8.6.61 - Finish This Report is now a true state-driven owner workflow. Only the active step's controls are shown; AI guidance is embedded in Step 1, AI results are revealed on demand, Owner/Management Comment is part of Step 2, and approval/finalize/send actions appear only at their stages.
8.6.59 - 2026-10-01
- Converted Finish This Report from a shortcut dashboard into a true staged owner workflow. Only the actions for the current step are shown.
- Step 1 Review + AI shows AI/review/return actions; a successful AI run advances to Step 2 automatically.
- Step 2 Apply Results shows Summary, Signoff Package, optional Owner Comment, and requires Summary + Signoff Package before Continue to Approval is enabled.
- Step 3 shows only Approve / Return actions; Step 4 only Finalize; Step 5 only recipient review/send.
- AI Policy & Training Review is the Step 1 detail area and collapses after AI completes; Owner / Management Comment is shown with Step 2 instead of competing with every stage.
- Saving an already-applied draft can resume at Step 3 by detecting the saved Summary and report-specific signoff package, preserving workflow continuity across reloads.
- Re-running AI resets the Apply Results completion state so newly generated recommendations must be deliberately applied again.

8.6.58 - 2026-10-01
- Finish This Report now spans the full report width instead of occupying one metadata-grid column.
- Removed the visible duplicate Owner / Management Review block below the AI area; its existing forms remain hidden as the workflow backend.
- Return for More Work now opens an inline request box directly inside Finish This Report and submits through the existing return-for-edits logic.
- Unsaved-change approval warning now appears in Finish This Report and disables its Approve button until the draft is saved.
- Kept the detailed AI review and editable Owner / Management Comment area because they contain review content, not duplicate workflow navigation.

8.6.57 - 2026-10-01
- Enforced a hard employee-specific AI highlighting rule: only the selected Signoff Employee may receive AI bold/color highlighting in the factual report.
- Dedicated highlight recovery now returns the smallest exact selected-employee quote instead of whole timeline lines, preventing another employee on the same line from being colored.
- Server-side validation rejects any AI highlight quote containing another employee's name in an employee-specific report.

8.6.56 - 2026-09-25
- Added a single Owner "Finish This Report" workflow panel that keeps review/AI, applying results, approval, finalization, and sending in one predictable place.
- Added proxy actions for AI run/re-run, Copy Summary, Apply Signoff Package, optional Owner Comment, Approve, Return for More Work, Finalize, and Send without changing the underlying safeguards or report-state logic.
- Owner approval/finalization redirects now return to the Finish This Report area instead of dropping the owner elsewhere on the page.
- Existing detailed AI, return-for-edits, recipient, approval, finalization, and send controls remain in place as the authoritative controls; the new panel orchestrates them rather than bypassing them.

8.6.54 - 2026-09-25
- Fixed AI timeline highlighting being silently discarded when the AI quote differed from the editor only by whitespace/line-break boundaries.
- Server now resolves each AI finding back to the exact canonical observation text before returning it to the browser.
- AI completion status now reports both findings identified and passages actually highlighted, so a highlight failure is visible instead of silent.
- Owner Comments continue to be included on rerun and the in-progress status says so when comments exist.

8.6.52 - 2026-09-25
- Before applying AI findings, Bloodhound now removes legacy/manual color and bold emphasis from the observation portion of the report, then applies only the AI-determined problem highlights for the selected Signoff Employee.
- Existing Summary formatting is left alone until the owner chooses Copy Summary to Report; observation normalization does not use Lissette's prior highlighting as evidence.
- Selected-employee highlight color is now determined server-side from the saved Signoff Employee, with a fallback for older/copied subject lines that lost the comma between employee/role entries.
- The in-progress AI status now explicitly says Owner Comments are being considered whenever comments are present.
- Owner Comments continue to be sent to the AI evaluation on every run/re-run; checked standing guidance remains available for future relevant reports.

8.6.51 - 2026-09-25
- AI evaluation now explicitly ignores all pre-existing report colors/highlighting when deciding what conduct violates handbook or training/standard work procedures.
- Employee-specific AI review is strictly limited to the selected Signoff Employee in summary, highlights, policy references, warning, and management review.
- Successful AI evaluation automatically applies AI-generated highlighting to exact problem passages in the original report using the selected employee's AI color.
- Re-running AI removes only prior AI-generated highlight wrappers before applying the new findings; manual report formatting is otherwise preserved.
- Highlight application now handles findings spanning multiple text nodes/formatting spans more reliably.
- Proposed AI summary and warning use the same selected-employee color as the AI-applied report highlights.

8.6.50 - Signoff-employee policy precision, matching AI color, hoodie policy selection, and owner-approved future AI guidance.
8.6.49 - Signoff-Employee-Only AI Evaluation
- AI Summary, highlights, policies, signoff review, warning, management suggestion, prior history and follow-up are restricted to the selected Signoff Employee.
- Server-side filtering removes standalone summary bullets/findings about other employees even if AI returns them.
- Shared timeline findings are normalized to the selected employee's color; other-employee-only highlights are rejected.
- Summary HTML is rebuilt from the employee-specific bullet set to prevent another employee's findings from leaking into the copied Summary.
- Supporting policy section is labeled as policies applied to the selected Signoff Employee.