Pilot Readiness
A Lumen pilot proves measurable improvement while keeping academic review under institutional control.
Start with one constrained evaluation workflow and measure it carefully before expanding to more subjects, languages, or high-stakes rollout paths.
Minimum pilot inputs
- question paper
- model answer or marking scheme
- student answer sheets
- assessment metadata
- candidate roster or candidate identifiers
- teacher reviewers
- moderation owner
- export expectation
Recommended first scope
Use a narrow first pilot:
- one assessment
- one subject
- one language
- 20-50 answer sheets
- readable scans
- teacher-provided rubric
- no final result export without teacher approval
Success metrics
Use measurable operating metrics instead of unsupported accuracy claims:
- average time saved per answer sheet
- OCR/extraction confidence
- AI scoring confidence
- teacher override percentage
- average difference between suggested and final marks
- mandatory review cases
- cost per evaluated answer sheet
- teacher satisfaction
The safest sales frame is a pilot target of 30-50% evaluation time reduction, not a guaranteed promise.