24. Roster Reset — Roster Reset (dark, ops-room cyan)

Audience: planning & operations teams adopting AI scheduling · Artifacts: planner.html (2D) · 3d.html + planner.js (3D planning table) · shared data planner-data.js · Case study: index.html

In one line: an AI-copilot planner board: four weeks, one hidden flaw each — accept, edit or override, and log a reason; the debrief audits the REASONS, not the choices.

Why this kit exists

AI-assisted planning is sold as acceptance-rate optimization, but the actual skill is a judgment layer: when to trust, when to correct, and — the part nobody trains — putting WHY on the record. The four weeks are engineered so the same three buttons mean four different things, and copy-paste reasons are the designed failure.

What the artifact does

Four weekly boards with an AI proposal chip (94% occupancy and up) and one hidden constraint violation each: geographic impossibility, contract-hours breach, a moveable soft booking, a known no-show pattern. Learner picks ACCEPT / EDIT / OVERRIDE and must log a reason (chips or free text). Scoring rewards correct calls with auditable reasons; accept-with-no-reason and override-everything both cost.

The design decisions worth noticing

• The audit trail as the score — a right choice with a thin reason loses to a corrected board with a specific one, because the log is what the next planner and the model owner inherit
• Two edits, different reasons — weeks 34 and 36 both want EDIT, but blind-spot vs over-eagerness get different logs; the reason you cite has to be the evidence that decides each week, so the two edits need different citations
• ACCEPT as an on-the-record act — accepting a good roster with a logged reason builds earned trust; silence builds nothing

Honest limitations

Certum Opleidingen is a fictional composite training-provider group. The 'AI' is scripted scenario logic, disclosed in-app; real planner tools vary in what they surface and log.

Part of I Wayan Usada's learning-design portfolio · all clients and scenarios are fictional composites built from realistic domain patterns · drafted with AI assistance and fully rewritten, fact-checked and owned by a human.

Process deliverables

• action-map.md — Action map — the behaviors the artifact must produce, traced to the business goal
• storyboard.md — Text-based storyboard — the scene plan written before build
• visual-mockups.md — Visual mockups & style guide — mood, type, key screens, component decisions