WinzeWinze

AI superintendent for mining operations

A new superintendent for your mine.

Winze finds the hours your equipment runs without producing, prices them in dollars, and traces each one to its cause. It answers everyone on site, from the CEO to the mine superintendent, in plain language, so losses are fixed in the shift they happen, not discovered at month end.

Book a pilot
  • Read-only to the plant
  • Every figure opens to its readings
  • Nothing inbound on your firewall
Finding · Night shiftIllustrative

“Why was the primary crusher idle on night shift?”

$41Klost throughput · 2.4 h running without producing

Starved of feed. Motor drew power, belt scale read zero: 3 gaps, 01:00–04:30. No crusher trip or fault; feed from the ROM pad stopped.

What it checked

  1. Motor power against belt scale, 01:00–04:30
  2. Crusher trips and fault codes: none
  3. Planned stops and work orders: none covering it
  4. Upstream: feed from the ROM pad stopped
How a finding is traced, for superintendents

Illustrative conversation. Figures are examples, not customer data.

A Winze dashboard showing ball mill motor power, ventilation fan bearing vibration and primary crusher lube oil temperature, each marked Verified with its source
A real dashboard, built by describing it. Every tile says how sure Winze is and where the number came from.Demo tenant · simulated plant data

The problem

Mining productivity has halved since 1997, and the data to fix it goes unused.

–50%1

Mining productivity, 1997–2023: it fell by half while other sectors grew

  • Manufacturing: more than doubled1
  • Agriculture, forestry, fishing: up 1.5x1

41–44%2

average effectiveness of open-pit shovel and truck fleets, against 85% world class

<1%3

of the data miners collect from their equipment is used to generate insight

Sources

  1. 1. OECD data, cited in McKinsey & Company mining innovation report, Dec 2025.
  2. 2. Int. Journal of System Assurance Engineering and Management, open-pit fleet benchmarking study.
  3. 3. McKinsey & Company, How digital innovation can improve mining productivity, 2015.

The gap is operational

Geology drove part of the decline; Winze targets the part mines can control.

Outside the mine’s control

  • Harder geology

    Lower grades, deeper pits and longer hauls1

Inside the mine’s control

  • Equipment runs without producing

    Open-pit shovel and truck fleets average 41–44% effectiveness2

  • Losses surface too late

    Miners use under 1% of the data their equipment collects3

  • Spending outran output

    Boom-era capex and opex grew faster than production4

Why we built Winze

25%5

below mid-2000s productivity, even with geology stripped out

That gap is operational, so it can be fixed. Winze finds the hours equipment runs without producing, prices them and traces each one to its cause within the shift.

Sources

  1. 1. McKinsey & Company, mining innovation report, Dec 2025.
  2. 2. Int. Journal of System Assurance Engineering and Management, open-pit fleet benchmarking study.
  3. 3. McKinsey & Company, How digital innovation can improve mining productivity, 2015.
  4. 4. McKinsey, Productivity at the mine face, 2015.
  5. 5. McKinsey MineLens Productivity Index, in Navigating a decade of challenges, 2022.

Ready is not working

Equipment sits ready far more often than it works, and site reports hide the gap.

Ready vs working · open-pit haul fleet

97%

available

36%1

actually used

Trucks were ready; the operation left them idle.

Reported vs measured · opencast colliery

83%

reported by the site

71%2

measured on the ground

Site reports overstated utilisation by 12 points.

Where the hours go

  • Shift change

    Up to 30 minutes of every shift; one coal miner took up to an hour3

  • Queuing

    Mismatched fleets leave trucks waiting while still counted as available4

  • Underground cycles

    Blasting and shift breaks cut effective work to 64% of the day5

Sources

  1. 1. Korean limestone open pit, Ain Shams Engineering Journal, 2025.
  2. 2. Pasch & Uludag, opencast colliery study, SAIMM Journal, 2018.
  3. 3. Krause (2008), in Review of Operational Delays in Shovel-Truck Systems; Drummond case, World Coal, 2017.
  4. 4. Match-factor studies, in Optimization of cycle time for loading and hauling trucks in open-pit mining.
  5. 5. Cuiabá gold mine, Brazil, Mining, Metallurgy & Exploration, 2024.

Why losses persist

Losses persist because sites lack the people to turn data into answers in time.

How a loss reaches the people who can fix it

  1. 01

    Loss in the shift

    A truck queues at the shovel

  2. 02

    Data is scattered

    Across fleet, maintenance and plant systems

  3. 03

    Built by hand

    Engineers spend 2+ hours a day on reports1

  4. 04

    Seen too late

    Month-end close took up to 6 days1

Why sites can’t hire their way out

86%2

of mining executives find it harder to hire and keep data science and automation talent

71%2

say the talent shortage is holding back their production targets

47%3

name skills shortages as the biggest barrier to adopting new technology

Sources

  1. 1. Rockwell Automation, mining production-intelligence case study (single operation).
  2. 2. McKinsey & Company survey of mining executives, 2022.
  3. 3. KPMG Global Metals and Mining Outlook, 2024 (survey of 453 executives).

What the data can do

A proven method: a major producer lifted throughput 10% and halved planned capex with its data.

A major producer’s published result, achieved with its own team of data scientists. Not a Winze result.

Published results at a major copper mine more than 50 years old

+10%1

mill throughput, to 85,000 t a day

+1 pt1

copper recovery, while throughput rose

+5%1

copper production in the months after rollout

–50%1

capital spent versus the planned expansion

How they did it

  1. 01Questioned the expansion planThe site had planned a $200M mill expansion to lift output 20%2
  2. 02Built a model from existing dataSite teams and data scientists built it together over six months2
  3. 03Put the answers into each shiftThe model recommended new mill settings every 12 hours2

The catch

It took a dedicated team of data scientists, which most sites can’t hire.

Winze targets the same kind of result without the team, starting with finding and pricing the hours lost.

Winze targets the result without the data-science team or the data build.

A major producerWith Winze
Team
Site experts plus a dedicated team of data scientists2,3
With WinzeYour existing site team, with no analysts to hire
Data groundwork
Years of sensor rollout, a site network and a central data warehouse2,3
With WinzeConnects to the OPC-UA server your site already runs, with no migration project; historians, maintenance and fleet systems next
Speed of answers
New mill settings recommended every 12 hours1
With WinzeLosses flagged as they happen, around the clock
Result
+10% throughput and +5% copper, on half the planned capital1
With WinzeLost hours found, priced and traced to a cause within the shift

Sources

  1. 1. McKinsey & Company, How a copper major is leading AI in mining (2018).
  2. 2. McKinsey & Company, Inside a mining company’s AI transformation, 2020.
  3. 3. E&MJ, Agile approach to process optimization pays off for copper producer.

The engine

Winze pairs proven analytics with LLM reasoning, grounded in your plant.

  • Visualisations

    Charts, dashboards and shift reports, built on request

  • Industry knowledge

    Mining process expertise applied to every analysis

AI and analytics, as one

Analytics computes

statistics, correlation and anomaly detection

LLMs reason

plan the work and explain the result in plain language

The loop · Self-learning

Every investigation is recorded, so earlier runs and incidents feed the next one.

  • Digital twin and context

    Maps what feeds what and ties every signal to its asset

  • Library of connectivity

    OPC-UA today; historians, maintenance and fleet systems next

Without the team

Winze closes each gap without a data-science team.

  1. 01

    Verification and trust at the core

    Every answer links to the evidence behind it, so teams can check the numbers before they act.

    Ends overstated reports

  2. 02

    System integration simplified

    AI does the integration work, so no one builds it by hand: OPC-UA today; historians, maintenance and fleet systems next.

    No data build needed

  3. 03

    Learns as it interacts

    Gets sharper with every question your team asks and every answer they confirm.

    No analysts to hire

  4. 04

    Proactive, continuous monitoring

    AI watches your data around the clock, checks its quality and flags losses as they happen.

    No month-end surprises

Live today on OPC-UA: what is live, and what comes next

How it works

Works like a superintendent, not a dashboard.

Continuously, in small budgeted steps. It is not a dashboard builder, and it is not an alert-rule tool.
  1. 01

    Watch

    every signal, every shift

    Connected to the OPC-UA server already on site, it reads continuously, in small budgeted steps, without anyone asking.

  2. 02

    Find

    hours that ran without producing

    Equipment drawing power with nothing moving through it: the loss a month-end report shows without a cause.

  3. 03

    Price

    each hour in dollars

    Hours running without producing, times the value of lost output, split into explained and unexplained.

  4. 04

    Explain

    plain language, evidence attached

    Each finding traced to its cause in words a superintendent uses, every figure opening to its readings.

Everyone on site

One plant. Every question, in plain words.

Illustrative conversation. Figures are examples, not customer data.

Chief Executive Officer

Where did we lose the most money last month, and do we know why?

$740K

unexplained, of $1.2M lost last month

62% of the unexplained loss: SAG mill below rate, night shift.

Site General Manager

Draft my summary for Monday’s ops meeting.

$286K

unexplained last week, top three ranked

  1. 1. SAG mill below rate on night shift$162K
  2. 2. Primary crusher starved of feed$74K
  3. 3. Flotation cell 4 air flow low$50K

Mine Superintendent

Why was the primary crusher idle on night shift?

2.4 h

running without producing · $41K lost throughput

Starved of feed. Motor drew power, belt scale read zero: 3 gaps, 01:00–04:30. No crusher trip or fault; feed from the ROM pad stopped.

Next step: a pilot on one site.

See it on your own plant. Book a pilot and we’ll run it with you on one site: the unexplained loss on your plant, priced, with its evidence.

  1. 01

    Connect one site

    OPC-UA, read-only, one command on site

  2. 02

    Run it for 90 days

    Chat, shift reports and Scheduler tasks

  3. 03

    Review the findings together

    The unexplained loss, priced, with its evidence

Book a pilot
A conveyor running up from a processing plant at a mine site