AI9GM readiness assessment

Judge current maturity and a target for each of the six layers. You get a gap per layer. No number stands for the organization.

What you do

If you do not know the five levels yet, start with the maturity descriptors. Then answer each question below with one of those levels. A layer's current level is the lowest of its three answers. After the eighteen, choose a target for each layer.

Read the five maturity levels

Limitation

Three questions per layer against five descriptors is an indicative profile. A reader who carries it into an organization as an audit has been misled by the instrument.

The eighteen questions

Each answer is one of Initial, Managed, Defined, Quantitatively Managed or Optimizing.

Layer 1. Foundation

  1. How does infrastructure currently learn that a production AI model exists?
  2. How are production models and datasets held as configuration items against a published capacity model?
  3. How is AI capacity provisioned relative to forecast?

Layer 2. Structural

  1. How do third-party model APIs currently enter the estate?
  2. How is the catalog of interfaces that AI systems consume maintained?
  3. How are interface specifications produced at release?

Layer 3. Intelligence

  1. How are production models listed today?
  2. Who authorizes deployment of a model the Model Owner validated?
  3. How is validation evidence produced at deployment?

Layer 4. Control

  1. How does AI policy currently attach to systems the organization can list?
  2. How is each production AI system registered and named to one accountable person?
  3. How does verification receive evidence about production AI systems?

Layer 5. Execution

  1. How do AI initiatives currently start?
  2. When is the Business Accountable Executive named relative to the first gate?
  3. How are production-gate entry conditions enforced?

Layer 6. Strategic

  1. What currently directs which AI work the organization pursues?
  2. How does a funding decision relate to a stated strategic outcome?
  3. How is strategy revised from measured outcomes?

Layer 1. Foundation

How does infrastructure currently learn that a production AI model exists?
How are production models and datasets held as configuration items against a published capacity model?
How is AI capacity provisioned relative to forecast?

Layer 2. Structural

How do third-party model APIs currently enter the estate?
How is the catalog of interfaces that AI systems consume maintained?
How are interface specifications produced at release?

Layer 3. Intelligence

How are production models listed today?
Who authorizes deployment of a model the Model Owner validated?
How is validation evidence produced at deployment?

Layer 4. Control

How does AI policy currently attach to systems the organization can list?
How is each production AI system registered and named to one accountable person?
How does verification receive evidence about production AI systems?

Layer 5. Execution

How do AI initiatives currently start?
When is the Business Accountable Executive named relative to the first gate?
How are production-gate entry conditions enforced?

Layer 6. Strategic

What currently directs which AI work the organization pursues?
How does a funding decision relate to a stated strategic outcome?
How is strategy revised from measured outcomes?

Answer all eighteen questions to see the profile.

If you cannot record answers here

The scored profile runs in the browser. Treat the questions as a worksheet. Match each answer to a level on the maturity page. If a layer sits at Level 1 or 2, read the minimum viable set.