Skip to main content

AIPCH05 — Discoverable

“Registered in AI Marketplace or Catalog”

What AIPCH05 is really asserting​

AIPCH05 is not asserting that:

“The AI Product is listed somewhere in a catalog.”

It is asserting that:

The AI Product is intentionally published into an enterprise discovery surface where unknown consumers can independently find, understand, evaluate, and adopt it — without relying on prior relationships, tribal knowledge, or direct interaction with the producer.

Discovery is about independent accessibility, not mere registration.

The Essence (HDIP + AIPS Interpretation)​

An AI Product is discoverable if and only if:

  1. It is published into a shared, searchable discovery surface
  2. It can be understood and evaluated without producer interaction
  3. It can be adopted by previously unknown consumers

If access depends on:

  • knowing the team that built it
  • informal communication
  • internal referrals
  • undocumented knowledge

then AIPCH05 is not met, even if a catalog exists.

Positive Criteria — When AIPCH05 is met​

AIPCH05 is met when all of the following are true:

1. The AI Product is intentionally published​

The product is:

  • registered in an enterprise AI Marketplace or Catalog
  • visible beyond the producing domain
  • treated as a product for discovery, not an internal artifact

Publication is explicit, not accidental.

2. Discovery is independent of relationships​

A consumer can:

  • find the AI Product without knowing the producer
  • search by:
    • capability (e.g., “fraud detection”, “credit scoring”)
    • domain
    • use case
    • modality or intelligence type
  • evaluate it without contacting the owner

This eliminates tribal dependency.

3. Discovery is capability-centric, not implementation-centric​

The AI Product is described in terms of:

  • what it does (capability)
  • what decisions it supports
  • who it is for

Not:

  • model types
  • pipelines
  • technical architecture

The product is discoverable as intelligence, not infrastructure.

4. Sufficient signals exist for evaluation​

The discovery surface exposes:

  • product purpose and usage context
  • trust signals (quality, risk, compliance)
  • ownership (AIPRO)
  • access methods (ports)
  • maturity indicators (PMDD score/level)

Consumers can make informed adoption decisions.

5. Discovery supports unknown future use​

The AI Product is:

  • not tied to a single predefined consumer
  • described generically enough to enable reuse
  • positioned as a reusable capability

Discovery is open-ended, not pre-wired.

Negative Criteria — When AIPCH05 is not met​

AIPCH05 is not met if any of the following are true:

❌ Product is known only through informal channels​

Examples:

  • “Ask the ML team”
  • “Check with the platform team”
  • knowledge spread through meetings or Slack

This is tribal discovery, not product discovery.

❌ Catalog exists but is not used or trusted​

Examples:

  • outdated entries
  • missing metadata
  • no trust signals
  • consumers ignore it and go directly to teams

This is a catalog in form, not in function.

❌ Product is described in technical terms only​

Examples:

  • “XGBoost model v3”
  • “LLM pipeline”
  • “Inference service”

This prevents business users from discovering it.

❌ Discovery requires prior access or permission​

Examples:

  • products only visible after onboarding
  • restricted visibility without reason
  • hidden behind team boundaries

This breaks discoverability.

Edge Cases (Important Guidance for Agents)​

Case 1: “Product listed but minimal metadata”​

⚠️ Partial

Rationale:

  • discoverable in form
  • not evaluable in practice
  • requires enhancement

Case 2: “Searchable but only by technical tags”​

❌ Not met

Rationale:

  • discoverable only by engineers
  • not accessible to business users

Case 3: “Marketplace with rich capability + trust signals”​

✅ Met

Rationale:

  • independent discovery
  • informed evaluation
  • reusable adoption

Case 4: “Private or restricted AI Products”​

⚠️ Context-dependent

Rationale:

  • acceptable for sensitive domains (e.g., regulatory, security)
  • but must still be:
    • discoverable within authorized scope
    • not hidden entirely

Evidence Signals an Agent Should Look For​

Authoritative evidence:​

  • AI Marketplace / Catalog entry exists
  • searchable by capability and domain
  • includes AIPRO, trust signals, and usage context

Supporting evidence:​

  • adoption by multiple consumers
  • search analytics showing discovery usage
  • structured metadata aligned to ontology

Red flags:​

  • reliance on personal communication
  • missing or outdated catalog entries
  • technical-only descriptions
  • no trust or usage signals

How an Agent Should Decide​

Decision rule (simplified):

If a consumer cannot independently discover, understand, and evaluate the AI Product without interacting with the producer, AIPCH05 is not met.

Why AIPCH05 Is Non-Negotiable​

Without AIPCH05:

  • reuse does not scale
  • duplication increases
  • domains operate in silos
  • AI capabilities remain hidden assets

AIPCH05 enables:

  • enterprise-wide reuse
  • capability marketplaces
  • composability at scale
  • informed consumer choice

Canonical Statement (for AIPS)​

AIPCH05 is satisfied only when an AI Product is intentionally published into an enterprise discovery surface where it can be independently found, understood, and evaluated by unknown consumers based on its declared capabilities, trust signals, and usage context, without reliance on prior relationships or tribal knowledge.