Skip to content
AnglinAI Docs

AI-Powered Self-Help System

add this email as a viewer to shared docs

your-service-account@your-project-id.iam.gserviceaccount.com

Overview

The help desk includes an intelligent self-help system that provides instant IT support using Google Gemini AI with a comprehensive FAQ database. Users can get immediate answers to common IT issues without creating tickets.

How It Works

1. FAQ-First Approach (Fastest)

  • Purpose: Instant responses for common questions
  • Database: 50+ IT-specific FAQs covering passwords, email, software, hardware, network issues
  • Matching: Uses fuzzy search (Fuse.js) with 70% confidence threshold
  • Speed: Millisecond response time
  • Cost: Free (no AI tokens used)

2. AI Fallback (When Needed)

  • Trigger: When no FAQ matches user’s question
  • Model: Google Gemini 1.5 Flash
  • Context: Uses company knowledge base + IT procedures
  • Processing:
    • Categorizes question (software/hardware/network/etc.)
    • Fetches relevant knowledge base articles
    • Generates personalized step-by-step solutions
  • Cost: ~$0.002-0.004 per query

3. Smart Features

Dynamic Content Replacement

  • Replaces {SUPPORT_PHONE} placeholders with actual support number from settings
  • Ensures consistent contact information across all responses

Context Preservation

  • Tracks conversation history
  • Maintains context between questions
  • Enables follow-up questions

Seamless Escalation

  • “Still Need Help?” button creates tickets with full conversation context
  • Pre-fills ticket with original question + AI suggestions
  • No lost information during escalation

User Experience

Self-Help Page (/self-help)

  1. AI Chat Interface: Interactive conversation with IT assistant
  2. Quick Actions: Direct ticket creation and phone support
  3. Common Issues: Categorized IT problems with chips for quick access
  4. Emergency Alerts: Clear escalation path for urgent issues

Response Types

  • FAQ Answers: Formatted with current date and follow-up suggestions
  • AI Solutions: Step-by-step instructions with safety warnings
  • Error Handling: Graceful fallbacks with support contact information

Technical Architecture

Core Components

  • aiService.ts: Main AI orchestration and Gemini integration
  • faqService.ts: FAQ database with intelligent matching
  • cacheService.ts: Token optimization for cost management
  • SelfHelpChat.tsx: Interactive UI component
  • SelfHelpPage.tsx: Full page layout with sidebar

Integration Points

  • Config System: Uses support phone, company settings
  • Theme System: Respects user’s accent color preferences
  • Auth System: Authenticated users only
  • Ticket System: Seamless escalation to traditional support

Cost Optimization

  • FAQ First: Eliminates 70-80% of AI calls
  • Context Caching: Reduces token usage for repeat questions
  • Efficient Model: Uses Gemini 1.5 Flash (lowest cost tier)
  • Smart Categorization: Targeted responses reduce token waste

Configuration

Environment Variables

VITE_GEMINI_API_KEY=your_api_key
VITE_ENABLE_AI_SELF_HELP=true
VITE_ENABLE_TOKEN_CACHE=false

Settings Integration

  • Support phone number automatically used throughout system
  • Company branding and contact info included in responses
  • Configurable via standard settings interface

Benefits

For Users

  • Instant Help: No waiting for ticket responses
  • 24/7 Availability: AI works outside business hours
  • Consistent Answers: Standardized solutions from knowledge base
  • Easy Escalation: Smooth transition to human support when needed

For IT Team

  • Reduced Ticket Volume: Common issues resolved automatically
  • Better Context: Escalated tickets include what user already tried
  • Knowledge Preservation: FAQ database captures institutional knowledge
  • Analytics: Track common issues and gaps in documentation

For Organization

  • Cost Effective: AI queries cost pennies vs. human support time
  • Scalable: Handles unlimited concurrent users
  • Consistent Service: Same quality help regardless of time or staff availability
  • Knowledge Building: System learns from interactions to improve over time

Future Enhancements

  • Integration with Google Docs: Dynamic knowledge base from company documentation
  • Usage Analytics: Track FAQ hits, AI query patterns, escalation rates
  • Learning System: Automatic FAQ generation from successful AI responses
  • Multi-language Support: Localized responses for global organizations