Skip to main content

Identity & Ownership

Every AI Product must have a clear logical identity, version-specific semantic identity, and declared ownership.

AIPS distinguishes the continuing Product identity from the immutable semantic description of an individual Product version. This permits a Product to evolve while preserving historical classifications and evidence.


Identity Requirements​

An AI Product must declare:

  • Product Name — a human-readable name;
  • Product ID — a persistent machine-readable logical identifier (URI/UUID);
  • Version — the exact Product version described by this AIPROD;
  • Versioned Descriptor ID — an immutable identifier for the AIPROD artifact;
  • Description — consumer-readable summary of the intelligence proposition;
  • AI Product Type (APT) — Primary or Derived for this qualified version;
  • Intelligence Capabilities (IC) — one or more declared intelligence capabilities;
  • Tags / Keywords — for search and discovery.

Product identity must not be conflated with Product Family classification. A logical Product can retain identity across versions even if a later version requires different intrinsic classification.


Ownership Requirements​

An AI Product must declare:

  • AI Product Owner (AIPRO) — the accountable Product authority;
  • Maintainers — technical/operational maintainers appropriate to the Product constitution;
  • Contact Information — support and escalation channels;
  • Providing Organization — the accountable legal or organizational entity.

Ownership includes accountability for:

  • material intelligence changes;
  • Product Family qualification and revalidation;
  • upstream dependency adoption;
  • model retraining where applicable;
  • knowledge/rule changes where applicable;
  • prohibited uses and policy compliance;
  • deprecation and retirement.

Provenance Linkage​

Identity must link to Lineage & Provenance, including as applicable:

  • AI Primitives and Model Products;
  • Data Products and data sources;
  • knowledge, rules, tools, and services;
  • upstream AI Products;
  • contributors and responsible teams;
  • creation/release dates;
  • superseded and derived Product versions.

Example​

AI Product: Enterprise Sentiment Intelligence

  • Product ID: urn:aiprod:enterprise-sentiment
  • Version: 1.2.0
  • AIPROD ID: urn:aiprod:enterprise-sentiment:1.2.0:aiprod
  • APT: Primary
  • IC: Classification + Detection
  • Owner: AI Product Owner, ACME Corp.
  • Maintainers: NLP Engineering Team
  • Qualification: Current; directly backed by a registered learned model primitive

Summary​

  • Logical Product identity persists across time.
  • AIPROD describes one immutable Product version.
  • Ownership establishes accountability for both Product evolution and AI qualification.

Principle: Identity tells us which Product this is; versioned ontology tells us what this Product version currently is.