18 / FOOD PROCESSING / ERP, CRM & ORDERING

An ERP, CRM, and WhatsApp ordering system that replaces three notebook checkpoints and a retail till with one live, auditable ledger.

Client: Poultry processing & retail company

Tags: Food Processing · Poultry · WhatsApp AI · Multilingual · Traceability

// Problem

Three separate paper notebooks tracked intake, slaughter, and cutting, with a retail till logging sales on top of that. Orders arriving over WhatsApp — from contract clients and walk-in customers alike — were free text with no structure, no yield accountability, and no way to trace a finished order back to the batch it came from.

// Our Solution

An ERP, CRM, and WhatsApp ordering system covering intake through order fulfillment as one system. B2B orders are parsed by an LLM directly from nightly WhatsApp messages in whichever language the client uses, matched to a registered contract by phone number, and run on a credit account; D2C orders go through a menu-driven WhatsApp flow that only finalizes once a Razorpay payment webhook confirms it; retail sales are logged in person at the till into the same central ledger. Every batch is yield-audited automatically and every order stays traceable back to the original vendor batch.

// Technical Details

Multi-Channel WhatsApp Ordering

B2B messages are matched to a registered contract by phone number and parsed by an LLM from code-mixed Hindi, Bengali, and English text; a low-confidence parse gets a clarifying question back to the client instead of a guess. D2C uses a menu-driven SKU selection flow and is only finalized once a Razorpay payment webhook confirms payment — never on the confirmation message alone.

Yield Auditing & Full Traceability

Every batch produces both a carcass yield % and a cut yield %, checked against an expected-yield reference table specific to that batch's weight grade. Any batch outside the threshold is flagged visibly in daily logs and on the dashboard. Every finished order stays traceable back through its processing lot, slaughter lot, weight grade, and original vendor batch.

Location-Aware Fulfillment

A customer's location is resolved however it arrives on WhatsApp — a location pin, a pasted Maps link, or a free-text address geocoded through a maps API. The system computes distance to every active outlet and proposes the nearest as the fulfillment point, with staff able to override before confirming.

Owner Dashboard

Nine live pages — Overview, Production & Yield, Inventory, Sales & Orders, Finance & Reconciliation, Vendors & Waste, Outlets, B2B Clients, and the WhatsApp Order Log — with every dataset exportable to Excel.