AIAF learning pathway: current ten-course map
Status: This crosswalk connects the ten courses to framework decisions and outputs. This map does not replace a learner's organization-specific case record or decision authority.
Complete text equivalent: Example Retailer's continuing objective (OBJ-1) is correct, accessible purchase answers that help customers make suitable purchases. Course 1 frames the case. Course 2 classifies possible AI jobs and delivery modes, including a W1/Mode C staff productivity option and a W3/Mode B customer assistant. Course 3 sets consequence-based boundaries. Course 4 chooses applicable organizational capabilities and proportionate target levels. Course 5 compares AI, guide, human, and information-repair responses and selects work worth designing. Course 6 designs a fair guide/assistant test with owners and a response path. Course 7 evaluates versioned candidates and decides on bounded customer exposure. Course 8 uses actual pilot observations and complete cost for an investment and staged-release decision. Course 9 assesses the first operating record and accepts a restricted routine service. Course 10 uses several later reviews and a provider change to compare continuation, repair, replacement, and retirement, then verifies the chosen transition. Workload and delivery, capabilities and levels, evidence and decisions, and roadmap and observed outcomes remain rails across the entire story. New evidence can redirect, narrow, repeat, or stop a path; the arrows are this teaching case's direction, not automatic approval.
The nine lifecycle phases describe what the adoption case is doing. The ten courses each teach a workplace performance. The eight assessment steps are reused when relevant, and the ten output types are views of one evolving case record, not forms to complete in sequence. The five named checkpoints—Investigate, Select, Pilot, Expand, and Continue—are memorable decision landmarks; the lifecycle decision map says what they are and are not. An accountable team member can make a bounded choice within delegated limits; wider leadership or specialist participation follows the consequence and authority involved.
| Course | Workplace action | Principal lifecycle position | Case record moved forward |
|---|---|---|---|
| 1. Frame an AI Adoption Case | Name the customer/business objective, affected people, uncertainty, and next investigation. | Envision into Discover | Objective, initial evidence, boundaries, and decision. |
| 2. Classify AI Work and Delivery | Pair each AI job with its delivery relationship and authority. | Discover lens | W1/C staff assistance, W3/B customer assistant, deferred W4 authority, responsibility questions. |
| 3. Set Proportionate Boundaries | Set limits, detection, response, recovery, and human recourse. | Discover lens, revisited later | Consequence and boundary record. |
| 4. Build the Capability Profile | Choose critical capabilities, applicability, evidence, and appropriate levels. | Discover lens, revisited later | Focused current/target profile and prerequisite gaps. |
| 5. Select a Response Worth Designing | Compare content repair, guide, people, staff AI assistance, and customer assistant against the same objective. | Discover through Prioritize into Design | Response comparison and bounded Select decision. |
| 6. Design an Accountable Candidate | Use existing support aids and prototype work to design a fair guide/assistant experiment, ownership, and response. | Design and Acquire/build preparation | Candidate design, responsibility, measures, and experiment criteria. |
| 7. Evaluate a Candidate for a Bounded Pilot | Compare observed versioned test results, failures, accessibility, and handoffs. | Acquire/build through Evaluate | Evaluation evidence and bounded pilot-or-guide-first choice. |
| 8. Use Pilot Evidence to Make an Investment Decision | Read actual pilot customer outcomes, response, people work, and complete cost against competing uses of resources. | Deploy and adopt | Investment choice, staged-release scope, measures, spend, and review trigger. |
| 9. Accept an AI Service into Routine Operation | Test first-period service result, incident route, ownership, capacity, and value/cost. | Operate and improve | Restricted operating acceptance and next review. |
| 10. Respond to Material Change and Retire an AI Service | Recheck value and trace after a provider change, compare routes, move the customer work, and verify closure. | Operate/improve through Replace/retire | Change decision, customer transition, exit evidence, and continuing objective. |
Approved public series naming
Use AI Adoption Framework (n of 10): [course title] for course cards and landing-page titles. The number conveys the recommended learning order; it does not impose a technical prerequisite or prevent someone from opening an individual course. Keep the course-specific title and the series label consistent across the catalog and course page.
- AI Adoption Framework (1 of 10): Frame an AI Adoption Case
- AI Adoption Framework (2 of 10): Classify AI Work and Delivery
- AI Adoption Framework (3 of 10): Set Proportionate Boundaries
- AI Adoption Framework (4 of 10): Build the Capability Profile
- AI Adoption Framework (5 of 10): Select a Response Worth Designing
- AI Adoption Framework (6 of 10): Design an Accountable Candidate
- AI Adoption Framework (7 of 10): Evaluate a Candidate for a Bounded Pilot
- AI Adoption Framework (8 of 10): Use Pilot Evidence to Make an Investment Decision
- AI Adoption Framework (9 of 10): Accept an AI Service into Routine Operation
- AI Adoption Framework (10 of 10): Respond to Material Change and Retire an AI Service
Pair workloads and modes; do not compare their code order
W1-W4 answer what AI does. A-D answer how AI is obtained and controlled. The code sequences are not parallel and neither is a maturity ladder. For the same customer objective, support staff might use a general-purpose tool to update macros and cheat sheets (W1/C) without development support for the drafting task; an embedded support-product drafting feature would be W1/B. Developers using AI to build a guide or chatbot perform W2, even though the finished customer chatbot is W3. The customer-facing commerce-platform assistant is W3/B. If it gains connected live-order observation, the case adds W4 Observe without losing W3. Whether any of these paths is useful and supportable is decided by evidence, not the code.
Next learner synthesis
The ten cards can be used independently. The Pathway Companion brings the Example Retailer case and roadmap together across the series.