Policy Generator

from datetime import datetime, date

import re

import json

class MotorInsuranceAIEngine:

    def __init__(self):

        # Simulated Large Language Model prompts / intent structures

        self.supported_languages = ["English", "Hindi", "Spanish", "Telugu"]

        

    def ai_extract_policy_data(self, raw_unstructured_text: str) -> dict:

        """

        Simulates an Agentic AI OCR / Document Extraction layer (e.g., PDF -> CRM processing) 

        to extract key entities out of messy customer inputs or document text blocks.

        """

        print("[AI Mode] Processing unstructured layout text via NLP extraction...")

        

        # Regex entities mimicking ML-based token classification

        policy_pattern = re.search(r'(POL-\d{5,8}|\b\d{7,10}\b)', raw_unstructured_text)

        date_pattern = re.findall(r'(\d{4}-\d{2}-\d{2}|\d{2}/\d{2}/\d{4})', raw_unstructured_text)

        premium_pattern = re.search(r'(?:Rs\.|INR|\$)\s?(\d{1,3}(?:,\d{3})*(?:\.\d{2})?)', raw_unstructured_text)

        vehicle_pattern = re.search(r'(Model|Vehicle|Car|Bike):\s*([A-Za-z0-9 ]+)', raw_unstructured_text, re.IGNORECASE)

        

        # Default extraction structure

        extracted_data = {

            "policy_number": policy_pattern.group(0) if policy_pattern else "UNKNOWN_RECOVERY",

            "expiry_date": date_pattern[0] if date_pattern else None,

            "premium_amount": premium_pattern.group(1) if premium_pattern else "0.00",

            "vehicle_details": vehicle_pattern.group(2).strip() if vehicle_pattern else "Standard Vehicle"

        }

        

        # Dynamic normalization of formats

        if extracted_data["expiry_date"] and "/" in extracted_data["expiry_date"]:

            try:

                dt = datetime.strptime(extracted_data["expiry_date"], "%d/%m/%Y")

                extracted_data["expiry_date"] = dt.strftime("%Y-%m-%d")

            except ValueError:

                pass

                

        return extracted_data

    def calculate_risk_and_expiry(self, expiry_date_str: str) -> dict:

        """

        AI Evaluation Engine that processes dates to deduce lapse status,

        days remaining, and flags critical windows for multi-touch outreach campaigns.

        """

        today = date.today()

        expiry_date = datetime.strptime(expiry_date_str, "%Y-%m-%d").date()

        days_remaining = (expiry_date - today).days

        

        status = "Active"

        lapse_risk = "Low"

        action_recommended = "Monitor Only"

        

        if days_remaining < 0:

            status = "Lapsed"

            lapse_risk = "CRITICAL (Expired)"

            action_recommended = "Trigger Break-in AI Vehicle Inspection Module" # e.g., ICICI Lombard / Microsoft style

        elif days_remaining <= 7:

            status = "Urgent Renewal Required"

            lapse_risk = "High"

            action_recommended = "Launch Daily High-Priority WhatsApp & Automated Voice Call Flow"

        elif days_remaining <= 30:

            status = "Approaching Expiry"

            lapse_risk = "Medium"

            action_recommended = "Queue in Standard Multi-Touch Renewal Campaign (Email / SMS)"

            

        return {

            "days_remaining": days_remaining,

            "policy_status": status,

            "lapse_risk_level": lapse_risk,

            "next_best_action": action_recommended

        }

    def ai_draft_generator(self, customer_name: str, policy_data: dict, analytics: dict, preferred_lang: str = "English") -> str:

        """

        Generative AI dynamic draft component. Creates highly personalized 

        outreach copy matching contextual risks and chosen parameters.

        """

        lang = preferred_lang if preferred_lang in self.supported_languages else "English"

        

        templates = {

            "English": {

                "active": f"Hello {customer_name},\nYour motor insurance policy {policy_data['policy_number']} for your {policy_data['vehicle_details']} expires in {analytics['days_remaining']} days ({policy_data['expiry_date']}). Renew now to avoid lapse premiums! Recommended Action: {analytics['next_best_action']}.",

                "lapsed": f"Alert {customer_name},\nYour policy {policy_data['policy_number']} has LAPSED by {abs(analytics['days_remaining'])} days. Your vehicle is uninsured. {analytics['next_best_action']} immediately to resume safety grids."

            },

            "Hindi": {

                "active": f"नमस्ते {customer_name},\nआपकी वाहन बीमा पॉलिसी {policy_data['policy_number']} ({policy_data['vehicle_details']}) {analytics['days_remaining']} दिनों में समाप्त हो रही है। कृपया तुरंत नवीनीकरण करें।",

                "lapsed": f"चेतावनी {customer_name},\nआपकी पॉलिसी {policy_data['policy_number']} समाप्त हो चुकी है। {analytics['next_best_action']} का उपयोग कर तुरंत नया बीमा प्राप्त करें।"

            }

        }

        

        state_key = "lapsed" if analytics['days_remaining'] < 0 else "active"

        draft = templates.get(lang, templates["English"])[state_key]

        return draft

    def process_pipeline(self, raw_input: str, customer_name: str, preferred_lang: str = "English") -> str:

        """

        Executes the end-to-end processing pipeline mimicking agentic system intelligence.

        """

        # Step 1: Document Parsing / Entity extraction

        extracted = self.ai_extract_policy_data(raw_input)

        

        if not extracted["expiry_date"]:

            return json.dumps({"Error": "AI system failed to extract a viable policy expiration deadline date."}, indent=4)

            

        # Step 2: Predictive analytics & date handling

        analytics = self.calculate_risk_and_expiry(extracted["expiry_date"])

        

        # Step 3: Generative Content Copy production

        ai_message_pitch = self.ai_draft_generator(customer_name, extracted, analytics, preferred_lang)

        

        # Package unified workflow payload

        output_dashboard = {

            "Customer Metadata": {

                "Name": customer_name,

                "Language": preferred_lang

            },

            "AI Extracted Entities": extracted,

            "Predictive Risk Matrix": analytics,

            "Tailored AI Generated Pitch": ai_message_pitch

        }

        

        return json.dumps(output_dashboard, indent=4)

# =====================================================================

# Execution Demo

# =====================================================================

if __name__ == "__main__":

    ai_engine = MotorInsuranceAIEngine()

    

    # Mock text extracted from an agent uploading a raw digital scrap / screenshot string 

    raw_document_chunk = "Vehicle details: Honda Civic Car. Premium paid: INR 12,500. Policy Code Reference: POL-998823. Validity expiration date is 2026-10-25."

    

    # Run the dynamic processing stack

    print("--- Executing AI Automation Pipeline Pipeline ---")

    result_json = ai_engine.process_pipeline(

        raw_input=raw_document_chunk, 

        customer_name="Aditya Sharma", 

        preferred_lang="English"

    )

    

    print("\nFinal Integrated System Outputs:")

    print(result_json)