Proven Results

Case Studies

Results from real operations — validated by measurable production data.

Featured Case — Named Customer

AMG Brasil — Predictive Control of the Mibra Grinding Circuit

Client

AMG Brasil, Mibra mine (Minas Gerais) — part of AMG Critical Materials N.V., listed on Euronext Amsterdam. Name, brand and results used with permission.

Challenge

Sudden rises in circulating load forced the feed to be cut — often below 70 t/h — about seven times a day.

Solution

A predictive model of the circulating load drives the feed conveyor in real time: it anticipates the rise, holds a level the mill can absorb, then returns to maximum feed.

+20 to 35 t of ore a day · 80%+ of feed cuts avoidedFirst tests, May–July 2025, with the software still in development: +3.5 to 5 t of ore per event.

Full case details Visit AMG Lithium

Illustrative

Featured Case — Named Customer · Published

Mineração Aurizona — AI Autopilot on the Gold Grinding Circuit

Client

Mineração Aurizona (MASA), gold mine and plant in Maranhão, Brazil — part of CMOC Group since January 2026

Challenge

Manual process control limiting throughput and consistency. Operators unable to maintain optimal parameters across shifting ore conditions 24/7.

Solution

A predictive model of the grinding circuit running the AI Autopilot, sending setpoints to the plant’s automation system in real time. 25-day test (Jan 10 – Feb 5, 2024) alternating AI and manual control every 48 hours. Delivered with SoftwareOne on AWS.

+6% throughput: 435 t/h (AI) vs 412 t/h (manual)Published in Brasil Mineral #441 (July 2024); results confirmed in writing by Mineração Aurizona in 2026. Test report projection if all feed were fresh rock: ≈ 2,694 additional ounces a year, ≈ US$781K net of AISC (2024 assumptions). Results may vary with ore grade and operating conditions.

Full case details SoftwareOne case study Brasil Mineral #441 (Jul 2024)

Illustrative

Featured Case — Under NDA · Bioenergy

Predictive Models for Biodiesel Production

Client

Global agribusiness & bioenergy producer, South America. Client under NDA.

Engagement

Predictive models for biodiesel furnaces and reactors. Phase 1 delivered: 56.5M sensor readings over 10 months consolidated into a curated data pipeline, plus a virtual sensor with 96.3% R² and catalyst-margin insight.

Status

Phase 2 underway: yield, catalyst and quality models with continuous retraining and self-learning. Phase 2 results are preliminary.

Phase 1 delivered · Virtual sensor 96.3% R²Same AI stack beyond mining — predictive models with continuous retraining applied to bioenergy production.

Full case details

Illustrative

Featured Case — Under NDA · Heavy Equipment

Predictive Maintenance for Component Test Cells

Client

Global heavy-equipment manufacturer. Client under NDA.

Engagement

Predictive-maintenance model and dashboard for component test cells. Cells with ~70 sensors consolidated; per-variant baselines help separate bench infrastructure from component behavior.

Status

May–September 2026: data foundation delivered; model training underway, on schedule.

In progress · On schedule (May–Sep 2026)Predictive maintenance on discrete test benches — beyond continuous process industries.

Full case details

Products & Partnerships

SoftwareOne NextGen ISV program — AWS

Cloud consumption costs 20% lower on AWS

NextGen ISV program, with AWS (published case)

Your operation could be next.

A paid engagement, delivered as consulting or as the first stage of a project: under NDA, we study your process data and deliver a prioritised opportunity map and a measurement plan for a controlled AI-versus-manual test.

Book a plant assessment

AMG Brasil — Predictive Control of the Mibra Grinding Circuit

Client

AMG Brasil, Mibra mine (Minas Gerais) — part of AMG Critical Materials N.V., listed on Euronext Amsterdam. Name, brand and results used with permission.

Challenge

Sudden rises in circulating load forced the feed to be cut — often below 70 t/h — about seven times a day.

Solution

A predictive model of the circulating load drives the feed conveyor in real time: it anticipates the rise, holds a level the mill can absorb, then returns to maximum feed.

+20 to 35 t of ore a day · 80%+ of feed cuts avoidedFirst tests, May–July 2025, with the software still in development: +3.5 to 5 t of ore per event.

Mineração Aurizona — AI Autopilot on the Gold Grinding Circuit

Client

Mineração Aurizona (MASA), gold mine and plant in Maranhão, Brazil — part of CMOC Group since January 2026

Challenge

Manual process control limiting throughput and consistency. Operators unable to maintain optimal parameters across shifting ore conditions 24/7.

Solution

Predictive Models for Biodiesel Production

Client

Global agribusiness & bioenergy producer, South America. Client under NDA.

Engagement

Status

Phase 2 underway: yield, catalyst and quality models with continuous retraining and self-learning. Phase 2 results are preliminary.

Phase 1 delivered · Virtual sensor 96.3% R²Same AI stack beyond mining — predictive models with continuous retraining applied to bioenergy production.

Predictive Maintenance for Component Test Cells

Client

Global heavy-equipment manufacturer. Client under NDA.

Engagement

Predictive-maintenance model and dashboard for component test cells. Cells with ~70 sensors consolidated; per-variant baselines help separate bench infrastructure from component behavior.

Status

May–September 2026: data foundation delivered; model training underway, on schedule.

In progress · On schedule (May–Sep 2026)Predictive maintenance on discrete test benches — beyond continuous process industries.