Productize Field Learning

Turn delivery evidence into reusable capability only when recurrence and value are demonstrated. Keep customer-specific data, policy, thresholds, and accidental workarounds out of the shared layer.

Read first

1. Read the Applied AI delivery and operating model and the solution portfolio.
2. Use the field-learning register, change management, and the relevant pattern, control, or product artifact.
3. Read only a selected business-flow pattern, vertical profile, or horizontal foundation when evidence suggests that destination; do not load the portfolio as a pattern library to search for a predetermined answer.
4. Apply FDE-004, DEL-001, DEL-002, OPS-007, CTX-001, and SEC-005 from the control catalog.

Workflow

1. Capture the observation, correction, adoption barrier, incident, or repeated implementation cost with evidence, owner, confidentiality, affected workflow, comparable cohort, and target-specific delivery/support effort.
2. Separate customer-specific business rules, thresholds, data, permissions, identity, and operating context from portable contract shapes, failure classes, evaluation methods, UX patterns, and platform gaps.
3. Sanitize the candidate without moving customer data across boundaries. Record recurrence as metadata and owner validation, not copied evidence.
4. Choose investigate, configure, fix, productize, standardize, defer, reject, or retire. If reusable, classify the destination as a business-flow pattern, vertical profile, horizontal foundation, platform capability, control, pattern, or configuration, and state why it is better than local handling.
5. Define maturity, smallest useful slice, customer-specific work, productization and maintenance cost, expected effect on future delivery/support effort, non-claims, compatibility impact, owner, evaluation, migration, canary, rollback, adoption, and retirement plan for the selected destination.
6. Validate the reusable claim across independent contexts before calling it a standard or platform capability. Confirm that lower customer-specific effort does not weaken outcomes, safety, adoption, supportability, or local-policy validation, then feed the result back into the register.

Output contract

Return a sanitized field-learning record, recurrence and effort evidence, customer-versus-shared boundary, disposition, value hypothesis and productization cost, affected artifacts, validation plan, owner, and release path.

Do not generalize one customer's workaround, centralize confidential evidence, or treat repeated custom service work as a product until the reusable boundary and economics are proven.