One engine for the whole life
of a mineral asset.
Exploration to closure — one engine for the whole life of a mine: model the ground, plan the drilling, and build the resource, then plan, manage and operate the mine through to closure — while keeping your data, tenements, budgets, procurement and compliance in order. Every result flows into a single audit trail, carries a cryptographic signature, and is signed off by a human Competent Person.
(21 mapped)
(14 on-platform today)
fails included
results
replayable
One engine across the whole lifecycle — the same stages you work through in the product
A reasoning swarm, not a single black box.
Twelve production agents live today, twenty-one mapped across the lifecycle. Each agent owns one job — from data intake and QC through resource estimation, permitting, mine development, operations, and closure. They pass evidence forward, argue over the posterior, and hand the Competent Person a decision-ready result with every assumption on record.
Built so the
signature survives audit.
Every result is cryptographically signed and replayable.
Results are signed against production key prod-54f50a4053f3 with a public JWKS endpoint, written to an append-only ledger, and reproducible bit-for-bit. When a Competent Person stakes their name on a target, the provenance is the proof.
Ranked drill targets
Each target carries a probability of intersection plus a depth-to-top estimate with a confidence interval, ordered for the drill plan, not the slide deck.
Posteriors, not point estimates
Joint inversion across complementary physics — gravity, magnetics, seismic velocity, electromagnetics and magnetotellurics, and geochemistry coupled to geophysics — returns full distributions, so risk is quantified instead of hidden in a single deterministic answer.
A tool that defends a QP report, not one that replaces judgement.
Tolun produces the evidence chain a Competent Person needs to sign off under code: traceable inputs, documented assumptions, validation against known geology, and a tamper-evident record of exactly how every number was produced.
Watch it work.
One full run, start to finish, on our standing demo scenario: the Harrat Rahat volcanic field in Saudi Arabia, modelled entirely from public survey data. The engine loads the data, models the ground, ranks the targets and proposes the holes — then records the whole chain of custody and queues the result for a human Competent Person to sign off.
Prefer real studies to a demo scenario? The public showcase →
Blind studies,
fails included.
~24 blind, pre-registered validation studies on fully public data in about three months — every verdict recorded as it came out, fails alongside passes. The three below are cleared by our Competent Person and published in full; the rest stay internal until each one is signed off.
Verify any result without trusting us.
The signature, the public key, and the ledger entry are all independently checkable. The drill decision rests on math and cryptography, not on a vendor's word.
signing_key_idStart with a pilot.
Scale on results.
A scoped run on one project area: full joint inversion, ranked targets, and a signed, replayable result package your QP can audit.
Scope a pilotOngoing discovery support across multiple targets and survey types, with provenance retained for every deliverable across the campaign.
Talk to us →On-prem deployment, custom forward models, and integration with your existing exploration data estate. Pricing on a per-deployment basis.
Contact sales
Put a signature on your next drill decision.
Questions or feedback? Email us at founders@tolun.ai.