Problems Solved. Systems Built.

The Problem

The operations team was manually reviewing 200+ freight documents per day — bills of lading, customs declarations, and carrier invoices — to extract key fields and flag discrepancies. It was slow, error-prone, and expensive.

What We Built

We built a document ingestion pipeline using GPT-4 with custom prompt engineering to extract structured data from unstructured freight documents. Extracted data was validated against their ERP records and exceptions were routed to a human reviewer. Straight-through processing for clean documents.

The Problem

The company ran on a legacy ERP with no modern API. Inventory, order, and pricing data lived in a system that couldn't talk to anything else. Reporting required manual exports and spreadsheet assembly every Monday morning.

What We Built

We built a custom API wrapper around their legacy ERP using database-level read access, exposing structured endpoints that downstream tools could consume. We then built automated data pipelines that synced inventory and order data to a modern data warehouse, with a live executive dashboard on top.

The Problem

New client onboarding involved 12 manual steps across 5 systems — contract generation, CRM updates, project setup, folder creation, and welcome email sequences. It took 2–3 days and was inconsistently executed.

What We Built

We mapped the full onboarding workflow and automated every step using n8n. Contract generation via template merge, CRM record creation, project board setup, Google Drive folder provisioning, and a multi-step email sequence — all triggered by a signed contract webhook.