Skip to main content

AIPCH20 — Interoperable & Composable

“Can Be Composed and Orchestrated with Other Products”

What AIPCH20 is really asserting​

AIPCH20 is not asserting that:

“The AI Product can integrate with other systems or APIs.”

It is asserting that:

The AI Product can participate as a first-class component in larger compositions — where it can be combined with other AI Products and Data Products through its defined ports, semantics, and contracts, without requiring bespoke engineering or procedural orchestration.

Integration is not composability.
Composability is product-level interoperability.

The Essence (HDIP + AIPS Interpretation)​

An AI Product is interoperable and composable if and only if:

  1. It exposes standardized, semantically aligned ports (AIPCH11 + AIPCH12)
  2. It can be combined with other products declaratively (AIPCH04)
  3. Composition does not require custom engineering or tight coupling

If combining products requires:

  • writing pipelines
  • building custom integrations
  • manual orchestration logic
  • deep knowledge of internal implementation

then AIPCH20 is not met, even if integration is possible.

What Composability Means in AIPS​

Composability is the ability to:

1. Combine AI Products​

  • chain capabilities
  • delegate decisions
  • orchestrate reasoning

2. Combine AI and Data Products​

  • consume structured data
  • enrich decision-making
  • provide contextual inputs

3. Participate in Agent-Based Systems​

  • act as tools or capabilities within agents
  • support dynamic reasoning systems
  • integrate into multi-agent compositions

👉 This ensures:

AI Products operate as building blocks in an ecosystem

Positive Criteria — When AIPCH20 is met​

AIPCH20 is met when all of the following are true:

1. Ports are compatible and standardized​

The AI Product exposes:

  • well-defined interfaces (AIPCH11)
  • consistent contracts
  • stable input/output schemas

These enable:

  • plug-and-play consumption
  • predictable interaction

2. Semantics are aligned​

The AI Product:

  • uses shared ontology (AIPCH12)
  • ensures meaning is consistent across products

This enables:

safe composition without misinterpretation

3. Composition is declarative, not procedural​

Products can be combined by:

  • referencing capabilities
  • defining intent and dependencies

Not by:

  • writing orchestration code
  • designing pipelines

This aligns with:

AIPCH04 — Declarative Intelligence Composition

4. No bespoke integration is required​

Consumers or creators do not need to:

  • write custom adapters
  • transform data manually
  • handle special cases

The product works as-is within compositions.

5. Composition preserves product boundaries​

Each AI Product:

  • remains independently governed
  • retains its policies (AIPCH10)
  • exposes its own trust signals (AIPCH07)

Composition does not:

  • merge products into opaque systems
  • break accountability

Negative Criteria — When AIPCH20 is not met​

AIPCH20 is not met if any of the following are true:

❌ Integration requires custom engineering​

Examples:

  • writing glue code
  • building pipelines
  • manual data transformation

This is integration, not composability.

❌ Products are tightly coupled​

Examples:

  • one product depends on internal details of another
  • changes in one break others

This prevents independent evolution.

❌ Semantic mismatch exists​

Examples:

  • inconsistent definitions across products
  • unclear meaning of inputs/outputs

This makes composition unsafe.

❌ Composition logic is embedded in implementation​

Examples:

  • orchestration hidden in code
  • pipelines define composition

This violates declarative model.

❌ Governance is not preserved in composition​

Examples:

  • policies bypassed in combined systems
  • trust signals lost or ignored

This breaks system integrity.

Edge Cases (Important Guidance for Agents)​

Case 1: “API integration with custom adapters”​

❌ Not met

Rationale:

  • integration exists
  • composability absent

Case 2: “Products reusable but not composable”​

⚠️ Partial

Rationale:

  • reuse (AIPCH08) exists
  • but composition requires effort

Case 3: “Declarative composition via product references”​

✅ Met

Rationale:

  • no engineering required
  • platform handles orchestration

Case 4: “Agent-based composition using AI Products as tools”​

✅ Met

Rationale:

  • aligns with agent ontology
  • supports dynamic composition

Evidence Signals an Agent Should Look For​

Authoritative evidence:​

  • ability to reference other AI/Data Products in AIPROD
  • composition defined declaratively
  • no custom integration code

Supporting evidence:​

  • multiple composed AI Products
  • reuse in agent systems
  • consistent schema and semantic alignment

Red flags:​

  • reliance on pipelines or workflows
  • custom integration layers
  • semantic inconsistencies
  • tight coupling between products

How an Agent Should Decide​

Decision rule (simplified):

If combining the AI Product with other AI or Data Products requires procedural integration, custom engineering, or semantic interpretation, AIPCH20 is not met.

Why AIPCH20 Is Non-Negotiable​

Without AIPCH20:

  • AI Products remain isolated
  • reuse is limited
  • ecosystem cannot emerge
  • scalability is constrained

AIPCH20 enables:

  • AI ecosystems and marketplaces
  • agent-based architectures
  • scalable intelligence composition
  • network effects across products

Canonical Statement (for AIPS)​

AIPCH20 is satisfied only when an AI Product can be combined with other AI and Data Products through its defined ports, semantics, and contracts in a declarative manner, without requiring custom engineering or procedural orchestration, while preserving independent governance, trust, and product boundaries.