The desert tracker read the sand. QASSAS reads the data room.
An entire data room goes in — tables, maps and scanned reports in Arabic or English — and ranked targets come out, each with a verdict tied to the document it rests on.
Four independent lines of evidence
Geochemical analysis asks whether a cell's chemistry looks like known gold. Structural analysis asks whether it sits near the fault plumbing that gold moves through. Satellite analysis asks whether the surface looks altered. Terrain joins them, and geophysics when a client has it.
No line votes alone. They are blended with weights that live in configuration, not code, and millions of cells are scored in seconds for nothing — so money is only ever spent on the shortlist.
Then the verdict is written
Every target gets a written verdict that cites the document behind it — a target backed by a 1985 trenching report is not the same as a bare anomaly.
There are three verdicts and no more: confirmed, partial, insufficient. When the evidence is thin, the engine refuses to issue one. It does not guess.
What goes in, and what gets read
The data room arrives as it is — tens of gigabytes with no prior order: laboratory tables, scanned maps, handwritten reports from the 1980s, core-box photographs, GIS files and geophysical surveys.
Text is extracted from the scans, images are understood, tables are cleaned of unit errors, and coordinates are projected into the zone they belong to. Then millions of cells are scored in seconds, for nothing.
Figures from the engine's own database — not estimates.
Three verdicts, no more
Every target gets one of three verdicts, decided by a gate in the code rather than a human opinion.
- ConfirmedThe evidence crosses and suffices; the target is ready for a first-pass drill programme.
- PartialA real signal missing one line of evidence — and we say exactly which one.
- InsufficientNot enough for a verdict. It is not guessed: it is printed as Pending Verification.
What you receive
Ranked drill targets, coordinates your field team can carry to the site, bilingual reports, interactive maps and GIS files.
And a coverage receipt: exactly what we used of your data, and what we could not, and why. Not a consultant's report that hides its limits.
Drilling is the most expensive thing in exploration. Every hole not spent in the wrong place is a hole spent once, in the right one.