From 15 Mines to Nationwide Impact: How I Ended Up Building GeoMine AI from Gilgit Baltistan

I was standing on a ridge above Skardu in April 2022, holding a paper map that was older than me, watching my crew dig in the wrong spot for the third week straight.

That's the moment. That's when GeoMine AI actually started, even though I didn't call it that yet.

Let me back up.

I own 15 mines across Gilgit Baltistan. Some are producing — marble mostly, a bit of granite, one chromite lease that's been in the family for a while. Others are exploration blocks I picked up because the geology looked right and the price was fair. Running that many sites teaches you something fast: the biggest cost in mining isn't labour, isn't fuel, isn't even royalties. It's digging in the wrong place.

I've done it. More times than I want to admit.

The survey that broke me

In 2021 I paid a well-known consulting firm from Islamabad roughly 4.7 million rupees to survey a gold-prospective block near Chilas. They took eight months. Delivered a PDF report that looked beautiful — nice cross-sections, a few thin-section photos, some XRF numbers. The recommendation? Trench in three zones.

We trenched. We got almost nothing. Trace values, nothing economic.

And here's the thing that really got me — six months later I got access to some Sentinel-2 imagery through a friend doing his PhD in Germany. He ran a basic band ratio on the same block in about two hours. The alteration signatures were pointing us 600 metres northeast of where we'd been digging.

Six hundred metres. Two hours of satellite work vs eight months of field consulting.

I wasn't angry at the consultants. Honestly, most of them are doing their best with the tools they know. I was angry at the whole system. Pakistani mine owners were paying premium fees for methods that hadn't really changed since the 90s, while the satellite data sitting in ESA's free archive could've saved us crores.

That gap is what GeoMine AI exists to close.

Why Gilgit Baltistan is the right place to build this

People ask me why I didn't set up in Karachi or Lahore. The answer's simple — you can't build a mineral intelligence company from a city where nobody's ever set foot on an actual mine face.

Gilgit Baltistan is one of the most mineralised zones on the planet. The Kohistan Island Arc, the Karakoram suture, the Nanga Parbat massif — this is textbook geology for gold, copper, chromite, and gemstones. I've got aquamarine coming out of Shigar, ruby out of Hunza, and enough marble varieties across the region to keep Italy nervous.

But it's also brutal terrain. You can't just drive to a prospect. Some of my leases take two days of jeep and foot to reach. So every wrong guess about where to start work costs weeks, not hours.

That's exactly why satellite intelligence matters more here than almost anywhere else. When I run breeze geo mineral analysis on a block before sending a team in, I'm not just saving money. I'm saving my guys from climbing the wrong mountain in the wrong season.

What we actually built

GeoMine AI pulls together Sentinel-2 for surface alteration mapping, ASTER for mineral-specific band ratios (SWIR is where the magic happens for clay and iron oxides), SAR for structural analysis under snow and vegetation cover, and SRTM DEM for drainage and lineament work. Then we layer AI models trained on known Pakistani deposits to flag anomalies.

The output isn't a 200-page PDF. It's a targeting map. Here's where to look. Here's the confidence level. Here's what mineral signature we're seeing.

A report that used to take a consulting firm six months and cost 40+ lakh, we turn around in under two weeks for a fraction of that. And I say this as someone who's on both sides of the transaction — I'm a customer of my own platform. I run every new lease I evaluate through geomines before I spend a single rupee on field work now.

I got the pricing wrong at first, by the way. Priced it too high because I was thinking like a consulting firm. Then I realised the actual market — the guy in Chitral with two chromite leases who's never going to pay 15 lakh for a report but will absolutely pay 80,000 for a targeting map that tells him where to start his adit. That's who needs this. That's where nationwide impact actually happens.

What comes next

We're onboarding provincial mining departments now. Balochistan's the obvious one — Reko Diq gets the headlines but there are hundreds of copper and gold blocks across Chagai and Kharan that nobody has properly imaged. Khyber Pakhtunkhwa has a serious emerald and marble portfolio. Sindh has coal and industrial minerals that satellite work can absolutely help with.

And I want individual mine owners on the platform. Not just the big companies. If you own one lease in Muzaffarabad or three in Waziristan, you should have the same intelligence a multinational has. That's the whole point.

Look, I'm not going to pretend I had some grand vision from day one. I was a Pakistani mining entrepreneur who got tired of guessing. I built the tool I wished existed when I was standing on that ridge in Skardu with a bad map.

Turns out a lot of other people were waiting for the same thing.

So — if you're sitting on a lease and you don't know what's under it, what are you actually waiting for?