A Client Sent Me Rock Samples From Chitral. We Verified His Gold Claim From Space in 48 Hours.

A guy named Faridullah called me last March. He'd spent almost three years and — his words — "more money than my wife knows about" chasing a gold claim near Chitral. Everyone kept telling him the site was rich. A local geologist. Two consultants from Peshawar. Even a cousin who owns a jewelry shop and apparently now qualifies as a mineral expert.

He wanted a second opinion that didn't involve another crew showing up, eating his food for a week, and handing him a report full of maybe.

So we ran his site through GeoMine AI. Start to finish took 48 hours. Here's what actually happened, because I think the details matter more than the pitch.

What we had to work with

Faridullah gave us coordinates and about six photos of rock samples on a plastic chair. That's it. No drill logs, no assay reports, nothing you could take to a bank.

Good news is we don't need much. For a rapid mineral assessment, coordinates are enough to start pulling the satellite layers we care about.

We stacked four data sources over his claim. Sentinel-2 for surface mineralogy and vegetation stress. ASTER for the thermal and shortwave bands that pick up alteration minerals. SRTM DEM for the terrain and structural clues. And SAR to see through the cloud cover, which — if you've ever tried optical imaging over Chitral in spring — you know is a real problem. Those valleys stay socked in.

The first thing we look for with gold isn't gold. You can't see gold from orbit. Nobody can. What you can see is the neighborhood gold likes to live in.

Hydrothermal alteration. Iron oxide staining. Argillic and phyllic zones — clay and mica minerals that form when hot mineral-rich fluids cook the surrounding rock over thousands of years. Gold rides along with those fluids. So when the ASTER bands light up with sericite and the Sentinel-2 iron ratios spike in the same footprint, you're onto something.

And Faridullah's site? It lit up. But not where he thought.

The part he didn't want to hear

Here's the thing about his claim. The actual pit he'd been digging — the spot all three consultants pointed at — sat on the weakest signal in the whole area. Barely any alteration. The iron oxide anomaly was there but faint, maybe a 2 out of 10 on our internal scoring.

The strong signal was roughly 800 metres northeast. Up a slope, along a structural lineament we picked out of the DEM. A fault line, basically, running northwest to southeast, and gold-bearing systems love faults because that's the plumbing the fluids move through.

So the good news and the bad news arrived together. Yes, there's a real gold indication near your claim. No, you haven't been digging in the right place. For three years.

I won't lie, I sat on that call for a second before saying it.

We scored the northeast target at 71% prospectivity on our model. Not a guarantee — I'll come back to that — but strong enough that I'd put my own money on drilling it. The old pit came in at 23%.

Where I got it wrong (and why the ground still matters)

I used to oversell this stuff in the early days. Back when I first built geomine, I'd tell people satellite data could basically replace fieldwork. That was arrogant and honestly a bit stupid.

Satellite intelligence tells you where to look and where not to waste money. It does not put a number on grade. It can't tell you if you've got 2 grams per tonne or 12. Only a drill and an assay lab does that.

What we did for Faridullah was save him from spending another year and another few million rupees on the wrong hole. The gold verification in Chitral we handed him was a probability map plus a ranked drill target list, not a certificate that says "there's gold here, go get rich." Anyone selling you that second thing is lying.

He took our report to a proper crew in July. They ran a small trench program on the northeast target. Visible sulphide mineralization, iron staining exactly where the ASTER data said it'd be. Assays came back with values worth chasing. Not the biggest thing I've ever seen, but real, and — the point — real in the spot the geo mining analysis flagged, not the spot he'd been breaking his back on.

That's a GeoMine case study I actually like telling, because it's honest about the limits.

Why 48 hours and not two months

The old way of checking a claim like this: hire a team, wait for them to be free, get them to Chitral (the road alone eats two days each way if the weather's bad), let them survey, wait for the writeup. Two to three months, easy. Sometimes more if the assay lab backs up.

Our breeze geo mineral analysis engine pulls the satellite layers, runs them through the alteration and structural models, and generates the report while you sleep. Faridullah sent coordinates on a Tuesday afternoon. He had a 40-page report Thursday evening.

Cost him a fraction of one ground survey. And it saved a ground survey — he only sent a crew once he knew exactly where to send them.

Look, I own 15 mines in Gilgit Baltistan. I know the pain of paying for exploration that goes nowhere. Every dry hole is money you're never getting back and a season you can't rerun. If a satellite pass can shrink the search area from a whole valley to three tight targets, that's not magic — it's just physics and a lot of training data doing what they should.

Faridullah still calls. Last time he wanted to know if we could run his uncle's chromite site in Waziristan.

I told him send me the coordinates. What else were you gonna do — trust the cousin with the jewelry shop again?