Skip to main content

Introduction to AIPS

The AI Product Specification (AIPS) is an open, non-profit standard designed to define how AI capabilities are described, deployed, and governed in a self-service, trustworthy, and future-proof way.

AIPS sits within the KivaNura Foundation’s family of specifications, alongside the Base Product Specification (BPS). Where BPS provides the meta-framework for describing any type of product, AIPS specializes this foundation for AI Products.

Why AIPS?​

AI systems are rapidly evolving beyond standalone models or assets. To be treated as true products, they must be:

  • Self-service → easily consumable without bespoke integration.
  • Trustworthy → governed with ethics, risk management, and transparency.
  • Discoverable → searchable, cataloged, and interoperable across domains.
  • Evolvable → capable of continuous monitoring, retraining, and versioning.

AIPS formalizes these characteristics so that AI can be consumed responsibly in enterprises, open ecosystems, and future AI-driven societies.

Scope of AIPS​

AIPS covers two complementary specifications:

  • AIPDS (AI Product Deployment Specification) → how an AI Product is deployed, served, secured, monitored, and evolved.
  • AIPROD (AI Product Semantic Specification) → how an AI Product is described, cataloged, governed, and discovered.

Together, they provide a complete lifecycle view: from semantics and intent, to deployment and observability.

Guiding Principle​

The load-bearing principle of AIPS is self-service.
Every AI Product must be packaged and exposed so that others — humans, agents, or systems — can consume it without special arrangements. This mirrors the philosophy behind Data Products but extends it with AI-specific needs such as capability types, governance, and explainability.

Relationship to BPS​

  • BPS defines the meta-model (product, deployment, governance).
  • AIPS applies BPS concepts to the domain of AI, introducing new elements such as:
    • Capability Types (Model, Agent, AGI/ASI, Hybrid).
    • Risk & Prohibited Use declarations.
    • Explainability, Fairness, and Drift Monitoring.

This ensures AIPS is interoperable with other product specifications while addressing the unique requirements of AI.

Who Should Use AIPS?​

  • Practitioners → AI/ML engineers, product managers, data scientists deploying models or agents.
  • Governance bodies → compliance, ethics boards, regulatory alignment.
  • Researchers & Standards groups → extending AI specifications in an interoperable way.
  • Ecosystem builders → catalogs, marketplaces, and platforms hosting AI Products.

AIPS is an open specification under continuous evolution.
It invites collaboration and aims to help society benefit responsibly from AI — today and in the decades to come.