azure-ai-search-insurance-claims-agent.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Insurance Claim Analysis Scenario with Semantic Kernel and Azure OpenAI Assistants API\n",
    "\n",
    "## Context\n",
    "\n",
    "In May 2024, Hurricane Alicia strikes the Gulf Coast near Houston, Texas, causing extensive property damage. Fabrikam Insurance, a leading commercial insurer, now faces numerous complex claims. One such claim is from Contoso Electronics Inc. for their retail location #TX-230 in Houston.\n",
    "\n",
    "The claim involves severe roof damage, flooding of inventory storage, compromised electrical systems, and potential business interruption. The insurer’s underwriting and claims team uses Semantic Kernel agents to orchestrate a multi-step analysis:\n",
    "\n",
    "- **Review** the claim details from the adjuster’s report.\n",
    "- **Verify** policy coverage, deductibles, and limits.\n",
    "- **Assess** the severity of damage, detect any fraud indicators, and recommend next steps.\n",
    "\n",
    "## Multi-Agent Process Using Semantic Kernel's Agent Framework\n",
    "\n",
    "We orchestrate a multi-agent process:\n",
    "\n",
    "1. **Claims Analyst**: Reviews claim documents, identifies key details, and notes any missing information.\n",
    "2. **Policy Verifier**: Checks coverage limits, applies correct deductibles, and identifies endorsements or exclusions.\n",
    "3. **Risk Assessor**: Evaluates the severity of the incident against the weather data, assesses mitigation efforts, and provides forward-looking recommendations.\n",
    "\n",
    "The `AgentGroupChat` coordinates these agents until a final recommendation is formed."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "!pip install semantic-kernel==1.17.0 python-dotenv aiofiles nest_asyncio azure-search-documents pyperclip python-docx matplotlib seaborn\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Imports & Configuration"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "import asyncio\n",
    "import os\n",
    "from pathlib import Path\n",
    "from datetime import datetime, timedelta\n",
    "from dotenv import load_dotenv\n",
    "import logging\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from docx import Document\n",
    "from pathlib import Path\n",
    "\n",
    "from semantic_kernel.kernel import Kernel\n",
    "from semantic_kernel.agents.open_ai.azure_assistant_agent import AzureAssistantAgent\n",
    "from semantic_kernel.agents import AgentGroupChat, ChatCompletionAgent\n",
    "from semantic_kernel.connectors.ai.open_ai.services.azure_chat_completion import AzureChatCompletion\n",
    "from semantic_kernel.agents.strategies.selection.kernel_function_selection_strategy import KernelFunctionSelectionStrategy\n",
    "from semantic_kernel.agents.strategies.termination.kernel_function_termination_strategy import KernelFunctionTerminationStrategy\n",
    "from semantic_kernel.contents.utils.author_role import AuthorRole\n",
    "from semantic_kernel.contents.chat_message_content import ChatMessageContent\n",
    "from semantic_kernel.functions.kernel_function_from_prompt import KernelFunctionFromPrompt\n",
    "from semantic_kernel.contents.streaming_file_reference_content import StreamingFileReferenceContent\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import pandas as pd\n",
    "\n",
    "import nest_asyncio\n",
    "nest_asyncio.apply()\n",
    "\n",
    "load_dotenv()\n",
    "\n",
    "AZURE_OPENAI_API_KEY = os.getenv(\"AZURE_OPENAI_API_KEY\")\n",
    "AZURE_OPENAI_ENDPOINT = os.getenv(\"AZURE_OPENAI_ENDPOINT\")\n",
    "AZURE_OPENAI_DEPLOYMENT = os.getenv(\"AZURE_OPENAI_CHAT_COMPLETION_DEPLOYED_MODEL_NAME\")\n",
    "AZURE_OPENAI_API_VERSION = os.getenv(\"AZURE_OPENAI_API_VERSION\", \"2024-10-01-preview\")\n",
    "\n",
    "if not all([AZURE_OPENAI_API_KEY, AZURE_OPENAI_ENDPOINT, AZURE_OPENAI_DEPLOYMENT]):\n",
    "    raise ValueError(\"Missing Azure OpenAI configuration.\")\n",
    "\n",
    "logging.getLogger().setLevel(logging.WARNING)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Create Kernel and Documents"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_kernel():\n",
    "    kernel = Kernel()\n",
    "    kernel.add_service(\n",
    "        AzureChatCompletion(\n",
    "            deployment_name=AZURE_OPENAI_DEPLOYMENT,\n",
    "            endpoint=AZURE_OPENAI_ENDPOINT,\n",
    "            api_key=AZURE_OPENAI_API_KEY,\n",
    "            api_version=AZURE_OPENAI_API_VERSION,\n",
    "        )\n",
    "    )\n",
    "    return kernel\n",
    "\n",
    "def save_sample_documents():\n",
    "    # Claim Report (Houston scenario)\n",
    "    claim_report = f\"\"\"\n",
    "INSURANCE CLAIM REPORT\n",
    "----------------------\n",
    "Claim ID: HURTX-2024-0456\n",
    "Date Filed: {datetime.now().strftime('%B %d, %Y')}\n",
    "Claimant: Contoso Electronics Inc.\n",
    "Store Location: Contoso Electronics #TX-230\n",
    "Address: 1125 Main Street, Houston, TX 77002\n",
    "Contact: Gabriel Allen (Store Manager)\n",
    "Contact Phone: (713) 555-4422\n",
    "Contact Email: gabriel.allen@contoso.com\n",
    "\n",
    "DAMAGE ASSESSMENT\n",
    "-----------------\n",
    "Incident Type: Hurricane Alicia\n",
    "Date of Incident: May 15, 2024\n",
    "Time of Incident: ~02:30 AM CST\n",
    "\n",
    "DAMAGE DETAILS\n",
    "--------------\n",
    "- Severe roof damage and water intrusion on the south side\n",
    "- Flooded inventory storage (electronics submerged)\n",
    "- Electrical panels shorted due to water ingress\n",
    "- Potential structural compromise in south wing support beams\n",
    "\n",
    "Preliminary Damage Estimates:\n",
    "- Inventory Loss: $380,000\n",
    "- Structural Damage: $200,000\n",
    "- Equipment Replacement: $160,000\n",
    "- Business Interruption: TBD (further accounting needed)\n",
    "\n",
    "Mitigation Steps Taken:\n",
    "1. Secure site and prevent unauthorized entry\n",
    "2. Engage emergency restoration team for water extraction\n",
    "3. Move salvageable inventory to upper floors\n",
    "4. Temporary roof patch installed\n",
    "\n",
    "Photos and videos available. Awaiting professional structural engineer’s report.\n",
    "\n",
    "FILED BY\n",
    "--------\n",
    "Adjuster: Lauren Chen\n",
    "Fabrikam Insurance Adjuster ID: TX-ADJ-3112\n",
    "Agency: Gulf Coast Claims Solutions\n",
    "Initial On-Site Visit: {(datetime.now() + timedelta(days=1)).strftime('%B %d, %Y')}\n",
    "\"\"\".strip()\n",
    "\n",
    "    # Policy Details (Fabrikam Insurance)\n",
    "    policy_details = f\"\"\"\n",
    "FABRIKAM INSURANCE COMMERCIAL POLICY\n",
    "-------------------------------------\n",
    "Policy Number: CP-TX-2024087\n",
    "Policyholder: Contoso Electronics Inc.\n",
    "Policy Period: January 1, 2024 - December 31, 2024\n",
    "Policy Type: Comprehensive Commercial Property Insurance\n",
    "\n",
    "COVERAGE & LIMITS\n",
    "-----------------\n",
    "Total Insured Value: $2,000,000\n",
    "Hurricane/Wind Coverage: Included\n",
    "Flood Coverage: Included (Deductible: $50,000)\n",
    "Business Interruption: Included (sub-limit $500,000)\n",
    "Replacement Cost: Included\n",
    "\n",
    "Sublimits:\n",
    "- Building Structure: Up to $1,000,000\n",
    "- Business Property (Incl. Inventory): Up to $600,000\n",
    "- Electronic Equipment: Up to $200,000\n",
    "\n",
    "DEDUCTIBLES\n",
    "-----------\n",
    "Hurricane/Wind Deductible: 5% of total insured value ($100,000)\n",
    "Flood Deductible: $50,000\n",
    "Standard Deductible: $25,000\n",
    "\n",
    "ENDORSEMENTS & NOTES\n",
    "--------------------\n",
    "- Hurricane Preparedness Credit applied\n",
    "- Extended Replacement Cost Endorsement active\n",
    "- Business Interruption Rider active\n",
    "\n",
    "UNDERWRITING\n",
    "------------\n",
    "Underwriter: Fabrikam Insurance Underwriting Dept.\n",
    "Last Risk Assessment: April 5, 2024 (No major hazards noted at that time)\n",
    "Annual Premium: $42,000 paid in full on January 5, 2024\n",
    "    \"\"\".strip()\n",
    "\n",
    "    # Weather Report (Hurricane Alicia in Houston)\n",
    "    weather_report = f\"\"\"\n",
    "NOAA SEVERE WEATHER REPORT\n",
    "--------------------------\n",
    "Report Date: {datetime.now().strftime('%B %d, %Y')}\n",
    "Event: Hurricane Alicia (2024 Season)\n",
    "\n",
    "STORM DETAILS\n",
    "-------------\n",
    "Category: 3\n",
    "Max Sustained Winds: ~120 mph\n",
    "Min Central Pressure: ~950 mb\n",
    "\n",
    "HOUSTON AREA IMPACT\n",
    "-------------------\n",
    "Landfall: Near Galveston, TX\n",
    "Time: May 15, 2024, ~02:10 AM CST\n",
    "Duration of Severe Conditions: 6-8 hours over Houston\n",
    "\n",
    "METEOROLOGY\n",
    "-----------\n",
    "Winds: Sustained 100-120 mph, Gusts 130+ mph\n",
    "Rainfall: 10-14 inches total, peak 3 in/hr\n",
    "Storm Surge: 8-10 feet along coastal areas\n",
    "\n",
    "DAMAGE POTENTIAL\n",
    "----------------\n",
    "Infrastructure Risk: Very High\n",
    "Property Damage Likelihood: High\n",
    "Business Disruption Probability: Significant\n",
    "\n",
    "Issued By: Dr. Marisol Reyes\n",
    "National Hurricane Center - Houston Office\n",
    "\"\"\".strip()\n",
    "\n",
    "    base_directory = Path.cwd() / \"insurance_docs\"\n",
    "    base_directory.mkdir(exist_ok=True)\n",
    "\n",
    "    documents = {\n",
    "        \"claim_report.txt\": claim_report,\n",
    "        \"policy_details.txt\": policy_details,\n",
    "        \"weather_report.txt\": weather_report,\n",
    "    }\n",
    "\n",
    "    for fname, content in documents.items():\n",
    "        (base_directory / fname).write_text(content, encoding=\"utf-8\")\n",
    "\n",
    "    return [str(base_directory / f) for f in documents.keys()]\n",
    "\n",
    "document_files = save_sample_documents()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Creating Agents\n",
    "\n",
    "We will have three agents:\n",
    "\n",
    "1. **Claims Analyst (with File Search)**: Reviews claim, policy, and weather documents.\n",
    "2. **Policy Verifier**: Checks coverage and compliance.\n",
    "3. **Risk Assessor**: Evaluates risks and provides recommendations."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_output_directories():\n",
    "    # Create base directory for insurance documents\n",
    "    base_directory = Path.cwd() / \"insurance_docs\"\n",
    "    base_directory.mkdir(exist_ok=True)\n",
    "\n",
    "    # Create subdirectories for different types of outputs\n",
    "    reports_dir = base_directory / \"generated_reports\"\n",
    "    reports_dir.mkdir(exist_ok=True)\n",
    "\n",
    "    visualizations_dir = base_directory / \"visualizations\"\n",
    "    visualizations_dir.mkdir(exist_ok=True)\n",
    "\n",
    "    letters_dir = base_directory / \"letters\"\n",
    "    letters_dir.mkdir(exist_ok=True)\n",
    "\n",
    "    return {\n",
    "        \"base\": str(base_directory),\n",
    "        \"reports\": str(reports_dir),\n",
    "        \"visualizations\": str(visualizations_dir),\n",
    "        \"letters\": str(letters_dir),\n",
    "    }\n",
    "\n",
    "\n",
    "async def create_agents(kernel: Kernel, document_files: list[str]):\n",
    "    output_dirs = create_output_directories()\n",
    "\n",
    "    claims_analyst = await AzureAssistantAgent.create(\n",
    "        kernel=kernel,\n",
    "        deployment_name=AZURE_OPENAI_DEPLOYMENT,\n",
    "        endpoint=AZURE_OPENAI_ENDPOINT,\n",
    "        api_key=AZURE_OPENAI_API_KEY,\n",
    "        name=\"Claims_Analyst\",\n",
    "        instructions=\"\"\"You are an experienced Insurance Claims Analyst evaluating claim HURTX-2024-0456 for hurricane damage.\n",
    "\n",
    "YOUR PROCESS:\n",
    "1. First Response: Initial analysis only\n",
    "- Review claim details and identify missing information\n",
    "- DO NOT make recommendations yet\n",
    "\n",
    "2. Final Response: Only after Policy_Verifier and Risk_Assessor input\n",
    "- Consider their input for coverage and risk assessment\n",
    "- Make final recommendation\n",
    "- Structure in JSON format:\n",
    "{\n",
    "    \"final_recommendation\": \"approve\"|\"deny\"|\"partial_approve\",\n",
    "    \"rationale\": [...],\n",
    "    \"approved_amount\": number,\n",
    "    \"deductions\": {...},\n",
    "    \"conditions\": [...]\n",
    "}\n",
    "\n",
    "KEY REQUIREMENTS:\n",
    "- Wait for both other agents before final recommendation\n",
    "- Verify damage aligns with weather data\n",
    "- Flag any inconsistencies\n",
    "- Consider policy limits and deductibles\"\"\",\n",
    "        enable_file_search=True,\n",
    "        vector_store_filenames=document_files,\n",
    "    )\n",
    "\n",
    "    policy_verifier = ChatCompletionAgent(\n",
    "        service_id=AZURE_OPENAI_DEPLOYMENT,\n",
    "        kernel=kernel,\n",
    "        name=\"Policy_Verifier\",\n",
    "        instructions=\"\"\"You are a Policy Verification Specialist examining claim HURTX-2024-0456.\n",
    "\n",
    "RESPOND IMMEDIATELY AFTER Claims_Analyst's initial review with:\n",
    "{\n",
    "    \"coverage_verification\": {\n",
    "        \"hurricane_coverage\": {...},\n",
    "        \"flood_coverage\": {...},\n",
    "        \"business_interruption\": {...}\n",
    "    },\n",
    "    \"applicable_deductibles\": [...],\n",
    "    \"available_limits\": {...},\n",
    "    \"coverage_gaps\": [...],\n",
    "    \"policy_compliance\": \"compliant\"|\"non_compliant\",\n",
    "    \"special_conditions\": [...]\n",
    "}\"\"\",\n",
    "    )\n",
    "\n",
    "    risk_assessor = ChatCompletionAgent(\n",
    "        service_id=AZURE_OPENAI_DEPLOYMENT,\n",
    "        kernel=kernel,\n",
    "        name=\"Risk_Assessor\",\n",
    "        instructions=\"\"\"You are a Risk Assessment Specialist analyzing claim HURTX-2024-0456.\n",
    "\n",
    "RESPOND AFTER Policy_Verifier with:\n",
    "{\n",
    "    \"risk_analysis\": {\n",
    "        \"weather_correlation\": {...},\n",
    "        \"mitigation_effectiveness\": {...},\n",
    "        \"future_risks\": [...]\n",
    "    },\n",
    "    \"damage_validation\": {...},\n",
    "    \"mitigation_assessment\": {...},\n",
    "    \"recommendations\": [...],\n",
    "    \"risk_score\": number\n",
    "}\"\"\",\n",
    "    )\n",
    "\n",
    "    data_analyzer = await AzureAssistantAgent.create(\n",
    "        kernel=kernel,\n",
    "        deployment_name=AZURE_OPENAI_DEPLOYMENT,\n",
    "        endpoint=AZURE_OPENAI_ENDPOINT,\n",
    "        api_key=AZURE_OPENAI_API_KEY,\n",
    "        name=\"Data_Analyzer\",\n",
    "        instructions=f\"\"\"You are a Data Analysis Specialist who creates visualizations after the final claim decision.\n",
    "        \n",
    "Wait for Claims_Analyst's final recommendation, then:\n",
    "1. Import required packages:\n",
    "```python\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import os\n",
    "```\n",
    "\n",
    "2. Create and save visualizations to {output_dirs['visualizations']}:\n",
    "```python\n",
    "# Example\n",
    "plt.figure(figsize=(10, 6))\n",
    "plt.savefig('{output_dirs['visualizations']}/chart_name.png')\n",
    "plt.close()\n",
    "```\n",
    "\n",
    "3. Required visualizations:\n",
    "   - damage_cost_breakdown.png\n",
    "   - deductions_applied.png\n",
    "   - risk_assessment_metrics.png\n",
    "\n",
    "IMPORTANT: Always use plt.close() after saving each figure!\n",
    "\n",
    "Output Format:\n",
    "{{\n",
    "    \"analysis_type\": \"visualization\",\n",
    "    \"generated_files\": [...],\n",
    "    \"insights\": [...]\n",
    "}}\"\"\",\n",
    "        enable_code_interpreter=True\n",
    "    )\n",
    "\n",
    "    doc_generator = await AzureAssistantAgent.create(\n",
    "        kernel=kernel,\n",
    "        deployment_name=AZURE_OPENAI_DEPLOYMENT,\n",
    "        endpoint=AZURE_OPENAI_ENDPOINT,\n",
    "        api_key=AZURE_OPENAI_API_KEY,\n",
    "        name=\"Doc_Generator\",\n",
    "        instructions=f\"\"\"You are a Document Generation Specialist who creates formal reports.\n",
    "        \n",
    "Wait for Data_Analyzer's visualizations, then:\n",
    "1. Import required packages:\n",
    "```python\n",
    "from docx import Document\n",
    "import os\n",
    "from pathlib import Path\n",
    "```\n",
    "\n",
    "2. Save documents to these paths:\n",
    "   - Summary: '{output_dirs['reports']}/HURTX-2024-0456_Summary.docx'\n",
    "   - Letter: '{output_dirs['letters']}/HURTX-2024-0456_Letter.docx'\n",
    "   - Internal: '{output_dirs['reports']}/HURTX-2024-0456_Internal.docx'\n",
    "\n",
    "3. Create documents using:\n",
    "```python\n",
    "doc = Document()\n",
    "doc.add_heading('Title', level=1)\n",
    "doc.add_paragraph('Content')\n",
    "doc.save('path/to/document.docx')\n",
    "```\n",
    "\n",
    "IMPORTANT: Always use full paths when saving files!\n",
    "\n",
    "Output Format:\n",
    "{{\n",
    "    \"document_type\": \"report|letter|summary\",\n",
    "    \"generated_files\": [...],\n",
    "    \"content_summary\": {{...}}\n",
    "}}\"\"\",\n",
    "        enable_code_interpreter=True\n",
    "    )\n",
    "\n",
    "    return claims_analyst, policy_verifier, risk_assessor, data_analyzer, doc_generator"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Setting Up Group Chat\n",
    "We’ll define a selection strategy to determine which agent responds next and a termination strategy to know when the conversation is complete. For demo purposes, we might simplify the logic or leave it as-is."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [],
   "source": [
    "async def setup_group_chat(kernel: Kernel, claims_analyst, policy_verifier, risk_assessor, data_analyzer, doc_generator):\n",
    "    selection_function = KernelFunctionFromPrompt(\n",
    "        function_name=\"selection\",\n",
    "        prompt=\"\"\"You determine which agent speaks next in the claims analysis conversation.\n",
    "        \n",
    "Previous messages: {{{$history}}}\n",
    "\n",
    "Strict speaking order:\n",
    "1. Claims_Analyst first (initial review)\n",
    "2. Then Policy_Verifier \n",
    "3. Then Risk_Assessor\n",
    "4. Then Claims_Analyst again (final decision)\n",
    "5. Then Data_Analyzer creates visualizations\n",
    "6. Finally Doc_Generator creates reports\n",
    "\n",
    "Rules:\n",
    "- Output ONLY the exact name: Claims_Analyst, Policy_Verifier, Risk_Assessor, Data_Analyzer, or Doc_Generator\n",
    "- No explanations or other text\n",
    "- Must follow the order above exactly\n",
    "- If the last message was from Claims_Analyst and didn't mention a final recommendation, output \"Policy_Verifier\"\n",
    "- If the last message was from Policy_Verifier, output \"Risk_Assessor\"\n",
    "- If the last message was from Risk_Assessor, output \"Claims_Analyst\"\n",
    "- If the last message was from Claims_Analyst with final recommendation, output \"Data_Analyzer\"\n",
    "- If the last message was from Data_Analyzer, output \"Doc_Generator\"\n",
    "- If no messages yet, output \"Claims_Analyst\"\n",
    "\"\"\")\n",
    "\n",
    "    termination_function = KernelFunctionFromPrompt(\n",
    "        function_name=\"termination\",\n",
    "        prompt=\"\"\"You check if the claims analysis conversation is truly complete.\n",
    "\n",
    "Previous messages: {{{$history}}}\n",
    "\n",
    "Requirements for COMPLETE status:\n",
    "1. Claims_Analyst gave initial review\n",
    "2. THEN Policy_Verifier analyzed coverage\n",
    "3. THEN Risk_Assessor evaluated risks\n",
    "4. THEN Claims_Analyst gave final recommendation\n",
    "5. THEN Data_Analyzer created visualizations\n",
    "6. FINALLY Doc_Generator created reports\n",
    "\n",
    "Output only one word:\n",
    "- \"COMPLETE\" if ALL steps happened IN ORDER\n",
    "- \"INCOMPLETE\" if any steps are missing or out of order\n",
    "\n",
    "Special rules:\n",
    "- The conversation can't be complete if less than 6 messages\n",
    "- Must have documents from Doc_Generator as last message\n",
    "- If any step is missing, it's INCOMPLETE\n",
    "\"\"\")\n",
    "\n",
    "    group_chat = AgentGroupChat(\n",
    "        agents=[claims_analyst, policy_verifier, risk_assessor, data_analyzer, doc_generator],\n",
    "        selection_strategy=KernelFunctionSelectionStrategy(\n",
    "            function=selection_function,\n",
    "            kernel=kernel,\n",
    "            result_parser=lambda r: str(r.value[0]) if r.value else \"Claims_Analyst\",\n",
    "            agent_variable_name=\"agents\",\n",
    "            history_variable_name=\"history\",\n",
    "        ),\n",
    "        termination_strategy=KernelFunctionTerminationStrategy(\n",
    "            agents=[claims_analyst],\n",
    "            function=termination_function,\n",
    "            kernel=kernel,\n",
    "            result_parser=lambda r: \"complete\" in str(r.value[0]).lower(),\n",
    "            history_variable_name=\"history\",\n",
    "            maximum_iterations=30,\n",
    "        ),\n",
    "    )\n",
    "\n",
    "    return group_chat"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Processing a Claim\n",
    "This function executes the group chat workflow, starting with a user message (the claim scenario). "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [],
   "source": [
    "async def process_claim(group_chat: AgentGroupChat, claim_id: str):\n",
    "    print(f\"🔄 Starting analysis for claim {claim_id}\")\n",
    "    \n",
    "    # Reset the chat completion state\n",
    "    group_chat.is_complete = False\n",
    "    \n",
    "    # Add the initial message\n",
    "    await group_chat.add_chat_message(\n",
    "        ChatMessageContent(\n",
    "            role=AuthorRole.USER, \n",
    "            content=f\"Please analyze claim {claim_id} from the provided documents. Each agent should follow their specific instructions and respond in turn.\"\n",
    "        )\n",
    "    )\n",
    "\n",
    "    current_agent = None\n",
    "    message_count = 0\n",
    "    file_ids = []\n",
    "    \n",
    "    try:\n",
    "        async for response in group_chat.invoke():\n",
    "            if response.name != current_agent:\n",
    "                current_agent = response.name\n",
    "                print(f\"\\n👤 {current_agent}:\")\n",
    "                message_count += 1\n",
    "            \n",
    "            if response.content:\n",
    "                print(response.content.strip())\n",
    "                \n",
    "            # Collect any generated files\n",
    "            if hasattr(response, 'items'):\n",
    "                file_ids.extend([\n",
    "                    item.file_id for item in response.items \n",
    "                    if isinstance(item, StreamingFileReferenceContent)\n",
    "                ])\n",
    "\n",
    "            # Prevent premature termination\n",
    "            if message_count < 6:  # Updated for 6 agents\n",
    "                group_chat.is_complete = False\n",
    "\n",
    "        if message_count < 6:\n",
    "            print(\"\\n⚠️ Warning: Not all agents completed their analysis\")\n",
    "        else:\n",
    "            print(\"\\n✅ Claim analysis complete!\")\n",
    "            \n",
    "        # Handle any generated files\n",
    "        if file_ids:\n",
    "            print(\"\\n📁 Generated files:\")\n",
    "            for file_id in file_ids:\n",
    "                print(f\"- {file_id}\")\n",
    "            \n",
    "    except Exception as e:\n",
    "        print(f\"\\n❌ Error during analysis: {str(e)}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Run the Entire Workflow\n",
    "\n",
    "In this final cell, we:\n",
    "\n",
    "- Initialize kernel & documents\n",
    "- Create agents\n",
    "- Setup group chat\n",
    "- Process the claim\n",
    "\n",
    "Run this cell and review the outputs. You can add additional Markdown cells afterwards to comment on the results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "🔄 Starting analysis for claim HURTX-2024-0456\n",
      "\n",
      "👤 Claims_Analyst:\n",
      "### Initial Analysis of Claim HURTX-2024-0456\n",
      "\n",
      "**Claim Details:**\n",
      "- Claim ID: HURTX-2024-0456\n",
      "- Claimant: Contoso Electronics Inc., Store #TX-230\n",
      "- Date of Incident: May 15, 2024\n",
      "- Type of Incident: Hurricane Alicia\n",
      "\n",
      "**Damage Assessment:**\n",
      "- **Severe roof damage** and **water intrusion** noted on the south side.\n",
      "- **Flooding** impacted inventory with electronics submerged.\n",
      "- **Electrical panels** damaged due to water ingress.\n",
      "- Potential **structural issues** identified in support beams of the south wing.\n",
      "\n",
      "**Preliminary Damage Estimates:**\n",
      "1. Inventory Loss: $380,000\n",
      "2. Structural Damage: $200,000\n",
      "3. Equipment Replacement: $160,000\n",
      "4. Business Interruption: TBD\n",
      "\n",
      "**Mitigation Steps Taken:**\n",
      "- Secured the site\n",
      "- Engaged emergency restoration for water extraction\n",
      "- Moved salvageable inventory to upper floors\n",
      "- Installed temporary roof patch\n",
      "\n",
      "**Policy Details:**\n",
      "- Total Insured Value: $2,000,000\n",
      "- Coverage includes Hurricane/Wind and Flood\n",
      "- Deductibles:\n",
      "  - Hurricane/Wind: $100,000 (5% of Total Insured Value)\n",
      "  - Flood: $50,000\n",
      "- Business Interruption: $500,000 limit\n",
      "\n",
      "**Weather Impact:**\n",
      "- Hurricane Alicia was a Category 3 storm with winds up to 120 mph and significant rainfall over the affected area【4:2†source】.\n",
      "\n",
      "### Missing Information:\n",
      "- The total for Business Interruption is still to be determined (TBD).\n",
      "- A professional structural engineer's report is pending, which may help in assessing the extent of structural damage.\n",
      "- Verification of actual inventory loss against policy limits for Business Property and Electronic Equipment needs confirmation.\n",
      "\n",
      "The analysis will require input from the Policy Verifier and Risk Assessor before a final recommendation can be made.\n",
      "\n",
      "👤 Policy_Verifier:\n",
      "{\n",
      "    \"coverage_verification\": {\n",
      "        \"hurricane_coverage\": {\n",
      "            \"coverage_status\": \"covered\",\n",
      "            \"applicable_conditions\": \"Standard policy conditions apply for hurricane damage\",\n",
      "            \"deductibles\": \"$100,000\"\n",
      "        },\n",
      "        \"flood_coverage\": {\n",
      "            \"coverage_status\": \"covered\",\n",
      "            \"applicable_conditions\": \"Flood coverage applies to damages caused by rising water\",\n",
      "            \"deductibles\": \"$50,000\"\n",
      "        },\n",
      "        \"business_interruption\": {\n",
      "            \"coverage_status\": \"covered\",\n",
      "            \"applicable_conditions\": \"Business Interruption covers loss of income during restoration period\",\n",
      "            \"limits\": \"$500,000\"\n",
      "        }\n",
      "    },\n",
      "    \"applicable_deductibles\": [\n",
      "        \"$100,000 for hurricane damage\",\n",
      "        \"$50,000 for flood damage\"\n",
      "    ],\n",
      "    \"available_limits\": {\n",
      "        \"hurricane_coverage\": \"$2,000,000\",\n",
      "        \"flood_coverage\": \"$2,000,000\",\n",
      "        \"business_interruption\": \"$500,000\"\n",
      "    },\n",
      "    \"coverage_gaps\": [\n",
      "        \"Potential limits on Business Interruption claims if not substantiated with income loss documentation.\"\n",
      "    ],\n",
      "    \"policy_compliance\": \"compliant\",\n",
      "    \"special_conditions\": [\n",
      "        \"Claim must be substantiated with full inventory assessment and business loss documentation.\",\n",
      "        \"Pending structural engineering assessment could affect claims payout for structural repairs.\"\n",
      "    ]\n",
      "}\n",
      "\n",
      "👤 Risk_Assessor:\n",
      "{\n",
      "    \"risk_analysis\": {\n",
      "        \"weather_correlation\": {\n",
      "            \"hurricane_category\": \"3\",\n",
      "            \"wind_speed\": \"120 mph\",\n",
      "            \"rainfall\": \"significant\",\n",
      "            \"historical_data\": \"The area has experienced similar hurricane patterns in the last decade.\"\n",
      "        },\n",
      "        \"mitigation_effectiveness\": {\n",
      "            \"response_time\": \"Immediate response noted post-storm\",\n",
      "            \"measures_taken\": [\n",
      "                \"Site secured and emergency restoration engaged\",\n",
      "                \"Salvageable inventory relocated\",\n",
      "                \"Temporary roof patch installed\"\n",
      "            ],\n",
      "            \"comment\": \"Initial mitigation steps were effective in limiting further damage.\"\n",
      "        },\n",
      "        \"future_risks\": [\n",
      "            \"Increased frequency of severe weather events due to climate change\",\n",
      "            \"Potential vulnerability of infrastructure in future storms\",\n",
      "            \"Need for enhanced preventive measures and building resilience\"\n",
      "        ]\n",
      "    },\n",
      "    \"damage_validation\": {\n",
      "        \"assessed_damage\": {\n",
      "            \"total_estimated_damage\": \"$740,000\",\n",
      "            \"inventory\": \"$380,000\",\n",
      "            \"structural\": \"$200,000\",\n",
      "            \"equipment\": \"$160,000\"\n",
      "        },\n",
      "        \"pending_items\": [\n",
      "            \"Business interruption claim estimation\",\n",
      "            \"Final engineer's report on structural integrity\"\n",
      "        ],\n",
      "        \"verification_status\": \"Initial validation completed; further documentation required.\"\n",
      "    },\n",
      "    \"mitigation_assessment\": {\n",
      "        \"current_measures\": [\n",
      "            \"Emergency restoration team engaged\",\n",
      "            \"Temporary protective measures in place\"\n",
      "        ],\n",
      "        \"effectiveness_rating\": \"Good\",\n",
      "        \"recommendations_for_improvement\": [\n",
      "            \"Implement flood-proofing measures for the inventory area\",\n",
      "            \"Conduct a full risk assessment to improve future resilience\"\n",
      "        ]\n",
      "    },\n",
      "    \"recommendations\": [\n",
      "        \"Complete the structural engineering assessment promptly\",\n",
      "        \"Accurate accounting of business interruption losses should be compiled for claim submission\",\n",
      "        \"Invest in comprehensive flood and hurricane mitigation strategies going forward\"\n",
      "    ],\n",
      "    \"risk_score\": 85\n",
      "}\n",
      "\n",
      "👤 Claims_Analyst:\n",
      "{\n",
      "    \"final_recommendation\": \"partial_approve\",\n",
      "    \"rationale\": [\n",
      "        \"The damage caused by Hurricane Alicia is covered under the comprehensive commercial property insurance policy.\",\n",
      "        \"Claim estimates for inventory, structural damage, and equipment replacement align with the policy coverage limits.\",\n",
      "        \"The claim for business interruption remains pending due to the need for further documentation and estimation, which may rеsult in a limit to coverage.\"\n",
      "    ],\n",
      "    \"approved_amount\": 590000,\n",
      "    \"deductions\": {\n",
      "        \"hurricane_deduction\": 100000,\n",
      "        \"flood_deduction\": 50000\n",
      "    },\n",
      "    \"conditions\": [\n",
      "        \"Approval for business interruption coverage pending final assessment of income loss due to operational downtime.\",\n",
      "        \"Claim must be finalized with a comprehensive structural assessment report.\",\n",
      "        \"Documentation substantiating inventory loss must be submitted for review.\"\n",
      "    ]\n",
      "}\n",
      "\n",
      "👤 Data_Analyzer:\n",
      "import matplotlib.pyplot as plt\n",
      "import seaborn as sns\n",
      "import os\n",
      "\n",
      "# Create directory if it does not exist\n",
      "output_dir = r'c:\\Dev\\azure-ai-search-python-playground\\insurance_docs\\visualizations'\n",
      "os.makedirs(output_dir, exist_ok=True)\n",
      "\n",
      "# Sample data for visualizations\n",
      "damage_data = {\n",
      "    'Damage Type': ['Inventory', 'Structural', 'Equipment', 'Business Interruption'],\n",
      "    'Cost ($)': [380000, 200000, 160000, 0]  # Business Interruption is TBD\n",
      "}\n",
      "\n",
      "deduction_data = {\n",
      "    'Deduction Type': ['Hurricane Deduction', 'Flood Deduction'],\n",
      "    'Amount ($)': [100000, 50000]\n",
      "}\n",
      "\n",
      "risk_metrics = {\n",
      "    'Metric': ['Risk Score', 'Potential Future Risks'],\n",
      "    'Value': [85, 'Increased frequency of storms']  # Qualitative for demonstration\n",
      "}\n",
      "\n",
      "# Visualization 1: Damage Cost Breakdown\n",
      "plt.figure(figsize=(10, 6))\n",
      "sns.barplot(x='Damage Type', y='Cost ($)', data=damage_data)\n",
      "plt.title('Damage Cost Breakdown')\n",
      "plt.xlabel('Type of Damage')\n",
      "plt.ylabel('Cost ($)')\n",
      "plt.savefig(os.path.join(output_dir, 'damage_cost_breakdown.png'))\n",
      "plt.close()\n",
      "\n",
      "# Visualization 2: Deductions Applied\n",
      "plt.figure(figsize=(10, 6))\n",
      "sns.barplot(x='Deduction Type', y='Amount ($)', data=deduction_data)\n",
      "plt.title('Deductions Applied')\n",
      "plt.xlabel('Type of Deduction')\n",
      "plt.ylabel('Amount ($)')\n",
      "plt.savefig(os.path.join(output_dir, 'deductions_applied.png'))\n",
      "plt.close()\n",
      "\n",
      "# Visualization 3: Risk Assessment Metrics\n",
      "plt.figure(figsize=(10, 6))\n",
      "sns.barplot(x='Metric', y='Value', data=risk_metrics)\n",
      "plt.title('Risk Assessment Metrics')\n",
      "plt.xlabel('Metric')\n",
      "plt.ylabel('Value')\n",
      "plt.savefig(os.path.join(output_dir, 'risk_assessment_metrics.png'))\n",
      "plt.close()\n",
      "\n",
      "{\n",
      "    \"analysis_type\": \"visualization\",\n",
      "    \"generated_files\": [\n",
      "        \"damage_cost_breakdown.png\", \n",
      "        \"deductions_applied.png\", \n",
      "        \"risk_assessment_metrics.png\"\n",
      "    ],\n",
      "    \"insights\": [\n",
      "        \"The majority of the damage costs are due to inventory loss, indicating high vulnerability.\",\n",
      "        \"The approved deductions significantly reduce the total claim amount.\",\n",
      "        \"The risk assessment suggests a concerning risk score, highlighting potential future vulnerabilities.\"\n",
      "    ]\n",
      "}\n",
      "import pandas as pd\n",
      "\n",
      "# Sample data for visualizations\n",
      "damage_data = {\n",
      "    'Damage Type': ['Inventory', 'Structural', 'Equipment', 'Business Interruption'],\n",
      "    'Cost ($)': [380000, 200000, 160000, None]  # Business Interruption is TBD\n",
      "}\n",
      "deduction_data = {\n",
      "    'Deduction Type': ['Hurricane Deduction', 'Flood Deduction'],\n",
      "    'Amount ($)': [100000, 50000]\n",
      "}\n",
      "risk_metrics = {\n",
      "    'Metric': ['Risk Score', 'Potential Future Risks'],\n",
      "    'Value': [85, 1]  # To plot a numeric value, we can just use a placeholder (0, 1) for \"Potential Future Risks\"\n",
      "}\n",
      "\n",
      "# Convert to DataFrames\n",
      "damage_df = pd.DataFrame(damage_data)\n",
      "deduction_df = pd.DataFrame(deduction_data)\n",
      "risk_metrics_df = pd.DataFrame(risk_metrics)\n",
      "\n",
      "# Visualization 1: Damage Cost Breakdown\n",
      "plt.figure(figsize=(10, 6))\n",
      "sns.barplot(x='Damage Type', y='Cost ($)', data=damage_df)\n",
      "plt.title('Damage Cost Breakdown')\n",
      "plt.xlabel('Type of Damage')\n",
      "plt.ylabel('Cost ($)')\n",
      "plt.savefig(os.path.join(output_dir, 'damage_cost_breakdown.png'))\n",
      "plt.close()\n",
      "\n",
      "# Visualization 2: Deductions Applied\n",
      "plt.figure(figsize=(10, 6))\n",
      "sns.barplot(x='Deduction Type', y='Amount ($)', data=deduction_df)\n",
      "plt.title('Deductions Applied')\n",
      "plt.xlabel('Type of Deduction')\n",
      "plt.ylabel('Amount ($)')\n",
      "plt.savefig(os.path.join(output_dir, 'deductions_applied.png'))\n",
      "plt.close()\n",
      "\n",
      "# Visualization 3: Risk Assessment Metrics\n",
      "plt.figure(figsize=(10, 6))\n",
      "sns.barplot(x='Metric', y='Value', data=risk_metrics_df)\n",
      "plt.title('Risk Assessment Metrics')\n",
      "plt.xlabel('Metric')\n",
      "plt.ylabel('Value')\n",
      "plt.xticks(ticks=[0, 1], labels=['Risk Score', 'Potential Future Risks'])\n",
      "plt.savefig(os.path.join(output_dir, 'risk_assessment_metrics.png'))\n",
      "plt.close()\n",
      "\n",
      "{\n",
      "    \"analysis_type\": \"visualization\",\n",
      "    \"generated_files\": [\n",
      "        \"damage_cost_breakdown.png\", \n",
      "        \"deductions_applied.png\", \n",
      "        \"risk_assessment_metrics.png\"\n",
      "    ],\n",
      "    \"insights\": [\n",
      "        \"The majority of the damage costs are due to inventory loss, indicating high vulnerability.\",\n",
      "        \"The approved deductions significantly reduce the total claim amount.\",\n",
      "        \"The risk assessment suggests a concerning risk score, highlighting potential future vulnerabilities.\"\n",
      "    ]\n",
      "}\n",
      "{\n",
      "    \"analysis_type\": \"visualization\",\n",
      "    \"generated_files\": [\n",
      "        \"damage_cost_breakdown.png\", \n",
      "        \"deductions_applied.png\", \n",
      "        \"risk_assessment_metrics.png\"\n",
      "    ],\n",
      "    \"insights\": [\n",
      "        \"The majority of the damage costs are due to inventory loss, indicating high vulnerability.\",\n",
      "        \"The approved deductions significantly reduce the total claim amount.\",\n",
      "        \"The risk assessment suggests a concerning risk score, highlighting potential future vulnerabilities.\"\n",
      "    ]\n",
      "}\n",
      "\n",
      "👤 Doc_Generator:\n",
      "from docx import Document\n",
      "import os\n",
      "\n",
      "# Define file paths for the documents\n",
      "summary_path = r'c:\\Dev\\azure-ai-search-python-playground\\insurance_docs\\generated_reports\\HURTX-2024-0456_Summary.docx'\n",
      "letter_path = r'c:\\Dev\\azure-ai-search-python-playground\\insurance_docs\\letters\\HURTX-2024-0456_Letter.docx'\n",
      "internal_path = r'c:\\Dev\\azure-ai-search-python-playground\\insurance_docs\\generated_reports\\HURTX-2024-0456_Internal.docx'\n",
      "\n",
      "# Create Summary Document\n",
      "summary_doc = Document()\n",
      "summary_doc.add_heading('Claim Summary for HURTX-2024-0456', level=1)\n",
      "summary_doc.add_paragraph('Claimant: Contoso Electronics Inc., Store #TX-230')\n",
      "summary_doc.add_paragraph('Date of Incident: May 15, 2024')\n",
      "summary_doc.add_paragraph('Type of Incident: Hurricane Alicia')\n",
      "summary_doc.add_paragraph('Total Estimated Damage: $740,000')\n",
      "summary_doc.add_paragraph('Damage Breakdown:')\n",
      "summary_doc.add_paragraph('1. Inventory Loss: $380,000')\n",
      "summary_doc.add_paragraph('2. Structural Damage: $200,000')\n",
      "summary_doc.add_paragraph('3. Equipment Replacement: $160,000')\n",
      "summary_doc.add_paragraph('Conditions for Approval: Pending further documentation on Business Interruption.')\n",
      "summary_doc.save(summary_path)\n",
      "\n",
      "# Create Letter Document\n",
      "letter_doc = Document()\n",
      "letter_doc.add_heading('Insurance Claim Decision Letter', level=1)\n",
      "letter_doc.add_paragraph('To: Contoso Electronics Inc.')\n",
      "letter_doc.add_paragraph('Claim ID: HURTX-2024-0456')\n",
      "letter_doc.add_paragraph('Date: [Insert Date]')\n",
      "letter_doc.add_paragraph('Dear Claimant,')\n",
      "letter_doc.add_paragraph('We have reviewed your claim regarding the damages incurred during Hurricane Alicia on May 15, 2024.')\n",
      "letter_doc.add_paragraph('Based on our findings, the claim has been preliminarily approved for an amount of $590,000.')\n",
      "letter_doc.add_paragraph('Please provide documentation for business interruption losses to finalize the claim.')\n",
      "letter_doc.add_paragraph('Sincerely,')\n",
      "letter_doc.add_paragraph('[Your Name]')\n",
      "letter_doc.add_paragraph('[Your Position]')\n",
      "letter_doc.save(letter_path)\n",
      "\n",
      "# Create Internal Document\n",
      "internal_doc = Document()\n",
      "internal_doc.add_heading('Internal Review Document for Claim HURTX-2024-0456', level=1)\n",
      "internal_doc.add_paragraph('Claimant: Contoso Electronics Inc.')\n",
      "internal_doc.add_paragraph('Total Damage Assessment: $740,000')\n",
      "internal_doc.add_paragraph('Status: Pending documentation for business interruption.')\n",
      "internal_doc.add_paragraph('Recommendations:')\n",
      "internal_doc.add_paragraph('1. Complete structural assessment for final approval.')\n",
      "internal_doc.add_paragraph('2. Ensure all necessary documentation for claim verification is collected.')\n",
      "internal_doc.add_paragraph('3. Evaluate risk mitigation strategies for future claims.')\n",
      "internal_doc.save(internal_path)\n",
      "\n",
      "# Output results\n",
      "{\n",
      "    \"document_type\": \"summary|letter|internal\",\n",
      "    \"generated_files\": [\n",
      "        \"HURTX-2024-0456_Summary.docx\", \n",
      "        \"HURTX-2024-0456_Letter.docx\", \n",
      "        \"HURTX-2024-0456_Internal.docx\"\n",
      "    ],\n",
      "    \"content_summary\": {\n",
      "        \"summary\": \"Summary document provides an overview of the claim, damage assessment, and conditions for approval.\",\n",
      "        \"letter\": \"The letter communicates the claim's preliminary approval and requests further documentation.\",\n",
      "        \"internal\": \"The internal document outlines the claim review status and recommendations for follow-up actions.\"\n",
      "    }\n",
      "}\n",
      "{\n",
      "    \"document_type\": \"summary|letter|internal\",\n",
      "    \"generated_files\": [\n",
      "        \"HURTX-2024-0456_Summary.docx\", \n",
      "        \"HURTX-2024-0456_Letter.docx\", \n",
      "        \"HURTX-2024-0456_Internal.docx\"\n",
      "    ],\n",
      "    \"content_summary\": {\n",
      "        \"summary\": \"Summary document provides an overview of the claim, damage assessment, and conditions for approval.\",\n",
      "        \"letter\": \"The letter communicates the claim's preliminary approval and requests further documentation.\",\n",
      "        \"internal\": \"The internal document outlines the claim review status and recommendations for follow-up actions.\"\n",
      "    }\n",
      "}\n",
      "The analysis and document generation for claim HURTX-2024-0456 has been successfully completed. Here are the key outcomes and generated documents:\n",
      "\n",
      "### Claim Overview\n",
      "- **Claim ID**: HURTX-2024-0456\n",
      "- **Claimant**: Contoso Electronics Inc.\n",
      "- **Date of Incident**: May 15, 2024\n",
      "- **Type of Incident**: Hurricane Alicia\n",
      "- **Total Estimated Damage**: $740,000\n",
      "- **Preliminary Approval Amount**: $590,000 (pending further documentation)\n",
      "\n",
      "### Generated Documents\n",
      "1. **Summary Document**: Provides an overview of the claim and conditions for approval.\n",
      "2. **Letter**: Communicates the claim's preliminary approval and requests for further documentation.\n",
      "3. **Internal Review Document**: Outlines the claim review status and recommendations for follow-up actions.\n",
      "\n",
      "### Document File Paths\n",
      "- Summary: `c:\\Dev\\azure-ai-search-python-playground\\insurance_docs\\generated_reports\\HURTX-2024-0456_Summary.docx`\n",
      "- Letter: `c:\\Dev\\azure-ai-search-python-playground\\insurance_docs\\letters\\HURTX-2024-0456_Letter.docx`\n",
      "- Internal: `c:\\Dev\\azure-ai-search-python-playground\\insurance_docs\\generated_reports\\HURTX-2024-0456_Internal.docx`\n",
      "\n",
      "If there are any additional requests or modifications needed, please let me know!\n",
      "If there are any further analyses, document modifications, or additional information required regarding claim HURTX-2024-0456, please feel free to specify, and I'll be glad to assist you!\n",
      "If you need any further analyses, adjustments to the documents, or have additional requests regarding claim HURTX-2024-0456, please let me know how I can assist you further!\n",
      "If you have any additional requests or need further analysis related to claim HURTX-2024-0456, please let me know how I can assist you!\n",
      "If there's anything more you need regarding claim HURTX-2024-0456, whether it's further analysis, changes to the documents, or any other request, just let me know! I'm here to help.\n",
      "If you have any additional requests or need further analysis related to claim HURTX-2024-0456, please let me know how I can assist you!\n",
      "If you need any further assistance regarding claim HURTX-2024-0456, whether it’s additional analyses, adjustments to the generated documents, or something specific you have in mind, please feel free to let me know!\n",
      "If you have any further requests or need additional assistance regarding claim HURTX-2024-0456, please don't hesitate to let me know. I'm here to help with any analysis, document modifications, or questions you may have!\n",
      "If you need any further assistance or have additional requests related to claim HURTX-2024-0456, please let me know. I'm here to help with any analysis, document adjustments, or any other information you might require!\n",
      "If you have any more requests or need further analysis concerning claim HURTX-2024-0456, please let me know how I can assist you! I'm here to help with any document modifications or analyses you may need.\n",
      "If you have any additional requests or further analysis needed regarding claim HURTX-2024-0456, please don't hesitate to ask. I'm here to assist with any document changes or supplementary information you might require!\n",
      "If there are any additional requests, analyses, or modifications needed regarding claim HURTX-2024-0456, please feel free to let me know. I'm here to assist with anything you require!\n",
      "If there are any further requests or analyses needed regarding claim HURTX-2024-0456, please let me know! I'm here to assist you with any document adjustments or additional information you might require.\n",
      "If you have any further requests, analyses, or modifications needed regarding claim HURTX-2024-0456, please feel free to specify. I'm here to assist you with any requirements you may have!\n",
      "If you have any additional requests or need further analysis related to claim HURTX-2024-0456, please let me know how I can assist you! I'm here to help.\n",
      "If you have any additional requests or need further analysis related to claim HURTX-2024-0456, please let me know how I can assist you!\n",
      "If there's anything more you need regarding claim HURTX-2024-0456, whether it's analysis, document modifications, or specific requests, just let me know! I'm available to help.\n",
      "\n",
      "👤 Claims_Analyst:\n",
      "If you need any further assistance regarding claim HURTX-2024-0456, whether it’s additional analyses, changes to the documents, or any other request, just let me know! I'm here to help.\n",
      "\n",
      "✅ Claim analysis complete!\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 76,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "kernel = create_kernel()\n",
    "claims_analyst, policy_verifier, risk_assessor, data_analyzer, doc_generator = asyncio.run(create_agents(kernel, document_files))\n",
    "group_chat = asyncio.run(setup_group_chat(kernel, claims_analyst, policy_verifier, risk_assessor, data_analyzer, doc_generator))\n",
    "\n",
    "asyncio.run(process_claim(group_chat, \"HURTX-2024-0456\"))\n",
    "\n",
    "# Cleanup\n",
    "asyncio.run(claims_analyst.delete())\n",
    "asyncio.run(data_analyzer.delete())\n",
    "asyncio.run(doc_generator.delete())"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": ".venv",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}