fix: localise update reasoning state

.claude/ux-guidelines.md

@@ -197,18 +197,35 @@ The workspace is interactive.
 
 ---
 
-### Reasoning Mode
+### Reasoning Mode (initial analysis)
 
-The engine is analysing the available evidence.
+The engine is constructing the first investigation from nothing.
+
+A full primary loading state appears:
+
+- prominent overlay with spinner, rotating status messages, elapsed timer;
+- the entire workspace is replaced until reasoning completes;
+- no partial or changing content is visible during processing.
+
+---
+
+### Reasoning Mode (subsequent answers — localised)
+
+The investigation already exists.
+
+Only the active response panel is replaced by the loading card:
+
+- Current investigation question remains visible for context;
+- Current understanding, Original situation, and Investigation history persist;
+- Terminal state cards are suppressed during loading;
+- The workspace layout remains stable and recognisable;
+- Recovery states appear in place of the loading card if reasoning fails.
 
 The interface should:
 
-- clearly indicate that reasoning is in progress
-- temporarily suspend the workspace
-- reassure the user that their input has been accepted
-- avoid displaying partial or changing reasoning
-
-The workspace is paused until reasoning completes.
+- clearly indicate that reasoning is in progress via the response-panel overlay;
+- reassure the user that their answer has been accepted;
+- avoid displaying partial or changing reasoning outside the response panel.
 
 ---
 
@@ -218,37 +235,10 @@ Every submission follows the same lifecycle:
 
 User submits information
 ↓
-Workspace pauses
+Loading card appears (full-page for initial analysis, localised for updates)
 ↓
-Reasoning mode
-↓
-Updated workspace appears
+Updated workspace returns
 
-The interaction should be identical whether the submission is:
-
-- the initial situation
-- an investigation answer
-- a future uploaded document
-- any other evidence
+The interaction is consistent in intent — both modes confirm input acceptance and pause the active response area — but the page-level behaviour differs because one constructs from nothing while the other refines existing context.
 
 Users should never wonder whether their input has been accepted or whether the engine is still reasoning.
-
----
-
-## Consistent Submission Lifecycle
-
-Every submission follows the same lifecycle regardless of context.
-
-1. User submits → workspace pauses immediately, stale content disappears
-2. Reasoning mode appears with loading overlay (spinner, rotating status messages, elapsed timer)
-3. Updated workspace returns or error state appears
-
-The loading card must appear before any async request begins and remain visible until the result arrives. 
-
-During reasoning mode:
-- The workspace is fully hidden (not just disabled)
-- No stale question, response form, or active indicators are visible
-- The Investigation Summary Panel shows "Reasoning" rather than "Investigation in progress"
-- Recovery states render on top of the loading overlay when applicable
-
-This applies equally to initial analysis and all update submissions.