Automated Inventory & Procurement System for Roofing Co

Hostetler Roofing was running their entire inventory and procurement operation manually. Incoming invoices were processed by hand, inventory counts were updated one line at a time, pick tickets were created from scratch each morning, and reorder decisions depended on someone remembering to check stock levels. Two full-time data entry positions were dedicated to keeping this machine running — and errors still slipped through.

I built a closed-loop automation system using Python, Airtable, OpenAI, and Claude-powered AI agents that took over the entire inventory lifecycle:

1. Invoice Ingestion AI agents scan incoming emails for vendor invoices, extract line items, quantities, and pricing, then upload everything directly into the SOS Inventory system — no manual data entry.

2. Inventory Verification The system automatically verifies that received inventory counts match what was invoiced and updates stock levels in real time.

3. Work Order Generation Each evening, the system generates the next day's worklist and creates pick tickets for products that need to be pulled from inventory — ready for the warehouse crew before they clock in.

4. Accounts Receivable Once product is picked and shipped, the system generates an invoice and sends it directly to accounts receivable. No handoffs, no delays.

5. Automated Reordering A watcher agent monitors inventory par levels across all products. When stock hits a reorder point, it generates a purchase order and routes it to the manager for approval. Once approved, the PO fires directly to the vendor.

The entire pipeline runs autonomously — from email inbox to vendor PO — with human oversight only where it matters (manager approval on reorders).

Eliminated 2 full-time data entry positions, saving $70,000/year in labor costs

Reduced inventory loss and mismanagement by 7% in the first year

Zero manual invoice processing — AI agents handle extraction, validation, and upload

Automated reorder cycle prevents stockouts without anyone watching spreadsheets

Daily pick tickets generated overnight, ready before the warehouse opens

What was a 3-person manual operation is now a single manager reviewing AI-generated purchase orders on their phone.