← All work
FIELD OPS · DIGITALIZATION

Elevator Installation Tracker

Role
Solo · stakeholder discovery, workflow modeling and implementation
Status
Core functionality implemented
Stack
Next.js · Prisma · Postgres · Telegram Bot
Field digitalization for a Singapore elevator installer: the site progress that was scattered across WhatsApp groups, pulled into one dashboard the owner can read at a glance — crews report from site through a Telegram bot.

A company that ran on WhatsApp

A Singapore elevator-installation company: an owner (also head of sales), one chief engineer, one admin, one salesperson, and thirty-odd workers. No office systems at all — one admin holding it together in Office software, site progress living across a handful of WhatsApp groups. Between site visits and phone calls, the owner spent the rest of his day scrolling those groups trying to piece progress together — and when the chat got noisy, he couldn’t.

This wasn’t a template job. I walked the real install workflow with the owner and chief engineer, step by step, and modeled it the way they actually work.

Put progress in one dashboard

I built a dashboard around the owner’s need to check progress: sites, lifts, completion and recent activity in one view. The images below show the implemented interfaces and workflows.

The owner dashboard groups sites, lifts, completion and activity in one view.
The owner dashboard groups sites, lifts, completion and activity in one view.

From group messages to records

The Telegram bot supports registration, clock-in and clock-out, step reporting and photo uploads. It is designed to turn information scattered across WhatsApp groups into structured records with timestamps, people and photos, without building a separate native app.

Telegram reporting interface: steps, manpower and photos become structured records.
Telegram reporting interface: steps, manpower and photos become structured records.

Every lift, every step, every photo

Each lift breaks into the real install steps — one record per step, photo evidence captured on site. The steps aren’t uniform: some need only a note and a photo, some are tracked floor by floor up a high-rise, and others can’t be marked done until a QC checklist is complete. That variation isn’t invented — it’s built from how they actually work.

One record per lift per step; some steps gate on a QC checklist before they close.
One record per lift per step; some steps gate on a QC checklist before they close.
Reporting flow modeled on installation steps: select a step, upload a photo, and record completed levels and attachments.
Reporting flow modeled on installation steps: select a step, upload a photo, and record completed levels and attachments.

Clock in on site, or not at all

Clock-in runs through the same bot — but it has to happen on site. The worker shares location, the bot measures the distance to the nearest registered site, and anything off-site is refused outright. On the office side every shift lands in one ledger, and an odd one — an off-site clock-out, an over-long shift — is flagged for review instead of quietly booked. Edits leave the before/after, who changed it, and why.

Bilingual attendance flow: the bot calculates distance to the site, validates clock-in and records attendance exceptions.
Bilingual attendance flow: the bot calculates distance to the site, validates clock-in and records attendance exceptions.
Every shift in one ledger; an off-site or over-long one is flagged “Needs Review” rather than silently saved.
Every shift in one ledger; an off-site or over-long one is flagged “Needs Review” rather than silently saved.

What I built

Implemented the progress dashboard, Telegram reporting, GPS attendance and photo records, connecting sites, lifts and installation steps in a traceable workflow. Attendance exceptions enter review, with before-and-after values and reasons retained for changes.