OES.Mine Fleet Management System Deployed in West Africa Gold Mine

Six months ago, we entered a large #gold mining operation in West Africa with a pilot project to deploy our fleet management system OES.Mine.

Due to NDA restrictions, we cannot disclose the company name, country, exact fleet size or operational details. But we can share what matters most: the pilot was successful, and the client has now signed a contract for full-scale implementation.

This was not a laboratory test. The operating environment was a serious challenge for both hardware and software: heat, dust, heavy rain season, LTE coverage limitations, site-specific workflows and a completely new operational context for our team.

The pilot scope was deliberately limited: several haul trucks, one excavator, and one month of test operation. During the pilot, we confirmed several critical things:

• The hardware works in African mining conditions, with installation adapted to the client’s environment, including cooled tablets, antenna positioning and equipment-specific mounting logic.
• The system was deployed in the cloud from scratch in two weeks.
• Local LTE connectivity was sufficient for real-time fleet visibility and dispatching.
• The system recognized more than 95% of haul truck cycles, validated against manual records and onboard truck data.
• The client worked with the system in daily operations and gave strong feedback on the core functionality: live map, haulage cycles, delays, downtime, reports and dashboards.

As a result, we are now moving into full-scale deployment in two stages.

1️⃣ Stage 1: core mining fleet We will deploy a full fleet management system for excavators, haul trucks and loaders, including automatic haulage and cycle tracking, real-time dispatching, downtime control, live fleet map and digital ore accounting instead of paper-based records.

2️⃣ Stage 2: auxiliary equipment and road/speed management We will connect dozers, graders, water trucks and drill rigs. The next layer will include road quality and speed control, where every haul truck becomes a sensor of haul road condition, and auxiliary equipment can be managed online based on real operational needs.

This is an important step for OES Ventures. Africa, gold, new climate, new equipment, new language, new operating culture. Full-scale deployment will not be easy. But that is exactly why the pilot mattered: to test the system in real conditions before scaling.

We will continue sharing practical lessons from the field.

Read more — in our LinkedIn www.linkedin.com/company/oesventures/

Ghana Impresses OES Ventures at WAMPEX 2026

WAMPEX 2026 was an important stop for OES Ventures in West Africa. We didn’t come with a booth this time. We came to listen, learn, meet people and understand the market better.

And Ghana impressed us.

The quality of conversations, the openness to technology, the strength of the local mining ecosystem and the level of ambition across the sector all point in one direction: Ghana is not just one of Africa’s leading gold producers. It is also one of the most promising markets for the next generation of mining technology.

For OES, the opportunity is clear. Across gold and other resources, the challenge is no longer only about having more equipment or more data. The real question is how to control complex open-pit operations in real time: reduce truck queues, prevent excavator idle time, manage losses during the shift and help teams make better operational decisions every hour.

That is exactly where AI-native operational control can create value. During WAMPEX, we had many meaningful conversations with operators, contractors, technology companies and potential partners, including Ramjack Technology Solutions, Rocksure International Limited, Maxmass, Asante Gold, Cardinal Namdini Mining Ltd. (A Shandong Gold Company), Zijin Mining Group and many others.

A special thanks to the The Ghana Chamber of Mines and Michael Edem Akafia - MCIArb for the role they play in strengthening the industry dialogue and bringing the ecosystem together.

We leave Accra with stronger conviction: Ghana has the assets, the people, the partners and the business landscape to become one of the key markets for OES in Africa.

Now the real work starts.

Read more — in our LinkedIn www.linkedin.com/company/oesventures/

Real-Time Telemetry Boosts Mine Productivity

We talked about technical availability as the foundation of mine productivity. If trucks and excavators are down, no dispatch logic or truck-shovel balancing can save the shift.

Telemetry is what makes that foundation manageable.

Modern mining equipment constantly generates data: oil pressure, coolant temperature, transmission vibration, engine load, brake condition, tyre pressure and many other parameters. In theory, this should give operations and maintenance teams a live view of machine health.

In practice, too often this data is stored and reviewed only after a failure. It becomes a tool for investigation, not prevention.

Real-time telemetry changes that. If a component temperature starts drifting over several shifts, if vibration patterns change, or if oil pressure behaves differently from similar machines, the team can act before the truck stops in the middle of the pit.

That means planning maintenance at the right time, preparing parts in advance, redistributing load across the fleet and avoiding emergency downtime.

The challenge is that telemetry is not a universal language. Every OEM, model and even generation of equipment can send data differently. At OES, we collect telemetry every 1-5 seconds, decode it for each vendor and model, and turn raw machine data into clear signals for the people making operational decisions.

But collecting and visualising data is only the first step. The real value starts when there is enough data for the machine to tell us what is likely to happen next.

Read more — in our LinkedIn www.linkedin.com/company/oesventures/

Fleet Technical Availability — The Foundation of Mine Productivity

When people talk about mine productivity, they usually start with volumes, speed and logistics: how many cubic metres were moved, how fast haul trucks are running, how efficiently excavators are loaded.

All of that matters. But behind these metrics sits one fundamental factor: the technical availability of the fleet.

A machine under repair produces nothing. Every hour that a 130-tonne haul truck spends in the workshop instead of the pit is not just a maintenance cost. It is lost production that usually cannot be recovered within the same shift.

Mining is a continuous flow. When one element drops out, the impact spreads across the system: excavators wait, trucks are reallocated, cycles become unstable, and the plan starts to slip.

At a large open-pit mine, the fleet may include hundreds of units. Each machine is a complex system with thousands of components exposed to wear, fatigue and degradation. Breakdowns will always happen. The real question is how predictable they are, and how prepared the operation is before they happen.

That is what technical availability really measures: how much of the time the equipment is actually available for work.

The difference between 85% and 92% availability at the scale of a large mine is not just seven percentage points. It can mean tens of thousands of machine hours per year. Those hours either become production and revenue, or disappear into downtime.

With today’s cost of equipment and spare parts, the gap between reactive and proactive maintenance can easily become hundreds of millions in lost value.

But to move from reactive to proactive, a mine first needs to learn how to listen to its equipment.

Read more — in our LinkedIn www.linkedin.com/company/oesventures/

ROC Coordinators: the people behind remote mine control

We often talk about AI agents, truck-shovel balancing and multi-agent systems.

But none of this works properly without the people behind the control loop.

At OES, we call them ROC Coordinators.

ROC is our Remote Operations Centre. From one location, the team works with open-pit mines hundreds or thousands of kilometres away. On their screens, they see the same fleet, the same queues, the same delays and the same losses as the site dispatch team.

The AI layer does a lot by itself.

It monitors the fleet, detects deviations, calculates losses, sends directives to operators and recalculates truck-shovel allocation every few minutes. It can forecast queues ahead of time and redirect trucks before an excavator starts waiting.

But automation still needs control.

ROC Coordinators configure the agents for each specific mine: thresholds, rules, priorities, shift logic, fleet behaviour and escalation scenarios. A good algorithm without proper configuration creates noise. A well-configured algorithm becomes a production control system.

They also step in when the system has done its job, but action has not followed. If a directive is ignored, if a machine is not connected, or if a situation requires human judgement, the coordinator takes over and makes sure the shift keeps moving.

This is not “AI replacing people”.

It is AI taking over routine monitoring and recommendations, while experienced people handle configuration, discipline, exceptions and safety-critical decisions.

Our ROC team brings together former dispatchers, production specialists and operational efficiency experts. Because they work across different mines, they see patterns that are hard to notice from one site only.

That is where the real value is.

The platform finds the losses.

The coordinator makes sure they are acted on. And the mine becomes more controllable during the shift, not after the report is written.

Read more — in our LinkedIn www.linkedin.com/company/oesventures/

«We’re not ready for this yet. We only have 30 trucks»

This is one of the weakest objections in mining tech.

AI-based truck balancing does not start at 100 trucks. It does not even start at 50. In practice, once you have around 20-25 trucks working across multiple excavators, the system is already too dynamic for manual allocation to compete with an algorithm recalculating every few seconds.

Let’s be precise. Fleet size affects the size of the prize. It does not determine whether the problem exists.

A 30-truck mine may generate less absolute upside than a 150-truck one. Obviously. But the underlying losses are usually the same: truck-shovel imbalance, hidden shift-change losses, badly timed refuelling, queueing, idle equipment, unstable haulage flow.

The difference is that smaller operations usually feel those losses more sharply. Large mines often have buffer. They can survive inefficiency. Smaller ones often cannot. When your fleet is small, every lost trip matters more, not less.

So “we’re too small for AI” usually means something else: “We’re too small to buy an overpriced, heavy, consultant-fed famous system.”

Fair enough. But that has nothing to do with whether AI balancing is relevant. It has everything to do with the kind of solution being offered.

We have seen smaller sites improve fast, because the operational losses were already there. The mine was not too small for AI.

It was just running below its potential.

Read more — in our LinkedIn www.linkedin.com/company/oesventures/

The first hour of every shift is disappearing. Twice a day.

In theory, a shift change takes fifteen to twenty minutes. One crew hands over, another takes over, work continues. In practice, up to sixty minutes evaporate every single shift — in chaos so familiar that nobody counts it as a loss anymore.

Here's what it looks like without automation. The outgoing crew starts winding down early — drivers park wherever is convenient for them, not where production needs the truck. One leaves his machine at the fuel station, another at the far dump, another in a pit face that takes the incoming driver twenty minutes to reach. The crew bus drops the new shift at the central yard. Then the quest begins: find your truck, get to it, start up, radio dispatch, get an assignment, pull onto the haul road. Meanwhile, excavators sit with empty buckets.

And then there are the early quits. A driver with ninety minutes left starts "preparing" — slower speeds, the nearest dump, maybe just parking up a little early. On paper, he's still on shift. In practice, the last hour runs at sixty percent of normal output. Nobody sees it because the report shows a full shift.

We measured this on one of our projects: the actual time from crew bus arrival to the last excavator running at full load was 47 minutes. Twice a day. Three hundred and sixty-five days a year. Do the math on how many haul cycles that is.

Our shift change module rebuilds this process from scratch. The system knows where every truck is, knows the assignment plan for the incoming shift, and knows which drivers are coming on. Before the new shift starts, every truck is automatically assigned an optimal handover point — so the incoming driver sits down, sees the route on the tablet, and is moving toward the excavator within three minutes. The crew bus gets a routing plan calculated around assignments, not convenience.

Early quits are visible in real time. If a driver starts parking forty minutes before end of shift, the agent flags it and calculates the cost of that loss — not to issue a penalty, but to make it visible. When people know every early quit is tracked and priced in real money, behavior changes without a single written warning.

Results where the module is running:
🔹 –25% early quits and late starts
🔹 –15% downtime during shift change windows
🔹 +$55–75K additional revenue per month

Same fleet. Same people. The first hour of the shift stopped being lost — simply because every driver now knows exactly where to go before they ever get in the cab.

Read more — in our LinkedIn www.linkedin.com/company/oesventures/

Autonomous Haulage: Separating Hype from Reality in Mining

Autonomous haul trucks are one of those topics that guarantee applause at any mining conference. The future has arrived, at least if you listen to the presentations.

But let’s look at it without the hype.

Today, there are around 4,000 autonomous haul trucks operating worldwide. Sounds impressive, until you realise the total global fleet is about 50,000. Less than 8%. And almost all of them are concentrated at a few of the world’s largest operations like Rio Tinto, BHP, and Suncor, companies that can afford to spend $500 million to $1 billion on the infrastructure behind autonomy.

Because the real cost is not just the truck. Retrofitting a single 200-tonne machine for autonomous operation can cost millions. Then come the networks, control rooms, sensors, road redesign, and system changes. Payback is usually 3 to 5 years, assuming everything goes to plan. But here is the question that is asked far too rarely: where exactly is that autonomous truck going?

An autonomous truck does not think. It executes. It goes where the system sends it. And if the system is making poor decisions, the truck will follow them just as obediently as a human driver, only at a much higher cost.

For most mines, the problem is not who is behind the wheel. The problem is that 10,000+ deviations per shift remain unmanaged, balancing is still done manually, and most losses are never captured in time. Spend a billion dollars on autonomous trucks, and they can still end up autonomously queuing, autonomously idling, and autonomously losing the same tonnes.

Our view is different. Before replacing drivers, mines need to replace manual control logic.

Autonomous haulage answers the question of who drives the truck. AI-native dispatch answers the more important one: where it should go, when, and what should happen next.

That second problem is more valuable. And much cheaper to solve.

Read more — in our LinkedIn www.linkedin.com/company/oesventures/

AI Shift Management Boosts Productivity in Eurasia's Largest Coal Mine

In the previous two posts we broke down why dispatching exists and why the traditional model hits a ceiling. Now — how we think shift management should actually work.

Today's typical dispatcher spends most of their shift on routine: tracking equipment, rerouting trucks, filling in logs, reacting to things that have already gone wrong. Proactive management — finding inefficiencies before they compound — gets maybe 10–15% of their attention. That's exactly where the real gains are.

At one of our flagship operations — largest coal mine in Eurasia — we built it differently. The core idea: take the cognitive load of managing hundreds of small situations per shift off the dispatcher, and hand it to the system.

An AI balancer continuously recalculates truck assignments and sends directives straight to drivers' tablets in the cab. Not through the dispatcher, not over the radio — directly. The driver sees the route, confirms, and moves. No more calling eighty people one by one.

For everything else, we built MAS: a multi-agent system of thirty-plus AI agents, each responsible for its own zone, each taking a decision all the way to the specific person who needs to act on it.

🟢 Excavator losing productivity? The agent identifies the cause and sends the mine supervisor a specific task. 🟢 Trucks slowing on a haul road? The agent generates a maintenance task automatically — no calls, no waiting.
🟢 Driver stretching a refueling stop? A reminder goes straight to his tablet.

And so it goes — for every deviation, every piece of equipment, around the clock.

The results:

✅ +17% excavator productivity
✅ 15–25% downtime
✅ $900K/month in savings

These numbers didn't appear because we replaced dispatchers. They appeared because we gave them the space to do what actually matters. The dispatcher of the future isn't someone replaced by AI. It's someone AI gave their time back to.

If you've ever walked into the dispatch center of a large open-pit mine, you've seen the same setup: screens with a map, a radio, a logbook, and a person trying to hold the balance between the plan and reality. The model took shape decades ago and, to its credit, it works.

But it runs on people more than it runs on systems. Every experienced dispatcher develops their own logic — their own priorities, their own habits. When they leave, it turns out the knowledge wasn't in the system. It was in their head.

Traditional dispatching is reactive by nature. A truck stops — reroute it. An excavator sits idle — find someone to send over. Honest work, but almost entirely made up of fighting fires that didn't have to start.

The advantages are real: battle-tested, familiar, understood by everyone on the floor. But the disadvantages compound with scale. Losses spread across small things — no single one looks critical, but added up over a year, the number is uncomfortable.

When a model hits its ceiling, something has to change.

Read more — in our LinkedIn www.linkedin.com/company/oesventures/