Terrain intelligence
A seeded LiDAR point cloud rendered in an isometric view you can rotate, zoom and drill into, with the sector, the point count and the provenance label carried in the panel header rather than implied by the picture.
ClearGlass Inc. › Products › CLEARMINE
ClearGlass // Mine Site Operating Picture
A working mine generates more signal than anyone can hold: LiDAR and magnetics from the survey team, torque and RPM off the rig, gas readings from the sensor mesh, assay certificates from the lab, drone links that drop, and a model somewhere that claims to know where the ore continues. CLEARMINE puts those streams into one operating picture — terrain, drill telemetry, geochemistry, predictive models, alerts, equipment risk and scenario economics — and holds each of them to the same standard: say where the number came from, say how uncertain it is, and refuse anything you cannot actually do.
That last part is what makes this surface different from a demo. The workspace below runs a deterministic simulator, and it never pretends otherwise. Selecting a live data source returns an explicit refusal instead of a fake connection. Every equipment command resolves to a dry-run preview with its preconditions, its simulated impact and its rollback path. A grade carries its uncertainty interval and its detection limit. A model result carries its inputs, its validation score and the sentence this is a model inference, not a reserve estimate.
Coverage
Seven surfaces over one simulated site, each carrying its own provenance line rather than borrowing credibility from the panel next to it.
A seeded LiDAR point cloud rendered in an isometric view you can rotate, zoom and drill into, with the sector, the point count and the provenance label carried in the panel header rather than implied by the picture.
Torque against an EWMA envelope, RPM stability, penetration energy and a methane channel — each with its sensor freshness, its source identifier and a quality flag. The gas readout is explicitly marked as a user interface, not a safety instrumented system.
Simulated ICP-MS batches with sample counts, chain-of-custody state, detection limits and a stated uncertainty on every grade — 1.24 ± 0.12 ppm rather than a bare number with implied precision.
Each inference opens its card: version, inputs, pipeline stages, inference time, holdout F1, calibrated confidence and the timestamp it was validated at — followed by the line that a target extension probability is a model output and not a reserve estimate.
Severity-filtered anomalies with the threshold that fired them, plus a 72-hour equipment risk matrix naming the component, the observed deviation and the recommended inspection — an advisory list, not an autonomous maintenance actuator.
Volume, bulk density, a lognormal grade distribution, recovery and cut-off are shown as editable inputs, so the Monte Carlo output is traceable to assumptions you can argue with instead of to a black box.
Commands are disabled by default and require confirmation and authorisation. Each resolves to a preview stating target, impact, preconditions and rollback, and the queue states its own lifecycle: queued → authorized → dispatched → acknowledged → completed.
Governed by construction
Most industrial dashboards are generous with certainty and quiet about limits. A pulsing green dot implies a connection that does not exist; a grade appears to four decimal places with no uncertainty; a "dispatch" button suggests an authority the page does not hold. Each of those is a small fabrication that becomes a large one the moment somebody spends money on it.
CLEARMINE inverts the default. The workspace holds three refusals in code, and states each of them on screen where the action is attempted:
This is the same operating rule the rest of the ClearGlass estate runs on — read-only analysis, then a draft, then a human decision, then execution — applied to an industrial surface rather than a commercial one. Where the ORELOGIC exploration workspace seals its exports into an RFED hash chain, CLEARMINE holds the line one step earlier: it declines to produce the action at all until the preconditions for it are real.
The workspace is deterministic on purpose. It runs a seeded generator (mulberry32, seed 42) across seven streams, so two people looking at the same panel are looking at the same numbers, and a screenshot can be reproduced rather than argued about.
Provenance and limits
Seven streams, one fixed seed, zero sensors. No LiDAR unit, drill rig, gas mesh, drone link or laboratory is connected to this page, and the status bar reads SIMULATED rather than showing a synthetic sync time.
The workspace makes no network call carrying site data. The WebSocket layer is an abstraction running against the simulator, not an open socket, and the footer names it as such.
The methane channel is a user-interface readout. It is not certified gas detection and does not substitute for any life-safety control on a mine site. The panel carries that sentence beside the reading, not in a footnote.
The workspace bundle and its stylesheet load under Subresource Integrity, so a modified asset fails to execute rather than running unnoticed. The published digests are recorded in data/clearmine/manifest.json.
Scenario output is a probabilistic estimate from stated assumptions. A resource estimate intended for disclosure must be prepared and signed by a Qualified Person under the applicable code — NI 43-101 in Canada, the JORC Code in Australia, or the local equivalent.
A deployment connects your own survey, telemetry and laboratory systems behind an administrator-scoped connector profile, moves command authorisation server-side under the governed approval gate, and scopes access per site and per role. Until those exist, the answer stays SIMULATED.
Questions
No. Every stream in the public workspace is generated in your browser by a seeded pseudo-random generator (mulberry32, seed 42). There is no sensor, LiDAR unit, drill rig, laboratory or telemetry link behind it. The status bar reads SIMULATED rather than showing a synthetic sync time.
No. Every equipment command in the public workspace is disabled by default and resolves to a dry-run preview stating the target, the simulated impact, the preconditions and the rollback path. Nothing is dispatched to hardware, and the hardware adapter feature flag is off.
Because the preconditions for a live source are not met in a public page. Selecting LIVE returns an explicit refusal: a live source requires administrator RBAC, a signed connector profile and a verified mTLS endpoint. The workspace stays in SIMULATED mode rather than pretending the switch succeeded.
No. The scenario tool produces a probabilistic estimate from assumptions you can see and edit — volume, density, grade distribution, recovery and cut-off — and labels the output as not a mineral resource statement. A disclosable resource estimate must be prepared and signed by a Qualified Person under the applicable code, such as NI 43-101 in Canada or the JORC Code in Australia.
No, and it says so on the panel. The gas channel is a user-interface readout, not a safety instrumented system. Nothing in CLEARMINE substitutes for certified gas detection, ground-control monitoring or any other life-safety control on a mine site.
Three points on the same chain. CLEARMINE is the site surface — what the ground, the rig and the sensors are doing right now. ORELOGIC is the programme surface — holes, assays, resource model and budget for an exploration campaign. Minerals Intelligence is the market surface — supply concentration, trade, policy and risk across minerals and countries.