# Omniarch

> Omniarch provides a REST API and bulk export of structured U.S. case law data — semantic triples, citation networks, and judicial speech acts — extracted from federal and California appellate decisions, for use in AI applications and legal research tools.

Omniarch builds legal knowledge graphs from court opinions using a 14-layer NLP pipeline. Outputs include: subject-action-object triples with modality and polarity labels, citation treatment relationships (followed, distinguished, overruled), entity resolution across opinions, and judicial speech act classification (holding, dicta, order, reasoning).

Data is delivered via REST API (JSON) or bulk JSONL export. Primary use cases: AI hallucination detection, legal grounding layers, citation graph analysis, and structured precedent research.

Coverage: U.S. federal circuits 1951 to present, California appellate 2000 to present. Additional state coverage available on request.

## Product

- [Live Demo](/demo/): Interactive citation verification demo — select a real attorney misconduct case (Mata v. Avianca, Fivehouse v. DoD, Mavy v. Commissioner, Johnson v. Dunn, Chevron overruled authority), click any citation, and see the KG-verified proof chain. Runs locally in-browser using SQLite WASM against a verified snapshot of the federal case law database.
- [Data Overview](/data.html): Data architecture, pipeline description, accuracy metrics, and coverage details
- [Use Cases](/use-cases.html): Primary use cases for AI vendors and law firms
- [Request Access](/request-access.html): Contact form for API access, data samples, and technical walkthroughs

## Research

- [Why RAG on Legal Documents Isn't Legal Verification](/research/rag-is-not-legal-verification.html): Six structural failure modes of RAG for legal authority verification. Cites Stanford empirical data (Westlaw 33% hallucination rate, Lexis+ AI 17%), arXiv research on citation grounding and graph-augmented retrieval. Explains why semantic similarity cannot substitute for structured assertion lookup.
- [Research Index](/research.html): All published and forthcoming research on legal AI verification, extraction methodology, and graph architecture.

## Company

- [About](/about.html): Team, mission, and what has been built
- [Terms of Service](/terms.html): Data licensing and API use terms
- [Privacy Policy](/privacy.html): Data handling practices

## API

- [OpenAPI Specification](/openapi.json): Machine-readable REST API specification (JSON, OpenAPI 3.1)