Customer Case Study · Procurement · Commodity Risk
A GenAI-built copper intelligence cockpit for the AC manufacturer's procurement team. It shows every price driver on one world map, puts 18 months of events on a 3D timeline, forecasts six months ahead with what-if levers, and traces where each kilogram of copper sits inside an air conditioner.
Figures indicative; validated metrics available on request.
Exchange prices, stocks, spreads, currency and demand gauges side by side. The tag on each card shows which way that reading is pushing copper prices. Click a card to see where it sits on the signal map.
Indicative snapshot for early October 2026, compiled from public market reports; the cockpit refreshes these from data feeds. Verify against live sources before acting.
Copper is one of the AC manufacturer's largest raw-material costs, in heat-exchanger tubes, motor windings and interconnecting pipes. In 2025–26 it set record after record, driven by mine accidents, US tariff moves and the dollar. Buying decisions were still made from a weekly email and gut feel.
LME copper went from under $11,000 to nearly $14,900 a tonne in twelve months, with daily swings of 3–4%.
A mudslide in Indonesia, a strike in Chile, a tariff memo in Washington, a PMI in Beijing. Nobody had the full picture.
Tubes and cathode were bought to the production calendar, so the AC manufacturer often paid above the monthly LME average in peak AC season.
Sales and finance asked what a $500 move meant per AC and per quarter. The answer took days.
Built with the AC manufacturer's procurement and finance teams, using public market data plus the AC manufacturer's own consumption and contract terms.
Select a step to see what happened.
We traced copper from mine to the AC manufacturer's coil lines.
Deliverable: Copper exposure map and signal long-list.
Three layers: global supply and demand, money and policy, and the AC manufacturer's own demand and terms.
Deliverable: Factor model and hedging policy draft.
Claude summarised daily news into factor signals; a transparent model turned them into a forecast.
Deliverable: Copper intelligence cockpit (single HTML, refreshed daily).
A 30-minute Monday copper call replaced scattered emails.
Deliverable: Weekly copper call and decision log.
The same cockpit pattern now covers more commodities.
Deliverable: Multi-commodity roadmap.
Four connected views. Explore each one: the factors on the map feed the forecast, and the forecast flows through to the cost of every AC.
Macro and micro factors pinned where they happen. Arrows show the direction each one is pushing price, and pulse size shows its weight in the model. Animated lines are the main copper trade flows. Click any pin.
Manufacturing clusters shown are indicative. Micro factors use the AC manufacturer's own data in the live cockpit; values here are illustrative.
The copper bar is the monthly LME price; the translucent band ahead is the forecast range. Pins are the events that moved copper, and dashed pins are scheduled events still to come. Drag to orbit, scroll to zoom, or press Fly through.
Past events from public market reports (Cochilco, LME, CME, SHFE, ICSG, company releases and news coverage); monthly prices are indicative averages. Future pins are scheduled or expected events, not predictions.
A transparent factor model: start from today's price, add each driver's effect over six months, and show the range. Change a lever and the forecast, the 3D timeline and the buying advice all update.
Illustrative factor model for decision support, not investment advice. Sensitivities are calibrated on 2024–26 history and reviewed quarterly.
An inverter split AC with the copper parts glowing. Turn on X-ray to see inside, click a part to isolate it, and watch refrigerant move through the copper loop. Change the size or switch to an aluminium condenser and the copper bill updates.
Copper weights are typical industry values for an inverter split AC and vary by design and efficiency rating. The 3D model is a schematic, not an actual product drawing.
With a weekly call and clear hedge rules, the AC manufacturer's average buy price moved from above the monthly LME range midpoint to below it, and the six-month view became good enough to plan on.
Monthly LME low–high range and the AC manufacturer's average buy price, $/t
Mean absolute % error, back-test Jan 2024 – Sep 2026
Figures indicative; validated metrics available on request.
When everyone can see the mine, the tariff and the PMI on one screen, the debate moves from opinions to weights.
No model calls the exact price. A range plus a hedge rule beats a single number and a gut call.
Showing copper per AC turned a market story into a pricing and margin conversation everyone could join.