CadetPrep
LMS for an officer-selection academy, with a mentor that keeps a senior register and knows which chapter is open.
Overview
AOSB, AIB and OASC preparation with a three-column study screen, a mock PDF reader and a mentor whose answers name the chapter they came from. Mock tests produce a weak-topic analysis the cadet can accept into today's plan — capped at the hours they actually study, because a plan nobody follows is worse than no plan.
Specifications
USE CASE
Structured exam preparation with an in-context mentor
WORKFLOW
Plan, study, test, analyse, revise
OUTPUT
Weak-topic analysis and a revision plan the cadet accepts into their day
SECURITY
UK-only residency, academy-authored content, 36-month retention
Features
- 01Three-column study screen with a zoomable reader from 60% to 240%
- 02Mentor answers that open by naming the chapter and keep a disciplined register
- 03Mock history with per-attempt weak-topic analysis and accuracy per topic
- 04Accepted revision plans land directly in today's plan, audit-logged
- 05Instructor console with roster, engagement scoring and course analytics
Architecture
CADETPREP PIPELINE
────────────────────────────────────────────
┌───────┐
│ Cadet │
└───────┘
↓
╭┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈╮
┊ AI mentor — persona ┊
╰┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈╯
↓
╭┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈╮
┊ Chapter context retrieval ┊
╰┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈╯
↓
┌──────────────┐ ┌────────────┐ ┌───────┐
│ Today's plan │ │ Mock tests │ │ Notes │
└──────────────┘ └────────────┘ └───────┘
↓
╭┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈╮
┊ Weak-topic analysis ┊
╰┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈╯
↓
┌────────────────────┐
│ Instructor console │
└────────────────────┘
╭┈┈╮ dashed = model step ┌──┐ deterministic stepWhat it will not do
Every screen carries one honest failure record with its recovery path. This is the one buyers remember. A run produced an eleven-hour daily plan for a cadet who is also in full-time college. It addressed every weak topic and nobody would have followed it. Plans are now capped at the hours actually logged in the last fortnight. A second run declined to predict a board outcome — it has no access to the board's criteria, and a confident answer either way would be a guess dressed as guidance.
