Sample report. The credit union, people and figures are fictional and illustrative. No cohort has run yet.

This is the report a sponsor and leadership receive at day 30 of the AI Team Pilot.

Sample · Day-30 report · 30-Day AI Team Pilot

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

Directional

Fewer than 20 usable time observations in at least one phase (18 baseline, 23 trial). The comparison is directional, not measured.

MeasureBaselineSupervised trial
Observations (excluded)18 (0)25 (1)
Usable time observations1823
Median active labor minutes (task + review + rework)19.012.0
Source accuracy17 pass · 1 fail · 0 unknown · 0 not applicable23 pass · 0 fail · 1 unknown · 0 not applicable
Completeness17 pass · 1 fail · 0 unknown · 0 not applicable24 pass · 0 fail · 0 unknown · 0 not applicable
Escalation correct18 pass · 0 fail · 0 unknown · 0 not applicable23 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.

Who scores the work

An Innorve reviewer scores each person's coached work and unseen check against rubric CU-AIW-R1. Your facilitator records the workflow observations. The workflow owner makes the recommendation, and your credit union's own authorizer approves any change in its own systems.

What you keep

The Day-30 report and the day-60 check, as documents you hold after the academy deletes your data, by default 90 days after the engagement ends.

Where it goes

To your sponsor and board as evidence for a documented decision, and into your vendor file alongside our trust packet.

How to read each measure

Participation
Who completed the required activities, counted against everyone enrolled at the start, including anyone who withdrew.
Demonstrated capability
Scored against the published rubric CU-AIW-R1. A pass needs every rubric dimension scored, at least 16 of 20, and no critical failure. A critical failure can't be offset by other points.
Median active minutes
Task, review and rework time together, on comparable tasks, before and during the trial. Missing any part makes that observation unusable.
Source accuracy, completeness, escalation
Whether the work matched the named source, covered what the procedure requires, and escalated when it should.
Material errors and boundary breaches
Errors that would matter to a member or the credit union, and any use outside the agreed boundary. Breaches count even on excluded rows.
Excluded
An observation that isn't comparable, such as a case that needed a manager from the start. It is left out of the time and quality comparison but still counted for boundary results.
Unknown
A quality result nobody could establish yet. It is shown as unknown and never guessed.
Measured or directional
A comparison is measured only with at least 20 usable time observations in each phase. Fewer, and the report says it is directional.

Pilot summary for procurement Talk to us