Automated steel quotation platform + HRC cutting module
Drawing to draft quote in about 31 seconds (recorded demo)
- Anthropic Claude API
- React
- Google Apps Script
- Google Sheets
- Vercel serverless
Before vs after
Before
Manual quoting
~200 minper manual quote (Johnny's observed estimate)
- Read the customer drawing by eye
- Retype every plate and section
- Missed items, wrong numbers copied
- About 200 min per quote (observed)
After
With the platform
~31 sdrawing to draft quote (recorded demo)
- Upload a photo or drawing
- AI extracts the steel line items
- Weights and prices computed automatically
- Draft quote in ~31 s (recorded demo)
Manual time is Johnny's observed estimate from watching sales staff quote, not a timed study. The ~31 s comes from the on-screen timer in the recorded demo.
Module 1: automated quoting
01Step 1 of 5
Drawing in
A customer photo, blueprint or drawing is uploaded.
02Step 2 of 5, Anthropic Claude API
AI extraction
Claude API, called from a serverless function, extracts plates and sections.
03Step 3 of 5
Barem weights
Multi-mill barem (steel weight) tables compute the weight of each item.
04Step 4 of 5, Microsoft Excel
Price and export
Pricing applies automatically; the quote exports as PDF, image or Excel.
05Step 5 of 5, Google Apps Script
Close order
Revenue, customer and product group flow to a live dashboard and Google Sheet.
Module 2: HRC plate cutting
01Step 1 of 4, Anthropic Claude API
Parts list
Enter a parts list, or let AI extract it from a drawing.
02Step 2 of 4
Nesting
Parts are nested onto the hot-rolled coil (HRC) sheet size.
03Step 3 of 4
Cutting diagram
The platform draws an SVG cutting diagram.
04Step 4 of 4
Yield and cost
Sheets required, uncoiling and CNC scrap, yield, weight and cost per kg.
Tech stack in this project
- Anthropic Claude APIReads drawings and extracts steel line items
- Vercel serverlessServerless function that calls the Claude API
- ReactQuoting and cutting screens
- TypeScriptLanguage of the app codebase
- ViteBuild tooling for the web app
- Tailwind CSSStyling of the interface
- RechartsCharts on the real-time dashboard
- Google Apps ScriptPushes closed orders to Google Sheets
- Google SheetsShared record of revenue, customer, product group
Why it matters for supply chain
- Quote automation = faster order-to-cash, fewer order errors
- Nesting and yield = material planning and scrap reduction
- Closed-order data = clean input for demand forecasting
Linked to my Master of Logistics & SCM, University of Sydney
Read the full case study
Problem
Sales staff read customer blueprints and drawings by hand, translated them into steel items (plates, sections), looked up weights and prices and typed the quote. Johnny estimates, from watching sales staff, that this took around 200 minutes per quote including double-checking, and errors included missing items and wrong numbers copied across.
Approach
- Module 1, automated quoting: AI extraction of line items from a photo or drawing (Anthropic Claude via a serverless API function), multi-mill barem tables to compute weights, automatic pricing, and export to PDF, image or Excel.
- "Close order" pushes revenue, customer and product group to a real-time dashboard and to a Google Sheet via Google Apps Script.
- Module 2, HRC plate cutting: a parts list (or AI extraction from a drawing) is nested on the HRC sheet size, with an SVG cutting diagram, sheets required, uncoiling and CNC scrap, yield, weight and cost per kg.
- Result: in the recorded demo, a sample drawing became a draft quote in about 31 seconds (on-screen timer), against a manual process Johnny estimates from observing sales staff.
Figures are the confirmed dashboard figures or Johnny's own estimates, labelled as such.