Skip to content
Ryedore Logo
Theme
ContactRequest Demo
A physics digital twin of industrial equipment

Specialists Lab · test before you commit

Six AI specialists run a controlled trial on a virtual copy of your machine — before you spend on the change.

Say “test whether a ceramic bearing on the CNC spindle reduces vibration.”

The Lab designs the experiment, runs baseline vs intervention on a physics twin, does the statistics, and hands you an auditable report. It never touches the plant.

How an experiment runsBring a question to a demo
6 specialists
Specialists per experiment
attested
100% on‑premises
Where it runs
attested
one experiment, end to end

Baseline versus intervention on a virtual copy — then real statistics

Hundreds of virtual trials in minutes, constraint violations flagged, the physics tier reported, and a report you can audit.

“test a ceramicbearing on the CNC”1·2understand& design3 · PHYSICS DIGITAL TWINbaselinesteel bearinginterventionceramic bearinghundreds of virtual trialsconstraint violations flaggedtier: physics → causal → heuristic4optimisewithin limits5analysestatistics6reportread‑only: the Lab simulates and advises — it never touches the plant · report = p‑value · effect size · confidence interval · provenance
Baseline versus intervention on a virtual copy of the machine — then real statistics, not a hunch.

swipe → to see the whole diagram

Engineers beside a robotic production line
six roles

A team, not a model

Each specialist does one job well and hands off to the next — understand, design, simulate, optimise, analyse, report.

You can see which specialist produced every number.

Product screen: the physics laboratory with first-principles models for eight equipment types
The physics specialists — one model per equipment family · Screenshot from the product — demonstration data.
Product screen: the experiment results gallery, each experiment keeping its hypothesis, run and measured result
Experiments keep their receipts · Screenshot from the product — demonstration data.
six roles

The six specialists

1/6
The Understander

Turns your plain‑English request into a precise experiment — asset, change, outcome — enriched from the knowledge base.

2/6
The Planner

Designs the methodology: phases, controls, duration, number of simulated runs.

3/6
The Simulator

Runs baseline vs intervention on a physics twin — and reports which physics tier it used.

4/6
The Optimiser

Finds the operating point that best achieves the goal within the machine’s limits.

5/6
The Analyst

Significance, effect size, a confidence interval, and which factors drove the result.

6/6
The Reporter

A plain‑language report with a provenance chip on every result.

Honest about the physics.

The Simulator uses a physics engine first, a causal model if the physics profile is incomplete, and an analytic estimate last — and tells you which.

We only cite statistics from physics‑tier runs.

The worked example below is in the format every Lab report takes — and it is labelled illustrative until customers attribute results.

a report, drawn

What comes back: two distributions and an interval

The worked example above as a picture — baseline versus intervention, the confidence interval, and the statistics chips every report carries.

Baseline · steel200 runs100Ceramic intervention200 runs8695 % CI −11…−17 %vibration −14 %95 % CI −11 to −17p < 0.001effect size 0.62tier: physicsvibration index, lower is better · the Lab never touches the plant — this ran on the physics twin
What a Lab report looks like as a picture: two distributions, an interval — not a hunch. (Worked scenario, illustrative.)

swipe → to see the whole diagram

how it relates to Ryekronix

One agent that acts. One team that experiments.

Ryekronix answers and acts — with a human’s approval. The Specialists Lab tests before you commit — and never touches the plant. Ryekronix can commission the Lab; the Lab reports back.

RYEKRONIX · ANSWERS & ACTS · HUMAN‑APPROVED1plan2execute3observe4revise5answer→ approvalSPECIALISTS LAB · TESTS BEFORE YOU COMMIT · NEVER ACTS1understand2design3simulate4analyse5report→ p · effect · CIcommissionsreports back
Ryekronix answers and acts — with a human’s approval. The Specialists Lab tests before you commit — and never touches the plant.
Meet Ryekronix →

swipe → to see the whole diagram

questions people ask

Four answers

Is this a digital twin?
It runs on one. A virtual copy of the machine, built from its physics, is where the Lab runs baseline versus intervention — hundreds of trials in minutes, with any constraint violation flagged.
Does the Lab ever change a setting on the plant?
Never. It is read‑only by design: it simulates and advises. Any real action goes through Ryekronix, six checks and a human approval.
What exactly comes back?
A plain‑language report: the effect with a confidence interval, its significance and effect size, which factors drove the result, and a provenance chip on every number showing which specialist produced it and which physics tier was used.
How is this different from a what‑if question?
Three depths, chosen by the stakes. Quick What‑if inside Ryekronix ranks set‑point moves in seconds. The What‑if Lab solves the physics for a checked range. The Specialists Lab runs a controlled experiment with statistics and provenance — for decisions with a budget attached.
where next

The Lab in the wider platform

what the round builds here

Where the Lab goes next

One program: the Lab’s experiments become the interventions that let causes be discovered rather than inferred. roadmap · not shipped

program 5

Experiments as interventions — causal discovery at scale

today · measured
Six specialists run baseline versus intervention on a physics twin and report statistics with provenance; the physics tier used is always disclosed.
next
Every Lab run and every approved action becomes an intervention for causal discovery, so cause chains are learned from what was actually changed — not inferred from correlation; counterfactual remaining life (‘had we re‑aligned in March…’) with intervals.
proof point
Cause‑chain precision measured against confirmed root causes on the label flywheel; counterfactual intervals signed on public rigs.

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

Have a change you’re weighing?

Bring the question to a demo and we’ll run it through the Lab.

Request a demoMeet Ryekronix