Eridian

Eridian Legal

Start contract review, clause extraction, and policy mapping.

Eridian Legal is the contracts and policy vertical. Parsing, clause extraction, PDPL/GDPR mapping, precedent search, and redlines run through GPT and Gemini on the same gateway as the rest of Eridian. Client matter data is Customer Data. Residency and PII redaction are not optional on Enterprise legal projects.

This page gets a matter to first parse. Detailed contracts live on the pages linked below.

What you get

WorkflowWhat it doesDocs
MattersContainer for documents, residency, and privilege flagsMatters
Contract reviewParse parties, obligations, dates, capsContract Review
Clause extractionSearch and compare clauses across the libraryClause Extraction
Policy mappingMap obligations to PDPL, GDPR, and internal catalogsPolicy Mapping
PDPL and GDPRDual-regime control planePDPL and GDPR
PrecedentRetrieve similar deals and positionsPrecedent Search
RedlinesFirst redline against playbooksRedlines

POST /v1/legal/* is a typed surface on Inference. There is no unmanaged model path.

Prerequisites

  1. Project prj_legal_001 (vertical legal).
  2. Key eridian_sk_ with legal:write, inference:write, models:read.
  3. Region pin (me-central-1 for UAE PDPL work, eu-west-1 for GDPR work, or separate projects). See Regions.
  4. PII redaction enabled. See PII Redaction.

First parse

POST /v1/legal/contracts
Host: api.geteridian.com
Authorization: Bearer eridian_sk_...
Content-Type: application/json
Idempotency-Key: msa-renedor-4821
X-Eridian-Project: prj_legal_001
{
  "matter_id": "mtr_4821",
  "filename": "msa-acme-2026.pdf",
  "document_id": "doc_msa_4821",
  "model": "auto",
  "routing_policy": "quality_optimized",
  "features": ["pii_redaction", "structured_output"]
}

Upload the bytes first with POST /v1/legal/documents (multipart). Then create the contract object that points at document_id.

{
  "id": "ctr_4821",
  "object": "eridian.legal.contract",
  "status": "parsed",
  "model": "gpt",
  "parties": ["ACME Holdings Ltd", "RENEDOR LLC"],
  "clause_count": 47,
  "eridian": {
    "request_id": "axm_req_legal_4821",
    "route": "gpt",
    "pii_redacted": 12,
    "region": "me-central-1",
    "cost_usd": 0.42
  }
}

Model policy

IDTypical Legal use
autoPolicy selects GPT or Gemini
gptReview memos, redlines, mapping narratives
geminiLong MSA and exhibit packs

Pin GPT for any workflow a committee will read. Run Evaluations before you require a template in production.

Privilege and logging

Set privilege: true on the matter. Audit events keep counts, route, and cost. Prompt text follows the project retention clock. ZDR legal projects do not retain clause bodies after parse. Examiners still get Evidence Export packs without raw client names when PII is masked.

Next steps

Troubleshooting

ErrorFix
403 insufficient_scopeGrant legal:write
413 document_too_largeSplit exhibits; see limits
403 residency_violationDocument region must match the matter pin

Counsel and API questions: contact@geteridian.com.

Production API credentials are issued with an institution workspace. Contact sales if you need access.