ZapSight logoZapSight
← Back to Proof

An African hardwood mill ran at half the global recovery benchmark — with no per-log data to understand why.

A tropical hardwood sawmill needed AI-driven yield prediction and order-mix optimisation to close a ten-point recovery gap worth seven figures annually.

Client

African & Tropical Hardwood Sawmill

Duration

6 months

Industry

Construction

+5 ptsRecovery Rate Uplift
7-figureAnnual Value at 100K m³
35→50%Recovery Rate Trajectory
EUDRCompliance-Ready Architecture

Client problem

In their words

African and tropical hardwood mills run at 35–45% recovery against a 50–65% global benchmark, leaving seven-figure value on the table annually. The mill lacked per-log data to identify where value was lost, approved orders without verifying if the available log mix was profitable, and faced a December 2026 EUDR deadline requiring per-log geolocation and traceability.

What ZapSight built

One operating system around the risk pattern

  • Deployed yield prediction using log dimensions and defect data to forecast recovered volume and grade outturn
  • Instrumented mill and team telemetry to measure OEE, throughput, and downtime per production line
  • Built an order-mix optimisation engine using MILP to generate accept/reject/quote signals against live log inventory
  • Designed EUDR-compliant per-log traceability architecture covering geolocation and chain-of-custody

Technologies Used

PythonXGBoostMILP SolverIoT SensorsGeospatial DB