Proposal Review

Use Decide to bring saved evidence and proposed limits together for a lot decision. Start with an analysis that contains the limits you want reviewed. Changing limits in Analyze does not create a proposal automatically.

If Loading proposal… appears when opening a proposal, wait for it to finish before editing.

Check that proposal changes were saved

Evidence changes, summary edits, and review-status changes require Create Proposal or Save Proposal. Sign Off does not save automatically. An older-build issue can leave saving disabled after an edit even though Approved appears on screen. The fix was verified on the hosted demo on 7 September 2026; other installations need their own rollout check. Reopen the saved version and verify its evidence, summary, status, and sign-off before distributing it. See the fixed builds and save-state precautions.

Attach evidence

1. Save the analysis or notebook you want to cite. A source is a particular saved version, not its unsaved working copy. See What-If Limits and Saved Work And Sharing.
2. Open Decide, choose New Proposal, and select the lot. Give the proposal a title and describe the recommendation in Proposal Summary.
3. Under Evidence Sources, choose the saved source and version. Check its snapshot, selected-die count, and limits, then choose Attach Source.
4. Inspect Limit Proposals. A source can provide evidence without providing limits; exact impact requires at least one extracted limit. If sources disagree about a test’s limits, remove or replace the conflicting source before resolving impact.
5. After adding, removing, or replacing evidence, use Create Proposal or Save Proposal, then reopen the saved version and check Evidence Sources. A disabled save button does not prove the change persisted; follow the caution above if a source or edit is missing.

The first source attached to a new proposal sets its snapshot. Subsequent sources must match it; saving the proposal also locks its lot and snapshot. Attaching a source does not make the proposal follow future source versions.

Compare the lot impact

Choose Resolve Exact Impact. The baseline is the pinned lot snapshot without the proposal’s merged limits. The proposed result uses that same snapshot with those limits applied.

This is a full-lot comparison, not a calculation restricted to the brushed dies in the attached analysis. Inspect baseline and proposed yield, rescued and newly failing dies, affected tests and wafers, and Changed Die Review.

The changed-die table is a bounded preview, not necessarily the complete list. Export Changed Dies CSV is intended to export the full changed-die set, but download completeness has not yet been verified in this alpha workflow. Check the file’s scope and distinct-die count against the exact impact totals before using it as decision evidence. Print / Save PDF creates a review report; neither action publishes limits or changes source measurements. Changing the proposal inputs can make cached impact stale. Resolve it again before relying on the result.

Record the review

Create Proposal creates the server version; Save Proposal saves later edits. Saving, review status, and lifecycle are separate:

Control	Meaning
Ready for Review	Moves the working proposal from Draft to In Review once the required evidence and current impact are available.
Sign Off	Adds the signed-in editor and time to the working proposal and shows Approved. It is not an automatic decision by the analysis engine.
Reopen	Returns the working proposal to Draft and clears its sign-off.
Promote to controlled, in the management menu	Permanently freezes a saved artifact. This is distinct from Approved.

Save after Ready for Review, Sign Off, or Reopen to record that state in a new version. An earlier saved version keeps its original status and sign-off. Controlled promotion has not yet been verified end to end in this alpha workflow. Do not promote an artifact just to test saving; the controlled state is intended to be irreversible.

Review status does not replace your organization’s release process. It does not publish specification overrides, rewrite tester results, or release a lot to manufacturing. Follow the save verification steps above; save important work to the server rather than relying on browser recovery.