LynIP: Invention to Production

1. A brief becomes requirements with IDs and acceptance criteria.
2. A CAD kernel builds the part; the viewer reads a GLB tessellated server-side.
3. A slicer runs on that solid — print time and filament are measured, not guessed.
4. The BOM rolls up at 1, 10 and 500 units, so prototype and production economics diverge visibly.
5. Requirements meet evidence on one design revision at a time; the gate names every blocker.
6. A fix is a new revision with its own test — and only that test can open the production gate.

Why a failed test beats four analyses

GripLite's fingertip has four analyses saying it is stiff enough and one bench test saying it deflects 0.71 mm at 40 N. The refutation wins outright, and the production gate stays shut on that revision. The fix is a new revision with its own bench test — the failure stays on the record of the part it measured, because the point is that evidence changed the design, not that the failure went away.