Statutory Infrastructure for AI & Analytics
Born from a passion for legal technology and compliance, Keystone compiles and sanitizes vast public law corpora for enterprise use. Our founder, John Dawson, combined his legal operations and data engineering background to solve the "last mile" problem of legal data ingestion. By handling the time consuming ingestion, validation, and schema mapping of statutory frameworks, we allow data scientists and legal engineering teams to focus strictly on building the future of legal intelligence.
Corporate Profile & Engineering Mandate
While the raw text of enacted United States federal, state, and territorial statutes belongs to the public domain under the government edicts doctrine, official legislative publications are fragmented across 57 incompatible publishing systems ranging from legacy SGML/RTF dumps and deeply nested HTML trees to bespoke XML dialects. Worse still, many online legislative portals interleave public-domain statutory text with third-party copyrighted publisher annotations, case notes, and editorial commentaries.
Keystone Data Services LLC was founded in Yuma, Arizona in 2026 to solve this bottleneck. Rather than wrapping a search engine around raw scrapes and charging recurring per-query API tolls, Keystone delivers the actual underlying statutory datasets normalized to a single universal schema, validated against international legislative XML standards, shredded to the individual paragraph and clause level, and sealed with end-to-end SHA-256 cryptographic lineage.
Every dataset purchase grants a perpetual, worldwide commercial snapshot license to the Bronze, Silver, and Gold archives to deploy however you see fit. You are not paying for a subscription, you are owning the data.
Zero Copyrighted Publisher Annotations or UCC Comments
Raw statutory portals frequently embed copyrighted Uniform Commercial Code (UCC) Official Comments (© by ALI and NCCUSL) as well as commercial publisher case notes and treatise references (West, LexisNexis, Matthew Bender, ALR). Keystone's deterministic hygiene engine strips your data of third-party copyrighted commentary so your LLM training and commercial products remain free of copyright contamination.
Strict Separation of Enacted Law & Legislative History
Non-statutory state Revisor, Compiler, and Code Commissioner commentaries are isolated and stripped from statutory body text, while authentic legislative session-law enactment and amendment history citations (e.g., Pub. L., Acts, Ch.) are extracted into a dedicated history_note column.
How Keystone Compares to Non-Commercial Datasets & Metered APIs
| Evaluation Dimension | Open Non-Commercial Datasets (CC BY-NC) | Metered Commercial Legal APIs | Keystone Legal Data (10/2026 Medallion Edition) |
|---|---|---|---|
| Commercial License & AI Rights | Restricted to Non-Commercial (CC BY-NC 4.0) use. Prohibited in commercial LegalTech products, law firm AI, or enterprise RAG without negotiating a separate commercial license. |
Subscription or per-call credit metering; typically prohibits bulk caching, offline indexing, or model training/fine-tuning. | Perpetual Commercial Dataset License ($2,000 single jurisdiction / $25,000 all 57). Full rights for commercial RAG, vector indexing, LLM training, and client-facing applications. It is all yours from source to finish with verifiable provenance. |
| Statutory Completeness | Often incomplete; commonly drops entire titles, unlinked containers, or territorial codes where scraper URLs fail. | Opaque coverage; dependent on vendor endpoint availability and rate limits. | Enacted statutory sections across all 57 / 57 US jurisdictions with 100% active container coverage (4,048 active titles, codes, and divisions). |
| Structural Granularity | Flat section-level text blobs; subclauses (a)(1)(A) are collapsed into single strings, forcing teams to write brittle chunkers. |
JSON text responses at the section or document level. | Both section-level (sections.parquet) and 12,613,059 hierarchically shredded paragraphs (subsections.parquet) with explicit enum_label, indent_level (1–6), and deterministic sequence order. |
| Legislative XML Standard | None (flat Parquet/CSV or raw HTML). | Proprietary JSON wrapper. | XSD-validated OASIS Akoma Ntoso 3.0 (LegalDocML) XML files delivered in the Silver archive alongside the Bronze raw sources and Gold Parquet tables. |
| Citation Graph Resolution | Typically limited to basic regex extraction of federal U.S.C./C.F.R. strings without intra-state target validation. | Rarely exposed as a bulk downloadable graph topology. | Pre-resolved intra-jurisdiction directed edges (edges.parquet) with 100.00% target node closure against sections.parquet. |
| IP & Editorial Hygiene | Minimal filtering; high risk of residual UCC Official Comments or publisher notes leaking into embeddings. | Varies by endpoint. | Zero-Trust Hygiene Audited: Stripped of copyrighted UCC Official Comments (ALI/ULC), West/Lexis annotations, and Revisor commentaries. |
| Deployment & Cost Predictability | Free for academic/non-commercial exploration; commercial teams are funneled into recurring API contracts. | Recurring monthly/annual platform fees plus per-request credit consumption and external network latency. | One-Time Capital Purchase, Zero API Lock-In. Deploy directly inside your own Snowflake, Databricks, DuckDB, Postgres, or air-gapped VPC environment. |
Why Comparing Keystone to Open Non-Commercial Compilations Is an Apples-to-Oranges Comparison
Prospective buyers occasionally ask how Keystone Legal Data differs from open-access projects. While open compilations are useful for non-commercial exploration, they serve a fundamentally different purpose and licensing model than an enterprise Medallion corpus:
1. Non-Commercial Restriction vs. Perpetual Commercial Ownership: Some companies license dataset compilations and schemas under CC BY-NC 4.0 (Non-Commercial), serving as a lead-in to a commercial API and credit-metered platform. Commercial engineering teams cannot legally deploy a CC BY-NC dataset in production software or commercial AI pipelines. Keystone sells a direct, perpetual commercial license to the complete underlying data archives for a flat one-time fee.
2. Coverage & Structural Depth: Whereas some others publish their statutory sections in a flat table (explicitly omitting sections where source URLs could not be linked), Keystone's 57-jurisdiction corpus delivers 2,129,709 enacted statutory sections across 100% of active statutory containers, plus 2,129,709 OASIS Akoma Ntoso 3.0 XML files, 12,613,059 hierarchically shredded subsection records, and 2,556,423 pre-resolved intra-state citation graph edges.