Gen AI - SMC AI Relationship Manager Workflow

AI-powered relationship management system for SMC's customer portfolio with cross-selling and retention capabilities

Back

Workflow Information

ID: smc_ai_relationship_manager_genai

Namespace: smc_finance

Version: 1.0

Created: 2025-07-17

Updated: 2025-07-17

Tasks: 14

Quick Actions

Manage Secrets

Inputs

Name	Type	Required	Default
`client_id`	string	Required	None
`analysis_depth`	string	Optional	`standard`
`enable_cross_selling`	boolean	Optional	`True`
`enable_retention_alerts`	boolean	Optional	`True`
`sub_broker_id`	string	Optional	None
`smc_api_base_url`	string	Optional	`https://api.smcindiaonline.com`

Outputs

Name	Type	Source
`client_analysis`	object	Complete client analysis results
`risk_assessment`	object	Client risk assessment and recommendations
`learning_outcomes`	object	Model learning and optimization results
`performance_metrics`	object	Workflow execution metrics and insights
`recommended_actions`	string	Recommended actions based on client profile

Tasks

collect_client_data

script

Gather client data from multiple SMC systems

Python Script:

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

# Simulate processing time
time.sleep(1)

# Get environment variables or use defaults
client_id = os.getenv("CLIENT_ID", "C001")
sub_broker_id = os.getenv("SUB_BROKER_ID", f"SB{random.randint(100, 999)}")

# Generate realistic dummy data based on client_id
client_profiles = {
    "C001": {
        "name": "Rajesh Kumar",
        "city": "Mumbai",
        "age": 45,
        "occupation": "Business Owner",
        "risk_profile": "MODERATE",
        "aum": 850000,
        "preferred_segments": ["equity", "mutual_funds"]
    },
    "C002": {
        "name": "Priya Sharma",
        "city": "Delhi",
        "age": 32,
        "occupation": "Software Engineer",
        "risk_profile": "AGGRESSIVE",
        "aum": 350000,
        "preferred_segments": ["equity", "derivatives"]
    },
    "C003": {
        "name": "Amit Patel",
        "city": "Ahmedabad",
        "age": 55,
        "occupation": "Doctor",
        "risk_profile": "CONSERVATIVE",
        "aum": 1200000,
        "preferred_segments": ["mutual_funds", "insurance"]
    }
}

# Default profile for unknown client_id
default_profile = {
    "name": f"Client_{client_id}",
    "city": "Bangalore",
    "age": 40,
    "occupation": "Professional",
    "risk_profile": "MODERATE",
    "aum": 500000,
    "preferred_segments": ["equity", "mutual_funds"]
}

profile = client_profiles.get(client_id, default_profile)

# Simulate comprehensive client data
client_data = {
    "client_id": client_id,
    "basic_info": {
        "name": profile["name"],
        "city": profile["city"],
        "age": profile["age"],
        "occupation": profile["occupation"],
        "registration_date": "2020-01-15",
        "kyc_status": "VERIFIED",
        "risk_profile": profile["risk_profile"],
        "sub_broker_id": sub_broker_id
    },
    "portfolio": {
        "equity_holdings": int(profile["aum"] * 0.6),
        "mutual_funds": int(profile["aum"] * 0.25),
        "insurance_policies": random.randint(1, 4),
        "active_loans": random.randint(0, 2),
        "total_aum": profile["aum"],
        "cash_balance": random.randint(10000, 100000)
    },
    "transaction_history": {
        "last_transaction_date": (datetime.now() - timedelta(days=random.randint(1, 30))).strftime("%Y-%m-%d"),
        "avg_monthly_volume": random.randint(25000, 150000),
        "preferred_segments": profile["preferred_segments"],
        "trading_frequency": random.choice(["low", "medium", "high"]),
        "total_trades_ytd": random.randint(15, 120)
    },
    "behavioral_data": {
        "login_frequency": random.choice(["daily", "weekly", "monthly"]),
        "mobile_app_usage": random.choice(["high", "medium", "low"]),
        "support_interactions": random.randint(0, 5),
        "webinar_attendance": random.randint(0, 10),
        "research_report_downloads": random.randint(5, 25)
    },
    "communication_preferences": {
        "channel": random.choice(["email", "sms", "whatsapp", "phone"]),
        "frequency": random.choice(["daily", "weekly", "monthly"]),
        "content_type": random.choice(["research_reports", "market_updates", "product_offers"])
    }
}

result = {
    "status": "success",
    "client_data": client_data,
    "data_sources": ["trading_system", "crm", "support_system", "mobile_app"],
    "collection_timestamp": datetime.now().isoformat(),
    "data_quality_score": round(random.uniform(0.8, 1.0), 2)
}

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

analyze_client_profile

script

AI-powered client profile analysis and segmentation

Python Script:

import json
import time
import random
from datetime import datetime

# Simulate AI processing time
time.sleep(1)

client_data = ${collect_client_data.client_data}

# Determine primary segment based on portfolio and behavior
aum = client_data["portfolio"]["total_aum"]
equity_ratio = client_data["portfolio"]["equity_holdings"] / aum
trading_freq = client_data["transaction_history"]["trading_frequency"]

# Segment classification logic
if aum >= 1000000:
    primary_segment = "HNI_CLIENT"
    segment_description = "High Net Worth Individual"
elif trading_freq == "high" and equity_ratio > 0.6:
    primary_segment = "ACTIVE_TRADER"
    segment_description = "Active Equity Trader"
elif "mutual_funds" in client_data["transaction_history"]["preferred_segments"]:
    primary_segment = "MF_INVESTOR"
    segment_description = "Mutual Fund Investor"
elif client_data["portfolio"]["insurance_policies"] >= 2:
    primary_segment = "INSURANCE_CLIENT"
    segment_description = "Insurance Focused Client"
elif client_data["portfolio"]["active_loans"] >= 1:
    primary_segment = "LOAN_CLIENT"
    segment_description = "Active Borrower"
else:
    primary_segment = "REGULAR_CLIENT"
    segment_description = "Regular Retail Client"

# Generate behavioral patterns
behavioral_patterns = [
    "Prefers long-term investments",
    "Responds well to research-based recommendations",
    "Price-sensitive to fees and charges",
    "Technology-savvy user",
    "Seeks regular market updates"
]

# Engagement level assessment
login_score = {"daily": 0.9, "weekly": 0.6, "monthly": 0.3}
app_usage_score = {"high": 0.8, "medium": 0.5, "low": 0.2}

engagement_score = (
    login_score.get(client_data["behavioral_data"]["login_frequency"], 0.5) * 0.4 +
    app_usage_score.get(client_data["behavioral_data"]["mobile_app_usage"], 0.5) * 0.3 +
    min(client_data["behavioral_data"]["webinar_attendance"] / 10, 1.0) * 0.3
)

result = {
    "analysis_timestamp": datetime.now().isoformat(),
    "primary_segment": primary_segment,
    "segment_description": segment_description,
    "secondary_characteristics": [
        f"Risk Profile: {client_data['basic_info']['risk_profile']}",
        f"AUM Tier: {'Premium' if aum >= 500000 else 'Standard'}",
        f"Trading Activity: {trading_freq.title()}"
    ],
    "behavioral_patterns": random.sample(behavioral_patterns, 3),
    "engagement_level": {
        "score": round(engagement_score, 2),
        "rating": "High" if engagement_score >= 0.7 else "Medium" if engagement_score >= 0.4 else "Low"
    },
    "investment_preferences": {
        "preferred_products": client_data["transaction_history"]["preferred_segments"],
        "risk_appetite": client_data["basic_info"]["risk_profile"],
        "investment_horizon": random.choice(["Short-term", "Medium-term", "Long-term"])
    },
    "confidence_score": round(random.uniform(0.75, 0.95), 2)
}

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

assess_client_risk

script

Comprehensive risk assessment for client relationship

Python Script:

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

# Simulate risk analysis processing
time.sleep(1)

client_data = ${collect_client_data.client_data}
profile_analysis = ${analyze_client_profile}

# Calculate risk factors
last_transaction = datetime.strptime(client_data["transaction_history"]["last_transaction_date"], "%Y-%m-%d")
days_since_last_transaction = (datetime.now() - last_transaction).days

# Churn risk calculation
activity_risk = min(days_since_last_transaction / 30, 1.0)  # 30 days = max risk
engagement_risk = 1.0 - profile_analysis["engagement_level"]["score"]
support_risk = min(client_data["behavioral_data"]["support_interactions"] / 5, 1.0)

churn_risk_score = (activity_risk * 0.4 + engagement_risk * 0.4 + support_risk * 0.2)

# Determine risk level
if churn_risk_score >= 0.7:
    risk_level = "HIGH"
    risk_description = "Immediate attention required"
elif churn_risk_score >= 0.4:
    risk_level = "MEDIUM"
    risk_description = "Monitor closely"
else:
    risk_level = "LOW"
    risk_description = "Stable relationship"

# Portfolio risk assessment
portfolio_concentration = max(client_data["portfolio"]["equity_holdings"] / client_data["portfolio"]["total_aum"], 0.3)
portfolio_risk = "HIGH" if portfolio_concentration > 0.8 else "MEDIUM" if portfolio_concentration > 0.6 else "LOW"

# Generate risk factors
risk_factors = []
if activity_risk > 0.5:
    risk_factors.append("Reduced trading activity")
if engagement_risk > 0.6:
    risk_factors.append("Low platform engagement")
if support_risk > 0.4:
    risk_factors.append("Frequent support requests")
if portfolio_concentration > 0.7:
    risk_factors.append("High portfolio concentration")

# Recommended interventions
interventions = {
    "HIGH": [
        "Immediate relationship manager contact",
        "Personalized retention offer",
        "Portfolio review meeting",
        "Competitive analysis"
    ],
    "MEDIUM": [
        "Enhanced engagement campaign",
        "Educational content delivery",
        "Regular market updates",
        "Product demonstrations"
    ],
    "LOW": [
        "Routine check-ins",
        "Cross-selling opportunities",
        "Loyalty program enrollment",
        "Referral incentives"
    ]
}

result = {
    "assessment_timestamp": datetime.now().isoformat(),
    "risk_level": risk_level,
    "risk_description": risk_description,
    "churn_risk_score": round(churn_risk_score, 2),
    "risk_factors": risk_factors if risk_factors else ["No significant risk factors identified"],
    "portfolio_risk": {
        "level": portfolio_risk,
        "concentration_ratio": round(portfolio_concentration, 2),
        "diversification_score": round(1 - portfolio_concentration, 2)
    },
    "relationship_health": {
        "score": round(1 - churn_risk_score, 2),
        "status": "Healthy" if churn_risk_score < 0.3 else "At Risk" if churn_risk_score < 0.7 else "Critical"
    },
    "recommended_interventions": interventions[risk_level],
    "next_review_date": (datetime.now() + timedelta(days=30 if risk_level == "LOW" else 14 if risk_level == "MEDIUM" else 7)).strftime("%Y-%m-%d")
}

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

risk_based_router

conditional_router

No description

Conditional Router

Router Type: condition
Default Route: standard_path

generate_retention_strategy

script

Create personalized retention strategy for high-risk clients

Python Script:

import json
import time
import random
from datetime import datetime

# Simulate strategy generation
time.sleep(1)

client_data = ${collect_client_data.client_data}
risk_assessment = ${assess_client_risk}

# Personalized retention offers based on client profile
aum = client_data["portfolio"]["total_aum"]

# Fee reduction offers
fee_reduction = {
    "brokerage_discount": "50% reduction for 6 months",
    "annual_fee_waiver": "Complete waiver for current year",
    "premium_services": "Free access to premium research"
}

# Service enhancements
service_enhancements = [
    "Dedicated relationship manager assignment",
    "Priority customer support",
    "Exclusive market insights",
    "Personalized portfolio review sessions",
    "Direct access to research analysts"
]

# Competitive analysis
competitive_offers = [
    "Rate matching guarantee",
    "Better margin rates",
    "Enhanced product offerings",
    "Technology platform improvements"
]

# Communication strategy
communication_plan = {
    "immediate_actions": [
        "Personal call within 24 hours",
        "Email with retention offer",
        "SMS with priority support number"
    ],
    "follow_up_schedule": [
        "Day 3: Check satisfaction with offer",
        "Week 1: Portfolio review meeting",
        "Week 2: Market insights sharing",
        "Month 1: Relationship health check"
    ],
    "escalation_triggers": [
        "No response within 48 hours",
        "Rejection of initial offer",
        "Continued low activity"
    ]
}

result = {
    "strategy_timestamp": datetime.now().isoformat(),
    "client_id": client_data["client_id"],
    "retention_priority": "CRITICAL",
    "personalized_offers": {
        "fee_reductions": fee_reduction,
        "value_adds": random.sample(service_enhancements, 3),
        "competitive_matching": random.sample(competitive_offers, 2)
    },
    "communication_strategy": communication_plan,
    "success_metrics": {
        "target_engagement_increase": "40%",
        "retention_probability": "75%",
        "timeline_to_stability": "30 days"
    },
    "budget_allocation": {
        "fee_concessions": f"₹{random.randint(5000, 25000)}",
        "service_costs": f"₹{random.randint(2000, 10000)}",
        "total_investment": f"₹{random.randint(7000, 35000)}"
    },
    "approval_required": True,
    "assigned_rm": f"RM{random.randint(100, 999)}"
}

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

create_engagement_campaign

script

Design engagement campaign for medium-risk clients

Python Script:

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

# Simulate campaign creation
time.sleep(1)

client_data = ${collect_client_data.client_data}
profile_analysis = ${analyze_client_profile}

# Educational content based on client interests
content_library = {
    "equity": [
        "Market Analysis Webinar",
        "Stock Picking Strategies",
        "Technical Analysis Basics",
        "Sector Rotation Insights"
    ],
    "mutual_funds": [
        "SIP Benefits Masterclass",
        "Fund Selection Guide",
        "Tax-Saving Opportunities",
        "Portfolio Rebalancing"
    ],
    "insurance": [
        "Life Insurance Planning",
        "Health Insurance Essentials",
        "Retirement Planning",
        "Tax Benefits of Insurance"
    ]
}

# Select relevant content
preferred_segments = client_data["transaction_history"]["preferred_segments"]
selected_content = []
for segment in preferred_segments:
    if segment in content_library:
        selected_content.extend(random.sample(content_library[segment], 2))

# Engagement activities
activities = [
    {
        "type": "webinar_invitation",
        "title": random.choice(selected_content),
        "date": (datetime.now() + timedelta(days=random.randint(3, 14))).strftime("%Y-%m-%d"),
        "priority": "HIGH"
    },
    {
        "type": "market_insights",
        "title": "Weekly Market Commentary",
        "frequency": "Weekly",
        "priority": "MEDIUM"
    },
    {
        "type": "product_demo",
        "title": "New Trading Platform Features",
        "date": (datetime.now() + timedelta(days=random.randint(7, 21))).strftime("%Y-%m-%d"),
        "priority": "LOW"
    },
    {
        "type": "portfolio_review",
        "title": "Quarterly Portfolio Health Check",
        "date": (datetime.now() + timedelta(days=random.randint(30, 60))).strftime("%Y-%m-%d"),
        "priority": "MEDIUM"
    }
]

# Communication schedule
communication_schedule = {
    "week_1": [
        "Welcome to enhanced engagement program",
        "Personalized market insights"
    ],
    "week_2": [
        "Webinar invitation",
        "Educational content delivery"
    ],
    "week_3": [
        "Portfolio optimization tips",
        "New product announcements"
    ],
    "week_4": [
        "Month-end review",
        "Satisfaction survey"
    ]
}

result = {
    "campaign_timestamp": datetime.now().isoformat(),
    "client_id": client_data["client_id"],
    "campaign_type": "ENGAGEMENT_BOOST",
    "duration": "30 days",
    "engagement_activities": activities,
    "content_recommendations": selected_content,
    "communication_schedule": communication_schedule,
    "success_metrics": {
        "target_engagement_increase": "25%",
        "webinar_attendance_goal": "2 sessions",
        "platform_usage_increase": "30%",
        "content_consumption_target": "5 pieces"
    },
    "personalization_elements": {
        "preferred_communication_time": "10:00 AM - 12:00 PM",
        "content_format": "Video + PDF summary",
        "interaction_style": "Educational focus"
    },
    "escalation_conditions": [
        "No engagement within 7 days",
        "Declined webinar invitations",
        "Reduced platform usage"
    ]
}

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

identify_upsell_opportunities

script

Analyze cross-selling and upselling opportunities for stable clients

Python Script:

import json
import time
import random
from datetime import datetime

# Simulate opportunity analysis
time.sleep(1)

client_data = ${collect_client_data.client_data}
profile_analysis = ${analyze_client_profile}

# Current product usage
current_products = []
portfolio = client_data["portfolio"]

if portfolio["equity_holdings"] > 0:
    current_products.append("equity_trading")
if portfolio["mutual_funds"] > 0:
    current_products.append("mutual_funds")
if portfolio["insurance_policies"] > 0:
    current_products.append("insurance")
if portfolio["active_loans"] > 0:
    current_products.append("loans")

# Cross-selling opportunities matrix
cross_sell_matrix = {
    "equity_trading": [
        {"product": "derivatives", "priority": "HIGH", "reason": "Natural progression for equity traders"},
        {"product": "mutual_funds", "priority": "MEDIUM", "reason": "Portfolio diversification"},
        {"product": "insurance", "priority": "LOW", "reason": "Risk protection"}
    ],
    "mutual_funds": [
        {"product": "sip_plus", "priority": "HIGH", "reason": "Systematic investment enhancement"},
        {"product": "insurance", "priority": "MEDIUM", "reason": "Complete financial planning"},
        {"product": "equity_trading", "priority": "LOW", "reason": "Direct market participation"}
    ],
    "insurance": [
        {"product": "investment_insurance", "priority": "HIGH", "reason": "Insurance + investment combo"},
        {"product": "mutual_funds", "priority": "MEDIUM", "reason": "Long-term wealth creation"},
        {"product": "loans", "priority": "LOW", "reason": "Against policy loans"}
    ],
    "loans": [
        {"product": "investment_products", "priority": "HIGH", "reason": "Wealth building alongside borrowing"},
        {"product": "insurance", "priority": "MEDIUM", "reason": "Loan protection coverage"},
        {"product": "savings_accounts", "priority": "LOW", "reason": "Better cash management"}
    ]
}

# Identify opportunities
opportunities = []
for product in current_products:
    if product in cross_sell_matrix:
        opportunities.extend(cross_sell_matrix[product])

# Remove duplicates and prioritize
unique_opportunities = []
seen_products = set()
for opp in opportunities:
    if opp["product"] not in seen_products:
        unique_opportunities.append(opp)
        seen_products.add(opp["product"])

# Product gap analysis
all_smc_products = [
    "equity_trading", "derivatives", "mutual_funds", "sip_plus",
    "insurance", "investment_insurance", "loans", "investment_products",
    "savings_accounts", "demat_services", "research_advisory"
]

missing_products = [p for p in all_smc_products if p not in current_products]

# Value propositions
value_propositions = {
    "derivatives": "Leverage your equity knowledge for enhanced returns",
    "mutual_funds": "Professional fund management for diversified growth",
    "sip_plus": "Automated investing with higher return potential",
    "insurance": "Protect your wealth while building more",
    "investment_insurance": "Dual benefit of protection and growth",
    "loans": "Unlock liquidity without selling investments",
    "investment_products": "Systematic wealth building solutions",
    "savings_accounts": "Higher returns on your idle cash",
    "demat_services": "Seamless investment management",
    "research_advisory": "Expert guidance for better decisions"
}

# Timing strategies
timing_recommendations = {
    "immediate": [opp for opp in unique_opportunities if opp["priority"] == "HIGH"],
    "next_quarter": [opp for opp in unique_opportunities if opp["priority"] == "MEDIUM"],
    "future": [opp for opp in unique_opportunities if opp["priority"] == "LOW"]
}

result = {
    "analysis_timestamp": datetime.now().isoformat(),
    "client_id": client_data["client_id"],
    "current_products": current_products,
    "primary_opportunities": unique_opportunities[:3],
    "product_gap_analysis": {
        "missing_products": missing_products,
        "penetration_score": round(len(current_products) / len(all_smc_products), 2),
        "expansion_potential": "HIGH" if len(missing_products) >= 5 else "MEDIUM"
    },
    "value_propositions": {
        opp["product"]: value_propositions.get(opp["product"], "Enhanced financial services")
        for opp in unique_opportunities[:3]
    },
    "timing_strategy": timing_recommendations,
    "revenue_potential": {
        "estimated_additional_aum": f"₹{random.randint(100000, 500000)}",
        "fee_income_potential": f"₹{random.randint(5000, 25000)} annually",
        "cross_sell_probability": f"{random.randint(35, 75)}%"
    },
    "recommended_approach": {
        "method": "Consultative selling with needs analysis",
        "timeline": "3-6 months phased approach",
        "success_factors": [
            "Demonstrate clear value proposition",
            "Address specific client needs",
            "Provide trial periods where applicable"
        ]
    }
}

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

consolidate_strategy_results

script

Gather and consolidate all strategy outputs from different paths

Python Script:

import json
import time
from datetime import datetime

# Simulate consolidation processing
time.sleep(1)

# Get the selected route
selected_route = "${risk_based_router.selected_route}"

# Initialize consolidated data structure
consolidated_data = {
    "consolidation_timestamp": datetime.now().isoformat(),
    "selected_route": selected_route,
    "all_strategies_generated": True,
    "consolidation_summary": {
        "total_strategies": 3,
        "active_strategy": selected_route,
        "processing_status": "completed"
    }
}

# Conditionally include strategy data based on availability
try:
    # Always try to include all strategy results for comprehensive analysis
    retention_strategy = ${generate_retention_strategy}
    consolidated_data["retention_strategy"] = retention_strategy
    consolidated_data["retention_available"] = True
except:
    consolidated_data["retention_strategy"] = None
    consolidated_data["retention_available"] = False

try:
    engagement_campaign = ${create_engagement_campaign}
    consolidated_data["engagement_campaign"] = engagement_campaign  
    consolidated_data["engagement_available"] = True
except:
    consolidated_data["engagement_campaign"] = None
    consolidated_data["engagement_available"] = False

try:
    upsell_opportunities = ${identify_upsell_opportunities}
    consolidated_data["upsell_opportunities"] = upsell_opportunities
    consolidated_data["upsell_available"] = True
except:
    consolidated_data["upsell_opportunities"] = None
    consolidated_data["upsell_available"] = False

# Determine primary strategy based on route
if selected_route == "retention_path":
    consolidated_data["primary_strategy"] = consolidated_data.get("retention_strategy", {})
    consolidated_data["strategy_type"] = "retention"
    consolidated_data["priority_level"] = "CRITICAL"
elif selected_route == "engagement_path":
    consolidated_data["primary_strategy"] = consolidated_data.get("engagement_campaign", {})
    consolidated_data["strategy_type"] = "engagement"
    consolidated_data["priority_level"] = "HIGH"
elif selected_route in ["upsell_path", "standard_path"]:
    consolidated_data["primary_strategy"] = consolidated_data.get("upsell_opportunities", {})
    consolidated_data["strategy_type"] = "upsell"
    consolidated_data["priority_level"] = "MEDIUM"
else:
    consolidated_data["primary_strategy"] = {}
    consolidated_data["strategy_type"] = "unknown"
    consolidated_data["priority_level"] = "LOW"

# Generate execution readiness assessment
execution_readiness = {
    "ready_for_execution": True,
    "primary_strategy_available": consolidated_data["primary_strategy"] is not None,
    "route_determined": selected_route != "unknown",
    "data_quality_score": 95.0,
    "next_steps": [
        "Execute campaign notifications",
        "Prepare client communications", 
        "Update client profiles",
        "Monitor response tracking"
    ]
}

consolidated_data["execution_readiness"] = execution_readiness

# Summary metrics
consolidated_data["metrics"] = {
    "strategies_processed": 3,
    "successful_generations": sum([
        consolidated_data["retention_available"],
        consolidated_data["engagement_available"], 
        consolidated_data["upsell_available"]
    ]),
    "consolidation_success_rate": "100%",
    "processing_time_seconds": 1
}

result = {
    "consolidation_completed": True,
    "consolidated_strategies": consolidated_data,
    "selected_route": selected_route,
    "primary_strategy_type": consolidated_data["strategy_type"],
    "execution_ready": execution_readiness["ready_for_execution"],
    "consolidation_timestamp": datetime.now().isoformat()
}

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

notify_sub_broker

script

Send notifications to sub-broker based on risk assessment

Python Script:

import json
import time
import random
from datetime import datetime

# Simulate notification processing
time.sleep(1)

sub_broker_id = "${sub_broker_id}" if "${sub_broker_id}" else "DEFAULT_BROKER"
client_id = "${client_id}" if "${client_id}" else "DEFAULT_CLIENT"

# Get consolidated strategy results
consolidated_data = ${consolidate_strategy_results.consolidated_strategies}
route = consolidated_data["selected_route"]
primary_strategy = consolidated_data["primary_strategy"]
strategy_type = consolidated_data["strategy_type"]
priority_level = consolidated_data["priority_level"]

# Generate notification based on route
if route == "retention_path":
    notification_type = "RETENTION_ALERT"
    priority = "CRITICAL"
    message = "Immediate client retention action required"
    recommendations = consolidated_data.get("retention_strategy", {})
    expected_response_time = "4 hours"
elif route == "engagement_path":
    notification_type = "ENGAGEMENT_OPPORTUNITY"
    priority = "HIGH"
    message = "Client engagement campaign ready for execution"
    recommendations = consolidated_data.get("engagement_campaign", {})
    expected_response_time = "24 hours"
else:
    notification_type = "UPSELL_OPPORTUNITY"
    priority = "MEDIUM"
    message = "Cross-selling opportunities identified"
    recommendations = consolidated_data.get("upsell_opportunities", {})
    expected_response_time = "3 days"

notification = {
    "notification_id": f"N{random.randint(10000, 99999)}",
    "sub_broker_id": sub_broker_id,
    "client_id": client_id,
    "notification_type": notification_type,
    "priority": priority,
    "message": message,
    "recommendations": recommendations,
    "expected_response_time": expected_response_time,
    "notification_channels": ["email", "sms", "dashboard"],
    "timestamp": datetime.now().isoformat(),
    "status": "SENT",
    "tracking_id": f"TRK{random.randint(100000, 999999)}"
}

result = {
    "notification_sent": True,
    "notification_details": notification,
    "delivery_status": "SUCCESS",
    "estimated_reach_time": "2 minutes"
}

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

prepare_client_communication

script

Prepare personalized client communications based on strategy

Python Script:

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

# Simulate communication preparation
time.sleep(1)

client_data = ${collect_client_data.client_data}

# Get consolidated strategy results
consolidated_data = ${consolidate_strategy_results.consolidated_strategies}
route = consolidated_data["selected_route"]

# Prepare communication based on route and client preferences
client_name = client_data["basic_info"]["name"]
preferred_channel = client_data["communication_preferences"]["channel"]

if route == "retention_path":
    message_type = "RETENTION_OFFER"
    subject = f"Exclusive benefits for you, {client_name}"
    content = f"Dear {client_name}, We value your relationship and have special offers designed just for you."
    urgency = "HIGH"
    call_to_action = "Call us at 1800-XXX-XXXX for immediate assistance"
elif route == "engagement_path":
    message_type = "ENGAGEMENT_INVITE"
    subject = f"Exciting opportunities await, {client_name}"
    content = f"Hi {client_name}, Discover new investment insights and exclusive content tailored for you."
    urgency = "MEDIUM"
    call_to_action = "Login to your account to explore"
else:
    message_type = "PRODUCT_RECOMMENDATION"
    subject = f"New ways to grow your wealth, {client_name}"
    content = f"Hello {client_name}, Based on your portfolio, we have identified opportunities for growth."
    urgency = "LOW"
    call_to_action = "Schedule a consultation with your advisor"

communication = {
    "communication_id": f"COM{random.randint(10000, 99999)}",
    "client_id": client_data["client_id"],
    "message_type": message_type,
    "channel": preferred_channel,
    "subject": subject,
    "content": content,
    "urgency": urgency,
    "call_to_action": call_to_action,
    "scheduled_time": (datetime.now() + timedelta(minutes=random.randint(30, 180))).isoformat(),
    "personalization_data": {
        "client_name": client_name,
        "preferred_products": client_data["transaction_history"]["preferred_segments"],
        "risk_profile": client_data["basic_info"]["risk_profile"],
        "city": client_data["basic_info"]["city"]
    },
    "tracking_enabled": True,
    "delivery_confirmation": True
}

result = {
    "communication_prepared": True,
    "communication_details": communication,
    "estimated_delivery_time": "15 minutes",
    "personalization_score": round(random.uniform(0.8, 1.0), 2)
}

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

update_client_profile

script

Update client profile with new insights and recommendations

Python Script:

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

# Simulate profile update processing
time.sleep(1)

client_id = "${client_id}" if "${client_id}" else "DEFAULT_CLIENT"

# Get consolidated strategy results
consolidated_data = ${consolidate_strategy_results.consolidated_strategies}
route = consolidated_data["selected_route"]

risk_assessment = ${assess_client_risk}
profile_analysis = ${analyze_client_profile}

# Update client profile with new insights
profile_updates = {
    "client_id": client_id,
    "last_analysis_date": datetime.now().isoformat(),
    "risk_level": risk_assessment["risk_level"],
    "risk_score": risk_assessment["churn_risk_score"],
    "segment": profile_analysis["primary_segment"],
    "engagement_score": profile_analysis["engagement_level"]["score"],
    "recommended_actions": route,
    "analysis_version": "1.0",
    "next_review_date": (datetime.now() + timedelta(days=30)).isoformat(),
    "tags": [
        f"risk_{risk_assessment['risk_level'].lower()}",
        f"segment_{profile_analysis['primary_segment'].lower()}",
        f"engagement_{profile_analysis['engagement_level']['rating'].lower()}"
    ],
    "ai_insights": {
        "confidence_score": profile_analysis["confidence_score"],
        "behavioral_patterns": profile_analysis["behavioral_patterns"],
        "investment_preferences": profile_analysis["investment_preferences"]
    },
    "action_history": {
        "last_action": route,
        "action_timestamp": datetime.now().isoformat(),
        "success_probability": f"{random.randint(60, 90)}%"
    }
}

result = {
    "profile_updated": True,
    "update_details": profile_updates,
    "update_timestamp": datetime.now().isoformat(),
    "fields_updated": 12,
    "data_quality_improvement": "+5%"
}

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

simulate_response_tracking

script

Track and analyze client responses to campaigns

Python Script:

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

# Simulate response tracking
time.sleep(1)

# Get consolidated strategy results
consolidated_data = ${consolidate_strategy_results.consolidated_strategies}
route = consolidated_data["selected_route"]
client_id = "${client_id}" if "${client_id}" else "DEFAULT_CLIENT"

# Response probabilities based on strategy type
response_probabilities = {
    "retention_path": {"response_rate": 0.75, "positive_rate": 0.65},
    "engagement_path": {"response_rate": 0.55, "positive_rate": 0.80},
    "upsell_path": {"response_rate": 0.35, "positive_rate": 0.45},
    "standard_path": {"response_rate": 0.40, "positive_rate": 0.50}
}

probs = response_probabilities.get(route, response_probabilities["standard_path"])

# Simulate response
has_response = random.random() < probs["response_rate"]
is_positive = random.random() < probs["positive_rate"] if has_response else False

# Response details
response_channels = ["email", "phone", "sms", "mobile_app", "web_portal"]
response_types = ["inquiry", "acceptance", "rejection", "deferral", "request_more_info"]

response_data = {
    "client_id": client_id,
    "campaign_route": route,
    "has_response": has_response,
    "is_positive": is_positive,
    "response_timestamp": datetime.now().isoformat(),
    "response_channel": random.choice(response_channels) if has_response else None,
    "response_type": random.choice(response_types) if has_response else None,
    "response_time_minutes": random.randint(30, 1440) if has_response else None,
    "engagement_score": round(random.uniform(0.3, 0.9), 2) if has_response else 0,
    "follow_up_required": has_response and not is_positive,
    "next_action": "schedule_call" if has_response and is_positive else "send_reminder" if has_response else "escalate"
}

# Additional metrics
tracking_metrics = {
    "email_open_rate": round(random.uniform(0.4, 0.8), 2),
    "click_through_rate": round(random.uniform(0.1, 0.3), 2),
    "time_to_response": f"{random.randint(1, 48)} hours" if has_response else "No response",
    "interaction_depth": random.choice(["surface", "moderate", "deep"]) if has_response else "none",
    "sentiment_score": round(random.uniform(0.2, 0.9), 2) if has_response else 0.5
}

result = {
    "response_tracked": True,
    "response_data": response_data,
    "tracking_metrics": tracking_metrics,
    "analysis_timestamp": datetime.now().isoformat(),
    "recommendation": "Continue monitoring" if is_positive else "Escalate to manager" if has_response else "Retry with different approach"
}

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

generate_performance_metrics

ai_agent

AI-powered comprehensive workflow performance metrics analysis

AI Agent Configuration:

Model Configuration

Model Client ID: gpt4_analytics
Agent Type: analyst

Input Format: json
Output Format: json

update_learning_model

script

Feed results back into AI learning system for continuous improvement

Python Script:

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

# Simulate model learning process
time.sleep(1)

# Get AI response and clean it to extract JSON
ai_response = """${generate_performance_metrics.ai_response}"""
# Remove markdown formatting
clean_response = ai_response.replace('```json', '').replace('```', '')
# Parse the JSON
try:
    ai_data = json.loads(clean_response.strip())
    performance_metrics = ai_data.get('performance_metrics', {})
except:
    performance_metrics = {}

client_data = ${collect_client_data.client_data}

# Prepare learning data
learning_data = {
    "client_segment": "${analyze_client_profile.primary_segment}",
    "risk_level": "${assess_client_risk.risk_level}",
    "actions_taken": "${risk_based_router.selected_route}",
    "response_outcome": ${simulate_response_tracking.response_data},
    "success_metrics": performance_metrics,
    "client_characteristics": {
        "aum": client_data["portfolio"]["total_aum"],
        "age": client_data["basic_info"]["age"],
        "risk_profile": client_data["basic_info"]["risk_profile"],
        "engagement_level": "${analyze_client_profile.engagement_level}"
    },
    "learning_timestamp": datetime.now().isoformat()
}

# Model improvement insights
model_insights = {
    "patterns_identified": [
        f"High AUM clients ({client_data['portfolio']['total_aum']}+) show {random.randint(15, 25)}% better response to retention offers",
        f"Clients in {client_data['basic_info']['city']} prefer {client_data['communication_preferences']['channel']} communication",
        f"{client_data['basic_info']['risk_profile']} risk profile clients respond best to {random.choice(['educational', 'promotional', 'advisory'])} content"
    ],
    "accuracy_improvements": {
        "risk_assessment_model": f"+{round(random.uniform(1.5, 4.2), 1)}%",
        "segmentation_model": f"+{round(random.uniform(0.8, 2.9), 1)}%",
        "recommendation_engine": f"+{round(random.uniform(2.1, 5.3), 1)}%",
        "response_prediction": f"+{round(random.uniform(1.2, 3.7), 1)}%"
    },
    "recommendations_refined": [
        "Adjust risk thresholds for better sensitivity",
        "Enhance personalization algorithms",
        "Improve timing prediction models",
        "Optimize communication channel selection"
    ]
}

# Algorithm updates
algorithm_updates = {
    "updated_algorithms": [
        "risk_assessment_model_v2.1",
        "client_segmentation_v1.8",
        "recommendation_engine_v3.2",
        "response_prediction_v1.5"
    ],
    "new_features": [
        "Dynamic risk scoring",
        "Behavioral pattern recognition",
        "Predictive engagement scoring",
        "Optimal timing recommendations"
    ],
    "model_performance": {
        "overall_accuracy": f"{round(random.uniform(87.5, 94.2), 1)}%",
        "precision": f"{round(random.uniform(82.3, 89.7), 1)}%",
        "recall": f"{round(random.uniform(78.9, 86.4), 1)}%",
        "f1_score": f"{round(random.uniform(80.1, 87.8), 1)}%"
    }
}

# Future optimization plans
optimization_plans = {
    "next_optimization_cycle": (datetime.now() + timedelta(days=7)).isoformat(),
    "focus_areas": [
        "Improve low-engagement client predictions",
        "Enhance cross-selling opportunity identification",
        "Optimize communication timing",
        "Refine retention strategy effectiveness"
    ],
    "data_requirements": [
        "More historical response data",
        "Enhanced behavioral tracking",
        "Market condition correlations",
        "Competitor analysis integration"
    ]
}

result = {
    "learning_completed": True,
    "learning_data": learning_data,
    "model_insights": model_insights,
    "algorithm_updates": algorithm_updates,
    "optimization_plans": optimization_plans,
    "learning_timestamp": datetime.now().isoformat(),
    "learning_confidence": round(random.uniform(0.85, 0.95), 2)
}

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

YAML Source

id: smc_ai_relationship_manager_genai
name: Gen AI - SMC AI Relationship Manager Workflow
retry:
  retryOn:
  - TEMPORARY_FAILURE
  - TIMEOUT
  - NETWORK_ERROR
  maxDelay: 60s
  maxAttempts: 3
  initialDelay: 2s
  backoffMultiplier: 2
tasks:
- id: collect_client_data
  name: Collect Client Data
  type: script
  script: "import json\nimport time\nimport random\nimport os\nfrom datetime import\
    \ datetime, timedelta\n\n# Simulate processing time\ntime.sleep(1)\n\n# Get environment\
    \ variables or use defaults\nclient_id = os.getenv(\"CLIENT_ID\", \"C001\")\n\
    sub_broker_id = os.getenv(\"SUB_BROKER_ID\", f\"SB{random.randint(100, 999)}\"\
    )\n\n# Generate realistic dummy data based on client_id\nclient_profiles = {\n\
    \    \"C001\": {\n        \"name\": \"Rajesh Kumar\",\n        \"city\": \"Mumbai\"\
    ,\n        \"age\": 45,\n        \"occupation\": \"Business Owner\",\n       \
    \ \"risk_profile\": \"MODERATE\",\n        \"aum\": 850000,\n        \"preferred_segments\"\
    : [\"equity\", \"mutual_funds\"]\n    },\n    \"C002\": {\n        \"name\": \"\
    Priya Sharma\",\n        \"city\": \"Delhi\",\n        \"age\": 32,\n        \"\
    occupation\": \"Software Engineer\",\n        \"risk_profile\": \"AGGRESSIVE\"\
    ,\n        \"aum\": 350000,\n        \"preferred_segments\": [\"equity\", \"derivatives\"\
    ]\n    },\n    \"C003\": {\n        \"name\": \"Amit Patel\",\n        \"city\"\
    : \"Ahmedabad\",\n        \"age\": 55,\n        \"occupation\": \"Doctor\",\n\
    \        \"risk_profile\": \"CONSERVATIVE\",\n        \"aum\": 1200000,\n    \
    \    \"preferred_segments\": [\"mutual_funds\", \"insurance\"]\n    }\n}\n\n#\
    \ Default profile for unknown client_id\ndefault_profile = {\n    \"name\": f\"\
    Client_{client_id}\",\n    \"city\": \"Bangalore\",\n    \"age\": 40,\n    \"\
    occupation\": \"Professional\",\n    \"risk_profile\": \"MODERATE\",\n    \"aum\"\
    : 500000,\n    \"preferred_segments\": [\"equity\", \"mutual_funds\"]\n}\n\nprofile\
    \ = client_profiles.get(client_id, default_profile)\n\n# Simulate comprehensive\
    \ client data\nclient_data = {\n    \"client_id\": client_id,\n    \"basic_info\"\
    : {\n        \"name\": profile[\"name\"],\n        \"city\": profile[\"city\"\
    ],\n        \"age\": profile[\"age\"],\n        \"occupation\": profile[\"occupation\"\
    ],\n        \"registration_date\": \"2020-01-15\",\n        \"kyc_status\": \"\
    VERIFIED\",\n        \"risk_profile\": profile[\"risk_profile\"],\n        \"\
    sub_broker_id\": sub_broker_id\n    },\n    \"portfolio\": {\n        \"equity_holdings\"\
    : int(profile[\"aum\"] * 0.6),\n        \"mutual_funds\": int(profile[\"aum\"\
    ] * 0.25),\n        \"insurance_policies\": random.randint(1, 4),\n        \"\
    active_loans\": random.randint(0, 2),\n        \"total_aum\": profile[\"aum\"\
    ],\n        \"cash_balance\": random.randint(10000, 100000)\n    },\n    \"transaction_history\"\
    : {\n        \"last_transaction_date\": (datetime.now() - timedelta(days=random.randint(1,\
    \ 30))).strftime(\"%Y-%m-%d\"),\n        \"avg_monthly_volume\": random.randint(25000,\
    \ 150000),\n        \"preferred_segments\": profile[\"preferred_segments\"],\n\
    \        \"trading_frequency\": random.choice([\"low\", \"medium\", \"high\"]),\n\
    \        \"total_trades_ytd\": random.randint(15, 120)\n    },\n    \"behavioral_data\"\
    : {\n        \"login_frequency\": random.choice([\"daily\", \"weekly\", \"monthly\"\
    ]),\n        \"mobile_app_usage\": random.choice([\"high\", \"medium\", \"low\"\
    ]),\n        \"support_interactions\": random.randint(0, 5),\n        \"webinar_attendance\"\
    : random.randint(0, 10),\n        \"research_report_downloads\": random.randint(5,\
    \ 25)\n    },\n    \"communication_preferences\": {\n        \"channel\": random.choice([\"\
    email\", \"sms\", \"whatsapp\", \"phone\"]),\n        \"frequency\": random.choice([\"\
    daily\", \"weekly\", \"monthly\"]),\n        \"content_type\": random.choice([\"\
    research_reports\", \"market_updates\", \"product_offers\"])\n    }\n}\n\nresult\
    \ = {\n    \"status\": \"success\",\n    \"client_data\": client_data,\n    \"\
    data_sources\": [\"trading_system\", \"crm\", \"support_system\", \"mobile_app\"\
    ],\n    \"collection_timestamp\": datetime.now().isoformat(),\n    \"data_quality_score\"\
    : round(random.uniform(0.8, 1.0), 2)\n}\n\nprint(f\"__OUTPUTS__ {json.dumps(result)}\"\
    )\n"
  description: Gather client data from multiple SMC systems
  timeout_seconds: 120
- id: analyze_client_profile
  name: Analyze Client Profile
  type: script
  script: "import json\nimport time\nimport random\nfrom datetime import datetime\n\
    \n# Simulate AI processing time\ntime.sleep(1)\n\nclient_data = ${collect_client_data.client_data}\n\
    \n# Determine primary segment based on portfolio and behavior\naum = client_data[\"\
    portfolio\"][\"total_aum\"]\nequity_ratio = client_data[\"portfolio\"][\"equity_holdings\"\
    ] / aum\ntrading_freq = client_data[\"transaction_history\"][\"trading_frequency\"\
    ]\n\n# Segment classification logic\nif aum >= 1000000:\n    primary_segment =\
    \ \"HNI_CLIENT\"\n    segment_description = \"High Net Worth Individual\"\nelif\
    \ trading_freq == \"high\" and equity_ratio > 0.6:\n    primary_segment = \"ACTIVE_TRADER\"\
    \n    segment_description = \"Active Equity Trader\"\nelif \"mutual_funds\" in\
    \ client_data[\"transaction_history\"][\"preferred_segments\"]:\n    primary_segment\
    \ = \"MF_INVESTOR\"\n    segment_description = \"Mutual Fund Investor\"\nelif\
    \ client_data[\"portfolio\"][\"insurance_policies\"] >= 2:\n    primary_segment\
    \ = \"INSURANCE_CLIENT\"\n    segment_description = \"Insurance Focused Client\"\
    \nelif client_data[\"portfolio\"][\"active_loans\"] >= 1:\n    primary_segment\
    \ = \"LOAN_CLIENT\"\n    segment_description = \"Active Borrower\"\nelse:\n  \
    \  primary_segment = \"REGULAR_CLIENT\"\n    segment_description = \"Regular Retail\
    \ Client\"\n\n# Generate behavioral patterns\nbehavioral_patterns = [\n    \"\
    Prefers long-term investments\",\n    \"Responds well to research-based recommendations\"\
    ,\n    \"Price-sensitive to fees and charges\",\n    \"Technology-savvy user\"\
    ,\n    \"Seeks regular market updates\"\n]\n\n# Engagement level assessment\n\
    login_score = {\"daily\": 0.9, \"weekly\": 0.6, \"monthly\": 0.3}\napp_usage_score\
    \ = {\"high\": 0.8, \"medium\": 0.5, \"low\": 0.2}\n\nengagement_score = (\n \
    \   login_score.get(client_data[\"behavioral_data\"][\"login_frequency\"], 0.5)\
    \ * 0.4 +\n    app_usage_score.get(client_data[\"behavioral_data\"][\"mobile_app_usage\"\
    ], 0.5) * 0.3 +\n    min(client_data[\"behavioral_data\"][\"webinar_attendance\"\
    ] / 10, 1.0) * 0.3\n)\n\nresult = {\n    \"analysis_timestamp\": datetime.now().isoformat(),\n\
    \    \"primary_segment\": primary_segment,\n    \"segment_description\": segment_description,\n\
    \    \"secondary_characteristics\": [\n        f\"Risk Profile: {client_data['basic_info']['risk_profile']}\"\
    ,\n        f\"AUM Tier: {'Premium' if aum >= 500000 else 'Standard'}\",\n    \
    \    f\"Trading Activity: {trading_freq.title()}\"\n    ],\n    \"behavioral_patterns\"\
    : random.sample(behavioral_patterns, 3),\n    \"engagement_level\": {\n      \
    \  \"score\": round(engagement_score, 2),\n        \"rating\": \"High\" if engagement_score\
    \ >= 0.7 else \"Medium\" if engagement_score >= 0.4 else \"Low\"\n    },\n   \
    \ \"investment_preferences\": {\n        \"preferred_products\": client_data[\"\
    transaction_history\"][\"preferred_segments\"],\n        \"risk_appetite\": client_data[\"\
    basic_info\"][\"risk_profile\"],\n        \"investment_horizon\": random.choice([\"\
    Short-term\", \"Medium-term\", \"Long-term\"])\n    },\n    \"confidence_score\"\
    : round(random.uniform(0.75, 0.95), 2)\n}\n\nprint(f\"__OUTPUTS__ {json.dumps(result)}\"\
    )\n"
  depends_on:
  - collect_client_data
  description: AI-powered client profile analysis and segmentation
  previous_node: collect_client_data
- id: assess_client_risk
  name: Assess Client Risk
  type: script
  script: "import json\nimport time\nimport random\nfrom datetime import datetime,\
    \ timedelta\n\n# Simulate risk analysis processing\ntime.sleep(1)\n\nclient_data\
    \ = ${collect_client_data.client_data}\nprofile_analysis = ${analyze_client_profile}\n\
    \n# Calculate risk factors\nlast_transaction = datetime.strptime(client_data[\"\
    transaction_history\"][\"last_transaction_date\"], \"%Y-%m-%d\")\ndays_since_last_transaction\
    \ = (datetime.now() - last_transaction).days\n\n# Churn risk calculation\nactivity_risk\
    \ = min(days_since_last_transaction / 30, 1.0)  # 30 days = max risk\nengagement_risk\
    \ = 1.0 - profile_analysis[\"engagement_level\"][\"score\"]\nsupport_risk = min(client_data[\"\
    behavioral_data\"][\"support_interactions\"] / 5, 1.0)\n\nchurn_risk_score = (activity_risk\
    \ * 0.4 + engagement_risk * 0.4 + support_risk * 0.2)\n\n# Determine risk level\n\
    if churn_risk_score >= 0.7:\n    risk_level = \"HIGH\"\n    risk_description =\
    \ \"Immediate attention required\"\nelif churn_risk_score >= 0.4:\n    risk_level\
    \ = \"MEDIUM\"\n    risk_description = \"Monitor closely\"\nelse:\n    risk_level\
    \ = \"LOW\"\n    risk_description = \"Stable relationship\"\n\n# Portfolio risk\
    \ assessment\nportfolio_concentration = max(client_data[\"portfolio\"][\"equity_holdings\"\
    ] / client_data[\"portfolio\"][\"total_aum\"], 0.3)\nportfolio_risk = \"HIGH\"\
    \ if portfolio_concentration > 0.8 else \"MEDIUM\" if portfolio_concentration\
    \ > 0.6 else \"LOW\"\n\n# Generate risk factors\nrisk_factors = []\nif activity_risk\
    \ > 0.5:\n    risk_factors.append(\"Reduced trading activity\")\nif engagement_risk\
    \ > 0.6:\n    risk_factors.append(\"Low platform engagement\")\nif support_risk\
    \ > 0.4:\n    risk_factors.append(\"Frequent support requests\")\nif portfolio_concentration\
    \ > 0.7:\n    risk_factors.append(\"High portfolio concentration\")\n\n# Recommended\
    \ interventions\ninterventions = {\n    \"HIGH\": [\n        \"Immediate relationship\
    \ manager contact\",\n        \"Personalized retention offer\",\n        \"Portfolio\
    \ review meeting\",\n        \"Competitive analysis\"\n    ],\n    \"MEDIUM\"\
    : [\n        \"Enhanced engagement campaign\",\n        \"Educational content\
    \ delivery\",\n        \"Regular market updates\",\n        \"Product demonstrations\"\
    \n    ],\n    \"LOW\": [\n        \"Routine check-ins\",\n        \"Cross-selling\
    \ opportunities\",\n        \"Loyalty program enrollment\",\n        \"Referral\
    \ incentives\"\n    ]\n}\n\nresult = {\n    \"assessment_timestamp\": datetime.now().isoformat(),\n\
    \    \"risk_level\": risk_level,\n    \"risk_description\": risk_description,\n\
    \    \"churn_risk_score\": round(churn_risk_score, 2),\n    \"risk_factors\":\
    \ risk_factors if risk_factors else [\"No significant risk factors identified\"\
    ],\n    \"portfolio_risk\": {\n        \"level\": portfolio_risk,\n        \"\
    concentration_ratio\": round(portfolio_concentration, 2),\n        \"diversification_score\"\
    : round(1 - portfolio_concentration, 2)\n    },\n    \"relationship_health\":\
    \ {\n        \"score\": round(1 - churn_risk_score, 2),\n        \"status\": \"\
    Healthy\" if churn_risk_score < 0.3 else \"At Risk\" if churn_risk_score < 0.7\
    \ else \"Critical\"\n    },\n    \"recommended_interventions\": interventions[risk_level],\n\
    \    \"next_review_date\": (datetime.now() + timedelta(days=30 if risk_level ==\
    \ \"LOW\" else 14 if risk_level == \"MEDIUM\" else 7)).strftime(\"%Y-%m-%d\")\n\
    }\n\nprint(f\"__OUTPUTS__ {json.dumps(result)}\")\n"
  depends_on:
  - analyze_client_profile
  - collect_client_data
  description: Comprehensive risk assessment for client relationship
  previous_node: analyze_client_profile
- id: risk_based_router
  type: conditional_router
  conditions:
  - name: high_risk_client
    route: retention_path
    condition: ${assess_client_risk.risk_level} == 'HIGH'
  - name: medium_risk_client
    route: engagement_path
    condition: ${assess_client_risk.risk_level} == 'MEDIUM'
  - name: low_risk_client
    route: upsell_path
    condition: ${assess_client_risk.risk_level} == 'LOW'
  depends_on:
  - assess_client_risk
  default_route: standard_path
  previous_node: assess_client_risk
- id: generate_retention_strategy
  name: Generate Retention Strategy
  type: script
  script: "import json\nimport time\nimport random\nfrom datetime import datetime\n\
    \n# Simulate strategy generation\ntime.sleep(1)\n\nclient_data = ${collect_client_data.client_data}\n\
    risk_assessment = ${assess_client_risk}\n\n# Personalized retention offers based\
    \ on client profile\naum = client_data[\"portfolio\"][\"total_aum\"]\n\n# Fee\
    \ reduction offers\nfee_reduction = {\n    \"brokerage_discount\": \"50% reduction\
    \ for 6 months\",\n    \"annual_fee_waiver\": \"Complete waiver for current year\"\
    ,\n    \"premium_services\": \"Free access to premium research\"\n}\n\n# Service\
    \ enhancements\nservice_enhancements = [\n    \"Dedicated relationship manager\
    \ assignment\",\n    \"Priority customer support\",\n    \"Exclusive market insights\"\
    ,\n    \"Personalized portfolio review sessions\",\n    \"Direct access to research\
    \ analysts\"\n]\n\n# Competitive analysis\ncompetitive_offers = [\n    \"Rate\
    \ matching guarantee\",\n    \"Better margin rates\",\n    \"Enhanced product\
    \ offerings\",\n    \"Technology platform improvements\"\n]\n\n# Communication\
    \ strategy\ncommunication_plan = {\n    \"immediate_actions\": [\n        \"Personal\
    \ call within 24 hours\",\n        \"Email with retention offer\",\n        \"\
    SMS with priority support number\"\n    ],\n    \"follow_up_schedule\": [\n  \
    \      \"Day 3: Check satisfaction with offer\",\n        \"Week 1: Portfolio\
    \ review meeting\",\n        \"Week 2: Market insights sharing\",\n        \"\
    Month 1: Relationship health check\"\n    ],\n    \"escalation_triggers\": [\n\
    \        \"No response within 48 hours\",\n        \"Rejection of initial offer\"\
    ,\n        \"Continued low activity\"\n    ]\n}\n\nresult = {\n    \"strategy_timestamp\"\
    : datetime.now().isoformat(),\n    \"client_id\": client_data[\"client_id\"],\n\
    \    \"retention_priority\": \"CRITICAL\",\n    \"personalized_offers\": {\n \
    \       \"fee_reductions\": fee_reduction,\n        \"value_adds\": random.sample(service_enhancements,\
    \ 3),\n        \"competitive_matching\": random.sample(competitive_offers, 2)\n\
    \    },\n    \"communication_strategy\": communication_plan,\n    \"success_metrics\"\
    : {\n        \"target_engagement_increase\": \"40%\",\n        \"retention_probability\"\
    : \"75%\",\n        \"timeline_to_stability\": \"30 days\"\n    },\n    \"budget_allocation\"\
    : {\n        \"fee_concessions\": f\"\u20B9{random.randint(5000, 25000)}\",\n\
    \        \"service_costs\": f\"\u20B9{random.randint(2000, 10000)}\",\n      \
    \  \"total_investment\": f\"\u20B9{random.randint(7000, 35000)}\"\n    },\n  \
    \  \"approval_required\": True,\n    \"assigned_rm\": f\"RM{random.randint(100,\
    \ 999)}\"\n}\n\nprint(f\"__OUTPUTS__ {json.dumps(result)}\")\n"
  depends_on:
  - risk_based_router
  - collect_client_data
  - assess_client_risk
  description: Create personalized retention strategy for high-risk clients
  previous_node: risk_based_router
  execute_on_routes:
  - retention_path
- id: create_engagement_campaign
  name: Create Engagement Campaign
  type: script
  script: "import json\nimport time\nimport random\nfrom datetime import datetime,\
    \ timedelta\n\n# Simulate campaign creation\ntime.sleep(1)\n\nclient_data = ${collect_client_data.client_data}\n\
    profile_analysis = ${analyze_client_profile}\n\n# Educational content based on\
    \ client interests\ncontent_library = {\n    \"equity\": [\n        \"Market Analysis\
    \ Webinar\",\n        \"Stock Picking Strategies\",\n        \"Technical Analysis\
    \ Basics\",\n        \"Sector Rotation Insights\"\n    ],\n    \"mutual_funds\"\
    : [\n        \"SIP Benefits Masterclass\",\n        \"Fund Selection Guide\",\n\
    \        \"Tax-Saving Opportunities\",\n        \"Portfolio Rebalancing\"\n  \
    \  ],\n    \"insurance\": [\n        \"Life Insurance Planning\",\n        \"\
    Health Insurance Essentials\",\n        \"Retirement Planning\",\n        \"Tax\
    \ Benefits of Insurance\"\n    ]\n}\n\n# Select relevant content\npreferred_segments\
    \ = client_data[\"transaction_history\"][\"preferred_segments\"]\nselected_content\
    \ = []\nfor segment in preferred_segments:\n    if segment in content_library:\n\
    \        selected_content.extend(random.sample(content_library[segment], 2))\n\
    \n# Engagement activities\nactivities = [\n    {\n        \"type\": \"webinar_invitation\"\
    ,\n        \"title\": random.choice(selected_content),\n        \"date\": (datetime.now()\
    \ + timedelta(days=random.randint(3, 14))).strftime(\"%Y-%m-%d\"),\n        \"\
    priority\": \"HIGH\"\n    },\n    {\n        \"type\": \"market_insights\",\n\
    \        \"title\": \"Weekly Market Commentary\",\n        \"frequency\": \"Weekly\"\
    ,\n        \"priority\": \"MEDIUM\"\n    },\n    {\n        \"type\": \"product_demo\"\
    ,\n        \"title\": \"New Trading Platform Features\",\n        \"date\": (datetime.now()\
    \ + timedelta(days=random.randint(7, 21))).strftime(\"%Y-%m-%d\"),\n        \"\
    priority\": \"LOW\"\n    },\n    {\n        \"type\": \"portfolio_review\",\n\
    \        \"title\": \"Quarterly Portfolio Health Check\",\n        \"date\": (datetime.now()\
    \ + timedelta(days=random.randint(30, 60))).strftime(\"%Y-%m-%d\"),\n        \"\
    priority\": \"MEDIUM\"\n    }\n]\n\n# Communication schedule\ncommunication_schedule\
    \ = {\n    \"week_1\": [\n        \"Welcome to enhanced engagement program\",\n\
    \        \"Personalized market insights\"\n    ],\n    \"week_2\": [\n       \
    \ \"Webinar invitation\",\n        \"Educational content delivery\"\n    ],\n\
    \    \"week_3\": [\n        \"Portfolio optimization tips\",\n        \"New product\
    \ announcements\"\n    ],\n    \"week_4\": [\n        \"Month-end review\",\n\
    \        \"Satisfaction survey\"\n    ]\n}\n\nresult = {\n    \"campaign_timestamp\"\
    : datetime.now().isoformat(),\n    \"client_id\": client_data[\"client_id\"],\n\
    \    \"campaign_type\": \"ENGAGEMENT_BOOST\",\n    \"duration\": \"30 days\",\n\
    \    \"engagement_activities\": activities,\n    \"content_recommendations\":\
    \ selected_content,\n    \"communication_schedule\": communication_schedule,\n\
    \    \"success_metrics\": {\n        \"target_engagement_increase\": \"25%\",\n\
    \        \"webinar_attendance_goal\": \"2 sessions\",\n        \"platform_usage_increase\"\
    : \"30%\",\n        \"content_consumption_target\": \"5 pieces\"\n    },\n   \
    \ \"personalization_elements\": {\n        \"preferred_communication_time\": \"\
    10:00 AM - 12:00 PM\",\n        \"content_format\": \"Video + PDF summary\",\n\
    \        \"interaction_style\": \"Educational focus\"\n    },\n    \"escalation_conditions\"\
    : [\n        \"No engagement within 7 days\",\n        \"Declined webinar invitations\"\
    ,\n        \"Reduced platform usage\"\n    ]\n}\n\nprint(f\"__OUTPUTS__ {json.dumps(result)}\"\
    )\n"
  depends_on:
  - risk_based_router
  - collect_client_data
  - analyze_client_profile
  description: Design engagement campaign for medium-risk clients
  previous_node: risk_based_router
  execute_on_routes:
  - engagement_path
- id: identify_upsell_opportunities
  name: Identify Upsell Opportunities
  type: script
  script: "import json\nimport time\nimport random\nfrom datetime import datetime\n\
    \n# Simulate opportunity analysis\ntime.sleep(1)\n\nclient_data = ${collect_client_data.client_data}\n\
    profile_analysis = ${analyze_client_profile}\n\n# Current product usage\ncurrent_products\
    \ = []\nportfolio = client_data[\"portfolio\"]\n\nif portfolio[\"equity_holdings\"\
    ] > 0:\n    current_products.append(\"equity_trading\")\nif portfolio[\"mutual_funds\"\
    ] > 0:\n    current_products.append(\"mutual_funds\")\nif portfolio[\"insurance_policies\"\
    ] > 0:\n    current_products.append(\"insurance\")\nif portfolio[\"active_loans\"\
    ] > 0:\n    current_products.append(\"loans\")\n\n# Cross-selling opportunities\
    \ matrix\ncross_sell_matrix = {\n    \"equity_trading\": [\n        {\"product\"\
    : \"derivatives\", \"priority\": \"HIGH\", \"reason\": \"Natural progression for\
    \ equity traders\"},\n        {\"product\": \"mutual_funds\", \"priority\": \"\
    MEDIUM\", \"reason\": \"Portfolio diversification\"},\n        {\"product\": \"\
    insurance\", \"priority\": \"LOW\", \"reason\": \"Risk protection\"}\n    ],\n\
    \    \"mutual_funds\": [\n        {\"product\": \"sip_plus\", \"priority\": \"\
    HIGH\", \"reason\": \"Systematic investment enhancement\"},\n        {\"product\"\
    : \"insurance\", \"priority\": \"MEDIUM\", \"reason\": \"Complete financial planning\"\
    },\n        {\"product\": \"equity_trading\", \"priority\": \"LOW\", \"reason\"\
    : \"Direct market participation\"}\n    ],\n    \"insurance\": [\n        {\"\
    product\": \"investment_insurance\", \"priority\": \"HIGH\", \"reason\": \"Insurance\
    \ + investment combo\"},\n        {\"product\": \"mutual_funds\", \"priority\"\
    : \"MEDIUM\", \"reason\": \"Long-term wealth creation\"},\n        {\"product\"\
    : \"loans\", \"priority\": \"LOW\", \"reason\": \"Against policy loans\"}\n  \
    \  ],\n    \"loans\": [\n        {\"product\": \"investment_products\", \"priority\"\
    : \"HIGH\", \"reason\": \"Wealth building alongside borrowing\"},\n        {\"\
    product\": \"insurance\", \"priority\": \"MEDIUM\", \"reason\": \"Loan protection\
    \ coverage\"},\n        {\"product\": \"savings_accounts\", \"priority\": \"LOW\"\
    , \"reason\": \"Better cash management\"}\n    ]\n}\n\n# Identify opportunities\n\
    opportunities = []\nfor product in current_products:\n    if product in cross_sell_matrix:\n\
    \        opportunities.extend(cross_sell_matrix[product])\n\n# Remove duplicates\
    \ and prioritize\nunique_opportunities = []\nseen_products = set()\nfor opp in\
    \ opportunities:\n    if opp[\"product\"] not in seen_products:\n        unique_opportunities.append(opp)\n\
    \        seen_products.add(opp[\"product\"])\n\n# Product gap analysis\nall_smc_products\
    \ = [\n    \"equity_trading\", \"derivatives\", \"mutual_funds\", \"sip_plus\"\
    ,\n    \"insurance\", \"investment_insurance\", \"loans\", \"investment_products\"\
    ,\n    \"savings_accounts\", \"demat_services\", \"research_advisory\"\n]\n\n\
    missing_products = [p for p in all_smc_products if p not in current_products]\n\
    \n# Value propositions\nvalue_propositions = {\n    \"derivatives\": \"Leverage\
    \ your equity knowledge for enhanced returns\",\n    \"mutual_funds\": \"Professional\
    \ fund management for diversified growth\",\n    \"sip_plus\": \"Automated investing\
    \ with higher return potential\",\n    \"insurance\": \"Protect your wealth while\
    \ building more\",\n    \"investment_insurance\": \"Dual benefit of protection\
    \ and growth\",\n    \"loans\": \"Unlock liquidity without selling investments\"\
    ,\n    \"investment_products\": \"Systematic wealth building solutions\",\n  \
    \  \"savings_accounts\": \"Higher returns on your idle cash\",\n    \"demat_services\"\
    : \"Seamless investment management\",\n    \"research_advisory\": \"Expert guidance\
    \ for better decisions\"\n}\n\n# Timing strategies\ntiming_recommendations = {\n\
    \    \"immediate\": [opp for opp in unique_opportunities if opp[\"priority\"]\
    \ == \"HIGH\"],\n    \"next_quarter\": [opp for opp in unique_opportunities if\
    \ opp[\"priority\"] == \"MEDIUM\"],\n    \"future\": [opp for opp in unique_opportunities\
    \ if opp[\"priority\"] == \"LOW\"]\n}\n\nresult = {\n    \"analysis_timestamp\"\
    : datetime.now().isoformat(),\n    \"client_id\": client_data[\"client_id\"],\n\
    \    \"current_products\": current_products,\n    \"primary_opportunities\": unique_opportunities[:3],\n\
    \    \"product_gap_analysis\": {\n        \"missing_products\": missing_products,\n\
    \        \"penetration_score\": round(len(current_products) / len(all_smc_products),\
    \ 2),\n        \"expansion_potential\": \"HIGH\" if len(missing_products) >= 5\
    \ else \"MEDIUM\"\n    },\n    \"value_propositions\": {\n        opp[\"product\"\
    ]: value_propositions.get(opp[\"product\"], \"Enhanced financial services\")\n\
    \        for opp in unique_opportunities[:3]\n    },\n    \"timing_strategy\"\
    : timing_recommendations,\n    \"revenue_potential\": {\n        \"estimated_additional_aum\"\
    : f\"\u20B9{random.randint(100000, 500000)}\",\n        \"fee_income_potential\"\
    : f\"\u20B9{random.randint(5000, 25000)} annually\",\n        \"cross_sell_probability\"\
    : f\"{random.randint(35, 75)}%\"\n    },\n    \"recommended_approach\": {\n  \
    \      \"method\": \"Consultative selling with needs analysis\",\n        \"timeline\"\
    : \"3-6 months phased approach\",\n        \"success_factors\": [\n          \
    \  \"Demonstrate clear value proposition\",\n            \"Address specific client\
    \ needs\",\n            \"Provide trial periods where applicable\"\n        ]\n\
    \    }\n}\n\nprint(f\"__OUTPUTS__ {json.dumps(result)}\")\n"
  depends_on:
  - risk_based_router
  - collect_client_data
  - analyze_client_profile
  description: Analyze cross-selling and upselling opportunities for stable clients
  previous_node: risk_based_router
  execute_on_routes:
  - upsell_path
  - standard_path
- id: consolidate_strategy_results
  name: Consolidate Strategy Results
  type: script
  script: "import json\nimport time\nfrom datetime import datetime\n\n# Simulate consolidation\
    \ processing\ntime.sleep(1)\n\n# Get the selected route\nselected_route = \"${risk_based_router.selected_route}\"\
    \n\n# Initialize consolidated data structure\nconsolidated_data = {\n    \"consolidation_timestamp\"\
    : datetime.now().isoformat(),\n    \"selected_route\": selected_route,\n    \"\
    all_strategies_generated\": True,\n    \"consolidation_summary\": {\n        \"\
    total_strategies\": 3,\n        \"active_strategy\": selected_route,\n       \
    \ \"processing_status\": \"completed\"\n    }\n}\n\n# Conditionally include strategy\
    \ data based on availability\ntry:\n    # Always try to include all strategy results\
    \ for comprehensive analysis\n    retention_strategy = ${generate_retention_strategy}\n\
    \    consolidated_data[\"retention_strategy\"] = retention_strategy\n    consolidated_data[\"\
    retention_available\"] = True\nexcept:\n    consolidated_data[\"retention_strategy\"\
    ] = None\n    consolidated_data[\"retention_available\"] = False\n\ntry:\n   \
    \ engagement_campaign = ${create_engagement_campaign}\n    consolidated_data[\"\
    engagement_campaign\"] = engagement_campaign  \n    consolidated_data[\"engagement_available\"\
    ] = True\nexcept:\n    consolidated_data[\"engagement_campaign\"] = None\n   \
    \ consolidated_data[\"engagement_available\"] = False\n\ntry:\n    upsell_opportunities\
    \ = ${identify_upsell_opportunities}\n    consolidated_data[\"upsell_opportunities\"\
    ] = upsell_opportunities\n    consolidated_data[\"upsell_available\"] = True\n\
    except:\n    consolidated_data[\"upsell_opportunities\"] = None\n    consolidated_data[\"\
    upsell_available\"] = False\n\n# Determine primary strategy based on route\nif\
    \ selected_route == \"retention_path\":\n    consolidated_data[\"primary_strategy\"\
    ] = consolidated_data.get(\"retention_strategy\", {})\n    consolidated_data[\"\
    strategy_type\"] = \"retention\"\n    consolidated_data[\"priority_level\"] =\
    \ \"CRITICAL\"\nelif selected_route == \"engagement_path\":\n    consolidated_data[\"\
    primary_strategy\"] = consolidated_data.get(\"engagement_campaign\", {})\n   \
    \ consolidated_data[\"strategy_type\"] = \"engagement\"\n    consolidated_data[\"\
    priority_level\"] = \"HIGH\"\nelif selected_route in [\"upsell_path\", \"standard_path\"\
    ]:\n    consolidated_data[\"primary_strategy\"] = consolidated_data.get(\"upsell_opportunities\"\
    , {})\n    consolidated_data[\"strategy_type\"] = \"upsell\"\n    consolidated_data[\"\
    priority_level\"] = \"MEDIUM\"\nelse:\n    consolidated_data[\"primary_strategy\"\
    ] = {}\n    consolidated_data[\"strategy_type\"] = \"unknown\"\n    consolidated_data[\"\
    priority_level\"] = \"LOW\"\n\n# Generate execution readiness assessment\nexecution_readiness\
    \ = {\n    \"ready_for_execution\": True,\n    \"primary_strategy_available\"\
    : consolidated_data[\"primary_strategy\"] is not None,\n    \"route_determined\"\
    : selected_route != \"unknown\",\n    \"data_quality_score\": 95.0,\n    \"next_steps\"\
    : [\n        \"Execute campaign notifications\",\n        \"Prepare client communications\"\
    , \n        \"Update client profiles\",\n        \"Monitor response tracking\"\
    \n    ]\n}\n\nconsolidated_data[\"execution_readiness\"] = execution_readiness\n\
    \n# Summary metrics\nconsolidated_data[\"metrics\"] = {\n    \"strategies_processed\"\
    : 3,\n    \"successful_generations\": sum([\n        consolidated_data[\"retention_available\"\
    ],\n        consolidated_data[\"engagement_available\"], \n        consolidated_data[\"\
    upsell_available\"]\n    ]),\n    \"consolidation_success_rate\": \"100%\",\n\
    \    \"processing_time_seconds\": 1\n}\n\nresult = {\n    \"consolidation_completed\"\
    : True,\n    \"consolidated_strategies\": consolidated_data,\n    \"selected_route\"\
    : selected_route,\n    \"primary_strategy_type\": consolidated_data[\"strategy_type\"\
    ],\n    \"execution_ready\": execution_readiness[\"ready_for_execution\"],\n \
    \   \"consolidation_timestamp\": datetime.now().isoformat()\n}\n\nprint(f\"__OUTPUTS__\
    \ {json.dumps(result)}\")\n"
  depends_on:
  - generate_retention_strategy
  - create_engagement_campaign
  - identify_upsell_opportunities
  - risk_based_router
  description: Gather and consolidate all strategy outputs from different paths
  previous_node:
  - generate_retention_strategy
  - create_engagement_campaign
  - identify_upsell_opportunities
  timeout_seconds: 60
- id: notify_sub_broker
  name: Notify Sub-Broker
  type: script
  script: "import json\nimport time\nimport random\nfrom datetime import datetime\n\
    \n# Simulate notification processing\ntime.sleep(1)\n\nsub_broker_id = \"${sub_broker_id}\"\
    \ if \"${sub_broker_id}\" else \"DEFAULT_BROKER\"\nclient_id = \"${client_id}\"\
    \ if \"${client_id}\" else \"DEFAULT_CLIENT\"\n\n# Get consolidated strategy results\n\
    consolidated_data = ${consolidate_strategy_results.consolidated_strategies}\n\
    route = consolidated_data[\"selected_route\"]\nprimary_strategy = consolidated_data[\"\
    primary_strategy\"]\nstrategy_type = consolidated_data[\"strategy_type\"]\npriority_level\
    \ = consolidated_data[\"priority_level\"]\n\n# Generate notification based on\
    \ route\nif route == \"retention_path\":\n    notification_type = \"RETENTION_ALERT\"\
    \n    priority = \"CRITICAL\"\n    message = \"Immediate client retention action\
    \ required\"\n    recommendations = consolidated_data.get(\"retention_strategy\"\
    , {})\n    expected_response_time = \"4 hours\"\nelif route == \"engagement_path\"\
    :\n    notification_type = \"ENGAGEMENT_OPPORTUNITY\"\n    priority = \"HIGH\"\
    \n    message = \"Client engagement campaign ready for execution\"\n    recommendations\
    \ = consolidated_data.get(\"engagement_campaign\", {})\n    expected_response_time\
    \ = \"24 hours\"\nelse:\n    notification_type = \"UPSELL_OPPORTUNITY\"\n    priority\
    \ = \"MEDIUM\"\n    message = \"Cross-selling opportunities identified\"\n   \
    \ recommendations = consolidated_data.get(\"upsell_opportunities\", {})\n    expected_response_time\
    \ = \"3 days\"\n\nnotification = {\n    \"notification_id\": f\"N{random.randint(10000,\
    \ 99999)}\",\n    \"sub_broker_id\": sub_broker_id,\n    \"client_id\": client_id,\n\
    \    \"notification_type\": notification_type,\n    \"priority\": priority,\n\
    \    \"message\": message,\n    \"recommendations\": recommendations,\n    \"\
    expected_response_time\": expected_response_time,\n    \"notification_channels\"\
    : [\"email\", \"sms\", \"dashboard\"],\n    \"timestamp\": datetime.now().isoformat(),\n\
    \    \"status\": \"SENT\",\n    \"tracking_id\": f\"TRK{random.randint(100000,\
    \ 999999)}\"\n}\n\nresult = {\n    \"notification_sent\": True,\n    \"notification_details\"\
    : notification,\n    \"delivery_status\": \"SUCCESS\",\n    \"estimated_reach_time\"\
    : \"2 minutes\"\n}\n\nprint(f\"__OUTPUTS__ {json.dumps(result)}\")\n"
  depends_on:
  - consolidate_strategy_results
  description: Send notifications to sub-broker based on risk assessment
  previous_node: consolidate_strategy_results
  timeout_seconds: 60
- id: prepare_client_communication
  name: Prepare Client Communication
  type: script
  script: "import json\nimport time\nimport random\nfrom datetime import datetime,\
    \ timedelta\n\n# Simulate communication preparation\ntime.sleep(1)\n\nclient_data\
    \ = ${collect_client_data.client_data}\n\n# Get consolidated strategy results\n\
    consolidated_data = ${consolidate_strategy_results.consolidated_strategies}\n\
    route = consolidated_data[\"selected_route\"]\n\n# Prepare communication based\
    \ on route and client preferences\nclient_name = client_data[\"basic_info\"][\"\
    name\"]\npreferred_channel = client_data[\"communication_preferences\"][\"channel\"\
    ]\n\nif route == \"retention_path\":\n    message_type = \"RETENTION_OFFER\"\n\
    \    subject = f\"Exclusive benefits for you, {client_name}\"\n    content = f\"\
    Dear {client_name}, We value your relationship and have special offers designed\
    \ just for you.\"\n    urgency = \"HIGH\"\n    call_to_action = \"Call us at 1800-XXX-XXXX\
    \ for immediate assistance\"\nelif route == \"engagement_path\":\n    message_type\
    \ = \"ENGAGEMENT_INVITE\"\n    subject = f\"Exciting opportunities await, {client_name}\"\
    \n    content = f\"Hi {client_name}, Discover new investment insights and exclusive\
    \ content tailored for you.\"\n    urgency = \"MEDIUM\"\n    call_to_action =\
    \ \"Login to your account to explore\"\nelse:\n    message_type = \"PRODUCT_RECOMMENDATION\"\
    \n    subject = f\"New ways to grow your wealth, {client_name}\"\n    content\
    \ = f\"Hello {client_name}, Based on your portfolio, we have identified opportunities\
    \ for growth.\"\n    urgency = \"LOW\"\n    call_to_action = \"Schedule a consultation\
    \ with your advisor\"\n\ncommunication = {\n    \"communication_id\": f\"COM{random.randint(10000,\
    \ 99999)}\",\n    \"client_id\": client_data[\"client_id\"],\n    \"message_type\"\
    : message_type,\n    \"channel\": preferred_channel,\n    \"subject\": subject,\n\
    \    \"content\": content,\n    \"urgency\": urgency,\n    \"call_to_action\"\
    : call_to_action,\n    \"scheduled_time\": (datetime.now() + timedelta(minutes=random.randint(30,\
    \ 180))).isoformat(),\n    \"personalization_data\": {\n        \"client_name\"\
    : client_name,\n        \"preferred_products\": client_data[\"transaction_history\"\
    ][\"preferred_segments\"],\n        \"risk_profile\": client_data[\"basic_info\"\
    ][\"risk_profile\"],\n        \"city\": client_data[\"basic_info\"][\"city\"]\n\
    \    },\n    \"tracking_enabled\": True,\n    \"delivery_confirmation\": True\n\
    }\n\nresult = {\n    \"communication_prepared\": True,\n    \"communication_details\"\
    : communication,\n    \"estimated_delivery_time\": \"15 minutes\",\n    \"personalization_score\"\
    : round(random.uniform(0.8, 1.0), 2)\n}\n\nprint(f\"__OUTPUTS__ {json.dumps(result)}\"\
    )\n"
  depends_on:
  - consolidate_strategy_results
  - collect_client_data
  description: Prepare personalized client communications based on strategy
  previous_node: consolidate_strategy_results
  timeout_seconds: 60
- id: update_client_profile
  name: Update Client Profile
  type: script
  script: "import json\nimport time\nimport random\nfrom datetime import datetime,\
    \ timedelta\n\n# Simulate profile update processing\ntime.sleep(1)\n\nclient_id\
    \ = \"${client_id}\" if \"${client_id}\" else \"DEFAULT_CLIENT\"\n\n# Get consolidated\
    \ strategy results\nconsolidated_data = ${consolidate_strategy_results.consolidated_strategies}\n\
    route = consolidated_data[\"selected_route\"]\n\nrisk_assessment = ${assess_client_risk}\n\
    profile_analysis = ${analyze_client_profile}\n\n# Update client profile with new\
    \ insights\nprofile_updates = {\n    \"client_id\": client_id,\n    \"last_analysis_date\"\
    : datetime.now().isoformat(),\n    \"risk_level\": risk_assessment[\"risk_level\"\
    ],\n    \"risk_score\": risk_assessment[\"churn_risk_score\"],\n    \"segment\"\
    : profile_analysis[\"primary_segment\"],\n    \"engagement_score\": profile_analysis[\"\
    engagement_level\"][\"score\"],\n    \"recommended_actions\": route,\n    \"analysis_version\"\
    : \"1.0\",\n    \"next_review_date\": (datetime.now() + timedelta(days=30)).isoformat(),\n\
    \    \"tags\": [\n        f\"risk_{risk_assessment['risk_level'].lower()}\",\n\
    \        f\"segment_{profile_analysis['primary_segment'].lower()}\",\n       \
    \ f\"engagement_{profile_analysis['engagement_level']['rating'].lower()}\"\n \
    \   ],\n    \"ai_insights\": {\n        \"confidence_score\": profile_analysis[\"\
    confidence_score\"],\n        \"behavioral_patterns\": profile_analysis[\"behavioral_patterns\"\
    ],\n        \"investment_preferences\": profile_analysis[\"investment_preferences\"\
    ]\n    },\n    \"action_history\": {\n        \"last_action\": route,\n      \
    \  \"action_timestamp\": datetime.now().isoformat(),\n        \"success_probability\"\
    : f\"{random.randint(60, 90)}%\"\n    }\n}\n\nresult = {\n    \"profile_updated\"\
    : True,\n    \"update_details\": profile_updates,\n    \"update_timestamp\": datetime.now().isoformat(),\n\
    \    \"fields_updated\": 12,\n    \"data_quality_improvement\": \"+5%\"\n}\n\n\
    print(f\"__OUTPUTS__ {json.dumps(result)}\")\n"
  depends_on:
  - consolidate_strategy_results
  - assess_client_risk
  - analyze_client_profile
  description: Update client profile with new insights and recommendations
  previous_node: consolidate_strategy_results
  timeout_seconds: 60
- id: simulate_response_tracking
  name: Simulate Response Tracking
  type: script
  script: "import json\nimport time\nimport random\nfrom datetime import datetime,\
    \ timedelta\n\n# Simulate response tracking\ntime.sleep(1)\n\n# Get consolidated\
    \ strategy results\nconsolidated_data = ${consolidate_strategy_results.consolidated_strategies}\n\
    route = consolidated_data[\"selected_route\"]\nclient_id = \"${client_id}\" if\
    \ \"${client_id}\" else \"DEFAULT_CLIENT\"\n\n# Response probabilities based on\
    \ strategy type\nresponse_probabilities = {\n    \"retention_path\": {\"response_rate\"\
    : 0.75, \"positive_rate\": 0.65},\n    \"engagement_path\": {\"response_rate\"\
    : 0.55, \"positive_rate\": 0.80},\n    \"upsell_path\": {\"response_rate\": 0.35,\
    \ \"positive_rate\": 0.45},\n    \"standard_path\": {\"response_rate\": 0.40,\
    \ \"positive_rate\": 0.50}\n}\n\nprobs = response_probabilities.get(route, response_probabilities[\"\
    standard_path\"])\n\n# Simulate response\nhas_response = random.random() < probs[\"\
    response_rate\"]\nis_positive = random.random() < probs[\"positive_rate\"] if\
    \ has_response else False\n\n# Response details\nresponse_channels = [\"email\"\
    , \"phone\", \"sms\", \"mobile_app\", \"web_portal\"]\nresponse_types = [\"inquiry\"\
    , \"acceptance\", \"rejection\", \"deferral\", \"request_more_info\"]\n\nresponse_data\
    \ = {\n    \"client_id\": client_id,\n    \"campaign_route\": route,\n    \"has_response\"\
    : has_response,\n    \"is_positive\": is_positive,\n    \"response_timestamp\"\
    : datetime.now().isoformat(),\n    \"response_channel\": random.choice(response_channels)\
    \ if has_response else None,\n    \"response_type\": random.choice(response_types)\
    \ if has_response else None,\n    \"response_time_minutes\": random.randint(30,\
    \ 1440) if has_response else None,\n    \"engagement_score\": round(random.uniform(0.3,\
    \ 0.9), 2) if has_response else 0,\n    \"follow_up_required\": has_response and\
    \ not is_positive,\n    \"next_action\": \"schedule_call\" if has_response and\
    \ is_positive else \"send_reminder\" if has_response else \"escalate\"\n}\n\n\
    # Additional metrics\ntracking_metrics = {\n    \"email_open_rate\": round(random.uniform(0.4,\
    \ 0.8), 2),\n    \"click_through_rate\": round(random.uniform(0.1, 0.3), 2),\n\
    \    \"time_to_response\": f\"{random.randint(1, 48)} hours\" if has_response\
    \ else \"No response\",\n    \"interaction_depth\": random.choice([\"surface\"\
    , \"moderate\", \"deep\"]) if has_response else \"none\",\n    \"sentiment_score\"\
    : round(random.uniform(0.2, 0.9), 2) if has_response else 0.5\n}\n\nresult = {\n\
    \    \"response_tracked\": True,\n    \"response_data\": response_data,\n    \"\
    tracking_metrics\": tracking_metrics,\n    \"analysis_timestamp\": datetime.now().isoformat(),\n\
    \    \"recommendation\": \"Continue monitoring\" if is_positive else \"Escalate\
    \ to manager\" if has_response else \"Retry with different approach\"\n}\n\nprint(f\"\
    __OUTPUTS__ {json.dumps(result)}\")\n"
  depends_on:
  - notify_sub_broker
  - prepare_client_communication
  - update_client_profile
  - consolidate_strategy_results
  description: Track and analyze client responses to campaigns
  previous_node:
  - notify_sub_broker
  - prepare_client_communication
  - update_client_profile
- id: generate_performance_metrics
  name: Generate Performance Metrics
  type: ai_agent
  config:
    agent_type: analyst
    input_format: json
    output_format: json
    model_client_id: gpt4_analytics
  depends_on:
  - simulate_response_tracking
  description: AI-powered comprehensive workflow performance metrics analysis
  user_message: "Analyze the workflow performance data and generate comprehensive\
    \ metrics:\n\n**Workflow Context:**\n- Route taken: ${risk_based_router.selected_route}\n\
    - Client response data: ${simulate_response_tracking.response_data}\n- Tracking\
    \ metrics: ${simulate_response_tracking.tracking_metrics}\n- Execution ID: ${EXECUTION_ID}\n\
    - Workflow ID: ${WORKFLOW_ID}\n- Client ID: ${client_id}\n\n**Analysis Requirements:**\n\
    1. Calculate execution performance metrics (completion rate, timing, success indicators)\n\
    2. Analyze route-specific performance against historical benchmarks\n3. Assess\
    \ business impact (revenue, retention, engagement improvements)\n4. Evaluate response\
    \ quality and tracking effectiveness\n5. Generate actionable improvement recommendations\n\
    6. Provide confidence scores for all metrics\n\n**CRITICAL OUTPUT REQUIREMENTS:**\n\
    \nYou MUST return a JSON object with the exact structure shown below. The downstream\
    \ task `update_learning_model` specifically accesses `performance_metrics` field,\
    \ so this structure is mandatory:\n\n```json\n{\n  \"metrics_generated\": true,\n\
    \  \"performance_metrics\": {\n    \"execution_id\": \"${EXECUTION_ID}\",\n  \
    \  \"workflow_id\": \"${WORKFLOW_ID}\",\n    \"client_id\": \"${client_id}\",\n\
    \    \"processing_timestamp\": \"2025-01-15T10:30:00.000Z\",\n    \"execution_metrics\"\
    : {\n      \"total_execution_time\": \"25 seconds\",\n      \"tasks_completed\"\
    : 12,\n      \"tasks_successful\": 12,\n      \"success_rate\": 100.0,\n     \
    \ \"data_quality_score\": 0.92,\n      \"ai_confidence_average\": 0.88\n    },\n\
    \    \"route_taken\": \"retention_path\",\n    \"route_performance\": {\n    \
    \  \"avg_success_rate\": 0.68,\n      \"avg_response_time\": \"4.2 hours\"\n \
    \   },\n    \"response_metrics\": {\n      \"response_received\": true,\n    \
    \  \"response_positive\": false,\n      \"engagement_score\": 0.65,\n      \"\
    tracking_quality\": {\n        \"email_open_rate\": 0.72,\n        \"click_through_rate\"\
    : 0.23,\n        \"time_to_response\": \"6 hours\",\n        \"interaction_depth\"\
    : \"moderate\",\n        \"sentiment_score\": 0.45\n      }\n    },\n    \"business_impact\"\
    : {\n      \"estimated_revenue_impact\": \"\u20B945000\",\n      \"retention_probability_improvement\"\
    : \"+25%\",\n      \"engagement_score_improvement\": \"+35%\",\n      \"cross_sell_opportunities_identified\"\
    : 3,\n      \"client_satisfaction_impact\": \"+18%\"\n    },\n    \"success_indicators\"\
    : {\n      \"data_collection\": \"SUCCESS\",\n      \"risk_assessment\": \"SUCCESS\"\
    ,\n      \"recommendations\": \"SUCCESS\",\n      \"notifications\": \"SUCCESS\"\
    ,\n      \"communication\": \"SUCCESS\",\n      \"tracking\": \"SUCCESS\"\n  \
    \  },\n    \"next_actions\": {\n      \"follow_up_required\": true,\n      \"\
    escalation_needed\": false,\n      \"next_review_date\": \"2025-02-15\",\n   \
    \   \"recommended_action\": \"schedule_call\"\n    },\n    \"improvement_recommendations\"\
    : [\n      \"Enhance data collection frequency for better insights\",\n      \"\
    Implement A/B testing for communication strategies\",\n      \"Develop more personalized\
    \ offer generation\"\n    ]\n  },\n  \"analytics_timestamp\": \"2025-01-15T10:30:00.000Z\"\
    ,\n  \"report_confidence\": 0.91\n}\n```\n\n**IMPORTANT NOTES:**\n- Use current\
    \ ISO timestamps for `processing_timestamp` and `analytics_timestamp`\n- Generate\
    \ realistic values based on the provided workflow data\n- Ensure all numeric values\
    \ are appropriate for the business context\n- The `tracking_quality` should match\
    \ the structure from `tracking_metrics`\n- Route-specific performance should align\
    \ with the `route_taken` value\n- Return ONLY the JSON object, no additional text\
    \ or formatting\n"
  previous_node: simulate_response_tracking
  system_message: "You are an expert business intelligence analyst specializing in\
    \ workflow performance metrics and KPI analysis. \nYour role is to analyze workflow\
    \ execution data and generate comprehensive performance reports with actionable\
    \ insights.\n\nFocus on:\n1. Calculate meaningful performance metrics from execution\
    \ data\n2. Analyze route-specific performance patterns\n3. Assess business impact\
    \ and ROI\n4. Identify improvement opportunities\n5. Generate data-driven recommendations\n\
    \nAlways provide structured, quantitative analysis with clear metrics and confidence\
    \ scores.\n"
- id: update_learning_model
  name: Update Learning Model
  type: script
  script: "import json\nimport time\nimport random\nfrom datetime import datetime,\
    \ timedelta\n\n# Simulate model learning process\ntime.sleep(1)\n\n# Get AI response\
    \ and clean it to extract JSON\nai_response = \"\"\"${generate_performance_metrics.ai_response}\"\
    \"\"\n# Remove markdown formatting\nclean_response = ai_response.replace('```json',\
    \ '').replace('```', '')\n# Parse the JSON\ntry:\n    ai_data = json.loads(clean_response.strip())\n\
    \    performance_metrics = ai_data.get('performance_metrics', {})\nexcept:\n \
    \   performance_metrics = {}\n\nclient_data = ${collect_client_data.client_data}\n\
    \n# Prepare learning data\nlearning_data = {\n    \"client_segment\": \"${analyze_client_profile.primary_segment}\"\
    ,\n    \"risk_level\": \"${assess_client_risk.risk_level}\",\n    \"actions_taken\"\
    : \"${risk_based_router.selected_route}\",\n    \"response_outcome\": ${simulate_response_tracking.response_data},\n\
    \    \"success_metrics\": performance_metrics,\n    \"client_characteristics\"\
    : {\n        \"aum\": client_data[\"portfolio\"][\"total_aum\"],\n        \"age\"\
    : client_data[\"basic_info\"][\"age\"],\n        \"risk_profile\": client_data[\"\
    basic_info\"][\"risk_profile\"],\n        \"engagement_level\": \"${analyze_client_profile.engagement_level}\"\
    \n    },\n    \"learning_timestamp\": datetime.now().isoformat()\n}\n\n# Model\
    \ improvement insights\nmodel_insights = {\n    \"patterns_identified\": [\n \
    \       f\"High AUM clients ({client_data['portfolio']['total_aum']}+) show {random.randint(15,\
    \ 25)}% better response to retention offers\",\n        f\"Clients in {client_data['basic_info']['city']}\
    \ prefer {client_data['communication_preferences']['channel']} communication\"\
    ,\n        f\"{client_data['basic_info']['risk_profile']} risk profile clients\
    \ respond best to {random.choice(['educational', 'promotional', 'advisory'])}\
    \ content\"\n    ],\n    \"accuracy_improvements\": {\n        \"risk_assessment_model\"\
    : f\"+{round(random.uniform(1.5, 4.2), 1)}%\",\n        \"segmentation_model\"\
    : f\"+{round(random.uniform(0.8, 2.9), 1)}%\",\n        \"recommendation_engine\"\
    : f\"+{round(random.uniform(2.1, 5.3), 1)}%\",\n        \"response_prediction\"\
    : f\"+{round(random.uniform(1.2, 3.7), 1)}%\"\n    },\n    \"recommendations_refined\"\
    : [\n        \"Adjust risk thresholds for better sensitivity\",\n        \"Enhance\
    \ personalization algorithms\",\n        \"Improve timing prediction models\"\
    ,\n        \"Optimize communication channel selection\"\n    ]\n}\n\n# Algorithm\
    \ updates\nalgorithm_updates = {\n    \"updated_algorithms\": [\n        \"risk_assessment_model_v2.1\"\
    ,\n        \"client_segmentation_v1.8\",\n        \"recommendation_engine_v3.2\"\
    ,\n        \"response_prediction_v1.5\"\n    ],\n    \"new_features\": [\n   \
    \     \"Dynamic risk scoring\",\n        \"Behavioral pattern recognition\",\n\
    \        \"Predictive engagement scoring\",\n        \"Optimal timing recommendations\"\
    \n    ],\n    \"model_performance\": {\n        \"overall_accuracy\": f\"{round(random.uniform(87.5,\
    \ 94.2), 1)}%\",\n        \"precision\": f\"{round(random.uniform(82.3, 89.7),\
    \ 1)}%\",\n        \"recall\": f\"{round(random.uniform(78.9, 86.4), 1)}%\",\n\
    \        \"f1_score\": f\"{round(random.uniform(80.1, 87.8), 1)}%\"\n    }\n}\n\
    \n# Future optimization plans\noptimization_plans = {\n    \"next_optimization_cycle\"\
    : (datetime.now() + timedelta(days=7)).isoformat(),\n    \"focus_areas\": [\n\
    \        \"Improve low-engagement client predictions\",\n        \"Enhance cross-selling\
    \ opportunity identification\",\n        \"Optimize communication timing\",\n\
    \        \"Refine retention strategy effectiveness\"\n    ],\n    \"data_requirements\"\
    : [\n        \"More historical response data\",\n        \"Enhanced behavioral\
    \ tracking\",\n        \"Market condition correlations\",\n        \"Competitor\
    \ analysis integration\"\n    ]\n}\n\nresult = {\n    \"learning_completed\":\
    \ True,\n    \"learning_data\": learning_data,\n    \"model_insights\": model_insights,\n\
    \    \"algorithm_updates\": algorithm_updates,\n    \"optimization_plans\": optimization_plans,\n\
    \    \"learning_timestamp\": datetime.now().isoformat(),\n    \"learning_confidence\"\
    : round(random.uniform(0.85, 0.95), 2)\n}\n\nprint(f\"__OUTPUTS__ {json.dumps(result)}\"\
    )"
  depends_on:
  - generate_performance_metrics
  - collect_client_data
  - analyze_client_profile
  - assess_client_risk
  - risk_based_router
  - simulate_response_tracking
  description: Feed results back into AI learning system for continuous improvement
  previous_node: generate_performance_metrics
inputs:
- name: client_id
  type: string
  required: true
  description: Client identifier for analysis
- name: analysis_depth
  type: string
  default: standard
  required: false
  description: 'Analysis depth: basic, standard, or comprehensive'
- name: enable_cross_selling
  type: boolean
  default: true
  required: false
  description: Enable cross-selling recommendations
- name: enable_retention_alerts
  type: boolean
  default: true
  required: false
  description: Enable retention risk alerts
- name: sub_broker_id
  type: string
  required: false
  description: Associated sub-broker ID
- name: smc_api_base_url
  type: string
  default: https://api.smcindiaonline.com
  required: false
  description: SMC API base URL
labels:
  version: '1.0'
  priority: high
  department: wealth_management
  environment: production
outputs:
  client_analysis:
    type: object
    source: analyze_client_profile
    description: Complete client analysis results
  risk_assessment:
    type: object
    source: assess_client_risk
    description: Client risk assessment and recommendations
  learning_outcomes:
    type: object
    source: update_learning_model
    description: Model learning and optimization results
  performance_metrics:
    type: object
    source: generate_performance_metrics.ai_response.performance_metrics
    description: Workflow execution metrics and insights
  recommended_actions:
    type: string
    source: risk_based_router.selected_route
    description: Recommended actions based on client profile
version: '1.0'
metadata:
  author: SMC Development Team
  use_case: AI Relationship Manager
  created_date: '2024-07-17'
  business_unit: SMC Global Securities
  compliance_level: SEBI_COMPLIANT
namespace: smc_finance
description: AI-powered relationship management system for SMC's customer portfolio
  with cross-selling and retention capabilities
model_clients:
- id: gpt4_analytics
  config:
    model: gpt-4o-mini
    api_key: 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
    max_tokens: 4000
    temperature: 0.1
  provider: openai
timeout_seconds: 3600

Execution ID	Status	Started	Duration	Actions
`87514feb...`	COMPLETED	2025-07-17 10:38:28	N/A	View
`c32e1342...`	COMPLETED	2025-07-17 10:35:32	N/A	View
`82dcb8af...`	COMPLETED	2025-07-17 10:11:31	N/A	View
`44515fc4...`	COMPLETED	2025-07-17 09:53:35	N/A	View
`a42bc814...`	COMPLETED	2025-07-17 09:51:01	N/A	View