Skip to content
A plant control room with analysis screens

Product

Twelve modules. One question each. All of it on your hardware.

This is the capability map: what is actually in the platform, module by module — each described by the question it answers, each linked to the page that proves it. Real screens throughout; statuses stated honestly.

Screenshot from the product — demonstration data
Product screen: the site view with KPIs, fleet health and at-risk assets
the capability map

What’s in the product

Concept level on purpose: the modules you operate, not the internals we protect. Everything here is in the product today.

Site monitoring

The whole site at a glance — KPIs against sector targets, fleet health, upcoming predictions, at‑risk assets and an action queue.

Command centerAlertsFleet intelligenceAction queue

Asset intelligence

One machine, all its answers: health and forecast, anomaly fingerprint, time to failure, sensor trends, and the asset’s own event history.

Asset overviewHealth & forecastAnomaly fingerprintsTime‑to‑failureHistorical patterns

Root cause

From symptom to cause: feature contributions, matched past patterns, a library of known failure modes with symptoms and fixes.

RCA overviewCorrelationsFailure patternsCause library

Maintenance intelligence

Turning predictions into work: schedules, work orders, parts prediction and cost views — the write‑back is a draft, after a human approves it.

ScheduleWork ordersParts predictionCost

Comparative analytics

Assets side by side: fleet benchmarks, overlays and rankings that show which units are the outliers and why.

Side‑by‑sideFleet benchmarkInsights

Reasoning & verification

Every conclusion carries its chain — and four ways to challenge it: evidence trail, root‑cause map, physics check, what‑if. Confirm or override; it learns from the verdict.

Reasoning chainChallenge & verifyConfirm / override

Ryekronix — the agent

A plain‑English question in; a plan, evidence, validation and a gated proposal out. Autonomous in thinking, human‑gated in acting.

ConsoleObjectivesApprovals

What‑if twin

A virtual copy you can question: move operating variables, read the sensitivity, and see whether physics and the model agree before touching the real machine.

Twin simulatorSensitivityPhysics agreement

Physics laboratory

First‑principles models per equipment family; predictions are held against the governing equations, and contradictions are flagged, never smoothed over.

Reference equationsEquipment physics models

Experiments & hypotheses

Hypotheses synthesised from live signals; experiments run on the virtual copy; every result keeps its hypothesis, run and measured outcome.

Hypothesis labExperiment galleryDecision optimiser

Industry signature views

Sector‑specific screens — a mine gets slope radar, blast compliance and tailings views; a refinery gets its own. Domains, targets and frameworks per sector.

Industry overviewSignature pagesPlan views

Trust & audit

The register of certifications with honest statuses, a signed tamper‑evident audit trail, gated promotions — and the signed public benchmarks.

Certification registerGate logSigned benchmarks
two of the screens

The site view — and one machine’s trajectory

Real screens, demonstration data. The rest are spread across the pages this map links to.

Product screen: site view with KPIs against targets, fleet health, upcoming predictions and at-risk assets
What needs attention this week — the site view · Screenshot from the product — demonstration data.
Product screen: health history and forecast with contributing factors
Where one machine is heading — and what is holding it down · Screenshot from the product — demonstration data.
what the round builds here

What is not in the product yet

Labelled the way everything not yet shipped is labelled on this site. roadmap · not shipped

program 1

Cross‑site learning without moving data

today · measured
Each site learns from its own labels inside its own boundary.
next
Sites contribute signed, gated improvements without any operational data leaving.
proof point
The first real two‑site round, published with its evidence.
program 2

Edge serving next to the controller

today · measured
Answers come from on‑premises servers inside your network.
next
A hardened profile that serves next to the controller for the fastest loops.
proof point
Measured latencies on reference hardware, in the registry with statuses.

Funded by the growth round — what the round builds, all ten programs →

questions buyers ask

Four answers

Is this everything in the product?
It is the capability map at concept level — every module a buyer operates, each linked to the page that explains it. What we deliberately do not publish is internals: model architectures, training mechanics and module wiring stay out of public view.
Are these screens real?
Yes. Every screenshot on this site is captured from the platform running on demonstration data, and says so in its caption. Nothing is a mock‑up.
Which parts are roadmap?
Everything above is in the product today. The roadmap items — cross‑site learning without moving data, edge serving next to the controller, the third‑party certification audits — are labelled roadmap wherever they appear, including here.
Can we see it live?
Yes — a 45‑minute demo on your data or ours, with the reasoning trace open. No pilots and no free trials; verify us first if you prefer.
where next

From the map to the proof

A map you can check, screen by screen.

Every module above answers a question someone on the plant actually asks.

If a capability cannot be shown on a real screen, it is not on this page.

See the modules on your own data.

A 45-minute demo — one machine, one question, the trace open. No pilots, no trials.