Implementation Methodology
Elevator pitch explains the gap: companies launch AI pilots, but without an operating model they rarely become a managed portfolio with confirmed impact.
This page does not repeat the full initiative lifecycle or the product platform. It is an implementation router: choose your starting situation and open the relevant documentation sections.
Three Levels
| Level | What it is | Where to go |
|---|---|---|
| Framework | The components that build the AI operating model: principles, roles, portfolios, business funnel, delivery tracks, stage gates, artifacts, and rituals. | Operating model, AI initiative portfolio, AI product portfolio. |
| Methodology | The sequence for implementing the framework in an organization: where to start, which decisions to make, and which practices to embed. | Methodology section and playbooks. |
| AI Conveyor | The product-platform where teams run portfolios, funnels, gates, artifacts, statuses, blockers, and analytics. | AI Conveyor and platform components. |
Choose Your Starting Situation
1. AI Adoption Is Just Starting
The goal is to assemble the minimum managed loop quickly: who owns initiatives, how ideas enter evaluation, how an AI product is selected, and what evidence is needed before delivery.
Practical route:
- Set the management rules: principles, roles, processes.
- Build the intake flow: idea mining and launch an AI initiative.
- Check that an idea has become a managed initiative: validate the use case, initiative card, use case document.
- Do not start from scratch until the product route is checked: select an AI product, AI product portfolio, AI product model.
The output of this route is the first AI initiatives with owners, a clear impact hypothesis, and a decision on which AI product or delivery track should carry the work forward.
2. Pilots Exist, but There Is No Managed Loop
The goal is to stop managing pilots as disconnected projects and bring them into a common decision loop: what continues, what needs rework, what closes, and where an impact owner is missing.
Practical route:
- Bring every pilot into the AI initiative portfolio and check the initiative model.
- Assign movement through the business funnel: evaluation, delivery, awaiting impact, support, closure, or rejection.
- Put stage gates before transitions: stage-gate model, decision framework.
- Connect initiatives to delivery tracks, artifacts, and playbooks: pilot, deploy the solution, handover to operations, measure impact.
The output of this route is not a pilot list. It is a managed delivery funnel with explicit decisions, owners, artifacts, and an impact review date.
3. There Are Many AI Efforts and the Portfolio Needs Order
The goal is to separate business demand from solution delivery: which initiatives matter most, which should use an existing AI product, and where there are duplicates, risks, or overloaded teams.
Practical route:
- Configure portfolio metrics: AI initiative portfolio metrics, initiative scoring, analytics layer.
- Link demand to the solution catalog: AI product portfolio, AI product model, AI product lifecycle.
- Remove duplication: use select an AI product, the product card artifact, and similar-initiative checks in artifacts.
- Confirm impact at portfolio level: impact confirmation, impact confirmation, impact report.
- Bring in the functions that scaling depends on: information security, architecture, data, risk, legal/compliance, project office.
The output of this route is a portfolio where decisions are based on value, risk, readiness, and reuse, not on the loudest request.
Questions
| Question | Where to look |
|---|---|
| Who owns what? | Operating model roles, initiative roles, AI product roles, committees. |
| How do we calculate and confirm impact? | Measuring impact, impact confirmation, impact report. |
| How do we prioritize initiatives? | Initiative scoring, AI initiative portfolio metrics, decision framework. |
| What should we do with duplicate solutions? | AI product portfolio, AI product model, select an AI product. |
| How does a pilot move into production use? | Pilot, deploy the solution, handover to operations. |
| When should we stop an initiative? | Business funnel, stage-gate model. |
| Where do security, architecture, data, risk, and legal/compliance participate? | Adjacent functions, information security, architecture, data, risk, legal/compliance. |
| Which artifacts are required? | Artifacts, checklists, initiative card, use case document, impact report. |
| Which system runs the work? | AI Conveyor, platform components, Architecture, analytics layer. |
Where to Go Next
- If you need the overall adoption sequence, open the Methodology section.
- If you need to launch a specific initiative, start with the launch an AI initiative playbook.
- If you need to organize the portfolio, open the AI initiative portfolio, AI product portfolio, and impact confirmation.