Integration patterns

What you can build on the Data API, which endpoints combine for each shape, and when to choose which. Parameter-level detail lives in the API reference.

Legislation and regulatory monitoring

A scheduled feed of new case law, regulator guidance, and legislative changes, scoped to defined fields of law and jurisdictions. Use this when your product tracks what changed over time.

Run keyword_search on a schedule: one saved query per topic you track, a rolling date window (from_date and until_date, always both), and scoping with jurisdictions, documentTypes, and fieldsOfLaw. Page through with page and num_results (up to 100 per page), and pull full text on demand with retrieve_document. Your integration keeps track of the identifiers it has already seen between runs; the API returns what matches each query inside the window.

Valid filter values come from jurisdictions_portals, documenttypes, and trees, so users can configure exactly what they track.

Compliance and regulatory research

Grounding compliance decisions in the text of regulations, enforcement actions, and supervisory guidance. Use this when the starting point is a natural-language business question, without a citation in hand.

Start with hybrid_search, scoped to the relevant jurisdictions and fields of law, and switch to hybrid_search_reranked where precision matters more than latency. retrieve_document pulls the full text and summary into the assessment. To confirm a provision is current law, check the inForce field on the retrieved document; date-sorting alone does not answer that question.

Cross-border legal research

One integration covering multiple European jurisdictions. Use this when the same question must be answered per country and compared.

Run one hybrid_search per jurisdiction and group the results. One rule decides whether this works: national jurisdictions are queried in the national language (Dutch for the Netherlands, German for Germany, French for France), while EU sources work in English. An English query against a national portal returns empty results, so the integration translates or maintains per-country query variants.

To compare at the same court level, list the issuing bodies per jurisdiction with sources and pass the chosen ids in the search call.

Reference graph analytics

Showing how a regulation is interpreted and how doctrine evolves over time. Use this when the product is built on the relationships between documents.

Break the regulation into article-level units with articles, then call reference_search with reference_identifier for the regulation as a whole or article_reference_id per article. reference_direction selects citing documents, cited documents, or both; sort_type=3 surfaces the most-cited interpreting decisions; facets=true adds the distribution by jurisdiction, source, and document type.

Embedded research assistant

A research assistant inside your own product, answering questions with grounded citations. Use this when you want finished answers delivered in your interface, without assembling the search calls yourself.

Open a session with initialize_chat, then send each question with ask_question, optionally scoped with jurisdictions. Answers stream over server-sent events, so your interface can render them token by token the way Luna does on the Moonlit platform.