Enterprise AI — Course 20: AI Architecture and Deployment Choices

Lesson list

1. Buy, assemble or buildCompare the capability purchased with the workflow, integration and operating work the organisation retains.
2. What actually needs customisingMatch the observed gap to an adaptation, compare it with an adequate simpler route and account for the maintenance it introduces.
3. Hosting and control boundariesSeparate physical location, network paths and administrative control, then test the actual data route against the supplied condition.
4. Compute and network feasibilityInterpret complete resource demand and measured service evidence for a stated workload, then choose a proportionate test or supported scope.
5. Operating capability and supportMatch provider duties, retained organisational work, staffed coverage and recovery evidence to the actual service promise.
6. Constraints before preferencesApply supplied mandatory conditions before preferences, keeping confirmed failures, unresolved evidence and supported scope distinct.
7. Comparing complete alternativesCompare complete arrangements for one requirement, repair unlike evidence and justify a conditional choice against a viable alternative.
8. Sensitivity to organisational changeUpdate a recommendation when demand, staffing or permissions change while retaining valid evidence and defining owned, feasible responses.
9. Staged deployment and handoverBound production claims to the actual pilot evidence, allocate handover and release roles, and define proportionate release, fallback and exit conditions.
10. Architecture and deployment assessmentSeven required groups assess seven distinct architecture decisions using complete supplied case packets. Each group contains three materially different variants; exactly one variant per group is served. Responses use one plain-text field. No new teaching or reading questions are introduced.