AI-Enabled Value Streams and Practices
Domain focus
AI-enabled value streams and practices examines where AI changes the flow of work from demand to value, and how practices such as incident, problem, change enablement, service desk, knowledge, and continual improvement are redesigned. The exam tests value-stream thinking, not isolated tools.
Core rules
- Map the current value stream first: trigger, steps, inputs, outputs, handoffs, controls, measures. Then decide whether AI should automate, augment, or advise at a step.
- AI belongs to a step only if it improves flow or outcomes. Typical insertions: triage and routing, summarization, knowledge retrieval, pattern detection, change risk assessment, and forecast-driven planning.
- Define control points: confidence thresholds, human review, exception paths, audit logs, and stop conditions.
- Treat models and data as dependencies with owners, versions, and service levels. A model update is a change.
- Close the loop: capture corrections, false positives, overrides, and outcomes, then feed them into improvement.
Worked example
An incident value stream runs from monitoring alert to resolution. The bottleneck is misrouted tickets. Reasoning: map the steps; insert AI triage at intake and summarization for the resolver. Keep the major incident manager as decision owner. Route low-confidence cases to a human queue. Measure reassignment rate, time to assign, MTTR, and false routing. Do not automate a broken escalation path; fix the process first.
Traps
- Automating a step before removing waste and unclear ownership.
- Designing only the happy path and ignoring exceptions.
- Letting AI outputs bypass change enablement or audit.
- Measuring model accuracy but not flow, service, or user outcomes.
- Failing to capture override data for retraining.