JBALL CASE STUDIES

CASE STUDY / Automations

Program Scheduler — the schedule on everyone's phone

FUGU GMBH · GLOBAL PHARMA · 2026

34people on one program
336-377events across three days
144coach and participant 1:1 sessions

SITUATION

A three-day leadership program where the top-reported problem was not the content, it was that nobody knew where they were meant to be. Schedules lived in documents, changes travelled by email and hallway relay, and every participant had a different version of the truth.

APPROACH

The program is built by uploading the documents that already exist, each one parsed by its own role-scoped importer so a prep-times document can only touch prep times. That path is deterministic by design: no model reads the schedule. A wrong room in a leadership program for people flown in from three continents is not a cost worth trading for convenience.

RESULT

A full-day schedule adjustment went from hours of manual editing to under a minute, and a change reaches every phone as one announcement. The time saved is my own estimate and is labelled as such; the counts above are read from the running program.

  • Six role-scoped document importers, none of which can overwrite another's data.
  • Personal live views on every phone: what is now, what is next, and a countdown.
  • An organizer console with cascade-aware bulk edits and a health panel for conflicts and drift.
  • A rebuild from zero verified through the real routes: 34 people, 377 events, 0 conflicts.

Next.js · TypeScript · Tailwind · Vercel Blob · Vitest · Playwright

Thesis page: a wrong room is not a cost worth trading for convenience.
A hand-drawn diagram of six role-scoped importers feeding a deterministic build, then going live on every phone.
Results page: the schedule stopped being an argument.
OPEN THE FULL PDF· 3 pages · 0.6 MB(opens in a new tab)
SEE THE ROLE ON THE TIMELINE →

Figures read 2026-08-28.