Clearbrook Learning Credit Union
Member services · card-dispute intake · November 2, 2026 to December 2, 2026 · day-60 check January 4, 2027
1. Participation
Denominator: everyone enrolled at the start (12), including 1 who withdrew.
- Completed all lessons: 10 of 12 (83%)
- Submitted coached work: 11 of 12 (92%)
- Withdrew: 1
2. Demonstrated capability
Released reviews only, scored by a person against the published rubric CU-AIW-R1. Reported separately from participation. No individual scores appear here.
- Coached work passed: 9 of 12 (75%)
- Unseen judgment check passed: 8 of 12 (67%)
- Skill demonstrated (both passed): 8 of 12 (67%)
- Asked to revise: 3 · reviews with a critical failure: 1 (each must be remediated before any relevant authorization)
- Not yet assessed: 1
3. Workflow value and boundary adherence
Workflow: Draft the internal intake checklist for a card-dispute call from the dispute procedure, with the source named and gaps listed. Owner: Member services supervisor.
Workflow value
DirectionalFewer than 20 usable time observations in at least one phase (18 baseline, 23 trial). The comparison is directional, not measured.
| Measure | Baseline | Supervised trial |
|---|---|---|
| Observations (excluded) | 18 (0) | 25 (1) |
| Usable time observations | 18 | 23 |
| Median active labor minutes (task + review + rework) | 19.0 | 12.0 |
| Source accuracy | 17 pass · 1 fail · 0 unknown · 0 not applicable | 23 pass · 0 fail · 1 unknown · 0 not applicable |
| Completeness | 17 pass · 1 fail · 0 unknown · 0 not applicable | 24 pass · 0 fail · 0 unknown · 0 not applicable |
| Escalation correct | 18 pass · 0 fail · 0 unknown · 0 not applicable | 23 pass · 1 fail · 0 unknown · 0 not applicable |
| Material errors (rows unknown) | 1 (0) | 1 (0) |
| Boundary breaches, including excluded rows (unknown) | 0 (0) | 0 (0) |
Median time change: 36.8% lower than baseline (directional).
Median improvement = 100 × (baseline median − trial median) ÷ baseline median. Not multiplied by volume, and not converted to money. Different task types or bands are not pooled without saying so.
Recommendation
Change, recommended by the member services supervisor. Keep the workflow, add a check that disputes needing a manager are escalated before the checklist is used, and re-measure at day 60.
- Evidence: Median active time fell on comparable tasks. The trial had one material error, a missed escalation, and no boundary breaches.
- Limits: The baseline has 18 usable time observations, fewer than 20, so the time comparison is directional.
4. Still unknown
- 1 of 12 learners has no released assessment yet. They are "not assessed", not zero.
- Fewer than 20 usable time observations in the baseline: the time comparison is directional.
- 1 trial observation has a source-accuracy result still unknown.
- 1 trial observation has no review time recorded, so it is not usable for the time comparison.
5. What this report does not say
It does not state that the credit union is compliant, predict any exam outcome, adopt any policy or approve any AI use. Course completion and assessment results do not change what any workflow is permitted to do. Time savings are not multiplied by volume or turned into money. Decisions are made and recorded in the credit union's own systems. Not legal advice.