Pilot Pattern

The 30-Day SDC Audit

A procurement-defensible pilot. Zero production impact. Board-readable deliverable. Empirical exit at day 30.

The procurement frame

The most common procurement objection to SDC is not technical and is not pricing. It is the fear of a six-month commitment ending in a disrupted production system and nothing to show for it.

Both halves of that fear are misplaced. The smallest viable SDC pilot takes 30 days, runs in parallel with production, and produces a deliverable a board can read.

The cost is engineer time for setup, typically one to two days, plus the audit at day 30. There is no vendor lock-in: sdcgovernance is Apache 2.0 on PyPI and runs anywhere Python runs. There is no data migration: the audit observes the decision flow you already have. There is no system replacement: sdcgovernance runs alongside production, not in it.

SDCStudio's generated artifacts are yours to keep. At the end of the on-ramp you download the composed data model, a template SDC4 instance, and an Apache 2.0 app stack scaffolded for the stack profile you pick: a fully FOSS profile (Apache Jena/Fuseki + PostgreSQL), an Enterprise profile (Graphwise GraphDB + SirixDB + Keycloak), or both. Every later app you generate imports into the original stack, so infrastructure is procured once across the portfolio rather than per pilot.

The technology overview

Where SDCStudio fits, what the generated app stack contains, and how sdcgovernance reads from the knowledge graph to emit the Receipt corpus.

SDC pilot technology overview: datasource flows through agents, SDCStudio composes domain plus ProvGov components, the generated app stack holds the data model, template SDC4 XML, app scaffolding, and knowledge graph (FOSS or Enterprise profile); sdcgovernance reads from the knowledge graph and emits a hash-chained Receipt corpus as the day-30 deliverable.
External input: your datasource, untouched.
SDC products: Agents, SDCStudio, sdcgovernance. Apache 2.0.
Generated artifacts: the app stack you walk away with.
Audit deliverable: the Receipt corpus at day 30.

What the pilot actually does

1

Set up the substrate

Pick a domain library (NIEM, FHIR, NIH CDE, Default), layer the ProvGov components on top, point an agent at one datasource, get validated SDC4 instances. One to two days for most pilots.

2

Install sdcgovernance

pip install sdcgovernance. Configure the DMN decision table that mirrors the rules your decision flow already follows.

3

Observe for 30 days

Every decision in the flow generates a tamper-evident Receipt. Hash-chained. Bound to the SDC instance that produced it. Read-only against your existing system.

4

Read the corpus

Three questions: complete provenance, gaps, would-have-been-refused. Either you exit with documented governance or with a specific Phase 2 scope.

Why this works as a pilot

Zero production impact

sdcgovernance runs alongside, not in. Receipts are observational. Nothing the existing system does changes.

Defined exit criteria

At day 30 you have a Receipt corpus. The exit is empirical, not narrative.

Procurement-defensible

No vendor lock-in. No data migration. No system replacement. Apache 2.0 means no paid commitment to Axius SDC.

Auditable artifact

The Receipt corpus is a board-readable deliverable. Hash-chained, structurally consistent, portable.

The substrate prerequisite, named honestly

sdcgovernance evaluates DMN decision tables against SDC instances and reads governance evidence from the ProvGov layer inside those instances. The Receipts it produces reference instance_id and instance_version on every decision, plus the provenance, attestation, party, and entity-state hashes the ProvGov components carry. The substrate is what gives the audit something to point at.

An SDC instance composed for governance has two component layers: domain (what the records are) and ProvGov (what makes them governable). Both ship as public libraries on SDCStudio. The substrate setup used to mean a modeling sprint. It does not anymore. Three affordances have collapsed that work into something a small team can complete in a day or two before the audit period begins.

A 30-day, zero-impact, board-readable pilot

should not be the kind of commitment that is hard to approve.

Start with the prerequisites