Review Problem Resolution
Changes on October 6, 2026
Complete change list
- Page and detail tabs now show the active view with readable foreground text, semibold weight and a cyan underline. Story editor and Changelog views use the same treatment.
- Story editor source linking now offers title/URL search when more than 50 unlinked sources exist and shows up to 50 matching choices. The chosen source remains selected when the search changes.
- Workspace lists, tables and boards now use the same Light/Dark canvas palette as Delivery: blue from the lower left and red from the upper right. Saved earlier standard palettes adopt the new defaults; customized combinations, disabled backgrounds and card/detail appearance remain intact.
- Standard detail headers keep long titles and the Brief/Technical group aligned beside the first title line. On narrow detail columns, the tab group moves intact below the title; menu controls retain their own space.
- Shared source and role menus appear above the containing popover and accept pointer selection.
- KPI sparklines and trend indicators now share green for favorable trends, yellow for neutral trends and red for unfavorable trends. Previously, unfavorable trends were yellow and neutral sparklines were blue. Compact labels remain
text-xs, smaller than the page tabs; headline values are unchanged. - Story editor view selection now uses transparent shared controls with a cyan active underline and visible keyboard focus, while Publish remains the primary action. Navigation, save actions, section headings and synopsis counts wrap at narrow widths; preview cards have opaque themed surfaces and wrapping titles.
- In the Story editor, artifact titles and the empty artifact action support keyboard activation, with external URLs retained as separate links. Source linking uses labelled dropdowns with the existing five roles and disables edits and repeat submission while linking. Read-only access and draft/publish data remain unchanged.
- Product attention replaces the long Drift Alerts list with up to five capabilities whose numeric Product Quality is below 80, ordered by lowest score and linked to Product. Unknown quality is reported separately. Larger portfolios load through bounded batches; a failed request shows an unavailable state and a Product link. Missing/deleted IDs preserve other available scores with incomplete-result metadata.
- The agent switcher and conversation avatars now use the same approved Light/Dark photographs as Crew for Nova, Polaris and Vega. Existing agent selection remains unchanged. Implemented with verified staging and production portrait assets.
- Read-only inline titles remain selectable text. Failed keyboard saves retain the draft without stealing focus after a later blur-save.
- Crew and member activity logs now group consecutive entries by actor and local calendar day, with a small theme-specific avatar beside each group. Event times, action types and chronological order remain visible; Nexus legacy names share the same identity and adjacent groups join across loaded pages.
- Crew now shows Nova, Polaris, Vega and Nexus in that order, placing the three human agents first. Nexus cards, member details and activity avatars use the user-approved mechanical portrait with dedicated Light and Dark backgrounds. Implemented with verified staging and production portrait assets.
- Command Control now keeps its header, signals and Crew filters visible while the content scrolls within one viewport. The themed canvas continues behind the content and scrollbar instead of ending partway down long activity history.
- Bridge now previews active competitor risks, placing triggered risks first and then sorting by severity. Up to five show their trigger, recorded response and a link to the competitor's risk workspace. Retired risks are excluded; loading and failures are distinct from an empty register.
- Agent activity has moved from Bridge to Crew and is also available in each member profile with Load more and Retry. Member histories are filtered before pagination; Nexus includes legacy System AI names.
- Bridge Crew Dark portraits for Nova, Polaris and Vega now use neutral studio lighting and quiet charcoal backgrounds without blue/red light cast on the subjects. Their Light portraits retain the soft backdrop gradients. Nexus now uses the user-approved mechanical portrait with dedicated Light/Dark backgrounds, superseding the temporary restoration of the original artwork after earlier generated replacements failed review. Cards, details and roster thumbnails use these final selections. Implemented with verified staging and production portrait assets.
- Competitor Risks is available in Configure Signals; new Bridge preferences default to risks, strategy progress, evidence coverage and capability health. Existing saved selections are preserved. The optional activity count is labelled Loaded Log Entries to clarify its scope.
- Command Center no longer offers the placeholder Deploy Member action or static Autonomy Model explanation. Crew cards and profiles no longer show seeded workload counters or a fabricated Pending Approvals total; actual capability proposal review remains available when proposals exist.
- Crew members appear once: featured portraits above, additional agents below. Cards use compact captions with name, role and status; profiles show a compact portrait, description and guiding question instead of oversized hero, repeated metrics and self-reflection panels. Search, status filters and profile links remain available in Light and Dark.
- Feedback items no longer hang indefinitely on "Extracting Evidence…" when analyzing new customer signals.
- Evidence extraction divides qualitative feedback into discrete, bite-sized evidence items with individual type icons and resolution checkboxes, preserving authentic customer quotes without artificial AI prefixes.
- Feedback Intelligence: Fixed an issue where running Nexus classification returned "TypeSafe API error (status 400)" due to invalid question primitive definitions and unparsed continuous score values.
- Feedback Intelligence: Fixed an issue where running Nexus classification on feedback items returned "Feedback not found or access denied" due to an internal column name mismatch.
- Feedback Intelligence: Fixed an issue where running Nexus classification on feedback items returned "TypeSafe API error (status 404)" due to an outdated API endpoint URL.
- Feedback Nexus Intelligence Rendering: Fixed an issue where clicking "Classify with Nexus" in the feedback detail view saved the classification to the server but did not immediately render the intelligence card in the UI.
- Nexus Intelligence Channel & Emotion Detection: Feedback classification now automatically identifies the raw qualitative feedback channel (User Research Interview, Customer Support Ticket, Sales Demo Call, Community/Social, In-App Widget, Agent Autonomous Audit) and accurately detects nuanced customer emotion (Delighted, Neutral, Confused, Frustrated/Angry, Mixed) with 1-click sync buttons to immediately update the parent feedback record's source and sentiment.
- Frustration Intensity Meter: A discrete 4-level customer distress indicator (Calm, Mild friction, Moderate frustration, Severe distress/Anger) highlights high-impact customer pain points with standard accessible progressbar indicators.
- Deep Transcript & Multi-Chunk Analysis: Qualitative transcripts (such as 30–60 minute user research interviews and sales calls) now evaluate evidence quotes across the entire conversation timeline by combining up to 10 child evidence chunks alongside an expanded 3,500-character raw text window, eliminating blunt head-truncation.
- Nexus Intelligence for Feedback: Introduced TypeSafe Jev AI analysis in Feedback Detail under the Technical tab, classifying feedback origin archetype, architectural target layer, operational friction pattern, candidate domain cluster, engineering effort, and agent-readiness blocker probability with on-demand re-classification.
- Problem resolution verdicts and closure notes are persisted in the database and automatically flagged as stale when linked changes, decisions, evidence links, or initiative links shift.
- The Signal Inbox displays resolved problem and evidence counters on feedback cards and problem review sections, surfacing customer friction resolution directly in the intake queue.
- Initiatives in both the slide-over detail drawer and Decision Brief display problem fulfillment indicators against delivered changes and decisions, and provide a one-click audit action to verify open problems in batch.
- Problem scope and customer evidence resolution verification now evaluates attributable Initiative Decisions alongside shipped Changes, recognizing deliberate strategic resolutions.
- Viewing and managing comments on Problems, Feedback, and Themes no longer triggers an unexpected application error boundary crash during page navigation or component remounts. Realtime comment updates continue to refresh automatically without channel collisions.