

A capable team can build a data-driven vertical agent. What cannot be compressed is what sits beneath it: a facility database assembled by hand over years, regulation read page by page, benchmarks that come from four decades of operating rather than from any dataset.

Operator classification is the clearest example. Nothing on a website states whether a venue is run for profit, for members or for pupils. Establishing that reliably — across countries, languages and legal forms — is interpretation, not extraction. It is exactly the kind of judgement that only exists where sector expertise and engineering sit in the same organisation.

Live data feeds

Authoritative connectors for demographics, economy, climate, catchment and accessibility — Eurostat, the World Bank, WorldPop, GHS/Copernicus, NASA POWER, OpenStreetMap and more — pulled at the moment a question needs them, sub-municipal where it exists.

Deep research

Fired per question, reaching wider than the base already holds whenever a topic calls for it.

Signals & sources

The paywalled, un-crawlable reporting plain search can't reach — monitored and refreshed daily, human-verified before it enters the base.

The right engine per question: some reason, some search, some summarise.

We use AI to open up knowledge — not to invent it.

The machine reads, verifies and connects what nobody could read in reasonable time. The statement at the end comes from the calculation and from the Institute’s benchmarks — not from the language model.

Calculation layer

The math runs off the model, on a dedicated engine over structured data. The answer reads the result; it never performs the arithmetic itself.

Grounded & tiered

Every answer sourced and ranked by evidence tier; where no curated source fits, it flags that instead of bluffing.

Self-critique

A review step re-checks evidence, numbers and rules — and corrects itself — before you ever see the answer.