5.1.3. Capability Gaps Across People, Process, Technology, and Governance
💡 First Principle: Transformation fails at its narrowest point. Naming the specific gaps (who lacks which skill, which process blocks which step, which technology is missing, which decision has no owner) turns a vague sense that "we are not ready" into a list of fixable problems with owners and costs.
The exam guide groups capability gaps into four categories:
| Category | Common gaps | Example symptom |
|---|---|---|
| People | AI literacy among business leaders; scarce technical skills (ML engineering, data engineering); missing roles (product owners for AI, data stewards); change leadership | Business teams cannot articulate use cases; AI team becomes a bottleneck |
| Process | No intake or prioritization process; no path from pilot to production; workflows not redesigned around AI; no feedback loop or value tracking | Pilots succeed and stall; benefits never measured |
| Technology | Data platforms and integration; scalable deployment and monitoring; security tooling; access to models | Every project rebuilds infrastructure; prototypes cannot connect to core systems |
| Governance | AI policy; risk classification; accountable owners; responsible AI processes; vendor and tool management | Approvals take months or never happen; shadow AI spreads; nobody owns incidents |
Identifying gaps works best by tracing a specific target outcome backwards. "To put AI-assisted claims review into production in six months, what do we need that we do not have?" produces concrete gaps (a data steward for claims documents, a production monitoring process, legal review of automated decision rules, adjuster training). Generic assessments ("we need more AI skills") are harder to act on.
Gaps interact. A technology gap (no monitoring tooling) often hides a process gap (no one defined what to monitor) and a governance gap (no one accountable for acting on alerts). Closing only the visible technology gap leaves the others in place.
⚠️ Exam Trap: When a scenario describes an AI transformation blocked by multiple issues, the best answer identifies gaps across all four categories (people, process, technology, governance) rather than treating the problem as purely technical. Hiring more data scientists does not fix missing process, ownership, or governance.
Reflection Question: An insurer's AI pilots repeatedly stall at the security review stage, which takes four months per project. Which capability categories might be involved in this gap, and how would you investigate whether it is a people, process, technology, or governance problem?
