PMDD Assessment Guidance for AIPCH

This section provides detailed guidance for evaluating AI Products against the AI Product Characteristics (AIPCH01–AIPCH21) using the Product Maturity Driven Development (PMDD) model.

PMDD treats product maturity as a continuously observable, measurable, and improvable system property, rather than a one-time certification.

What This Guidance Is​

This guidance is not a checklist to be completed once.

It is:

A structured interpretation layer that enables consistent, automated, and agent-driven evaluation of AI Product maturity over time.

Each AIPCH characteristic defines:

• what it truly asserts (beyond surface interpretation)
• when it is satisfied or not satisfied
• how an AI agent should evaluate it
• what evidence signals indicate maturity
• why it is non-negotiable for AI Productization

The Role of PMDD in AI Products​

In the AIPS model:

PMDD is the mechanism through which AI Products continuously prove that they are, and remain, products.

PMDD:

• evaluates all AIPCH characteristics continuously
• produces maturity scores and profiles
• detects regressions in productization
• drives backlog prioritization and evolution

Maturity Levels (AI Products)​

AI Products are classified into three maturity levels:

L2 — Production-Grade AI Product (≥ 80%)​

• Fully productized AI capability
• Governed, observable, trustworthy, and reusable
• Suitable for enterprise-scale and critical use cases

L1 — Evolving / MVP AI Product (50–79%)​

• Partial productization
• Core characteristics present but incomplete or inconsistent
• Requires improvement before scaling or critical usage

L0 — AI Asset (< 50%)​

• Not yet a true product
• Lacks key product characteristics (ownership, trust, governance, etc.)
• Typically model-centric, pipeline-centric, or experimental

Key Principle​

AI Products are not defined by their models, but by their product characteristics.

A highly sophisticated model can still be:

❌ an AI Asset (L0)

while a simpler capability can be:

✅ a Production-Grade AI Product (L2)

What Makes AI PMDD Different from Data PMDD​

AI Products introduce additional complexity beyond Data Products:

• Behavioral uncertainty (non-deterministic outputs)
• Risk tiers (R0–R4) influencing governance obligations
• Learning dynamics (drift, retraining, adaptation)
• Compositional intelligence (agents, multi-product orchestration)

Therefore, PMDD for AI evaluates:

• structure (productization)
• behavior (performance, drift, explainability)
• governance (policy, safety, compliance)
• trust (signals, evaluation, risk posture)

Continuous vs Static Assessment​

PMDD rejects static maturity assessment.

AIPCH evaluation must be:

• continuous (e.g., periodic automated evaluation)
• signal-driven (not opinion-based)
• machine-evaluable (not document-driven)

If maturity is:

• manually assessed
• periodically reviewed
• stored in documents

then PMDD is not being applied.

Relationship with AIPROD, AIPDS, and DPP​

PMDD evaluation is grounded in core AIPS artifacts:

• AIPROD → semantic definition of the AI Product
• AIPDS → deployment and execution specification
• DPP (Digital Product Passport) → trust, risk, and compliance signals

Together, these provide the evidence surface for PMDD.

How to Use This Guidance​

For each AIPCH characteristic:

1. Read “What it is really asserting”
2. Evaluate against:
   • positive criteria
   • negative criteria
   • edge cases
3. Inspect evidence signals
4. Apply the AI agent decision rule
5. Determine:
   • Met / Partial / Not Met
6. Feed results into:
   • PMDD scoring
   • maturity level classification
   • improvement recommendations

The List of AIPCH Characteristics​

Final Principle​

PMDD ensures that AI Product maturity is not declared, it is continuously proven.