Workshop Outcomes & Path Forward
Following our productive discovery workshop on February 10, 2026, this memorandum outlines the key outcomes and proposes a concrete path forward for the Blending Optimization System (BOS) pilot project at the Robinson Mine. The workshop successfully aligned all stakeholders on the strategic value of an AI-driven approach to ore blending, confirming the potential for significant improvements in recovery, throughput, and yield.
We are confident that by co-creating this solution, we can unlock substantial operational efficiencies and establish a new standard for data-driven decision-making at KGHM. We recommend proceeding with a phased pilot implementation to demonstrate value quickly and mitigate risk.
Proceed immediately with Phase 1 of the BOS pilot. airth.io is prepared to mobilize upon formal approval, with the statement of work ready for finalization with KGHM's designated project leads.
Robinson Mine Pilot: The Business Case
The workshop discussions solidified the business case for the BOS pilot. The core objective is to develop and deploy a machine learning system that optimizes mill feed by dynamically adjusting for geological variability and metallurgical performance.
The Core Challenge
Achieving these results hinges on addressing the core challenge identified in the workshop: balancing over 20 parameters to align with the mill budget while minimizing the impact of deleterious minerals — or "bad actors" — like iron and zinc.
Data Integration Architecture
Real-time data exchange → BOS System
Alignment Achieved
The workshop served as a critical forum for validating our shared understanding of the problem and the proposed solution. Three key outcomes emerged from the session.
Problem Statement Confirmation
All parties reached a consensus on the primary challenge: managing ore variability to ensure a stable and predictable feed to the concentrator. Specific areas of concern were identified, including clay and viscosity spikes, hardness variability, and the discrepancy between the block model's designated "Material Type" and the actual material arriving at the crusher.
Data Integration Strategy
We confirmed a robust and feasible data integration strategy that leverages KGHM's existing ecosystem. The BOS will ingest data in real-time from five key systems: AssayNet (LIMS), Hexagon MinePlan, acQuire, AVEVA PI, and CAT Minestar FMS. This ensures a comprehensive view of the value chain, from blast hole assays to mill operations.
Collaborative Framework
The workshop established a strong collaborative framework between KGHM, airth.io, and our respective partners — including WSP, GMOO, and Open Loop Energy. This partnership is essential for combining KGHM's deep operational knowledge with airth.io's AI and platform expertise to ensure the pilot's success.
Proposed Phased Implementation
To maintain momentum and deliver value expeditiously, we propose the following four-phase pilot roadmap. This approach allows for iterative development, continuous feedback, and regular value demonstration.
Data Integration & Foundation
2 – 3 Weeks- Finalize Data Sharing Agreements & IT/OT integration
- Establish secure data connections to all five source systems
- Develop and validate foundational data models
Predictive Model Development
5 – 6 Weeks- Train, test, and validate initial ML models for ore characterization
- Develop initial BOS dashboard for visualizing predictions
- Iterative refinement with KGHM geology and mill teams
Live Pilot & Shadow Mode
7 – 10 Weeks- Deploy BOS in non-interventional shadow mode alongside operations
- Refine models with live data and gather user feedback
- Document decision alignment vs. manual blend planning
Go-Live & Handover
2 – 3 Weeks- Transition BOS to active decision support for the Senior Planner
- Complete knowledge transfer and comprehensive KGHM training
- Establish ongoing support and continuous improvement plan
A Partnership Built for the Future of Mining
The discovery workshop reinforced our belief that a partnership between KGHM and airth.io can drive significant and sustainable operational improvements. The Robinson Mine Blending Optimization System pilot is a low-risk, high-reward initiative that will serve as a powerful demonstration of AI's potential to transform mining operations.
It directly aligns with the vision of a self-learning, continuously improving, and sustainable mining future — one where real-time intelligence replaces reactive, manual decision-making across the entire mine value chain.
We are excited to take this next step with you and are confident that this project will deliver tangible value to KGHM. We look forward to your feedback and formal approval to proceed.