Pharmeasy Autonomous Funnel Intelligence System

Pharmeasy Autonomous Funnel Intelligence System - Real-time funnel optimization through Perception-Reasoning-Action loop

Back

Workflow Information

ID: pharmeasy_funnel_intelligence_master

Namespace: pharmeasy

Version: 1.0

Created: 2025-07-28

Updated: 2025-07-28

Tasks: 9

Quick Actions

Manage Secrets

Inputs

Name	Type	Required	Default
`clevertap_account_id`	string	Required	`primary`
`clevertap_passcode`	string	Required	None
`clevertap_api_endpoint`	string	Optional	`https://api.clevertap.com/1`
`sales_drop_threshold`	float	Optional	`0.2`
`sales_spike_threshold`	float	Optional	`0.15`
`anomaly_sensitivity`	float	Optional	`2.5`
`analysis_window_days`	integer	Optional	`7`
`target_region`	string	Optional	`NCR`
`target_category`	string	Optional	`medicines`

Outputs

Name	Type	Source
`final_report`	object	Comprehensive funnel optimization report
`perception_data`	object	Combined data from CleverTap and internal systems
`campaign_results`	object	Recovery campaign execution results
`anomaly_detection`	object	Anomaly detection results and root cause analysis
`cross_sell_opportunities`	object	Identified cross-sell opportunities

Tasks

initialize_pra_loop

script

Initialize the Perception-Reasoning-Action loop

Python Script:

import os
import json
from datetime import datetime, timedelta

print("🤖 PHARMEASY FUNNEL INTELLIGENCE SYSTEM")
print("=" * 70)
print("🔄 Initializing Perception-Reasoning-Action Loop...")
print("")

# Get input parameters
execution_id = os.environ.get('EXECUTION_ID', 'unknown')
clevertap_account = "${clevertap_account_id}"
target_region = "${target_region}"
target_category = "${target_category}"
analysis_window = int("${analysis_window_days}")

# Calculate time ranges
end_date = datetime.now()
start_date = end_date - timedelta(days=analysis_window)

print("📊 Analysis Configuration:")
print(f"   Execution ID: {execution_id}")
print(f"   Region: {target_region}")
print(f"   Category: {target_category}")
print(f"   Analysis Window: {analysis_window} days")
print(f"   Period: {start_date.strftime('%Y-%m-%d')} to {end_date.strftime('%Y-%m-%d')}")
print("")

# Define hypothesis framework for multi-hypothesis testing
hypotheses = [
    "payment_gateway_issues",
    "inventory_stockouts", 
    "competitor_campaigns",
    "pricing_changes",
    "app_performance_issues",
    "delivery_delays",
    "user_experience_friction",
    "seasonal_patterns",
    "marketing_campaign_effectiveness",
    "new_user_onboarding_issues"
]

print(f"🔍 Multi-Hypothesis Testing Framework:")
print(f"   Total Hypotheses: {len(hypotheses)}")
print(f"   Approach: Parallel analysis across data silos")
print("")

result = {
    "status": "initialized",
    "execution_id": execution_id,
    "configuration": {
        "region": target_region,
        "category": target_category,
        "start_date": start_date.isoformat(),
        "end_date": end_date.isoformat(),
        "analysis_window_days": analysis_window
    },
    "hypotheses": hypotheses,
    "pra_loop_state": "PERCEPTION",
    "timestamp": datetime.now().isoformat()
}

print("✅ PRA Loop initialized. Starting PERCEPTION phase...")
print(f"__OUTPUTS__ {json.dumps(result)}")

perception_data_bridge

script

Fetch and combine data from CleverTap and internal dashboards

Python Script:

import json
import time
import random
from datetime import datetime, timedelta

print("")
print("👁️ PERCEPTION PHASE: Data Collection")
print("=" * 50)
print("")

config = json.loads('${initialize_pra_loop.configuration}')
region = config['region']
category = config['category']

# Simulate CleverTap API call
print("🔗 Fetching CleverTap behavioral data...")
time.sleep(1)

# Generate realistic CleverTap data
clevertap_data = {
    "events": {
        "app_opens": random.randint(50000, 80000),
        "searches": random.randint(30000, 50000),
        "product_views": random.randint(40000, 60000),
        "add_to_cart": random.randint(15000, 25000),
        "checkout_initiated": random.randint(8000, 12000),
        "orders_placed": random.randint(5000, 8000),
        "orders_completed": random.randint(4500, 7500)
    },
    "funnel_metrics": {
        "search_to_view": 0.75,
        "view_to_cart": 0.35,
        "cart_to_checkout": 0.65,
        "checkout_to_order": 0.70,
        "order_to_completion": 0.92
    },
    "user_segments": {
        "new_users": random.randint(5000, 8000),
        "chronic_patients": random.randint(12000, 15000),
        "regular_buyers": random.randint(20000, 25000),
        "dormant_users": random.randint(30000, 40000)
    },
    "cart_abandonment_rate": random.uniform(0.35, 0.45),
    "avg_session_duration": random.randint(180, 300),
    "bounce_rate": random.uniform(0.25, 0.35)
}

print("✓ CleverTap data fetched successfully")
print(f"  - Total events tracked: {sum(clevertap_data['events'].values()):,}")
print(f"  - Cart abandonment rate: {clevertap_data['cart_abandonment_rate']:.1%}")

# Simulate Internal Dashboard API call
print("")
print("📊 Fetching internal operational data...")
time.sleep(1)

# Generate realistic internal data with potential anomaly
base_sales = random.randint(800000, 1200000)
sales_drop = random.uniform(0.15, 0.25)  # 15-25% drop

internal_data = {
    "sales_metrics": {
        "current_period_sales": int(base_sales * (1 - sales_drop)),
        "previous_period_sales": base_sales,
        "sales_change_percentage": -sales_drop,
        "average_order_value": random.randint(450, 650),
        "total_orders": clevertap_data['events']['orders_completed']
    },
    "inventory_status": {
        "total_skus": random.randint(15000, 20000),
        "in_stock_skus": random.randint(12000, 18000),
        "stockout_skus": random.randint(500, 2000),
        "critical_stockouts": random.randint(50, 200)
    },
    "payment_metrics": {
        "success_rate": random.uniform(0.92, 0.98),
        "failed_transactions": random.randint(200, 500),
        "avg_processing_time": random.uniform(2.5, 4.5)
    },
    "logistics_metrics": {
        "on_time_delivery": random.uniform(0.85, 0.95),
        "avg_delivery_time_hours": random.randint(24, 48),
        "pending_deliveries": random.randint(500, 1500)
    }
}

print("✓ Internal dashboard data fetched successfully")
print(f"  - Sales change: {internal_data['sales_metrics']['sales_change_percentage']:.1%}")
print(f"  - Stockout SKUs: {internal_data['inventory_status']['stockout_skus']:,}")

# Combine data for unified analysis
print("")
print("🔄 Bridging data silos...")

unified_data = {
    "timestamp": datetime.now().isoformat(),
    "region": region,
    "category": category,
    "clevertap_behavioral": clevertap_data,
    "internal_operational": internal_data,
    "anomaly_detected": internal_data['sales_metrics']['sales_change_percentage'] < -0.15,
    "anomaly_severity": "HIGH" if sales_drop > 0.2 else "MEDIUM",
    "data_quality_score": random.uniform(0.92, 0.98),
    "pra_loop_state": "PERCEPTION_COMPLETE"
}

print("✅ PERCEPTION phase complete. Data silos bridged successfully.")
print(f"   Anomaly Detected: {'YES' if unified_data['anomaly_detected'] else 'NO'}")
if unified_data['anomaly_detected']:
    print(f"   Severity: {unified_data['anomaly_severity']}")

print(f"__OUTPUTS__ {json.dumps(unified_data)}")

detect_analyze_anomalies

script

Detect anomalies and perform multi-hypothesis root cause analysis

Python Script:

import json
import math
import random
from datetime import datetime

print("")
print("🧠 REASONING PHASE: Anomaly Detection & Analysis")
print("=" * 50)
print("")

# Parse perception data
perception_data = json.loads('${perception_data_bridge}')
clevertap = perception_data['clevertap_behavioral']
internal = perception_data['internal_operational']
hypotheses = json.loads('${initialize_pra_loop.hypotheses}')
sensitivity = float("${anomaly_sensitivity}")

print(f"📊 Analyzing sales metrics...")
sales_change = internal['sales_metrics']['sales_change_percentage']
print(f"   Sales Change: {sales_change:.1%}")
print(f"   Z-Score Sensitivity: {sensitivity}")

# Calculate Z-score for anomaly detection
# Simulate historical data for Z-score calculation
historical_changes = [random.uniform(-0.05, 0.10) for _ in range(30)]
mean_change = sum(historical_changes) / len(historical_changes)
std_dev = math.sqrt(sum((x - mean_change) ** 2 for x in historical_changes) / len(historical_changes))
z_score = (sales_change - mean_change) / std_dev if std_dev > 0 else 0

print(f"   Calculated Z-Score: {z_score:.2f}")
print(f"   Anomaly Threshold: {-sensitivity}")

is_anomaly = z_score < -sensitivity
print(f"   Anomaly Detected: {'YES' if is_anomaly else 'NO'}")

# Multi-hypothesis testing
print("")
print("🔍 Executing Multi-Hypothesis Testing...")
print("   Testing 10 hypotheses in parallel...")

hypothesis_results = {}

# Test each hypothesis based on data patterns
for hypothesis in hypotheses:
    score = 0.0
    evidence = []
    
    if hypothesis == "payment_gateway_issues":
        if internal['payment_metrics']['success_rate'] < 0.95:
            score = 0.7
            evidence.append(f"Payment success rate low: {internal['payment_metrics']['success_rate']:.1%}")
        if internal['payment_metrics']['failed_transactions'] > 300:
            score += 0.3
            evidence.append(f"High failed transactions: {internal['payment_metrics']['failed_transactions']}")
            
    elif hypothesis == "inventory_stockouts":
        stockout_rate = internal['inventory_status']['stockout_skus'] / internal['inventory_status']['total_skus']
        if stockout_rate > 0.08:
            score = 0.8
            evidence.append(f"High stockout rate: {stockout_rate:.1%}")
        if internal['inventory_status']['critical_stockouts'] > 100:
            score = min(score + 0.4, 1.0)
            evidence.append(f"Critical stockouts: {internal['inventory_status']['critical_stockouts']}")
            
    elif hypothesis == "user_experience_friction":
        if clevertap['cart_abandonment_rate'] > 0.40:
            score = 0.6
            evidence.append(f"High cart abandonment: {clevertap['cart_abandonment_rate']:.1%}")
        if clevertap['funnel_metrics']['cart_to_checkout'] < 0.60:
            score += 0.3
            evidence.append("Low cart-to-checkout conversion")
            
    elif hypothesis == "delivery_delays":
        if internal['logistics_metrics']['on_time_delivery'] < 0.90:
            score = 0.5
            evidence.append(f"Low on-time delivery: {internal['logistics_metrics']['on_time_delivery']:.1%}")
            
    else:
        # Simulate scores for other hypotheses
        score = random.uniform(0.1, 0.4)
        evidence.append("Pattern analysis in progress")
    
    hypothesis_results[hypothesis] = {
        "score": round(score, 2),
        "evidence": evidence,
        "confidence": "HIGH" if score > 0.7 else "MEDIUM" if score > 0.4 else "LOW"
    }

# Sort hypotheses by score
sorted_hypotheses = sorted(hypothesis_results.items(), key=lambda x: x[1]['score'], reverse=True)

print("")
print("📈 Top Root Causes Identified:")
for i, (hyp, result) in enumerate(sorted_hypotheses[:3]):
    print(f"   {i+1}. {hyp.replace('_', ' ').title()}")
    print(f"      Score: {result['score']}")
    print(f"      Confidence: {result['confidence']}")
    for evidence in result['evidence'][:2]:
        print(f"      - {evidence}")

# Determine primary root cause
primary_cause = sorted_hypotheses[0][0] if sorted_hypotheses[0][1]['score'] > 0.5 else "multiple_factors"

anomaly_analysis = {
    "timestamp": datetime.now().isoformat(),
    "anomaly_detected": is_anomaly,
    "z_score": round(z_score, 2),
    "sales_change_percentage": sales_change,
    "hypothesis_results": hypothesis_results,
    "primary_root_cause": primary_cause,
    "top_3_causes": [h[0] for h in sorted_hypotheses[:3]],
    "confidence_level": sorted_hypotheses[0][1]['confidence'] if sorted_hypotheses else "LOW",
    "recommended_actions": [],
    "pra_loop_state": "REASONING_COMPLETE"
}

# Generate recommended actions based on root cause
if primary_cause == "inventory_stockouts":
    anomaly_analysis["recommended_actions"] = [
        "trigger_stockout_notification_campaign",
        "activate_alternative_product_recommendations",
        "expedite_inventory_replenishment"
    ]
elif primary_cause == "payment_gateway_issues":
    anomaly_analysis["recommended_actions"] = [
        "trigger_payment_retry_campaign",
        "activate_cod_promotion",
        "notify_payment_team"
    ]
elif primary_cause == "user_experience_friction":
    anomaly_analysis["recommended_actions"] = [
        "trigger_cart_recovery_campaign",
        "simplify_checkout_flow",
        "offer_instant_discount"
    ]

print("")
print("✅ REASONING phase complete. Root cause analysis finished.")
print(f"__OUTPUTS__ {json.dumps(anomaly_analysis)}")

anomaly_router

conditional_router

Route based on anomaly detection results

Conditional Router

Router Type: condition
Default Route: normal_optimization

execute_recovery_actions

script

Execute recovery campaigns and interventions

Python Script:

import json
import time
import random
from datetime import datetime

print("")
print("🚀 ACTION PHASE: Executing Recovery Campaigns")
print("=" * 50)
print("")

# Parse anomaly analysis
anomaly_data = json.loads('${detect_analyze_anomalies}')
perception_data = json.loads('${perception_data_bridge}')
primary_cause = anomaly_data['primary_root_cause']
recommended_actions = anomaly_data['recommended_actions']

print(f"🎯 Primary Root Cause: {primary_cause.replace('_', ' ').title()}")
print(f"📋 Recommended Actions: {len(recommended_actions)}")
print("")

campaign_results = []

# Execute recommended actions
for action in recommended_actions:
    print(f"⚡ Executing: {action.replace('_', ' ').title()}")
    
    if action == "trigger_stockout_notification_campaign":
        # Simulate CleverTap campaign creation
        campaign = {
            "name": "Critical Stockout Alert",
            "type": "push_notification",
            "segment": "users_with_stockout_items_in_cart",
            "message": "Items in your cart are running out fast! Complete your order now.",
            "target_users": random.randint(5000, 8000),
            "expected_conversion": 0.15
        }
        print(f"   Creating campaign: {campaign['name']}")
        print(f"   Target users: {campaign['target_users']:,}")
        time.sleep(0.5)
        
        # Simulate campaign execution
        campaign['status'] = "executed"
        campaign['actual_sends'] = int(campaign['target_users'] * 0.95)
        campaign['opens'] = int(campaign['actual_sends'] * random.uniform(0.25, 0.35))
        campaign['conversions'] = int(campaign['opens'] * random.uniform(0.12, 0.18))
        campaign['revenue_impact'] = campaign['conversions'] * random.randint(500, 800)
        
        campaign_results.append(campaign)
        print(f"   ✓ Campaign executed successfully")
        print(f"   Conversions: {campaign['conversions']:,}")
        
    elif action == "trigger_cart_recovery_campaign":
        campaign = {
            "name": "Cart Abandonment Recovery",
            "type": "email_sms_combo",
            "segment": "cart_abandoners_last_24h",
            "discount_offered": "10%",
            "target_users": random.randint(8000, 12000),
            "expected_conversion": 0.08
        }
        print(f"   Creating campaign: {campaign['name']}")
        print(f"   Discount offered: {campaign['discount_offered']}")
        time.sleep(0.5)
        
        campaign['status'] = "executed"
        campaign['emails_sent'] = int(campaign['target_users'] * 0.90)
        campaign['sms_sent'] = int(campaign['target_users'] * 0.85)
        campaign['conversions'] = int(campaign['target_users'] * random.uniform(0.06, 0.10))
        campaign['revenue_impact'] = campaign['conversions'] * random.randint(400, 600)
        
        campaign_results.append(campaign)
        print(f"   ✓ Multi-channel campaign launched")
        
    elif action == "trigger_payment_retry_campaign":
        campaign = {
            "name": "Payment Retry Assistance",
            "type": "in_app_notification",
            "segment": "failed_payment_users",
            "cod_promotion": True,
            "target_users": random.randint(2000, 4000),
            "expected_conversion": 0.25
        }
        print(f"   Creating campaign: {campaign['name']}")
        time.sleep(0.5)
        
        campaign['status'] = "executed"
        campaign['notifications_shown'] = int(campaign['target_users'] * 0.98)
        campaign['payment_retries'] = int(campaign['notifications_shown'] * 0.30)
        campaign['successful_orders'] = int(campaign['payment_retries'] * 0.80)
        campaign['revenue_impact'] = campaign['successful_orders'] * random.randint(600, 900)
        
        campaign_results.append(campaign)
        print(f"   ✓ Payment assistance activated")
    
    print("")

# Calculate total impact
total_conversions = sum(c.get('conversions', c.get('successful_orders', 0)) for c in campaign_results)
total_revenue_impact = sum(c.get('revenue_impact', 0) for c in campaign_results)

print("📊 Campaign Execution Summary:")
print(f"   Total Campaigns: {len(campaign_results)}")
print(f"   Total Conversions: {total_conversions:,}")
print(f"   Revenue Impact: ₹{total_revenue_impact:,}")
print(f"   Recovery Rate: {(total_revenue_impact / abs(perception_data['internal_operational']['sales_metrics']['current_period_sales'] - perception_data['internal_operational']['sales_metrics']['previous_period_sales']) * 100):.1f}%")

action_results = {
    "timestamp": datetime.now().isoformat(),
    "campaigns_executed": len(campaign_results),
    "campaign_details": campaign_results,
    "total_conversions": total_conversions,
    "total_revenue_impact": total_revenue_impact,
    "execution_status": "SUCCESS",
    "pra_loop_state": "ACTION_COMPLETE"
}

print("")
print("✅ ACTION phase complete. Recovery campaigns executed.")
print(f"__OUTPUTS__ {json.dumps(action_results)}")

identify_cross_sell

script

Identify cross-sell opportunities during normal operations

Python Script:

import json
import random
from datetime import datetime

print("")
print("💡 CROSS-SELL OPPORTUNITY ANALYSIS")
print("=" * 50)
print("")

perception_data = json.loads('${perception_data_bridge}')

print("🔍 Analyzing cross-category behavior patterns...")

opportunities = []

# Medicine to Diagnostics cross-sell
opportunity1 = {
    "type": "medicine_to_diagnostics",
    "target_segment": "chronic_disease_medicine_buyers",
    "segment_size": random.randint(8000, 12000),
    "current_penetration": random.uniform(0.15, 0.25),
    "potential_penetration": random.uniform(0.35, 0.45),
    "recommended_action": "Bundle diabetes medicines with HbA1c tests",
    "expected_revenue_uplift": random.randint(200000, 400000)
}
opportunities.append(opportunity1)

# Diagnostics to Wellness cross-sell
opportunity2 = {
    "type": "diagnostics_to_wellness",
    "target_segment": "health_checkup_users",
    "segment_size": random.randint(5000, 8000),
    "current_penetration": random.uniform(0.10, 0.20),
    "potential_penetration": random.uniform(0.30, 0.40),
    "recommended_action": "Recommend vitamins based on test results",
    "expected_revenue_uplift": random.randint(150000, 250000)
}
opportunities.append(opportunity2)

print(f"✓ Identified {len(opportunities)} cross-sell opportunities")

for opp in opportunities:
    print(f"\n📌 {opp['type'].replace('_', ' ').title()}")
    print(f"   Target Size: {opp['segment_size']:,} users")
    print(f"   Penetration Potential: {opp['current_penetration']:.1%} → {opp['potential_penetration']:.1%}")
    print(f"   Revenue Uplift: ₹{opp['expected_revenue_uplift']:,}")

cross_sell_results = {
    "timestamp": datetime.now().isoformat(),
    "opportunities_identified": len(opportunities),
    "opportunities": opportunities,
    "total_revenue_potential": sum(o['expected_revenue_uplift'] for o in opportunities),
    "execution_status": "READY_FOR_CAMPAIGNS"
}

print("")
print("✅ Cross-sell analysis complete.")
print(f"__OUTPUTS__ {json.dumps(cross_sell_results)}")

ai_churn_prevention

ai_agent

AI-powered churn prediction and prevention strategies

AI Agent Configuration:

Model Configuration

Model Client ID: funnel_master
Agent Type: assistant

Input Format: text
Output Format: text

ai_journey_optimizer

ai_agent

AI-powered dynamic customer journey optimization

AI Agent Configuration:

Model Configuration

Model Client ID: campaign_creator
Agent Type: assistant

Input Format: text
Output Format: text

generate_optimization_report

ai_agent

Generate comprehensive funnel optimization report

AI Agent Configuration:

Model Configuration

Model Client ID: retention_optimizer
Agent Type: assistant

Input Format: text
Output Format: text

YAML Source

id: pharmeasy_funnel_intelligence_master
name: Pharmeasy Autonomous Funnel Intelligence System
tasks:
- id: initialize_pra_loop
  name: Initialize Perception-Reasoning-Action Loop
  type: script
  script: "import os\nimport json\nfrom datetime import datetime, timedelta\n\nprint(\"\
    \U0001F916 PHARMEASY FUNNEL INTELLIGENCE SYSTEM\")\nprint(\"=\" * 70)\nprint(\"\
    \U0001F504 Initializing Perception-Reasoning-Action Loop...\")\nprint(\"\")\n\n\
    # Get input parameters\nexecution_id = os.environ.get('EXECUTION_ID', 'unknown')\n\
    clevertap_account = \"${clevertap_account_id}\"\ntarget_region = \"${target_region}\"\
    \ntarget_category = \"${target_category}\"\nanalysis_window = int(\"${analysis_window_days}\"\
    )\n\n# Calculate time ranges\nend_date = datetime.now()\nstart_date = end_date\
    \ - timedelta(days=analysis_window)\n\nprint(\"\U0001F4CA Analysis Configuration:\"\
    )\nprint(f\"   Execution ID: {execution_id}\")\nprint(f\"   Region: {target_region}\"\
    )\nprint(f\"   Category: {target_category}\")\nprint(f\"   Analysis Window: {analysis_window}\
    \ days\")\nprint(f\"   Period: {start_date.strftime('%Y-%m-%d')} to {end_date.strftime('%Y-%m-%d')}\"\
    )\nprint(\"\")\n\n# Define hypothesis framework for multi-hypothesis testing\n\
    hypotheses = [\n    \"payment_gateway_issues\",\n    \"inventory_stockouts\",\
    \ \n    \"competitor_campaigns\",\n    \"pricing_changes\",\n    \"app_performance_issues\"\
    ,\n    \"delivery_delays\",\n    \"user_experience_friction\",\n    \"seasonal_patterns\"\
    ,\n    \"marketing_campaign_effectiveness\",\n    \"new_user_onboarding_issues\"\
    \n]\n\nprint(f\"\U0001F50D Multi-Hypothesis Testing Framework:\")\nprint(f\" \
    \  Total Hypotheses: {len(hypotheses)}\")\nprint(f\"   Approach: Parallel analysis\
    \ across data silos\")\nprint(\"\")\n\nresult = {\n    \"status\": \"initialized\"\
    ,\n    \"execution_id\": execution_id,\n    \"configuration\": {\n        \"region\"\
    : target_region,\n        \"category\": target_category,\n        \"start_date\"\
    : start_date.isoformat(),\n        \"end_date\": end_date.isoformat(),\n     \
    \   \"analysis_window_days\": analysis_window\n    },\n    \"hypotheses\": hypotheses,\n\
    \    \"pra_loop_state\": \"PERCEPTION\",\n    \"timestamp\": datetime.now().isoformat()\n\
    }\n\nprint(\"\u2705 PRA Loop initialized. Starting PERCEPTION phase...\")\nprint(f\"\
    __OUTPUTS__ {json.dumps(result)}\")\n"
  description: Initialize the Perception-Reasoning-Action loop
  timeout_seconds: 30
- id: perception_data_bridge
  name: 'PERCEPTION: Bridge Data Silos'
  type: script
  script: "import json\nimport time\nimport random\nfrom datetime import datetime,\
    \ timedelta\n\nprint(\"\")\nprint(\"\U0001F441\uFE0F PERCEPTION PHASE: Data Collection\"\
    )\nprint(\"=\" * 50)\nprint(\"\")\n\nconfig = json.loads('${initialize_pra_loop.configuration}')\n\
    region = config['region']\ncategory = config['category']\n\n# Simulate CleverTap\
    \ API call\nprint(\"\U0001F517 Fetching CleverTap behavioral data...\")\ntime.sleep(1)\n\
    \n# Generate realistic CleverTap data\nclevertap_data = {\n    \"events\": {\n\
    \        \"app_opens\": random.randint(50000, 80000),\n        \"searches\": random.randint(30000,\
    \ 50000),\n        \"product_views\": random.randint(40000, 60000),\n        \"\
    add_to_cart\": random.randint(15000, 25000),\n        \"checkout_initiated\":\
    \ random.randint(8000, 12000),\n        \"orders_placed\": random.randint(5000,\
    \ 8000),\n        \"orders_completed\": random.randint(4500, 7500)\n    },\n \
    \   \"funnel_metrics\": {\n        \"search_to_view\": 0.75,\n        \"view_to_cart\"\
    : 0.35,\n        \"cart_to_checkout\": 0.65,\n        \"checkout_to_order\": 0.70,\n\
    \        \"order_to_completion\": 0.92\n    },\n    \"user_segments\": {\n   \
    \     \"new_users\": random.randint(5000, 8000),\n        \"chronic_patients\"\
    : random.randint(12000, 15000),\n        \"regular_buyers\": random.randint(20000,\
    \ 25000),\n        \"dormant_users\": random.randint(30000, 40000)\n    },\n \
    \   \"cart_abandonment_rate\": random.uniform(0.35, 0.45),\n    \"avg_session_duration\"\
    : random.randint(180, 300),\n    \"bounce_rate\": random.uniform(0.25, 0.35)\n\
    }\n\nprint(\"\u2713 CleverTap data fetched successfully\")\nprint(f\"  - Total\
    \ events tracked: {sum(clevertap_data['events'].values()):,}\")\nprint(f\"  -\
    \ Cart abandonment rate: {clevertap_data['cart_abandonment_rate']:.1%}\")\n\n\
    # Simulate Internal Dashboard API call\nprint(\"\")\nprint(\"\U0001F4CA Fetching\
    \ internal operational data...\")\ntime.sleep(1)\n\n# Generate realistic internal\
    \ data with potential anomaly\nbase_sales = random.randint(800000, 1200000)\n\
    sales_drop = random.uniform(0.15, 0.25)  # 15-25% drop\n\ninternal_data = {\n\
    \    \"sales_metrics\": {\n        \"current_period_sales\": int(base_sales *\
    \ (1 - sales_drop)),\n        \"previous_period_sales\": base_sales,\n       \
    \ \"sales_change_percentage\": -sales_drop,\n        \"average_order_value\":\
    \ random.randint(450, 650),\n        \"total_orders\": clevertap_data['events']['orders_completed']\n\
    \    },\n    \"inventory_status\": {\n        \"total_skus\": random.randint(15000,\
    \ 20000),\n        \"in_stock_skus\": random.randint(12000, 18000),\n        \"\
    stockout_skus\": random.randint(500, 2000),\n        \"critical_stockouts\": random.randint(50,\
    \ 200)\n    },\n    \"payment_metrics\": {\n        \"success_rate\": random.uniform(0.92,\
    \ 0.98),\n        \"failed_transactions\": random.randint(200, 500),\n       \
    \ \"avg_processing_time\": random.uniform(2.5, 4.5)\n    },\n    \"logistics_metrics\"\
    : {\n        \"on_time_delivery\": random.uniform(0.85, 0.95),\n        \"avg_delivery_time_hours\"\
    : random.randint(24, 48),\n        \"pending_deliveries\": random.randint(500,\
    \ 1500)\n    }\n}\n\nprint(\"\u2713 Internal dashboard data fetched successfully\"\
    )\nprint(f\"  - Sales change: {internal_data['sales_metrics']['sales_change_percentage']:.1%}\"\
    )\nprint(f\"  - Stockout SKUs: {internal_data['inventory_status']['stockout_skus']:,}\"\
    )\n\n# Combine data for unified analysis\nprint(\"\")\nprint(\"\U0001F504 Bridging\
    \ data silos...\")\n\nunified_data = {\n    \"timestamp\": datetime.now().isoformat(),\n\
    \    \"region\": region,\n    \"category\": category,\n    \"clevertap_behavioral\"\
    : clevertap_data,\n    \"internal_operational\": internal_data,\n    \"anomaly_detected\"\
    : internal_data['sales_metrics']['sales_change_percentage'] < -0.15,\n    \"anomaly_severity\"\
    : \"HIGH\" if sales_drop > 0.2 else \"MEDIUM\",\n    \"data_quality_score\": random.uniform(0.92,\
    \ 0.98),\n    \"pra_loop_state\": \"PERCEPTION_COMPLETE\"\n}\n\nprint(\"\u2705\
    \ PERCEPTION phase complete. Data silos bridged successfully.\")\nprint(f\"  \
    \ Anomaly Detected: {'YES' if unified_data['anomaly_detected'] else 'NO'}\")\n\
    if unified_data['anomaly_detected']:\n    print(f\"   Severity: {unified_data['anomaly_severity']}\"\
    )\n\nprint(f\"__OUTPUTS__ {json.dumps(unified_data)}\")\n"
  depends_on:
  - initialize_pra_loop
  description: Fetch and combine data from CleverTap and internal dashboards
  timeout_seconds: 60
- id: detect_analyze_anomalies
  name: 'REASONING: Anomaly Detection & Multi-Hypothesis Testing'
  type: script
  script: "import json\nimport math\nimport random\nfrom datetime import datetime\n\
    \nprint(\"\")\nprint(\"\U0001F9E0 REASONING PHASE: Anomaly Detection & Analysis\"\
    )\nprint(\"=\" * 50)\nprint(\"\")\n\n# Parse perception data\nperception_data\
    \ = json.loads('${perception_data_bridge}')\nclevertap = perception_data['clevertap_behavioral']\n\
    internal = perception_data['internal_operational']\nhypotheses = json.loads('${initialize_pra_loop.hypotheses}')\n\
    sensitivity = float(\"${anomaly_sensitivity}\")\n\nprint(f\"\U0001F4CA Analyzing\
    \ sales metrics...\")\nsales_change = internal['sales_metrics']['sales_change_percentage']\n\
    print(f\"   Sales Change: {sales_change:.1%}\")\nprint(f\"   Z-Score Sensitivity:\
    \ {sensitivity}\")\n\n# Calculate Z-score for anomaly detection\n# Simulate historical\
    \ data for Z-score calculation\nhistorical_changes = [random.uniform(-0.05, 0.10)\
    \ for _ in range(30)]\nmean_change = sum(historical_changes) / len(historical_changes)\n\
    std_dev = math.sqrt(sum((x - mean_change) ** 2 for x in historical_changes) /\
    \ len(historical_changes))\nz_score = (sales_change - mean_change) / std_dev if\
    \ std_dev > 0 else 0\n\nprint(f\"   Calculated Z-Score: {z_score:.2f}\")\nprint(f\"\
    \   Anomaly Threshold: {-sensitivity}\")\n\nis_anomaly = z_score < -sensitivity\n\
    print(f\"   Anomaly Detected: {'YES' if is_anomaly else 'NO'}\")\n\n# Multi-hypothesis\
    \ testing\nprint(\"\")\nprint(\"\U0001F50D Executing Multi-Hypothesis Testing...\"\
    )\nprint(\"   Testing 10 hypotheses in parallel...\")\n\nhypothesis_results =\
    \ {}\n\n# Test each hypothesis based on data patterns\nfor hypothesis in hypotheses:\n\
    \    score = 0.0\n    evidence = []\n    \n    if hypothesis == \"payment_gateway_issues\"\
    :\n        if internal['payment_metrics']['success_rate'] < 0.95:\n          \
    \  score = 0.7\n            evidence.append(f\"Payment success rate low: {internal['payment_metrics']['success_rate']:.1%}\"\
    )\n        if internal['payment_metrics']['failed_transactions'] > 300:\n    \
    \        score += 0.3\n            evidence.append(f\"High failed transactions:\
    \ {internal['payment_metrics']['failed_transactions']}\")\n            \n    elif\
    \ hypothesis == \"inventory_stockouts\":\n        stockout_rate = internal['inventory_status']['stockout_skus']\
    \ / internal['inventory_status']['total_skus']\n        if stockout_rate > 0.08:\n\
    \            score = 0.8\n            evidence.append(f\"High stockout rate: {stockout_rate:.1%}\"\
    )\n        if internal['inventory_status']['critical_stockouts'] > 100:\n    \
    \        score = min(score + 0.4, 1.0)\n            evidence.append(f\"Critical\
    \ stockouts: {internal['inventory_status']['critical_stockouts']}\")\n       \
    \     \n    elif hypothesis == \"user_experience_friction\":\n        if clevertap['cart_abandonment_rate']\
    \ > 0.40:\n            score = 0.6\n            evidence.append(f\"High cart abandonment:\
    \ {clevertap['cart_abandonment_rate']:.1%}\")\n        if clevertap['funnel_metrics']['cart_to_checkout']\
    \ < 0.60:\n            score += 0.3\n            evidence.append(\"Low cart-to-checkout\
    \ conversion\")\n            \n    elif hypothesis == \"delivery_delays\":\n \
    \       if internal['logistics_metrics']['on_time_delivery'] < 0.90:\n       \
    \     score = 0.5\n            evidence.append(f\"Low on-time delivery: {internal['logistics_metrics']['on_time_delivery']:.1%}\"\
    )\n            \n    else:\n        # Simulate scores for other hypotheses\n \
    \       score = random.uniform(0.1, 0.4)\n        evidence.append(\"Pattern analysis\
    \ in progress\")\n    \n    hypothesis_results[hypothesis] = {\n        \"score\"\
    : round(score, 2),\n        \"evidence\": evidence,\n        \"confidence\": \"\
    HIGH\" if score > 0.7 else \"MEDIUM\" if score > 0.4 else \"LOW\"\n    }\n\n#\
    \ Sort hypotheses by score\nsorted_hypotheses = sorted(hypothesis_results.items(),\
    \ key=lambda x: x[1]['score'], reverse=True)\n\nprint(\"\")\nprint(\"\U0001F4C8\
    \ Top Root Causes Identified:\")\nfor i, (hyp, result) in enumerate(sorted_hypotheses[:3]):\n\
    \    print(f\"   {i+1}. {hyp.replace('_', ' ').title()}\")\n    print(f\"    \
    \  Score: {result['score']}\")\n    print(f\"      Confidence: {result['confidence']}\"\
    )\n    for evidence in result['evidence'][:2]:\n        print(f\"      - {evidence}\"\
    )\n\n# Determine primary root cause\nprimary_cause = sorted_hypotheses[0][0] if\
    \ sorted_hypotheses[0][1]['score'] > 0.5 else \"multiple_factors\"\n\nanomaly_analysis\
    \ = {\n    \"timestamp\": datetime.now().isoformat(),\n    \"anomaly_detected\"\
    : is_anomaly,\n    \"z_score\": round(z_score, 2),\n    \"sales_change_percentage\"\
    : sales_change,\n    \"hypothesis_results\": hypothesis_results,\n    \"primary_root_cause\"\
    : primary_cause,\n    \"top_3_causes\": [h[0] for h in sorted_hypotheses[:3]],\n\
    \    \"confidence_level\": sorted_hypotheses[0][1]['confidence'] if sorted_hypotheses\
    \ else \"LOW\",\n    \"recommended_actions\": [],\n    \"pra_loop_state\": \"\
    REASONING_COMPLETE\"\n}\n\n# Generate recommended actions based on root cause\n\
    if primary_cause == \"inventory_stockouts\":\n    anomaly_analysis[\"recommended_actions\"\
    ] = [\n        \"trigger_stockout_notification_campaign\",\n        \"activate_alternative_product_recommendations\"\
    ,\n        \"expedite_inventory_replenishment\"\n    ]\nelif primary_cause ==\
    \ \"payment_gateway_issues\":\n    anomaly_analysis[\"recommended_actions\"] =\
    \ [\n        \"trigger_payment_retry_campaign\",\n        \"activate_cod_promotion\"\
    ,\n        \"notify_payment_team\"\n    ]\nelif primary_cause == \"user_experience_friction\"\
    :\n    anomaly_analysis[\"recommended_actions\"] = [\n        \"trigger_cart_recovery_campaign\"\
    ,\n        \"simplify_checkout_flow\",\n        \"offer_instant_discount\"\n \
    \   ]\n\nprint(\"\")\nprint(\"\u2705 REASONING phase complete. Root cause analysis\
    \ finished.\")\nprint(f\"__OUTPUTS__ {json.dumps(anomaly_analysis)}\")\n"
  depends_on:
  - perception_data_bridge
  description: Detect anomalies and perform multi-hypothesis root cause analysis
  timeout_seconds: 90
- id: anomaly_router
  name: Anomaly Detection Router
  type: conditional_router
  conditions:
  - name: anomaly_detected
    route: handle_anomaly
    condition: ${detect_analyze_anomalies.anomaly_detected} == true
  depends_on:
  - detect_analyze_anomalies
  description: Route based on anomaly detection results
  default_route: normal_optimization
- id: execute_recovery_actions
  name: 'ACTION: Execute Recovery Campaigns'
  type: script
  script: "import json\nimport time\nimport random\nfrom datetime import datetime\n\
    \nprint(\"\")\nprint(\"\U0001F680 ACTION PHASE: Executing Recovery Campaigns\"\
    )\nprint(\"=\" * 50)\nprint(\"\")\n\n# Parse anomaly analysis\nanomaly_data =\
    \ json.loads('${detect_analyze_anomalies}')\nperception_data = json.loads('${perception_data_bridge}')\n\
    primary_cause = anomaly_data['primary_root_cause']\nrecommended_actions = anomaly_data['recommended_actions']\n\
    \nprint(f\"\U0001F3AF Primary Root Cause: {primary_cause.replace('_', ' ').title()}\"\
    )\nprint(f\"\U0001F4CB Recommended Actions: {len(recommended_actions)}\")\nprint(\"\
    \")\n\ncampaign_results = []\n\n# Execute recommended actions\nfor action in recommended_actions:\n\
    \    print(f\"\u26A1 Executing: {action.replace('_', ' ').title()}\")\n    \n\
    \    if action == \"trigger_stockout_notification_campaign\":\n        # Simulate\
    \ CleverTap campaign creation\n        campaign = {\n            \"name\": \"\
    Critical Stockout Alert\",\n            \"type\": \"push_notification\",\n   \
    \         \"segment\": \"users_with_stockout_items_in_cart\",\n            \"\
    message\": \"Items in your cart are running out fast! Complete your order now.\"\
    ,\n            \"target_users\": random.randint(5000, 8000),\n            \"expected_conversion\"\
    : 0.15\n        }\n        print(f\"   Creating campaign: {campaign['name']}\"\
    )\n        print(f\"   Target users: {campaign['target_users']:,}\")\n       \
    \ time.sleep(0.5)\n        \n        # Simulate campaign execution\n        campaign['status']\
    \ = \"executed\"\n        campaign['actual_sends'] = int(campaign['target_users']\
    \ * 0.95)\n        campaign['opens'] = int(campaign['actual_sends'] * random.uniform(0.25,\
    \ 0.35))\n        campaign['conversions'] = int(campaign['opens'] * random.uniform(0.12,\
    \ 0.18))\n        campaign['revenue_impact'] = campaign['conversions'] * random.randint(500,\
    \ 800)\n        \n        campaign_results.append(campaign)\n        print(f\"\
    \   \u2713 Campaign executed successfully\")\n        print(f\"   Conversions:\
    \ {campaign['conversions']:,}\")\n        \n    elif action == \"trigger_cart_recovery_campaign\"\
    :\n        campaign = {\n            \"name\": \"Cart Abandonment Recovery\",\n\
    \            \"type\": \"email_sms_combo\",\n            \"segment\": \"cart_abandoners_last_24h\"\
    ,\n            \"discount_offered\": \"10%\",\n            \"target_users\": random.randint(8000,\
    \ 12000),\n            \"expected_conversion\": 0.08\n        }\n        print(f\"\
    \   Creating campaign: {campaign['name']}\")\n        print(f\"   Discount offered:\
    \ {campaign['discount_offered']}\")\n        time.sleep(0.5)\n        \n     \
    \   campaign['status'] = \"executed\"\n        campaign['emails_sent'] = int(campaign['target_users']\
    \ * 0.90)\n        campaign['sms_sent'] = int(campaign['target_users'] * 0.85)\n\
    \        campaign['conversions'] = int(campaign['target_users'] * random.uniform(0.06,\
    \ 0.10))\n        campaign['revenue_impact'] = campaign['conversions'] * random.randint(400,\
    \ 600)\n        \n        campaign_results.append(campaign)\n        print(f\"\
    \   \u2713 Multi-channel campaign launched\")\n        \n    elif action == \"\
    trigger_payment_retry_campaign\":\n        campaign = {\n            \"name\"\
    : \"Payment Retry Assistance\",\n            \"type\": \"in_app_notification\"\
    ,\n            \"segment\": \"failed_payment_users\",\n            \"cod_promotion\"\
    : True,\n            \"target_users\": random.randint(2000, 4000),\n         \
    \   \"expected_conversion\": 0.25\n        }\n        print(f\"   Creating campaign:\
    \ {campaign['name']}\")\n        time.sleep(0.5)\n        \n        campaign['status']\
    \ = \"executed\"\n        campaign['notifications_shown'] = int(campaign['target_users']\
    \ * 0.98)\n        campaign['payment_retries'] = int(campaign['notifications_shown']\
    \ * 0.30)\n        campaign['successful_orders'] = int(campaign['payment_retries']\
    \ * 0.80)\n        campaign['revenue_impact'] = campaign['successful_orders']\
    \ * random.randint(600, 900)\n        \n        campaign_results.append(campaign)\n\
    \        print(f\"   \u2713 Payment assistance activated\")\n    \n    print(\"\
    \")\n\n# Calculate total impact\ntotal_conversions = sum(c.get('conversions',\
    \ c.get('successful_orders', 0)) for c in campaign_results)\ntotal_revenue_impact\
    \ = sum(c.get('revenue_impact', 0) for c in campaign_results)\n\nprint(\"\U0001F4CA\
    \ Campaign Execution Summary:\")\nprint(f\"   Total Campaigns: {len(campaign_results)}\"\
    )\nprint(f\"   Total Conversions: {total_conversions:,}\")\nprint(f\"   Revenue\
    \ Impact: \u20B9{total_revenue_impact:,}\")\nprint(f\"   Recovery Rate: {(total_revenue_impact\
    \ / abs(perception_data['internal_operational']['sales_metrics']['current_period_sales']\
    \ - perception_data['internal_operational']['sales_metrics']['previous_period_sales'])\
    \ * 100):.1f}%\")\n\naction_results = {\n    \"timestamp\": datetime.now().isoformat(),\n\
    \    \"campaigns_executed\": len(campaign_results),\n    \"campaign_details\"\
    : campaign_results,\n    \"total_conversions\": total_conversions,\n    \"total_revenue_impact\"\
    : total_revenue_impact,\n    \"execution_status\": \"SUCCESS\",\n    \"pra_loop_state\"\
    : \"ACTION_COMPLETE\"\n}\n\nprint(\"\")\nprint(\"\u2705 ACTION phase complete.\
    \ Recovery campaigns executed.\")\nprint(f\"__OUTPUTS__ {json.dumps(action_results)}\"\
    )\n"
  depends_on:
  - anomaly_router
  description: Execute recovery campaigns and interventions
  timeout_seconds: 120
  execute_on_routes:
  - handle_anomaly
- id: identify_cross_sell
  name: Cross-sell Opportunity Identification
  type: script
  script: "import json\nimport random\nfrom datetime import datetime\n\nprint(\"\"\
    )\nprint(\"\U0001F4A1 CROSS-SELL OPPORTUNITY ANALYSIS\")\nprint(\"=\" * 50)\n\
    print(\"\")\n\nperception_data = json.loads('${perception_data_bridge}')\n\nprint(\"\
    \U0001F50D Analyzing cross-category behavior patterns...\")\n\nopportunities =\
    \ []\n\n# Medicine to Diagnostics cross-sell\nopportunity1 = {\n    \"type\":\
    \ \"medicine_to_diagnostics\",\n    \"target_segment\": \"chronic_disease_medicine_buyers\"\
    ,\n    \"segment_size\": random.randint(8000, 12000),\n    \"current_penetration\"\
    : random.uniform(0.15, 0.25),\n    \"potential_penetration\": random.uniform(0.35,\
    \ 0.45),\n    \"recommended_action\": \"Bundle diabetes medicines with HbA1c tests\"\
    ,\n    \"expected_revenue_uplift\": random.randint(200000, 400000)\n}\nopportunities.append(opportunity1)\n\
    \n# Diagnostics to Wellness cross-sell\nopportunity2 = {\n    \"type\": \"diagnostics_to_wellness\"\
    ,\n    \"target_segment\": \"health_checkup_users\",\n    \"segment_size\": random.randint(5000,\
    \ 8000),\n    \"current_penetration\": random.uniform(0.10, 0.20),\n    \"potential_penetration\"\
    : random.uniform(0.30, 0.40),\n    \"recommended_action\": \"Recommend vitamins\
    \ based on test results\",\n    \"expected_revenue_uplift\": random.randint(150000,\
    \ 250000)\n}\nopportunities.append(opportunity2)\n\nprint(f\"\u2713 Identified\
    \ {len(opportunities)} cross-sell opportunities\")\n\nfor opp in opportunities:\n\
    \    print(f\"\\n\U0001F4CC {opp['type'].replace('_', ' ').title()}\")\n    print(f\"\
    \   Target Size: {opp['segment_size']:,} users\")\n    print(f\"   Penetration\
    \ Potential: {opp['current_penetration']:.1%} \u2192 {opp['potential_penetration']:.1%}\"\
    )\n    print(f\"   Revenue Uplift: \u20B9{opp['expected_revenue_uplift']:,}\"\
    )\n\ncross_sell_results = {\n    \"timestamp\": datetime.now().isoformat(),\n\
    \    \"opportunities_identified\": len(opportunities),\n    \"opportunities\"\
    : opportunities,\n    \"total_revenue_potential\": sum(o['expected_revenue_uplift']\
    \ for o in opportunities),\n    \"execution_status\": \"READY_FOR_CAMPAIGNS\"\n\
    }\n\nprint(\"\")\nprint(\"\u2705 Cross-sell analysis complete.\")\nprint(f\"__OUTPUTS__\
    \ {json.dumps(cross_sell_results)}\")\n"
  depends_on:
  - anomaly_router
  description: Identify cross-sell opportunities during normal operations
  timeout_seconds: 60
  execute_on_routes:
  - normal_optimization
- id: ai_churn_prevention
  name: AI Churn Prevention Strategist
  type: ai_agent
  config:
    model_client_id: funnel_master
  depends_on:
  - perception_data_bridge
  - detect_analyze_anomalies
  description: AI-powered churn prediction and prevention strategies
  user_message: 'Analyze churn risk based on this data:


    User Segments:

    - Chronic Patients: ${perception_data_bridge.clevertap_behavioral.user_segments.chronic_patients}

    - Regular Buyers: ${perception_data_bridge.clevertap_behavioral.user_segments.regular_buyers}

    - Dormant Users: ${perception_data_bridge.clevertap_behavioral.user_segments.dormant_users}


    Behavioral Metrics:

    - Cart Abandonment Rate: ${perception_data_bridge.clevertap_behavioral.cart_abandonment_rate}

    - Avg Session Duration: ${perception_data_bridge.clevertap_behavioral.avg_session_duration}s

    - Bounce Rate: ${perception_data_bridge.clevertap_behavioral.bounce_rate}


    Provide churn prevention strategies focusing on chronic patient retention.

    '
  system_message: "You are an expert in customer retention and churn prevention for\
    \ e-commerce pharmacy platforms.\n\nAnalyze the behavioral data and identify users\
    \ at risk of churning. Provide your analysis in JSON format:\n{\n  \"churn_risk_segments\"\
    : [\n    {\n      \"segment_name\": \"segment description\",\n      \"risk_level\"\
    : \"HIGH|MEDIUM|LOW\",\n      \"user_count\": estimated_number,\n      \"key_indicators\"\
    : [\"indicator1\", \"indicator2\"],\n      \"recommended_interventions\": [\"\
    action1\", \"action2\"]\n    }\n  ],\n  \"chronic_patient_retention\": {\n   \
    \ \"at_risk_patients\": number,\n    \"retention_strategies\": [\"strategy1\"\
    , \"strategy2\"],\n    \"personalization_approach\": \"description\"\n  },\n \
    \ \"overall_churn_rate\": percentage,\n  \"predicted_revenue_at_risk\": amount\n\
    }\n"
- id: ai_journey_optimizer
  name: AI Dynamic Journey Creator
  type: ai_agent
  config:
    model_client_id: campaign_creator
  depends_on:
  - execute_recovery_actions
  - ai_churn_prevention
  description: AI-powered dynamic customer journey optimization
  user_message: 'Create optimized customer journeys based on:


    Recent Campaign Performance: ${execute_recovery_actions.campaign_details}


    User Behavior Patterns:

    - Funnel Metrics: ${perception_data_bridge.clevertap_behavioral.funnel_metrics}

    - Cart Abandonment: ${perception_data_bridge.clevertap_behavioral.cart_abandonment_rate}


    Churn Risk Analysis: ${ai_churn_prevention.churn_risk_segments}


    Focus on creating journeys that improve conversion and retention.

    '
  system_message: "You are an expert in creating personalized customer journeys for\
    \ pharmaceutical e-commerce.\n\nBased on the campaign results and user behavior,\
    \ design optimized customer journeys in JSON format:\n{\n  \"journey_name\": \"\
    descriptive name\",\n  \"target_segments\": [\"segment1\", \"segment2\"],\n  \"\
    journey_stages\": [\n    {\n      \"stage\": \"stage_name\",\n      \"trigger\"\
    : \"event or condition\",\n      \"action\": \"campaign or intervention\",\n \
    \     \"expected_outcome\": \"metric improvement\",\n      \"personalization_rules\"\
    : [\"rule1\", \"rule2\"]\n    }\n  ],\n  \"success_metrics\": {\n    \"primary_kpi\"\
    : \"metric name\",\n    \"target_improvement\": percentage,\n    \"measurement_period\"\
    : \"timeframe\"\n  },\n  \"a_b_test_recommendations\": [\"test1\", \"test2\"]\n\
    }\n"
- id: generate_optimization_report
  name: AI Funnel Optimization Report Generator
  type: ai_agent
  config:
    model_client_id: retention_optimizer
  depends_on:
  - ai_journey_optimizer
  - identify_cross_sell
  description: Generate comprehensive funnel optimization report
  user_message: "Generate a comprehensive funnel optimization report based on:\n\n\
    Initial State:\n- Sales Change: ${perception_data_bridge.internal_operational.sales_metrics.sales_change_percentage}%\n\
    - Region: ${initialize_pra_loop.configuration.region}\n- Category: ${initialize_pra_loop.configuration.category}\n\
    \nAnomaly Analysis:\n- Primary Cause: ${detect_analyze_anomalies.primary_root_cause}\n\
    - Top Causes: ${detect_analyze_anomalies.top_3_causes}\n\nRecovery Actions:\n\
    - Campaigns Executed: ${execute_recovery_actions.campaigns_executed}\n- Revenue\
    \ Impact: \u20B9${execute_recovery_actions.total_revenue_impact}\n\nCross-sell\
    \ Opportunities: ${identify_cross_sell.opportunities}\n\nCustomer Journey Optimization:\
    \ ${ai_journey_optimizer}\n\nProvide actionable insights and strategic recommendations.\n"
  previous_node:
  - ai_journey_optimizer
  - identify_cross_sell
  system_message: "You are a senior analytics expert specializing in funnel optimization\
    \ for e-commerce platforms.\n\nGenerate a comprehensive executive report in JSON\
    \ format:\n{\n  \"executive_summary\": {\n    \"key_findings\": [\"finding1\"\
    , \"finding2\", \"finding3\"],\n    \"business_impact\": \"revenue and conversion\
    \ impact summary\",\n    \"recommended_priorities\": [\"priority1\", \"priority2\"\
    ]\n  },\n  \"funnel_performance\": {\n    \"conversion_improvements\": {\n   \
    \   \"stage\": \"improvement_percentage\"\n    },\n    \"bottlenecks_addressed\"\
    : [\"bottleneck1\", \"bottleneck2\"],\n    \"optimization_opportunities\": [\"\
    opportunity1\", \"opportunity2\"]\n  },\n  \"campaign_effectiveness\": {\n   \
    \ \"total_campaigns_executed\": number,\n    \"overall_roi\": percentage,\n  \
    \  \"most_effective_campaign\": \"campaign_name\",\n    \"learnings\": [\"learning1\"\
    , \"learning2\"]\n  },\n  \"predictive_insights\": {\n    \"next_7_days_forecast\"\
    : \"prediction\",\n    \"risk_factors\": [\"risk1\", \"risk2\"],\n    \"proactive_recommendations\"\
    : [\"recommendation1\", \"recommendation2\"]\n  },\n  \"action_plan\": {\n   \
    \ \"immediate_actions\": [\"action1\", \"action2\"],\n    \"weekly_optimizations\"\
    : [\"optimization1\", \"optimization2\"],\n    \"strategic_initiatives\": [\"\
    initiative1\", \"initiative2\"]\n  }\n}\n"
inputs:
- name: clevertap_account_id
  type: string
  default: primary
  required: true
  description: CleverTap account ID for API access
- name: clevertap_passcode
  type: string
  required: true
  description: CleverTap API passcode
- name: clevertap_api_endpoint
  type: string
  default: https://api.clevertap.com/1
  description: CleverTap API endpoint
- name: sales_drop_threshold
  type: float
  default: 0.2
  description: Threshold for sales drop detection (0.2 = 20% drop)
- name: sales_spike_threshold
  type: float
  default: 0.15
  description: Threshold for sales spike detection (0.15 = 15% increase)
- name: anomaly_sensitivity
  type: float
  default: 2.5
  description: Z-score sensitivity for anomaly detection
- name: analysis_window_days
  type: integer
  default: 7
  description: Days to analyze for trends
- name: target_region
  type: string
  default: NCR
  description: Region to analyze (NCR, Mumbai, Bangalore, or All)
- name: target_category
  type: string
  default: medicines
  description: Category to analyze (medicines, diagnostics, wellness, or all)
outputs:
  final_report:
    type: object
    source: generate_optimization_report
    description: Comprehensive funnel optimization report
  perception_data:
    type: object
    source: perception_data_bridge
    description: Combined data from CleverTap and internal systems
  campaign_results:
    type: object
    source: execute_recovery_actions
    description: Recovery campaign execution results
  anomaly_detection:
    type: object
    source: detect_analyze_anomalies
    description: Anomaly detection results and root cause analysis
  cross_sell_opportunities:
    type: object
    source: identify_cross_sell
    description: Identified cross-sell opportunities
version: 1.0
namespace: pharmeasy
description: Pharmeasy Autonomous Funnel Intelligence System - Real-time funnel optimization
  through Perception-Reasoning-Action loop
model_clients:
- id: funnel_master
  config:
    model: gpt-4o-mini
    api_key: ${OPENAI_API_KEY}
    max_tokens: 3000
    temperature: 0.3
  provider: openai
- id: anomaly_analyzer
  config:
    model: gpt-4o-mini
    api_key: ${OPENAI_API_KEY}
    max_tokens: 2000
    temperature: 0.1
  provider: openai
- id: retention_optimizer
  config:
    model: gpt-4o-mini
    api_key: ${OPENAI_API_KEY}
    max_tokens: 2500
    temperature: 0.2
  provider: openai
- id: campaign_creator
  config:
    model: gpt-4o-mini
    api_key: ${OPENAI_API_KEY}
    max_tokens: 2000
    temperature: 0.4
  provider: openai
timeout_seconds: 1800

No executions yet. Execute this workflow to see results here.