Update readmes and imporve UI

services/alert_processor/README.md

@@ -29,14 +29,17 @@ Services → RabbitMQ → [Alert Processor] → PostgreSQL
 
 ### Enrichment Pipeline
 
-1. **Message Generator**: Creates i18n keys and parameters from metadata
-2. **Orchestrator Client**: Queries AI orchestrator for context
-3. **Business Impact Analyzer**: Calculates financial and operational impact
-4. **Urgency Analyzer**: Assesses time sensitivity and deadlines
-5. **User Agency Analyzer**: Determines user's ability to act
-6. **Priority Scorer**: Calculates weighted priority score (0-100)
-7. **Smart Action Generator**: Creates contextual action buttons
-8. **Entity Link Extractor**: Maps metadata to entity references
+1. **Duplicate Detection**: Checks for duplicate alerts within 24-hour window
+2. **Message Generator**: Creates i18n keys and parameters from metadata
+3. **Orchestrator Client**: Queries AI orchestrator for context
+4. **AI Reasoning Extractor**: Extracts AI reasoning details and confidence scores
+5. **Business Impact Analyzer**: Calculates financial and operational impact
+6. **Urgency Analyzer**: Assesses time sensitivity and deadlines
+7. **User Agency Analyzer**: Determines user's ability to act
+8. **Priority Scorer**: Calculates weighted priority score (0-100)
+9. **Type Classifier**: Determines if action needed or issue prevented
+10. **Smart Action Generator**: Creates contextual action buttons
+11. **Entity Link Extractor**: Maps metadata to entity references
 
 ## Service Structure
 
@@ -177,12 +180,15 @@ await publisher.publish_alert(
 
 ### 2. Alert Processor Enriches
 
+- **Checks for duplicates**: Searches 24-hour window for similar alerts
 - Generates i18n: `alerts.critical_stock_shortage.title` with params
 - Queries orchestrator for AI context
+- Extracts AI reasoning and confidence scores (if available)
 - Analyzes business impact: €197.50 financial impact
 - Assesses urgency: 12 hours until consequence
 - Determines user agency: Can create PO, requires supplier
 - Calculates priority: Score 78 → "important"
+- Classifies type: `action_needed` or `prevented_issue`
 - Generates smart actions: [Create PO, Call Supplier, Dismiss]
 - Extracts entity links: `{ingredient: "..."}`
 
@@ -192,8 +198,12 @@ await publisher.publish_alert(
 {
   "id": "...",
   "event_type": "critical_stock_shortage",
+  "event_domain": "inventory",
+  "severity": "urgent",
+  "type_class": "action_needed",
   "priority_score": 78,
   "priority_level": "important",
+  "confidence_score": 95,
   "i18n": {
     "title_key": "alerts.critical_stock_shortage.title",
     "title_params": {"ingredient_name": "Flour"},
@@ -206,7 +216,10 @@ await publisher.publish_alert(
   "business_impact": {...},
   "urgency": {...},
   "user_agency": {...},
-  "smart_actions": [...]
+  "ai_reasoning_details": {...},
+  "orchestrator_context": {...},
+  "smart_actions": [...],
+  "entity_links": {"ingredient": "..."}
 }
 ```
 
@@ -249,30 +262,83 @@ Publishes to Redis channel `alerts:{tenant_id}` for real-time frontend updates.
 
 See [app/utils/message_templates.py](app/utils/message_templates.py) for complete list.
 
-Key alert types:
-- `critical_stock_shortage`
-- `low_stock_warning`
-- `production_delay`
-- `equipment_failure`
-- `po_approval_needed`
-- `temperature_breach`
-- `delivery_overdue`
-- `expired_products`
+### Standard Alerts
+- `critical_stock_shortage` - Urgent stock shortages
+- `low_stock_warning` - Stock running low
+- `production_delay` - Production behind schedule
+- `equipment_failure` - Equipment issues
+- `po_approval_needed` - Purchase order approval required
+- `temperature_breach` - Temperature control violations
+- `delivery_overdue` - Late deliveries
+- `expired_products` - Product expiration warnings
+
+### AI Recommendations
+- `ai_yield_prediction` - AI-predicted production yields
+- `ai_safety_stock_optimization` - AI stock level recommendations
+- `ai_supplier_recommendation` - AI supplier suggestions
+- `ai_price_forecast` - AI price predictions
+- `ai_demand_forecast` - AI demand forecasts
+- `ai_business_rule` - AI-suggested business rules
 
 ## Database Schema
 
 **events table** with JSONB enrichment:
-- Core: `id`, `tenant_id`, `created_at`, `event_type`
+- Core: `id`, `tenant_id`, `created_at`, `event_type`, `event_domain`, `severity`
 - i18n: `i18n_title_key`, `i18n_title_params`, `i18n_message_key`, `i18n_message_params`
 - Priority: `priority_score` (0-100), `priority_level` (critical/important/standard/info)
 - Enrichment: `orchestrator_context`, `business_impact`, `urgency`, `user_agency` (JSONB)
+- AI Fields: `ai_reasoning_details`, `confidence_score`, `ai_reasoning_summary_key`, `ai_reasoning_summary_params`
+- Classification: `type_class` (action_needed/prevented_issue)
 - Actions: `smart_actions` (JSONB array)
+- Entities: `entity_links` (JSONB)
 - Status: `status` (active/acknowledged/resolved/dismissed)
+- Metadata: `raw_metadata` (JSONB)
+
+## Key Features
+
+### Duplicate Alert Detection
+
+The service automatically detects and prevents duplicate alerts:
+- **24-hour window**: Checks for similar alerts in the past 24 hours
+- **Smart matching**: Compares `tenant_id`, `event_type`, and key metadata fields
+- **Update strategy**: Updates existing alert instead of creating duplicates
+- **Metadata preservation**: Keeps enriched data while preventing alert fatigue
+
+### Type Classification
+
+Events are classified into two types:
+- **action_needed**: User action required (default for alerts)
+- **prevented_issue**: AI already handled the situation (for AI recommendations)
+
+This helps the frontend display appropriate UI and messaging.
+
+### AI Reasoning Integration
+
+When AI orchestrator has acted on an event:
+- Extracts complete reasoning data structure
+- Stores confidence scores (0-100)
+- Generates i18n-friendly reasoning summaries
+- Links to orchestrator context for full details
+
+### Notification Service Integration
+
+Enriched events are automatically sent to the notification service for delivery via:
+- WhatsApp
+- Email
+- Push notifications
+- SMS
+
+Priority mapping:
+- `critical` → urgent priority
+- `important` → high priority
+- `standard` → medium priority
+- `info` → low priority
 
 ## Monitoring
 
 Structured JSON logs with:
 - `enrichment_started` - Event received
+- `duplicate_detected` - Duplicate alert found and updated
 - `enrichment_completed` - Enrichment pipeline finished
 - `event_stored` - Saved to database
 - `notification_sent` - Notification queued