Signalise

Same Plant. More Gold.

Signalise is software that continuously optimises your plant’s setpoints.More throughput. More recovery. More gold.No capex. No shutdown.

Small percentages. Serious ounces. Here’s the arithmetic.

170koz/yrBaseline5.0 Mtpa · 1.2 g/t · 88% rec.
7–14koz/yrThroughput+4–8%
2–6koz/yrRecovery+1–3 pp
179–190koz/yrWith Signalise+9–20 koz uplift
+54–120A$M/yrAdded revenueat A$6,000/oz

At this plant, every +1% throughput ≈ 1.7 koz ≈ A$10M/yr. Every +1 pp recovery ≈ 1.9 koz ≈ A$12M/yr.Reference plant only. Actual uplift depends on plant headroom and constraints. See benchmark for assumptions and method.

Onboard. Shadow. AutoPilot.

Signalise is plant optimisation delivered as software: three phases over ninety days.

No shutdownNo capex90 days to uplift

Onboard Connects to your historian and trains the Signalise AI to simulate your plant's dynamics in realtime.

  • Read-only
  • Historian connected
  • Flowsheet mapped
  • Model validated against your operating history

Shadow Connects to your DCS and runs live beside your operators, recommending setpoints, never dispatching.

  • Advisory
  • Ground-truth baseline established
  • Predictions verified against real outcomes
  • Expected uplift quantified

AutoPilot Bounded control inside setpoint limits you set, dispatching through the control systems you already have.

  • Supervisory
  • Uplift reconciled against baseline
  • Operator override, always
  • Authority expands only as evidence accrues

No authority before evidence.

Predictions score
62%tracking
Uplift by reward type
  • Throughput+0.0%
  • Recovery+0.0 pp
  • Energy0.0%
  • Realised0 oz/day
Upliftday 0A$0
Baseline
Expected uplift
Uplift
Onboardd1–30
Shadowd31–60
AutoPilotd61+
Realised uplift · oz/yrPredicted uplift

Signalise Research

Our team has been building production data systems across industries since the late 1990s, and specialising in machine learning control and optimisation for mining and mineral processing since the early 2010s. From operationalising ML research in real-time reconciliation of geometallurgical models from ball mill performance, to neural-network control of cyanide dosing in live leach circuits, we’ve earned control authority in environments where getting it wrong costs real metal.