GCOMME 2026: Wholesale & Truck Distribution ERP with WhatsApp Receipt OCR
An ERP built for Moroccan wholesale truck distribution — separating main depot stock from rolling truck inventory, tracking client credit and post-dated checks, and extracting handwritten delivery slips via WhatsApp AI vision.
Laravel 11PostgreSQLWhatsApp OCRERP
Context
In Moroccan wholesale distribution, each truck ('camion') leaves at dawn as a rolling warehouse, sells on cash, check, or credit, pays road expenses out of pocket, and returns at night with paper slips. Generic inventory tools fail because they cannot reconcile rolling truck stock against daily driver cash boxes.
Scope I disclose publicly
- Dual-inventory ledger tracking main warehouse stock, morning truck loading ('chargement camion'), route sales, returns, and evening stock reconciliation.
- Customer credit and collection engine managing credit ceilings, partial cash payments, and post-dated checks ('effets/chèques') with maturity alerts.
- WhatsApp and Gemini Vision OCR pipeline that extracts line items from photographed delivery slips and fuel receipts into draft ERP entries.
Decisions and trade-offs
- Modeling every truck as an isolated inventory location paired with a daily driver cash register so discrepancies surface the same evening.
- Requiring a human confirmation click on AI-extracted receipt totals before posting to accounting ledgers.
- The trade-off: mandatory end-of-day route reconciliation adds a daily checkpoint for dispatchers, in exchange for eliminating month-end stock mystery losses.
A live deployment is available at mbardouni.dev/youssef-erp/.
Check the public sourceDisclosure: Public demo credentials access synthetic data only; private distributor figures are never published.