Reliability intelligence for metals, mining & process plants

Find the real root causes of failures — not just the symptoms.

Subatix connects plant data across systems, logs, and engineering knowledge to identify why failures happen and what to fix next.

Built for mills, mines, and refineries running on legacy systems, Excel, and tribal knowledge.

Run a diagnosticSee sample output
↓50%

unplanned downtime, by attacking the recurring failures driving most of it

↓80%

RCA time — from weeks of manual stitching to hours of evidence-linked review

↑1–2 wk

first ranked findings on your data — no integration, no IT project

S/01·The problem

The data exists. The visibility doesn’t.

Industrial plants generate massive amounts of data. When failures happen, it still takes weeks to understand why — because the answer is split across systems that were never built to talk to each other.

  • HistorianProcess behavior — pressures, flows, temperatures, vibration
  • CMMSFailure history, work orders, maintenance records
  • MESProduction data, downtime classifications, batch records
  • ReportsRCA documents, inspection findings, daily logs
  • EmailDecisions, escalations, context the system never captures
  • PeopleOperators, engineers, the why behind the data
When systems don’t connect
  • →RCA takes weeks
  • →Root causes stay unclear
  • →Failures repeat
  • →Downtime compounds
FIG. 01 · A typical failure investigation
Disconnected
HISTORIANTIME-SERIESCMMSWORK ORDERSMESPRODUCTIONEXCELTRACKERSEMAILDECISIONSPEOPLEKNOWLEDGEWHY?FAILURE?SIX SOURCES · ZERO CONNECTION · NO ANSWER
S/02·The solution

A system that finds what others can’t see.

Subatix connects and contextualises operational systems, engineering data, unstructured inputs, and human knowledge — then turns it into clear, prioritised failure root causes with specific corrective actions.

FIG. 02 · The intelligence layer
HISTORIANCMMSMESREPORTS & LOGSEMAIL THREADSENGINEER NOTESCONTEXTUALISEDSubatix01Connect02Reason03PrescribeINTELLIGENCE LAYER01 · ROOT CAUSE→ corrective action02 · ROOT CAUSE→ corrective action03 · ROOT CAUSE→ corrective action04 · ROOT CAUSE→ corrective action05 · ROOT CAUSE→ corrective actionFRAGMENTED INPUTCONTEXTUALISEDRANKED OUTPUT
S/03·The output

Not dashboards. Clear priorities.

A ranked bad-actor list — equipment, system, downtime, loss, action. Every finding shows where the evidence came from — maintenance records, control system trends, lab data, shift logs, engineering reports. Defensible to leadership. Actionable in the field.

ReliabilityEngine · bad-actor ranking
Updated overnight · Ranked by severity
Copper concentrator · 5 active findings on this siteIndicative · sample output
Rank01

Copper concentrator · Flotation Row 4

System · Cell froth depth control / reagent dosing
Reagent dosing variability — recovery loss
Critical
Recovery
−2.1%
Lost Cu
$4.7M
Conf.
88%
Days
42
Evidence
Lab

LIMS — concentrate grade trending below spec for 6 weeks; assay variability tracked but not flagged to ops

Ops

DCS — collector dosing rate manually overridden across 3 shifts; never raised to MOC

Shift

Operator notes mention “froth looking off” repeatedly — never opened as work order

Action →

lock dosing setpoints under MOC · standardise froth assessment SOP · automated grade alarms with escalation rules

Rank02

Copper concentrator · SAG mill SAG-01

System · Discharge grate & liner
Ore hardness mismatch — uncoordinated feed
In RCA
Failures
5×
Downtime
384h
Loss
$3.4M
Conf.
92%
Evidence
Maint

CMMS — grate replacements cluster around hard-ore campaigns; PM strategy never updated for feed variability

Ops

Historian — mill power excursions exceed envelope on the same days; throughput pushed past safe operating window

Eng

Geology shift reports flag hard zones in advance — never linked to mill control or PM scheduling

Action →

coordinate ore feed with geology · widen grate aperture · add power-excursion trigger for pre-emptive inspection

Rank03

Copper concentrator · Ball mill BM-03

System · Trunnion bearing
Lubrication starvation — cold-start sequences
Open
Vibration
1.8× RMS
Downtime
168h
Loss
$890K
Conf.
78%
Evidence
Proc

Vibration trend — recurring spikes correlated with cold-start sequences after weekend shutdowns

Maint

CMMS — three bearing inspections in 14 months; lube oil sample records inconsistent

Shift

Start-up procedure executed inconsistently across shifts — pre-lube cycle skipped under time pressure

Action →

mandatory pre-lube interlock · lube sample protocol enforcement · revise start-up SOP

Rank04

Copper concentrator · Cyclone bank C-201

System · Apex liner / spigot
Slurry density excursions × silica content
Open
Wear rate
2.3×
Defer
24h/mo
Loss
$520K
Conf.
84%
Evidence
Ops

Density logger — running 8% above target during specific ore campaigns to maintain throughput

Lab

Metallurgy — silica content variation tracked, never linked to wear pattern across cyclones

Maint

CMMS — replacement clustering by ore type; spares unprepared for next high-silica campaign

Action →

density control retune · ore type feeds spares planning · campaign-based wear prediction model

Rank05

Copper concentrator · Conveyor CV-450

System · Belt tracking & chute geometry
Chute geometry × moisture variability — resolved
Closed
Spillage
0/mo
Down since
Aug
Saved
$310K/yr
Conf.
94%
Evidence
Shift

Daily reports flagged spillage during wet-ore periods; resolved after chute redesign in Q1

Maint

CMMS — belt tracking adjustments dropped from 12/mo to 1/mo post-fix

Eng

Engineering review confirmed root cause; chute liner replaced with curved profile

Action →

solution validated · monitoring sustained · case learnings filed for sister conveyors

S/04·Deployment

Value in weeks, not months.

We don’t need a data lake, a warehouse, or 12 months of integration work. Raw exports from your existing systems are enough to start. First insights in 1–2 weeks. Then the system stays.

  • Works on raw exports — CMMS, historian, downtime records. Whatever you already produce.

  • No heavy integration — we land before any IT project starts.

  • First insights in 1–2 weeks — ranked findings on your data, not a generic demo.

  • Then it becomes a continuous layer — updates as new data arrives, your team operates it.

FIG. 03 · The deployment timeline
26× faster to value
TYPICAL ENTERPRISE INTEGRATION· slow, expensive, uncertainDATA LAKEWAREHOUSEINTEGRATION?MAYBE INSIGHTSDAY 0MONTH 3MONTH 612+ MONTHSSUBATIX· ranked findings from week 1CONTINUOUS VALUE LAYERRAWRANKEDVALUE FROM WEEK 1–2From raw exports to ranked causes — in two weeks.
S/05·Why Subatix works

Built for real industrial environments.

01

Messy data ready

Works on brownfield reality — inconsistent failure coding, fragmented exports, decades of operational drift. The data your existing tools bounce off.

02

Cross-functional intelligence

Connects historian, CMMS, Excel, logs, documents, and human inputs — the structured and unstructured layers in one analysis.

03

A continuous system

Not a one-off diagnostic. Updates as new data arrives. Your team operates it; we don't come back every quarter to rebuild the analysis.

04

Learns over time

Every failure analysed makes the system sharper on your assets, your patterns, your failure modes. The longer it runs, the more valuable it becomes.

S/06·Expansion

Starts with reliability. Expands across operations.

TODAY · ENTRY

Reliability & root cause

Recurring failures, RCA compression, action prioritisation — the pain that's most visible and most quantifiable. The wedge.

NEXT · ADJACENT

Production losses & maintenance optimisation

Same data, same plant, same trust chain. Throughput loss, backlog risk, PM effectiveness — the next layer of pain, opened by the same engine.

LATER · FULL STACK

Quality, energy, cost

The full operational stack — not from a roadmap, but because every domain originates from the same data fragmentation problem we already solved.

One wedge. One trust chain. The operational intelligence layer compounds — without ever overpromising what we can deliver today.

S/07·Who built it

Built by operators who’ve seen the problem firsthand.

We spent years inside refineries, smelters, and concentrators running exactly this kind of analysis. We watched the same pattern every time — the findings were real, the engagement ended, and six months later the same plant was back to manual processes. Subatix exists to break that cycle.

  • Former McKinsey operational excellence practitioners
  • Engagements across copper, nickel, gold, aluminum, refining, and petrochemicals
  • Deep working knowledge of how failures actually happen — not how they're described in textbooks
  • Built from real operations, not theoretical frameworks
CopperNickelGoldZincAluminumRefiningPetrochemicals

Which failure on your reliability list has been there for two years — and what would it be worth to actually close it?

Run a diagnosticRequest a pilotTalk to us
What happens next

A 30-minute call. Two questions. No deck.