Overview
| Client | K-Mobility Export — Automotive parts exporter | approx. KRW 80bn revenue | 210 staff | written consent for name and logo [TBC] |
|---|---|
| Industry / Technology | Manufacturing / Automation |
| Period | Feb-May 2026 (14 weeks) |
| Scope | Process diagnosis | order-to-shipment workflow | handover to operations |
| Stack | Python, Temporal, PostgreSQL, Grafana |
| Disclosure grade | L1 Named |
What was the problem?
Overseas sales, production control and logistics each kept their own spreadsheets for buyer POs, production orders and shipments. One person consolidated them at month end: two days for the consolidation itself, and another half day reconciling differences by buyer and by currency. Almost every error came from re-typing.
Why this approach?
A full ERP replacement did not fit the budget or the timeline. We left the spreadsheets in place as the input surface and connected a workflow behind them that lands everything in a single store. Leaving overseas sales with the sheet formats their buyers expect is what decided adoption.
What we built
- 01Watch the shared drive and parse only the rows that changed
- 02Apply 12 normalisation rules for part and buyer codes, then load to staging
- 03Reconcile PO quantity against shipped quantity - mismatches bounce back to the owner on their messenger
- 04Convert to KRW at the published rate and generate a daily stock and shipment dashboard
Before to After (measured in-house)
Monthly export close
2.5 days
0.4 days
Entry errors per month
31
4
Shipment and stock data lag
1 day
15 min
Manual hours per month
96 h
18 h
Figures were measured on the same basis before the engagement and after go-live, and the calculation records are retained.
“The biggest thing was that nobody had to learn a new system. We still work in the buyer format, and the month-end numbers are simply there.”
Head of Business Support | quoted with consent