Mining at the crossroads of geology and algorithms

Illustration by Digital Journal

In mining, billion-dollar bets depend on how someone reads a rock. Exploration starts by drilling narrow tubes of rock called “cores” that reveal the mineral potential underground. The way those cores are logged can decide whether a project advances or stalls. 

Sometimes it’s an experienced geologist, and sometimes it’s a student. In both cases, the process comes down to what the human eye can see. 

For more than a century, geologists have examined core samples, comparing notes and trusting judgment to guide exploration and investment. 

“We miss things, and we miss a lot of things,” says Michelle Legat, vice-president of applied geosciences at GeologicAI. A geologist by background, Legat is not speaking about her company’s work, but about the traditional process of logging core that GeologicAI was built to fix.

In mining exploration, the process of drilling cores can cost $50 to $150 per metre, meaning a single 1,000-metre hole could cost more than $100,000.

GeologicAI replaces handwritten logs with mobile labs that scan cores on mining sites and applies AI to highlight patterns, creating digital records that can be analyzed and shared almost instantly.

Once the core is drilled, a geologist stands in a core shack which is almost always in a remote location with a pencil and clipboard, wetting down rock to bring out contrast and jotting observations by hand. It’s slow and manual, and the record of a deposit’s potential is built from that person’s notes.

To illustrate the shift in how GeologicAI does it, Legat pulls up her screen and shows me a full core that has already been logged. Instead of pencil marks on wet rock, the record is captured in high-resolution images and traced with computer analysis. Artificial intelligence highlights features that geologists once had to mark by hand, turning what was a manual exercise into something that can be reviewed, compared, and shared easily and quickly.

The images all look the same to me at first. They’re dark rock with yellow grease-pencil markings that loop around fractures and veins. But as Legat runs through the demo, I can see depth labels. Similar-looking rocks are differentiated with on-screen markup. Even without knowing how mining works in depth, you can see how patterns stand out and how a long story of rock becomes readable at a glance.

She clicks through the core images, and the software highlights features that could slip past the human eye. Veins, fractures, subtle mineral shifts, even the difference between real ore and fool’s gold can be flagged within seconds. Her tone stays even as she explains it, but the outcome is significant. A process once reliant on guesswork can now be backed by data.

What she is showing on screen is the same process GeologicAI brings into the field with its mobile labs.

The company transports scanning units to remote projects, where entire drill programs can be captured in real time and turned into digital records that anyone on the team can review from anywhere. Each scanning unit has a geologist on site who checks the scans for quality and accuracy. 

What once took weeks, months, or a full season to interpret can be checked in less than a day, with a permanent record that makes results easier to compare across teams, projects, and years.

The combination of speed, consistency, and evidence in the field has made the company hard to ignore. In July, GeologicAI raised $44 million in Series B funding from Breakthrough Energy Ventures, BHP Ventures, and Rio Tinto to scale these systems globally.

Why innovation delivers outsized returns in mining

Mining is living through the same wave of innovation shaping industries from energy to finance. The companies that adopt new approaches earliest are the ones that see the largest gains. 

In mining, those gains show up in faster discoveries, more reliable reporting, and stronger credibility with investors. 

The industry’s scale makes this visible.

According to the Mining Association of Canada, Canada’s broad mining sector (which includes extraction, services, primary and downstream manufacturing) contributed $117 billion to the country’s GDP in 2023. For those keeping score, that’s 4% of the national economy, with 430,000 direct jobs and another 281,000 in indirect roles.

Those numbers show how central mining already is to the economy, and underline why the stakes are high as demand shifts. The next wave of spending is chasing the critical minerals needed for electric vehicles, renewable energy, and advanced computing.

The trend is global. 

Australia is investing in digital exploration, Chile is pursuing technology partnerships to speed copper output, and China is deploying state-backed systems to secure long-term mineral supply. The shift shows how central data and technology have become to the future of mining, and it’s reshaping how Canada positions its own resources in global supply chains.

It’s also what drew John Mortimer to GeologicAI. 

Mortimer joined the company recently as chief technology officer after years of leading enterprise-scale data projects. What drew him in was the timing. Mining is under pressure to deliver the critical minerals the world needs, and he saw in GeologicAI a system ready to meet that demand at scale.

His focus now is preparing the company for its next phase, where digital records are not just useful in the field but trusted across the largest mining companies in the world.

“From the first byte of imaging to the display seen by an end user in a corporate office doing large-scale mine planning, each step brings different challenges,” says Mortimer. “If there is one thing that we need to care about, it’s the data. It’s the data platform. It’s how we ingest it, how we can train our models on it, how we can put new models on it for new inferences and interpretation.”

Mortimer’s focus on data is not just theoretical. In the field, the difference is already showing up.

At a silver mine on Canada’s west coast, GeologicAI’s scans spotted high-grade material that had been missed entirely. In the traditional process, the core had been logged and sampled, but no one flagged the richer section. When the digital record was reviewed, it stood out immediately.

“They had a high-grade sample at the top and they did not sample it whatsoever,” says Legat. “They had no idea that it was in there and they started pulling up the core photos. They never even gave us an indication to look there and now we can expand into the parts of the mine we did not know that we should be looking at.”

A missed section of core can mean leaving millions of dollars worth of metal in the ground. Catching it in the data changes the economics of a project and the confidence investors have in it.

From skepticism to standard practice

For a technology to take hold in mining, it has to win over the people who use it every day. Legat has seen that shift firsthand. 

When GeologicAI’s scanning units roll onto drill sites, veteran loggers are sometimes the toughest critics. Many have decades of experience and are initially reluctant to trust a machine with work they have done by eye their entire careers.

Legat remembers one team she worked with in particular.

“They had been doing this for decades and they were our biggest skeptics,” she says. “My favourite people to work with because once you win those ones you win everybody.”

Legat said a senior staffer with that project did a complete 180 after realizing the scans could capture more detail than he could by hand. The turnaround was so strong that he now insists on having the scanning data before logging.

That kind of turnaround speaks to how adoption spreads. Mining is conservative by nature, and people are not quick to abandon methods that have worked for generations. 

But once respected loggers change their minds, others tend to follow and peer-to-peer advocacy can spread faster than any formal rollout.

Legat also points to the issue of consistency, as no two geologists log a core in exactly the same way. Personal experience, training, even fatigue can change how the same rock is described. 

“We all have our own biases and our own ways of looking at things,” she says. “What we’re building is something that everybody can look at and get the same result.” That shared record gives companies a foundation they can compare across projects, years, and teams.

Legat says the shift is so widespread that on site, decisions now often start with the same question: “What does the scan data say?”

That shorthand shows how the scans have become part of routine practice.

The system’s Auto Lithology Logger is part of that push for consistency. By training on the way a particular site already classifies rock, the software can predict lithologies (the types and characteristics of rock) with confidence levels attached.

“Imagine being the first day on the job and you have no idea how they log their core,” says Legat. “Our software tells you what they have called it previously so you can train junior people very quickly on how to log at that site so it’s specific to that site.”

The technology is also changing how companies validate their work. 

Traditional logging often means sending rock samples to external labs for testing, which can be quite time consuming. 

“If you’re working up in the Arctic, you send your samples off to the lab and it’s months before you get anything back,” she says. “By the time you know what you’ve got, the season’s over and you can’t even go back to drill until the next year.”

Legat explained that one of the big “aha” moments comes when companies see GeologicAI’s scans line up with their own lab results. What really convinces people is when the system shows how the AI reached its conclusions. Rather than acting like a black box, it highlights the signals driving its predictions.

Beyond speed, the scans can also reveal features invisible to the human eye. 

For example, Legat explained that hyperspectral scanning can pick up details like subtle changes in minerals that hint at gold. The system produces a heat map showing exactly where those changes appear. 

“As soon as you see that, sample the crap out of it because we’re going to find gold,” she says. With the company’s hyperspectral scanning capabilities, the images tell you where to look. “It screams at you: ‘Sample here, because all conditions are right to have gold.’ It doesn’t get any easier in my mind.”

Mortimer sees this ability to merge speed, accuracy, and consistency as the bridge between the field and the enterprise. 

The logger in a shack and the engineer in a head office need to be looking at the same data, and that data has to be secure enough and consistent enough to stand up inside a global mining company. 

That enterprise lens is what brought him to GeologicAI. 

Mortimer says mining is ripe for digital transformation because of the industry’s reliance on data that is often still recorded by hand. Bringing those records into a digital platform, he argues, is the only way to meet the expectations of global operators and their investors.

Mortimer says creating permanent digital records that teams can review right away helps companies move faster without losing confidence. Even if investors never set foot in a core shack, the data they see has to carry the same weight.

The timing also matters.

Demand for critical minerals is rising at the same moment industrial AI is reaching maturity. Mortimer believes the overlap is set to reshape the industry. 

“I think the moment in time right now is that you’re seeing an intersection of the need for critical minerals in the world, and this is being driven by the electrification of everything,” he says. “It’s being driven by energy transition. It’s being driven by AI and super data centres and all of the things that you need for what I’ll call industrial AI. This is not ChatGPT. We’re applying AI with a true backing of science, and that’s where the frontier of the industry is moving.”

That transformation is cultural as much as it is technical. 

For Legat, it looks like skeptical geologists becoming advocates once they see results. 

For Mortimer, it looks like enterprise clients demanding systems that can deliver speed, scale, and security all at once. Put together, it signals that digital core logging is not just an experiment. It is on its way to being a new standard.

Turning science into a system the industry can trust

Proving the science was one step. Turning it into infrastructure that global mining companies can rely on is another. 

Mortimer says the company’s challenge now is making sure the system holds up under the weight of scale, where data is constantly pouring in.

Each scanning unit can generate up to 100 terabytes of data in a single month. With dozens already deployed, the company is managing petabytes of scans — what Mortimer calls the world’s highest resolution store of critical mineral information.

Managing that flood requires more than clever software.

The scanning units function as high-performance compute centres, processing and filtering raw scans before sending them over satellite connections. 

Even with systems like Starlink (which promise speeds of 200 megabits per second) bandwidth at remote sites is usually closer to 5-10 Mbps. That makes it impossible to move terabytes of raw data and forces GeologicAI to process and package information in the field before transmitting it to the cloud where mining majors can integrate it into their systems.

“You can’t treat this like a startup proof of concept,” says Mortimer. “We are delivering to global mining companies. Security, scale, and reliability are non-negotiable.”

That shift means tackling the tech stack and deciding what can scale and what has to be rebuilt. 

“Any company that’s more than a day old has legacy in it, and so you’re trying to understand what you can build upon and where you have technical debt,” he says. “You need to start to think about investing so that you can build scalability.

Trust in the system extends beyond geology, too. 

The records must be defensible to geologists, transparent enough for investors, and consistent enough for engineers planning multi-billion-dollar projects. Mortimer frames it as a shift from being a tool for geologists to becoming a data backbone for the enterprise.

“This is no longer just about showing geologists a better tool,” he says. “It is about giving an entire enterprise confidence in its most valuable records.”

For mining companies, the payoff is twofold. Having scanning units on site delivers faster decisions and fewer blind spots. But the greater value lies in the creation of a permanent, trusted record that can be compared, validated, and shared across time. 

If GeologicAI can scale that model globally, it will no longer be seen as a vendor. It will be treated as infrastructure, embedded in how the industry explores, plans, and invests.

The technical hurdles are there, but the opportunity is large and GeologicAI is  building a world where every metre of drilled core contributes to a unified, secure data platform that reduces risk and accelerates discovery. For mining companies and investors, the payoff is immediate: more consistent records, fewer surprises, and stronger confidence in the results.

In a sector defined by uncertainty, this foundation could change not just exploration itself, but the very decisions that determine which projects move forward.