ALTERLAYER · WORKING CHECKLIST

The Alterlayer 7-Week AI Governance Implementation Checklist

A practical working checklist for turning AI visibility into ownership, decisions, actions and durable governance records.

A Head of Data & AI, AI Governance Lead, AI Officer, CDAO or equivalent should be able to use this resource during the seven-week operating cycle.

What should we do? Who should be involved? What should exist when we finish this step?

WEEKS 1–2

DISCOVER

Objective

Establish what AI actually needs to be governed.

Checklist

Identify existing sources of AI information: inventories, spreadsheets, project portfolios, policies, assessments and relevant enterprise signals.
Collect an initial set of observations about AI applications, use cases, models, agents and other relevant AI activity.
Identify obvious duplicates and references to already-known AI.
Reconcile observations using: Ignore / Investigate / Merge / Create
Do not automatically convert every discovered signal into governance truth.
Create or confirm the first priority AI Objects.
Record important gaps and unknowns rather than hiding them.

People to involve

AI/Data Lead · Business Owners · Technology · Risk/Compliance where relevant

Exit condition

We have a usable starting view of the AI that matters enough to govern.

WEEKS 3–4

OWN & GOVERN

Objective

Make responsibility and decision-making explicit.

Checklist

Assign one accountable Owner to each priority AI Object.
Identify the appropriate Reviewer where review is required.
Establish who has authority for the consequential decision.
Record why each AI Object currently requires governance attention.
Identify relevant organizational controls or expectations.
Define the minimum review and decision path.
Avoid unnecessary approval-chain complexity.
Identify AI Objects where ownership or responsibility remains unresolved.

People to involve

AI Object Owners · AI/Data Lead · Relevant Reviewer · Decision authority

Exit condition

For each priority AI Object, we know who owns it and how a governance decision should happen.

WEEKS 5–6

DECIDE & RECORD

Objective

Run the operating model on real governance situations.

Checklist

Select a small number of real AI Objects requiring attention.
Capture the situation and relevant evidence.
Obtain the appropriate reviewer judgment.
Keep reviewer judgment distinct from the consequential decision.
Record the decision and decision actor.
Translate required follow-up into explicit Actions.
Assign Action owners where applicable.
Track whether Actions remain open or have been completed.
Retain the relevant Governance Record.

Governance Record check

Can we answer:

What was reviewed?
Who participated?
What was decided?
What actions followed?
When did it happen?

Exit condition

We can reconstruct a real governance decision and what happened afterwards.

WEEK 7

OPERATE

Objective

Turn the initial setup into a repeatable operating rhythm.

Checklist

Review outstanding governance Actions.
Identify AI Objects requiring future review.
Determine how relevant changes will be surfaced.
Establish who handles new observations.
Confirm that ownership remains current.
Define how material changes return to review.
Ensure subsequent decisions remain connected to the governance history.

Exit condition

The governance model can continue operating without restarting the project whenever AI changes.

7-WEEK COMPLETION CHECK

At the end of the cycle, ask:

Do we have a usable view of our priority AI Objects?
Does each priority AI Object have explicit ownership?
Do we know how reviews and decisions happen?
Are required governance Actions assigned and visible?
Can we retrieve the history of important decisions?
Do we know what happens when an AI Object materially changes?

If several answers are No, do not hide the gaps. They become the starting priorities for the next operating cycle.

The operating chain

DISCOVER

Observation
AI Object

GOVERN

Owner
Review
Decision
Actions

RECORD

Governance Record
Change / Monitoring