Reverse writting

InsurGenie_PPT_Content.txt

@@ -0,0 +1,43 @@
+=========================================
+INSURGENIE - PPT / PITCH DECK CONTENT
+=========================================
+
+SLIDE 1: TITLE SLIDE
+• Project Name: InsurGenie
+• Tagline: AI-Powered Dynamic Insurance & Reverse Underwriting
+
+SLIDE 2: THE PROBLEM WITH TRADITIONAL INSURANCE
+• Outdated Metrics: Relies heavily on static demographics (Age, Gender, Zip Code) instead of actual behavior.
+• Lack of Transparency: Users don't know exactly why their premiums are so high.
+• Zero Incentive: Safe drivers and healthy individuals are penalized to subsidize risky users, with no ongoing motivation to improve.
+
+SLIDE 3: THE INSURGENIE SOLUTION
+• Real-time AI Risk Engine: Ingests dynamic data points from wearables (Health) and vehicle telematics (Driving).
+• Dynamic Pricing Models: Calculates a continuously updating "Risk Score" (0.0 to 1.0) and assigns users to one of 7 tiers (Elite to Extreme).
+• Empowering the User: The safer they live, the less they pay.
+
+SLIDE 4: GAMIFIED DASHBOARD & REWARDS
+• Unified Ecosystem: Displays a comprehensive breakdown of Health, Driving, and Lifestyle risk metrics.
+• Financial Reinforcement: Positively framed UI highlighting "Insurance Savings" directly tied to lifestyle choices.
+• Milestone Tracking: Next savings milestones and progress bars incentivize users to stick to healthier goals.
+
+SLIDE 5: REVERSE UNDERWRITING (THE CROWN JEWEL)
+• A complete flip of conventional insurance practices.
+• Goal Setting: Users actively choose a target insurance plan tier (e.g., "I want to pay ₹799/month on the Elite Plan").
+• AI Roadmap: The AI analyzes their current data and generates a highly personalized, actionable roadmap to qualify for that rate (e.g., "Walk 8,000 steps daily," "Reduce hard braking by 15%").
+
+SLIDE 6: CONTEXT-AWARE AI CHATBOT
+• 24/7 intelligent insurance assistant built right into the dashboard.
+• Explainability: Helps users understand policy details, how their risk score was calculated, and interprets complex insurance jargon instantly.
+
+SLIDE 7: HOW IT WORKS (THE USER JOURNEY)
+1. Connect Data: Seamlessly link wearables and vehicle telematics.
+2. AI-Powered Assessment: Receive an instant, real-time risk evaluation.
+3. Monitor via Dashboard: Track health/driving scores and watch savings grow.
+4. Set Targets: Use Reverse Underwriting to chart a path to lower premiums.
+5. Re-Insure: Meet the goals and automatically unlock better rates.
+
+SLIDE 8: SUMMARY OF TECHNICAL WORK
+• Frontend: Built with React/Vite, featuring dark-mode sleek UI, dynamic routing, component-driven tables, and deep interactive states.
+• Backend/AI: FastAPI Python backend natively connected to Groq LLMs to handle complex logic securely.
+• Cloud & Data: Fully integrated Supabase environment tying user profiles with risk scores and real-time state management.

backend/app/auth.py

@@ -37,4 +37,18 @@ def create_or_get_user(user_info):
     }
 
     insert_res = supabase.table("users").insert(new_user).execute()
-    return insert_res.data[0]
+    return insert_res.data[0]
+
+def get_current_user(user_id: int):
+    res = (
+        supabase
+        .table("users")
+        .select("*")
+        .eq("id", user_id)
+        .execute()
+    )
+
+    if res.data:
+        return res.data[0]
+
+    return None

backend/app/controllers/chatbot.py

@@ -0,0 +1,45 @@
+import os
+from pydantic import BaseModel
+from typing import Optional, Dict, Any
+from fastapi import APIRouter
+from groq import Groq
+from dotenv import load_dotenv
+
+load_dotenv()
+
+router = APIRouter()
+
+GROQ_API_KEY = os.getenv("GROQ_API_KEY")
+client = Groq(api_key=GROQ_API_KEY) if GROQ_API_KEY else None
+
+class ChatMessage(BaseModel):
+    message: str
+    context: Optional[Dict[str, Any]] = None
+
+@router.post("/chat")
+def chat_endpoint(data: ChatMessage):
+    if not client:
+        return {"reply": "I am currently offline. Please check the GROQ API key configuration."}
+
+    system_prompt = (
+        "You are InsureGenie, an AI assistant for an insurance platform. "
+        "You help users understand their health, driving risk, and insurance policies. "
+        "Be concise, friendly, and helpful."
+    )
+    if data.context:
+        system_prompt += f"\nHere is the user's current context based on their dashboard: {data.context}"
+
+    try:
+        chat_completion = client.chat.completions.create(
+            messages=[
+                {"role": "system", "content": system_prompt},
+                {"role": "user", "content": data.message}
+            ],
+            model="llama-3.3-70b-versatile",
+            temperature=0.5,
+            max_tokens=1024,
+        )
+        reply = chat_completion.choices[0].message.content
+        return {"reply": reply}
+    except Exception as e:
+        return {"reply": f"Error communicating with the bot: {str(e)}"}

backend/app/controllers/reverse_underwriting.py

@@ -0,0 +1,242 @@
+import os
+import json
+from groq import Groq
+from dotenv import load_dotenv
+from app.database import supabase
+
+load_dotenv()
+
+GROQ_API_KEY = os.getenv("GROQ_API_KEY")
+client = Groq(api_key=GROQ_API_KEY) if GROQ_API_KEY else None
+
+PLAN_DETAILS = {
+    1: {"name": "Elite",    "risk_range": "Below 0.20",  "premium": 350, "label": "Best users"},
+    2: {"name": "Pro",      "risk_range": "0.20 - 0.30", "premium": 380, "label": "Very safe"},
+    3: {"name": "Advanced", "risk_range": "0.30 - 0.45", "premium": 420, "label": "Slight risk"},
+    4: {"name": "Guarded",  "risk_range": "0.45 - 0.60", "premium": 450, "label": "Moderate"},
+    5: {"name": "Risky",    "risk_range": "0.60 - 0.75", "premium": 500, "label": "Neutral"},
+    6: {"name": "Critical", "risk_range": "0.75 - 0.90", "premium": 550, "label": "Risk loading"},
+    7: {"name": "Extreme",  "risk_range": "0.90 - 1.00", "premium": 650, "label": "Very risky"},
+}
+
+
+def get_reverse_underwriting_advice(user_id: int, target_plan_level: int) -> dict:
+
+    # 1. Validate plan level
+    if target_plan_level not in PLAN_DETAILS:
+        return {"success": False, "error": "Invalid target plan level. Must be between 1 and 7."}
+
+    # 2. Fetch current premium / risk data
+    premium_resp = (
+        supabase.table("premium")
+        .select("*")
+        .eq("user_id", user_id)
+        .order("created_at", desc=True)
+        .limit(1)
+        .execute()
+    )
+    if not premium_resp.data:
+        return {"success": False, "error": "No policy data found. Please calculate your insurance quote first."}
+
+    premium = premium_resp.data[0]
+    current_plan_level = int(premium.get("policy_plan_id", 5))
+
+    if target_plan_level >= current_plan_level:
+        current_name = PLAN_DETAILS[current_plan_level]["name"]
+        return {
+            "success": False,
+            "error": (
+                f"Please select a plan better than your current plan "
+                f"({current_name}, Level {current_plan_level}). "
+                f"Choose a plan with a lower level number."
+            ),
+        }
+
+    # 3. Recent activity (7-day avg)
+    activity_resp = (
+        supabase.table("user_activity")
+        .select("steps, calories, active_minutes")
+        .eq("user_id", user_id)
+        .order("recorded_at", desc=True)
+        .limit(7)
+        .execute()
+    )
+    activity = activity_resp.data or []
+    avg_steps      = int(sum(a.get("steps", 0) for a in activity) / len(activity)) if activity else 0
+    avg_calories   = int(sum(a.get("calories", 0) for a in activity) / len(activity)) if activity else 0
+    avg_active_min = int(sum(a.get("active_minutes", 0) for a in activity) / len(activity)) if activity else 0
+
+    # 4. Wellness score
+    score_resp = (
+        supabase.table("user_score")
+        .select("score, reward")
+        .eq("user_id", user_id)
+        .execute()
+    )
+    score_data = score_resp.data[0] if score_resp.data else {"score": 0, "reward": 0}
+
+    # 5. Insurance inputs for BMI / heart rate (best-effort)
+    bmi        = "unknown"
+    heart_rate = "unknown"
+    try:
+        input_resp = (
+            supabase.table("insurance_inputs")
+            .select("bmi, heart_rate")
+            .eq("user_id", user_id)
+            .order("created_at", desc=True)
+            .limit(1)
+            .execute()
+        )
+        if input_resp.data:
+            bmi        = input_resp.data[0].get("bmi", "unknown")
+            heart_rate = input_resp.data[0].get("heart_rate", "unknown")
+    except Exception:
+        pass
+
+    # 6. Build variables
+    current_plan    = PLAN_DETAILS[current_plan_level]
+    target_plan     = PLAN_DETAILS[target_plan_level]
+    monthly_savings = current_plan["premium"] - target_plan["premium"]
+    current_risk    = premium.get("risk_score", 0.5)
+    health_risk     = premium.get("health_risk", "medium")
+    lifestyle_risk  = premium.get("lifestyle_risk", "medium")
+    driving_risk    = premium.get("driving_risk", "medium")
+
+    # 7. Build LLM prompt
+    prompt = f"""
+You are an expert insurance underwriter and health coach for InsureGenie, an AI-powered insurance platform.
+
+A user wants to upgrade their insurance plan from their current plan to a better (lower risk) plan.
+Your job is to generate a specific, data-driven, actionable improvement roadmap.
+
+CURRENT USER PROFILE:
+- Current Plan: {current_plan['name']} (Level {current_plan_level})
+- Current Monthly Premium: Rs {current_plan['premium']}/month
+- Current Risk Score: {current_risk:.2f} (scale: 0.0 = lowest risk, 1.0 = highest risk)
+- Health Risk Category: {health_risk}
+- Lifestyle Risk Category: {lifestyle_risk}
+- Driving Risk Category: {driving_risk}
+- Current BMI: {bmi}
+- Resting Heart Rate: {heart_rate} bpm
+- Wellness Points Earned: {score_data['score']}
+
+RECENT ACTIVITY (7-day average):
+- Daily Steps: {avg_steps}
+- Daily Calories Burned: {avg_calories}
+- Daily Active Minutes: {avg_active_min}
+
+TARGET PLAN:
+- Plan: {target_plan['name']} (Level {target_plan_level})
+- Monthly Premium: Rs {target_plan['premium']}/month
+- Required Risk Score Range: {target_plan['risk_range']}
+- Monthly Savings if Achieved: Rs {monthly_savings}
+
+Return ONLY a valid JSON object — no markdown, no code fences, no extra text.
+The JSON must exactly follow this structure:
+
+{{
+  "summary": "Two sentences summarizing what must change and how achievable this is given the current profile.",
+  "health_actions": [
+    "Specific health action with exact target numbers",
+    "Specific health action with exact target numbers",
+    "Specific health action with exact target numbers"
+  ],
+  "lifestyle_actions": [
+    "Specific daily activity target with exact numbers",
+    "Specific daily activity target with exact numbers",
+    "Specific daily activity target with exact numbers"
+  ],
+  "driving_actions": [
+    "Specific driving improvement action",
+    "Specific driving improvement action"
+  ],
+  "timeline": "Realistic estimate such as 3 to 6 months with consistent effort",
+  "targets": {{
+    "daily_steps": 10000,
+    "daily_calories_burned": 500,
+    "daily_active_minutes": 45,
+    "target_bmi_range": "18.5 - 24.9",
+    "risk_score_needed": 0.35
+  }},
+  "difficulty": "Easy or Moderate or Challenging"
+}}
+
+Focus improvements on the weakest areas: health is {health_risk}, lifestyle is {lifestyle_risk}, driving is {driving_risk}.
+Do not use emojis. Be specific with numbers. Be encouraging but realistic.
+""".strip()
+
+    # 8. Call Groq
+    if not client:
+        return {"success": False, "error": "AI service not configured. Please check GROQ_API_KEY in backend .env"}
+
+    try:
+        completion = client.chat.completions.create(
+            messages=[
+                {
+                    "role": "system",
+                    "content": (
+                        "You are a precise AI insurance underwriter. "
+                        "Always return only valid JSON with no markdown formatting, no emojis, no extra explanation."
+                    ),
+                },
+                {"role": "user", "content": prompt},
+            ],
+            model="llama-3.3-70b-versatile",
+            temperature=0.3,
+            max_tokens=1000,
+        )
+        raw = completion.choices[0].message.content.strip()
+
+        # Strip markdown code fences if model wraps them anyway
+        if raw.startswith("```"):
+            parts = raw.split("```")
+            raw = parts[1] if len(parts) > 1 else raw
+            if raw.startswith("json"):
+                raw = raw[4:]
+            raw = raw.strip()
+        if raw.endswith("```"):
+            raw = raw[:-3].strip()
+
+        advice = json.loads(raw)
+
+    except json.JSONDecodeError:
+        advice = {
+            "summary": "The AI returned an unstructured response. Key actions are listed below.",
+            "health_actions": ["Maintain a BMI between 18.5 and 24.9", "Keep resting heart rate below 80 bpm", "Get an annual health check-up"],
+            "lifestyle_actions": [f"Increase daily steps from {avg_steps} to 10,000", "Burn at least 400 calories per day through exercise", "Achieve 30 active minutes daily"],
+            "driving_actions": ["Avoid speeding violations for 12 consecutive months", "Maintain a clean driving record"],
+            "timeline": "3 to 6 months with consistent effort",
+            "targets": {
+                "daily_steps": 10000,
+                "daily_calories_burned": 400,
+                "daily_active_minutes": 30,
+                "target_bmi_range": "18.5 - 24.9",
+                "risk_score_needed": float(target_plan_level) * 0.13,
+            },
+            "difficulty": "Moderate",
+        }
+    except Exception as e:
+        return {"success": False, "error": f"AI service error: {str(e)}"}
+
+    return {
+        "success":         True,
+        "current_plan":    current_plan["name"],
+        "current_level":   current_plan_level,
+        "target_plan":     target_plan["name"],
+        "target_level":    target_plan_level,
+        "current_risk":    round(current_risk, 2),
+        "monthly_savings": monthly_savings,
+        "current_premium": current_plan["premium"],
+        "target_premium":  target_plan["premium"],
+        "advice":          advice,
+        "current_metrics": {
+            "health_risk":     health_risk,
+            "lifestyle_risk":  lifestyle_risk,
+            "driving_risk":    driving_risk,
+            "avg_steps":       avg_steps,
+            "avg_calories":    avg_calories,
+            "avg_active_mins": avg_active_min,
+            "bmi":             bmi,
+            "heart_rate":      heart_rate,
+        },
+    }