Standards and ecosystem

An assurance layer beside the systems that create and publish law.

LawVM sits beside drafting editors, XML standards, identifier vocabularies, publishing systems, and official portals. Through source-specific adapters, it can consume versioned artifacts, reconstruct supported transitions, and return inspectable evidence.

System fit

Preserve each layer’s job.

gazette / source archive→structure + identity→drafting / publication→LawVM shadow assurance→evidence / PIT state / review queue

Layer	Examples	LawVM relationship
Source capture	Gazettes, scans, PDFs, official archives	Preserves exact manifestations, provenance, extraction witnesses, and missing-source accounting.
Document structure	Akoma Ntoso, local XML	Consumes structured legal units where available; does not claim the format itself proves semantic correctness.
Identity and metadata	ELI, ELI impact metadata, local identifiers	Uses work/expression/manifestation identity, versions, relationships, and effects as typed evidence.
Drafting	LEOS, parliamentary editors	Can add pre-publication target, transition, cross-reference, temporal, and mutation-boundary checks when a concrete adapter exists.
Consolidation and publication	Indigo / Laws.Africa, official CMSs	Runs independently in no-write shadow mode and returns a classified evidence queue.
Public legal access	legislation.gov.uk, official portals and APIs	Can receive source-linked historical state, explicit residuals, and transition artifacts from a supported frontend.

Three entry paths, one model

Meet the source where it is.

Source bootstrap

Paper, scans, PDFs, and incomplete archives first yield preserved manifestations, extraction witnesses, structure evidence, and an explicit blocked-source map.

Pre-publication QA

A versioned drafting or publication export can receive declared target, reference, legal-time, language, and mutation checks through a concrete adapter.

Published-text audit

A bounded set of amendment sources and published states can support independent replay, classified differences, and a review queue.

Shared decision chain

Each route moves from declared sources and a model to QA checks, then a strict or observe-only disposition and a named human decision. LawVM does not automatically publish or correct text.

Future strict-authoring direction: legislation co-produced with a formally specified machine artifact could make consolidation automatic and machine-checkable within its accepted formal domain. This requires source closure, human-text/artifact correspondence, compiler and replay proofs, independent checking, and rejection or blocking of unsupported input. See the full conditional boundary.

Integration patterns

Start where structure and authority already exist.

Akoma Ntoso input

Map stable structural units and metadata into LawVM’s source-document boundary; keep jurisdiction-specific amendment idioms in the frontend.

ELI and effect feeds

Use identifiers and effect relationships as source witnesses and candidates. Promotion into executable transitions still requires target, payload, temporal, and mutation-boundary evidence.

Drafting-system checks

Validate targets, references, effective dates, expression parity, and unintended mutation before publication. This is an integration direction, not a current universal product claim.

Publisher shadow audit

Freeze a source-complete corpus, replay it independently, classify divergence, and return evidence without writing to production.

Relationship status: references to standards and projects describe ecosystem roles and possible interfaces. Treat integration, partnership, certification, or endorsement as present only when explicitly stated.

Next action

Choose a versioned artifact boundary.

Begin integration with one versioned input and one inspectable output: an AKN corpus, ELI effect set, drafting export, publisher transition bundle, or read-only API snapshot.

Define a bounded integration pilot Inspect the compiler boundaries