A legal-AI company (anonymized) · Legal · Text
Indic legal annotation for a legal-AI product
Expert clause-typing and entity annotation across multilingual Indian judgments and contracts for a legal-AI product that couldn't hallucinate.
By Cognegica Quality & Standards · QA & annotation-standards team
Clause & entity IAA 0.89
Languages: Hindi, Marathi, Tamil, English
Anonymised. Client identity is withheld at their request. The methods, gates, and metrics are real.
Challenge
The product's contract-review and case-research copilots failed on Indian legal text: multilingual judgments, code-mixed affidavits and scanned cause-lists. A wrong citation or misread clause was malpractice risk, so the model needed expert-grade ground truth — not crowd labels.
Approach
We assembled legally-trained annotators and a calibrated rubric:
- Clause typing, entity and obligation extraction, and structure labeling across contracts, judgments and pleadings.
- Guidelines co-written with the client's legal SMEs and piloted before scale; two-pass QA with adjudication on contested items.
- Access-controlled handling, de-identification and documented chain of custody for confidential matters.
Outcome
Annotation reached inter-annotator agreement of 0.89 on clause and entity labels across four languages, with a fully documented, rights-cleared and de-identified delivery the client could defend in a procurement and compliance review.
Representative outcome from an anonymized engagement.
How this maps to what we do
The services and data behind this engagement
This outcome was delivered with the same rights-cleared, documented services and datasets you can engage today.
-
Data Annotation
Domain-expert annotation with calibrated rubrics and two-pass QA.
Explore the service -
Legal
How we build trustworthy data for legal AI across Indian languages.
Explore the industry -
Indic Legal Judgments Corpus
Annotated Indian judgments and statutes in the catalog.
View data card
Build legal AI that holds up.
See how we structure engagements and indicative pricing, or tell us your languages, modalities and quality bar for a scoped quote.
Written by
Cognegica Quality & Standards
QA & annotation-standards team
Cognegica Quality & Standards is the internal team that defines and enforces our annotation guidelines, multi-layer QA, native-linguist review and inter-annotator agreement reporting. This is an editable team identity — a named reviewer with a public profile can be assigned to it later in the admin.