Overview — AI Product Semantic Specification (AIPROD)
The AI Product Semantic Specification (AIPROD) defines the versioned semantic description of an AI Product.
Where AIPDS explains how a qualified AI Product is realized and operated, AIPROD describes what the Product version means, why it qualifies as an AI Product, how its intelligence is constituted, and how consumers and governance systems should interpret it.
AIPROD specializes the AI Product Ontology and the general Product concepts inherited from BPS.
Purpose of AIPROD
AIPROD provides a structured semantic contract for the identity, purpose, AI qualification, intrinsic AI Product Type, intelligence characteristics, dependencies, governance, and lifecycle of an AI Product version.
It ensures that each version can be:
- Self-describing — with human- and machine-readable semantics;
- Qualification-backed — with evidence showing why it currently belongs to the AI Product Family;
- Discoverable — through catalogs and marketplaces using orthogonal ontology dimensions;
- Governed — through explicit risk, policy, agency, effect, and embodiment declarations;
- Comparable — without collapsing Product Type, capability, mode, and risk into one field;
- Composable — with versioned dependency and intelligence provenance semantics;
- Impact-analyzable — when an upstream Product or AI Primitive changes.
Scope of AIPROD
AIPROD covers the semantic backbone of an AI Product, including:
- Identity & Ownership
- Purpose & Intent
- AI Product Qualification & Evidence
- AI Product Type (APT)
- Intelligence Capability (IC)
- Inputs & Outputs
- Lineage & Provenance
- Metadata & Self-Description
- Governance & Policy
- Quality Metrics
- Observability
- Interoperability & Standards
- Economic Model
- Discoverability & Cataloging
- Lifecycle & Versioning
- Explainability & Transparency
- Composability
- Prohibited Uses
- Extension Schema
- Examples
Core Ontology Dimensions
AIPROD keeps the following dimensions separate:
| Dimension | Question |
|---|---|
| AQ / AQE | Why does this Product version qualify as an AI Product? |
| APT | Where does its productized intelligence originate? |
| IC | What intelligence does it provide? |
| CS | Who / where consumes it? |
| PU | Why is it consumed? |
| MO | How is it invoked or operated? |
| IE | What does it produce or cause? |
| AR | What authority does it exercise? |
| EM | How does it relate to the physical world? |
| RT | What AIPS risk tier applies? |
These dimensions are orthogonal. For example, Primary + Detection + Real-time + Recommend + Non-embodied + R3 is a coherent AI Product characterization.
Relationship to BPS and AIPDS
- BPS defines universal Product principles and Product Intrinsic Nature (PIN).
- AIPS AI Product Ontology establishes the conditions for AI Product Family membership and AI-specific ontology dimensions.
- AIPROD records the versioned semantic contract and qualification evidence for a qualified AI Product.
- AIPDS records deployment and execution realization for that Product version.
Together, AIPROD and AIPDS describe the semantic and operational realization of the AI Product without conflating Product identity with a particular model, runtime, or interface.
Principles of AIPROD
- Qualification before classification — APT is assigned only after AI Product qualification is established.
- Evidence-backed AI-ness — AI Product Family membership must be supported by current, version-specific evidence.
- Orthogonal semantics — Product Type, capability, agency, embodiment, effect, mode, and risk remain separate dimensions.
- Version immutability — a published AIPROD describes one immutable Product version; material change creates a new descriptor/version.
- Dependency truth — lineage records versioned Data Products, Model Products, AI Primitives, and upstream AI Products as distinct dependency kinds.
- Currentness — qualification and derived classifications must be revalidated when constitution changes materially.
- Interoperability — semantics must remain machine-readable and suitable for catalogs, policy engines, knowledge graphs, and orchestration.
Summary
- AIPROD = the versioned semantic contract of a qualified AI Product.
- It records identity, intent, AI qualification, APT, intelligence capabilities, lineage, governance, and interoperability.
- It complements AIPDS while remaining independent of deployment implementation.
Principle: AIPROD should make both the Product's meaning and the evidence for its AI Product classification computable.