HOW IT WORKS

A clear path from model to result.

Keep the editable project, the measured run and the review decision connected. Start with a small saved example.

ProjectSimulationMeasured resultReview & keep
  1. 01 / OPEN & EDIT

    Work from the actual sources.

    Open a native project. Inspect the mechanism, make a supported change and save your writable copy.

    Create your first project →
  2. 02 / RUN & INSPECT

    Measure what happened.

    Run a saved setup, play the recording and inspect available channels with their engineering units. A completed process is not automatically a successful engineering result.

    Read a simulation result →
  3. 03 / REVIEW & APPLY

    Make the decision explicit.

    Compare proposed changes before Apply. Preserve the project’s source identity and the evidence behind your decision.

    Review a candidate →
  4. 04 / KEEP A WAY BACK

    Recover without starting over.

    Return to an engineering version while retaining newer conversations and drafts. Keep your original project before attempting recovery.

    Preserve and recover work →

Try the complete loop.

The supplied slide example includes a native project, controller and inspectable recording. It is one fixture, not a performance guarantee for other mechanisms.

ENGINEERING, NOT JUST A PICTURE

A gantry.
Six parts.
A checkable result.

Open the editable cell, follow its controller and inspect measured transfer and fault evidence. Built with the private Workcell Agent; shared as an MIT-licensed example.

Public Client AI and checkout are not open. This example does not enable them.

All six measured part-centre errors are below the example’s 5 mm limit.
Measured native placements, not a product-wide accuracy claim. Explore the source and fault test.