Can use anomaly-flag triage with a defined task, suitable inputs, and a basic validation check.
augmented skill · Stage 09
Anomaly-flag triage
Anomaly-flag triage helps a designer read outcomes, investigate drift, and feed validated learning into the next cycle. It combines AI-enabled speed with explicit evidence checks, correction, and accountable human judgment.
Why it matters
This skill supports reliable work in Continuous Learning and prevents speed from being mistaken for quality.
Stage outcome
Monitor the shipped experience continuously and feed findings back into Stage 01 of the next cycle. Decoupled from Stage 07 because this is an always-on loop, not a one-time gate
What good looks like
01
Frames anomaly-flag triage against a clear goal, relevant evidence, and the stage quality gate.
02
Directs AI with scoped inputs, reviews uncertainty, and redirects weak or unsupported output.
03
Documents the decision, trade-off, correction, or override so another designer can follow the reasoning.
Checkpoint application
Apply this skill before the Stage 09 gate is approved. The accountable designer must be able to explain how it changed the work, what evidence was considered, and why the result is safe to advance.
Common failure patterns
- Optimising a positive metric without investigating trust, harm, external causes, or long-term behaviour.
- Accepting fluent or high-confidence output without testing its evidence, assumptions, exclusions, and downstream effects.
- Recording the final artefact while omitting the rejected option, correction, trade-off, or override that explains the decision.
Five-minute practice
Use it on the work you already have.
Take one current continuous learning artifact. Apply anomaly-flag triage for five minutes, then record one AI contribution, one human correction, and the evidence that justified the final decision.
Evidence of capability
- A reviewed experiment readout, anomaly investigation, or validated learning log showing the skill in use.
- A short rationale linking the decision to evidence, risk, and intended outcome.
- An example of an AI suggestion that was corrected, rejected, or deliberately accepted.
Leadership assessment questions
01
Can the designer demonstrate anomaly-flag triage on a real continuous learning decision?
02
What observable evidence distinguishes competent performance from confident explanation?
03
Which anchor skill depends on this capability, and how would a gap surface in delivery?
Maturity ladder
Can diagnose weak output, redirect the work, combine sources, and explain the resulting trade-off.
Can decide when not to delegate, own the human checkpoint, and remain accountable for the outcome.
Backlinks and relationships