Pikl Knowledge Base

This repository contains the Pikl Insurance website and comprehensive AI-powered knowledge base management system for Intercom.

Overview

Current Status:

• ✅ 1,441 FAQs uploaded to Intercom
• ✅ 100% policy coverage (26 policies processed)
• ✅ 135 customer knowledge gaps identified
• ✅ 22 gaps addressed (16.3% coverage)
• ✅ 8 category tags created for organization

Contents

Website

• Mock Pikl Insurance website with Intercom Fin agent integration
• Responsive design with brand assets
• Live chat support through Intercom

pikl-kb-processor

AI-powered knowledge base processing system that:

• Ingests customer interactions from emails and phone call transcripts
• Identifies knowledge gaps in the existing knowledge base
• Generates FAQs and articles using AI to fill those gaps
• Analyzes coverage of customer questions and policy information
• Provides visual dashboard with Streamlit for monitoring and analytics

See the pikl-kb-processor documentation for detailed usage instructions.

faq-system

Complete FAQ management and Intercom integration system:

Core Features:

• Generate comprehensive FAQs from 26 policy wording documents
• Upload internal and public articles to Intercom
• De-duplicate articles by title
• Organize articles into folders
• Create and apply category tags
• Analyze knowledge gap coverage

Available Scripts:

1. FAQ Generation & Upload

   • generate_faqs.py - Generate FAQs from policy documents
   • enhance_faqs_for_agents.py - Enhance FAQs with agent-friendly details
   • intercom_import.py - Upload FAQs to Intercom as internal articles
   • add_public_faq.py - Create public Help Center articles

2. Organization & Management

   • intercom_dedup.py - Remove duplicate articles from Intercom
   • move_to_folder.py - Organize articles into folders
   • tag_articles.py - Create tags and apply to articles

3. Analytics

   • analyze_gap_coverage.py - Analyze how FAQs address customer knowledge gaps
   • coverage_analysis.py - Analyze FAQ coverage of policy information

Categories:

• Exclusions
• Claims Requirements
• Eligibility
• Endorsements & Modifications
• Policy Definitions
• Coverage Limits
• Common Questions
• Comparisons

Data Files:

• data/internal_faqs.json - 1,441 FAQs uploaded to Intercom
• data/internal_faqs_enhanced.json - Enhanced versions with agent details
• data/deleted_duplicates_backup.json - Backup of removed duplicates

Getting Started

Website

Open index.html in a browser to view the Pikl Insurance website.

Knowledge Base Processor

See pikl-kb-processor/QUICKSTART.md for setup and usage instructions.

FAQ System

Requirements:

cd faq-system
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

Environment Variables:

export INTERCOM_ACCESS_TOKEN="your_token_here"
export ANTHROPIC_API_KEY="your_key_here"

Quick Commands:

# Generate FAQs from policies
cd scripts
python generate_faqs.py

# Upload to Intercom
python intercom_import.py

# De-duplicate articles
python intercom_dedup.py --execute

# Analyze gap coverage
cd ../pikl-kb-processor/scripts
python analyze_gap_coverage.py

# Run dashboard
cd ..
python -m streamlit run app.py

Dashboard

The Streamlit dashboard provides:

• Overview: Key metrics and recent activity
• Knowledge Gaps: Customer questions and coverage analysis
• Analytics: FAQ distribution by category, theme breakdown, gap coverage
• FAQ Review: Browse and search all FAQs with categories

Access at: http://localhost:8501

Intercom Integration

Internal Articles Folder: Internal FAQs

Help Center: Pikl Help Center

Tags Created:

• exclusions, claims-requirements, eligibility, endorsements
• definitions, coverage-limits, common-questions, comparisons

Links

• Pikl Help Center
• Intercom Internal FAQs Folder
• Intercom Integration