For developers

Build legal AI on models trained for it.

Documentation, API credits, and the same models behind our platform, so you can build on retrieval, enrichment, and structure that were trained for the domain from the start. Call the embedding and reranking APIs, graphitize a document, or wire the open chunker into your pipeline.

Request API credits Request documentation access

Retrieval endpoint

Embed and rerank in one call

A single request embeds the query, searches the corpus, and reranks the candidates, returning ranked authorities with scores. The retrieval pair does the recall and the precision together, so you get the case you needed rather than a list of plausibly related text.

The interface is the same wherever you run it. Point the request at the hosted platform, at a marketplace deployment inside your cloud, or at an air-gapped container on your own hardware.

brelyne · v1/retrieve

# Embed, search, and rerank a legal query
POST https://api.brelynetech.net/v1/retrieve

{
  "query": "duty to warn of latent defect",
  "corpus": "us-case-law",
  "top_k": 3,
  "rerank": true,
  "deployment": "air-gapped"
}
# 200 OK · 38 ms · retention: none
{
  "results": [
    { "rank": 1, "score": 0.94, "cite": "22 U.S. 1" },
    { "rank": 2, "score": 0.89, "cite": "17 U.S. 316" },
    { "rank": 3, "score": 0.81, "cite": "5 U.S. 137" }
  ],
  "status": "ranked"
}

Enrichment endpoint

Graphitize a document

Send a document of any length and receive a structured knowledge graph of its parties, clauses, dates, courts, citations, and holdings. Because it runs at sub-second latency, enrichment becomes something you do across a corpus rather than an overnight job.

The response is law you can traverse. Query a clause, follow a citation, or pull every party, working with a graph instead of a wall of text.

brelyne · v1/graphitize

# Turn a document into a knowledge graph
POST https://api.brelynetech.net/v1/graphitize

{
  "document_id": "lease-2041",
  "extract": ["parties", "dates", "citations"]
}
# 200 OK · 0.6 s · nodes: 42 · edges: 71
{
  "nodes": [
    { "type": "party", "label": "Lessor" },
    { "type": "clause", "label": "Indemnity" },
    { "type": "date", "label": "Effective" }
  ],
  "status": "graphitized"
}

The surface

What you can call

A compact API over the whole model family, plus the open chunker that runs beneath it. Every capability is available across all three deployment modes.

Endpoint

Retrieve

Embed a query, search a corpus, and rerank the candidates in one call. Returns ranked authorities with relevance scores.

EmbeddingReranking

Endpoint

Graphitize

Turn a document of any length into a knowledge graph of parties, clauses, citations, dates, courts, and holdings.

EnrichmentSub-second

Open source

Chunk

Split documents at semantic boundaries with the open-source algorithm behind more than two million downloads a month.

PackageSelf-serve

Getting started

From request to first call

Step 01

Request access. Tell us what you are building and which deployment fits, from a hosted trial to an air-gapped pilot.

Step 02

Receive credits and docs. We provision API credits and share documentation for the endpoints and the open chunker.

Step 03

Call the API. Embed and rerank a query, or graphitize a document, with a short request and a structured response.

Step 04

Ship where it runs. Keep the same interface as you move from the platform to a marketplace deployment or your own containers.

Get your credits.

Documentation and API credits are available on request across every deployment mode. Tell us what you are building and we will get you to a first call.

Request API credits