An initiative's path
From idea to production, every AI initiative follows the same path: the AI Strategic Operating Model (AI-SOM). Here is what is expected of you and what the team provides, phase by phase.
AI Strategic Operating Model (AI-SOM)
The marker found on every record: a pre-phase and four phases, with the expected documents under each. On a record, it shows where the initiative stands.
- Not started : Concept Case
- Not started : Statement of Sensitivity (SoS)
- Not started : Outcome-Based Roadmap
- Not started : High-level Architecture
- Not started : ATIP Assessment
- Not started : Cyber Assessments (SoS, SA&A, SIP, Audits)
- Not started : DERAI — Pilot Assess
- Not started : RISK — Cost, People, Technology
- Not started : DERAI — Pilot Address
- Not started : DERAI — Production Assess
- Not started : AI Production Assessment (AIR — final/approved)
- Not started : Production readiness assessment
- Not started : DERAI Monitoring plan / monitoring configuration
- Not started : Operational support and ownership documentation
- Not started : Approval records (e.g., CDOS CAB decision), if required
Each phase, unfolded: the steps and what you do in them, the documents you deliver, and the teams involved.
Pre-phase 4 steps
Initial steps before the initiative enters the registry: define the idea, submit it to the AI Registry and meet the AI Enablement Team.
What you do
-
PRE.1 Concept Development
Define your idea and, where applicable, validate it through an initial proof of concept.
- Define the idea and, where applicable, validate it through an initial proof of concept
-
PRE.2 AI Registry Submission
Complete the AI Registry Form to capture key project details and establish an official record.
- Complete the AI Registry Form
-
PRE.3 Engagement with the AI Enablement Team
Schedule a meeting to review your submission, receive guidance, and align on next steps.
- Meet the AI Enablement Team to review the submission
-
PRE.4 Ideation Framework (optional)
Optional preliminary step: the Ideation Framework helps teams identify, refine and structure potential AI use cases before entering the AI-SOM process.
- Use the Ideation Framework to structure the use case (optional)
What you deliver
No document expected at this phase.
Who is involved
You and the AI Enablement team.
Phase A — AI Strategy 1 step
Intake & Concept Case: establishes the foundation for an AI initiative by formally capturing the idea and assessing its potential value and feasibility.
What you do
-
A.1.1 Gather information and generate the Concept Case
- Meeting with AI Enablement Team
- AI Enablement Team creates the project folder structure in SharePoint
- Product Owner/Lead creates the Concept Case
- Product Owner/Lead engages with IT Security (start immediately with the SoS, as it is time-consuming)
What you deliver
- Concept Case
- Statement of Sensitivity (SoS)
Who is involved
- Product Owner
- Project Lead
- AI Enablement Team
- IT Security
Phase B — AI Discovery 5 steps
Helps teams shape an approved concept into a clear and feasible path forward by defining intended outcomes, technical direction, and overall readiness.
What you do
-
B.1.1 Outcome-Based Roadmap and feasibility validation
Develop required documentation and engage relevant stakeholder teams for initial alignment and validation.
- Meet with the EA team to review the High-level Architecture document
- Engage with OCDS for AI assessment
- Status with IT Security
- Contact ATIP for project review
- Drop the documents into the SharePoint folders and inform the AI Enablement Team
-
B.1.2 AI Technical Architecture Review
Contact the AI Enablement Team to schedule a formal review session with the ESAT team to evaluate the proposed solution and technical approach.
- Present the architecture document to EA ESAT for approval
-
B.2.1.1 AIRB presentation — Step 1: Deck preparation
- Product Owner / Project Lead verifies that all required Phase A and Phase B artifacts, assessments and supporting documentation are in the project folder structure
- AI Enablement Team reviews the completed artifacts and prepares a consolidated AIRB presentation deck
- AI Enablement Team notifies (email) AIRB members to review the deck and supporting artifacts
- Signed RRI
-
B.2.1.2 AIRB presentation — Step 2: Presentation
- AI Enablement Team opens the presentation with a summary of the initiative and the purpose of the review session
- Product Owner provides a brief overview of the business case and value delivered
- Project Lead provides a brief overview of the technical aspects
- AI Enablement Team presents the submitted artifacts and highlights key findings, assessments, risks, recommendations, dependencies and readiness indicators
-
B.2.1.3 AIRB presentation — Step 3: Discussion and approval
- AIRB members ask questions and request clarification
- AIRB members vote on the proposed initiative
- AI Enablement Team emails stakeholders announcing the project's approval
What you deliver
- Outcome-Based Roadmap
- High-level Architecture
- ATIP Assessment
- Cyber Assessments (SoS, SA&A, SIP, Audits)
- DERAI — Pilot Assess
- RISK — Cost, People, Technology
Who is involved
- Enterprise Architecture (EA)
- AI Enablement Team
- Enterprise Data Hub (EDH)
- Office of the Chief Data Stewards (OCDS)
- IT Security
Phase C — AI Development 2 steps
Builds and validates the AI solution through a structured pilot: development of the AI model or RAG solution according to the approved plan, followed by iterative testing, evaluation and fine-tuning. Governance continues through the next stages of the DERAI assessment.
What you do
-
C.1.1 AI Pilot Development
- DERAI — Pilot Address
-
C.1.2 AI model testing, evaluation and fine tuning
- Complete the DERAI Production Assess
What you deliver
- DERAI — Pilot Address
- DERAI — Production Assess
Who is involved
- AI Enablement Team
- Office of the Chief Data Stewards (OCDS)
Phase D — AI Delivery 4 steps
Safely and responsibly transitions the AI solution from pilot to production: deployment validation, formal AI review and approval, production release, then continuous monitoring. Ongoing governance (DERAI production address and monitoring, Value on AI) keeps the solution compliant, effective and sustainable.
What you do
-
D.1.1 Deployment Validation
- Deploy the AI solution to production
-
D.1.2 AI Review Approval
- Complete AI production approval activities
- Obtain required governance approvals (e.g., CDOS CAB, where applicable)
-
D.1.3 Production Release
- Transition the solution to steady-state operations and support
- Confirm ownership and operational accountability (Product / Platform / Ops)
-
D.1.4 Monitor Results
- Enable and validate monitoring and operational controls
What you deliver
- AI Production Assessment (AIR — final/approved)
- Production readiness assessment
- DERAI Monitoring plan / monitoring configuration
- Operational support and ownership documentation
- Approval records (e.g., CDOS CAB decision), if required
Who is involved
- Product Owner
- Project Lead
- AI Enablement Team
- Office of the Chief Data Stewards (OCDS)
- Enterprise Data Hub (EDH)
- Chief Data Officer Secretariat
- IT Security
Ready to start?
The pre-phase starts with you: an idea and a form. The team takes over from there.