Hear someone called “Segovian” in Colombia and you know they love gold. After 150 years of mining in the Segovia region of Antioquia, locals live and breathe the art of excavation.

GoldSpot Discoveries Corp., a technology company that leverages machine learning in resource exploration, set out to help Gran Colombia Gold piece together new geological insights – from a literal maze of historic mine data.

“Gran Colombia’s Segovia operations include three main producing mines, and numerous smaller mines operated by contract miners,” says Shawn Hood, Principal Geologist and Vice President of Technical Services at GoldSpot.

GoldSpot’s data integration and machine learning solutions allowed them to create robust interpretations from the massive amount of data covering the area’s network of tunnels and ore deposits. But, like all geological models, the AI results could only be as accurate as the inputted data.  

The controls on ore-shoots are discussed between structural geologists of GoldSpot and Gran Colombia. Understanding these, and representing them digitally, was fundamentally important to the success of the project. 

GoldSpot needed a way to bring geoscience experts together to refine the information first.

“What we didn’t have was a way to share data which complemented our collaborative company approach and philosophy. We’d done things in the traditional project-specific way of uploading and sharing data in difficult to manage cloud directory structures,” Shawn says.  

“Seequent Central helped us modernise our approach.”

A place for all filetypes 

GoldSpot data scientists receive modelled data from geoscientists, and use data features for machine learning propectivity mapping and target generation. 

Their globally distributed team needed to collaborate on the same geological model and share a variety of file types – but how? Shawn spent a week calling every geological data group he could find.

“Every single software offering I could find was outdated, inefficient, and really just legacy products that companies had hung on to because they had spent so much time developing it in the past,” Shawn says.

“And none of it really dealt well with unstructured data, or with a modern cloud-oriented storage and version tracking.”

Using Central’s Data Room, GoldSpot found a secure place to store and version control their files – from photos, meshes, and CAD files, to engineering designs and Excel files – whether the data was structured or not. Crucially, they could track their revisions and comments.

“We suddenly had a very simple and straightforward way and it saved us a tremendous amount of time. We knew the version and when it was made. We knew we had the most current geological interpretation, and we captured notes and caveats from our colleagues.”

3D models from the underground, up 

Shawn’s team spent weeks in the mines and countryside of Segovia, working collaboratively with Gran Colombia’s exploration geologists. Days were used to understand the possibilities and limitations of the data for machine and deep learning techniques.

In the evenings, they progressively built their initial geological models, constructing an understanding of the maze underfoot.

GoldSpot worked with Gran Colombia’s exploration geologists to understand regional trends, which are then mapped or digitised in Leapfrog. 

“Getting geological ideas into 3D, digitally, is faster with Leapfrog than with other methods I have used. Being on site for 16 days, we made observations to decide which data we could use from the client, where to verify uncertain areas, and where to add extra data points ourselves,” says Shawn.

Information came in by hand, hard drives, maps, and samples to reveal a much bigger picture of Gran Colombia’s resource.

“We were able to integrate diverse data in Leapfrog, and produce a representation of the area, which then became the foundation for the more advanced computational techniques that followed.”

The team stored these files and meshes in Central’s accessible cloud and continued to work together, even after the field team members returned to their respective regional offices.

Managing data without the doubt 

GoldSpot’s experts and AI algorithms worked together from Montreal, Toronto, Ottawa, Sudbury, Vancouver, Whitehorse, and even Belgium.

Their geological model was constantly evolving with new insights (from both human minds and AI) so they needed a way to update it in real-time – without losing track of changes.

Leapfrog is used to visualise data clouds, produced by data scientists working with geoscientists, for validation and interpretation by the multi-disciplinary team of geologists and geophysicists. Interpolants can then easily be shared using the Central platform.

“With Central, we don’t have to worry about replicated data or losing track of interpretations, because there’s a master file which is on the server. It eliminates the confusion and the doubt, especially working across time zones,” says Shawn. 

“Having an automated audit trail facilitates our workflow internally to a very noticeable degree. It makes us a lot faster. We can respond more quickly. But most importantly – we don’t make mistakes.” 

Show great results, faster 

Within their exploration target model of Segovia’s mines, GoldSpot is using Leapfrog‘s drillhole planning tool to recommend potential targets to explore.

Through well-managed data, AI, and a collaborative spirit, GoldSpot seeks to uncover metals in cutting-edge ways for their clients – fast. Back at the mine site, some might even call their team’s prospecting abilities “Segovian.”