Turning trade paperwork and sales visibility into a system that actually keeps up with international operations.
Built for A&D International, the platform combines a real-time sales dashboard with automated trade document extraction — two systems that work together to close the gap between what's happening and what the ERP knows.
Customer dashboard gives sales and operations a live view of RFQs, orders, and shipments without opening Odoo
Trade document extraction handles invoices, bills of lading, certificates, and packing lists using layered OCR + LLM parsing
Runs locally on Apple Silicon so sensitive trade documents never leave the building
Multi-currency conversion and real-time pipeline visibility keep sales decisions grounded in current data
Two connected problems. First: sales and operations had no shared view of where things stood — RFQs, orders, and shipments were tracked in separate systems and personal spreadsheets, not in one place. Second: the trade documents that feed the ERP (invoices, bills of lading, certificates of origin, packing lists) were manually entered, which meant slow processing, frequent errors, and a bottleneck that got worse with volume.
We built two systems that work together under one operational roof. The Customer Dashboard & Portal gives sales, operations, and customers a live view of RFQs, orders, and shipments — all pulled from Odoo, displayed in a clean interface that non-technical users can actually navigate. Multi-currency conversion to USD, real-time pipeline visibility, and enough filtering to answer questions without opening Odoo itself. The Trade Document Extraction system takes the paper side of international trade and turns it into structured, ERP-ready data. It combines template-based extraction with OCR and LLM parsing — not betting on one brittle method, but layering approaches so the system handles the messy real-world documents operations teams actually deal with. It runs locally on Apple Silicon, which matters when the documents contain sensitive pricing, supplier terms, and shipping details that shouldn't leave the building.
Odoo's data model is sprawling — mapping RFQs, orders, shipments, and customer records into a coherent dashboard meant understanding relationships Odoo itself doesn't make obvious
Trade documents come in every format: scanned PDFs, photos of faxes, Excel exports pretending to be invoices — the extraction system needed to handle all of them without a human selecting the right parser first
Running ML-based extraction locally on Apple Silicon (PaddleOCR + Vision) instead of cloud APIs required tuning for performance without losing accuracy
Multi-currency conversion needs to be live and trustworthy — sales decisions depend on it being right, not approximately right
A&D International's operations team now sees their entire pipeline in one place — from the first RFQ to the last shipment — without opening Odoo for routine questions. Trade documents that used to take 15–20 minutes of manual entry per document now flow through extraction in under a minute, with structured output ready for Odoo import. The sensitive data stays on-premises, the error rate dropped, and the team stopped losing time to the gap between paper and system.