Measure before
you promise
REPROF is testing whether structured, AI-augmented practical work can produce measurable human capability — and whether that capability can be evidenced in a way an outside person can inspect.
1. Capability before curriculum
The program is not the product. The intended product is a demonstrable capability and an evidence stack: externally-scored work, a blind-graded capstone, participant-owned artifacts and a record of how the person supervised AI.
2. Externalise the ruler
Where possible, REPROF uses third-party platforms for objective scoring and independent practitioners for blind grading. REPROF should not author the training, the test, the score and the marketing claim without external checks.
3. AI Supervision Record
For each meaningful AI-assisted work session, the method records whether the human accepted after verification, corrected, rejected, stopped or overruled the agent. The highest-risk event is a missed intervention: accepting output that should have been corrected or rejected.
4. Capability Delta
Capability Delta asks a narrow question: what changed between comparable Day-0 and Day-30 conditions? It is reported by construct rather than as one magic score.
- task completion
- false-positive control
- missed-intervention rate
- confidence calibration
- recovery after a bad hypothesis
- communication quality
5. Publish the limits
REPROF does not claim accreditation, job readiness, employment, salary, client work or legal authority from completion alone.