Skip to main content

Glossary — AI Product Deployment Specification (AIPDS)

This glossary provides definitions of deployment-specific concepts in AIPDS.
For general AI Product terms, see the AI Product Glossary.

AIPDS Terms​

Deployment Target​

The environment where an AI Product can be deployed (e.g., cloud, on-prem, edge, hybrid, agent runtime).
See Deployment Targets.

Packaging​

The technical delivery format of an AI Product (e.g., API, container, SDK, model package).
See Packaging & Interfaces.

Inference Mode​

The way an AI Product provides outputs, such as batch, online (real-time), streaming, or hybrid.
See Inference & Scaling.

Scaling Strategy​

The method by which an AI Product adapts to demand (e.g., horizontal, vertical, elastic, edge).
See Inference & Scaling.

Drift​

A change in data distribution, model performance, or system behavior that reduces reliability over time.
See Retraining & Continuous Learning.

Retraining Strategy​

The declared approach for updating an AI Product to address drift, including periodic, triggered, online, or human-in-the-loop retraining.
See Retraining & Continuous Learning.

Security Controls​

Mechanisms for authentication, authorization, entitlements, encryption, and abuse prevention.
See Security & Access Controls.

Risk Classification​

Categorization of AI Products into risk tiers (Minimal, Limited, High, Unacceptable), determining compliance requirements.
See Governance & Compliance.

Economic Model​

The declared cost and resource usage structure of an AI Product (e.g., consumption-based, provisioned, tiered).
See Economic & Resource Model.

Composability​

The ability of an AI Product to be chained, orchestrated, or combined with other products while preserving governance and policy inheritance.
See Integration & Composability.

Lifecycle State​

The phase of maturity of an AI Product: Experimental, Development, Production, Deprecated, Retired.
See Versioning & Lifecycle.

Cross-Reference​

  • AI Product Glossary → General terms for AI Products
  • AIPROD Glossary → Semantic concepts like identity, purpose, explainability, prohibited uses

Summary​

This glossary defines the deployment-specific vocabulary required to apply AIPDS.
Together with the AI Product and AIPROD glossaries, it ensures clarity across what an AI Product is and how it is deployed.