Setting Up Knowledge Sources

This guide covers how to set up and configure knowledge sources for your ept AI chatbot. Knowledge sources are the foundation of your AI's intelligence - they contain the information your chatbot will use to provide accurate, helpful responses to users.

Overview

Setting up knowledge sources involves:

• Content Preparation: Organizing and preparing your content for AI consumption
• Knowledge Source Creation: Adding your content to the ept AI system
• Knowledge Source Configuration (KSC): Grouping sources for different use cases
• Testing and Validation: Ensuring your AI can access and use the content effectively

Prerequisites

Before setting up knowledge sources, ensure you have:

• Administrator Access: You need admin privileges to configure knowledge sources
• Content Ready: Prepare your documentation, FAQs, product information, and other content
• Use Case Defined: Understand what your chatbot should help users accomplish
• Content Strategy: Plan how to organize content for different audiences and purposes

Step 1: Prepare Your Content

Content Organization Strategy

Organize your content into logical categories based on your chatbot's purpose:

For Customer Support Chatbots:

• Product Documentation: User guides, technical specifications, feature descriptions
• Troubleshooting Guides: Common issues, error messages, and solutions
• FAQs: Frequently asked questions and their answers
• How-to Guides: Step-by-step instructions for common tasks
• Contact Information: Support channels, escalation procedures

For Sales Chatbots:

• Product Information: Features, benefits, specifications
• Pricing Information: Plans, packages, pricing tiers
• Case Studies: Customer success stories and testimonials
• Competitive Information: Comparison with alternatives
• Sales Process: Qualification criteria, next steps

For Internal Knowledge Bots:

• Company Policies: HR policies, procedures, guidelines
• Process Documentation: Workflows, standard operating procedures
• Training Materials: Onboarding guides, best practices
• Resource Information: Tools, systems, and access procedures

Content Format Requirements

ept AI supports multiple content formats:

Document Files:

• PDF: User manuals, technical documentation, reports
• DOC/DOCX: Word documents, policies, procedures
• TXT: Plain text files, simple documentation
• CSV: Structured data, lists, tables

Web Content:

• URLs: Links to documentation, knowledge bases, websites
• Web Pages: Direct content from your website or intranet
• API Endpoints: Dynamic content from databases or systems

Images and Media:

• Images: Screenshots, diagrams, charts (with OCR processing)
• Scanned Documents: PDFs of printed materials

Content Quality Guidelines

Accuracy and Currency:

• Ensure all information is current and factually correct
• Update content regularly to reflect changes in products, policies, or procedures
• Remove outdated information that could mislead users

Clarity and Readability:

• Use clear, concise language appropriate for your target audience
• Avoid jargon unless your audience is technical
• Structure content with headings, lists, and clear sections
• Use consistent terminology throughout your content

Completeness and Relevance:

• Provide comprehensive information for each topic
• Include context and background information where helpful
• Focus on content that addresses common user questions and needs
• Ensure content is relevant to your chatbot's purpose

Step 2: Create Knowledge Sources

Adding Knowledge Sources

1. Navigate to Configuration: Go to Configuration > Knowledge Sources in your ept AI dashboard
2. Create New Source: Click "Create Knowledge Source"
3. Configure Basic Information:

{
  "basic_info": {
    "name": "Product User Guide",
    "description": "Complete user guide for our flagship product",
    "source_type": "PDF",
    "confidentiality": "public",
    "tags": ["product", "user-guide", "documentation"]
  }
}

Source Type Configuration

For File Uploads:

• Upload File: Select and upload your document
• File Validation: Ensure the file is readable and properly formatted
• Content Extraction: The system will extract text content automatically
• Processing Status: Monitor the processing status in the dashboard

For URL Sources:

{
  "url_config": {
    "url": "https://your-domain.com/documentation",
    "authentication": "none", // or "basic", "oauth"
    "credentials": {
      "username": "optional_username",
      "password": "optional_password"
    },
    "crawl_depth": 2, // How deep to crawl linked pages
    "include_patterns": ["*.html", "*.pdf"],
    "exclude_patterns": ["*/admin/*", "*/private/*"]
  }
}

For Database Connections:

{
  "database_config": {
    "connection_string": "your_connection_string",
    "query": "SELECT * FROM knowledge_base WHERE status = 'active'",
    "refresh_schedule": "daily",
    "authentication": "connection_string" // or "environment_variables"
  }
}

Organizing Knowledge Sources

Create Logical Groups:

• Public Content: Marketing materials, general product information
• Internal Documentation: Company policies, internal procedures
• Technical Documentation: API docs, technical specifications
• Support Resources: Troubleshooting guides, FAQs

Use Descriptive Names:

• Product-User-Guide-v2.1
• Support-FAQs-Q4-2024
• HR-Policies-Employee-Handbook
• Technical-API-Documentation

Add Metadata:

• Tags: Use consistent tags for easy filtering and organization
• Categories: Group sources by content type or audience
• Version Information: Track content versions and update dates
• Access Control: Mark sources as public, internal, or confidential

Step 3: Create Knowledge Source Configurations (KSCs)

Understanding KSCs

Knowledge Source Configurations (KSCs) are groups of knowledge sources that define what information your chatbot can access for specific use cases. They provide:

• Content Organization: Logical grouping of related knowledge sources
• Access Control: Different KSCs for different user types or channels
• Performance Optimization: Efficient content retrieval and processing
• Confidentiality Management: Control over sensitive information access

Creating KSCs

1. Navigate to KSC Configuration: Go to Configuration > Knowledge Source Configurations
2. Create New KSC: Click "Create Knowledge Source Configuration"
3. Configure the KSC:

{
  "ksc_config": {
    "name": "Customer Support KSC",
    "description": "Knowledge sources for customer support and troubleshooting",
    "confidentiality": "public",
    "knowledge_sources": [
      "Product-User-Guide-v2.1",
      "Support-FAQs-Q4-2024",
      "Troubleshooting-Guide",
      "Contact-Information"
    ],
    "source_priority": {
      "Product-User-Guide-v2.1": 1,
      "Support-FAQs-Q4-2024": 2,
      "Troubleshooting-Guide": 3,
      "Contact-Information": 4
    }
  }
}

Common KSC Types

Customer Support KSC:

{
  "customer_support_ksc": {
    "purpose": "Handle customer support inquiries and troubleshooting",
    "sources": [
      "product-documentation",
      "troubleshooting-guides",
      "faqs",
      "contact-information"
    ],
    "confidentiality": "public",
    "response_style": "helpful_and_supportive"
  }
}

Sales Team KSC:

{
  "sales_ksc": {
    "purpose": "Support sales conversations and lead qualification",
    "sources": [
      "product-specifications",
      "pricing-information",
      "case-studies",
      "competitive-analysis"
    ],
    "confidentiality": "internal",
    "response_style": "informative_and_persuasive"
  }
}

Internal Knowledge KSC:

{
  "internal_ksc": {
    "purpose": "Help employees with internal processes and policies",
    "sources": [
      "hr-policies",
      "process-documentation",
      "training-materials",
      "resource-information"
    ],
    "confidentiality": "internal",
    "response_style": "professional_and_efficient"
  }
}

KSC Best Practices

Purpose-Driven Design:

• Create KSCs for specific use cases and audiences
• Include only sources relevant to the KSC's purpose
• Consider the user's context and needs when selecting sources

Content Prioritization:

• Order sources by importance (most relevant first)
• Consider source freshness and accuracy
• Balance comprehensive coverage with performance

Confidentiality Management:

• Ensure KSC confidentiality matches content sensitivity
• Separate public and internal content clearly
• Review access permissions regularly

Performance Optimization:

• Limit KSC size for faster response times
• Use specific sources rather than broad categories
• Monitor and adjust based on usage patterns

Step 4: Test Your Knowledge Sources

Initial Testing

Basic Functionality Test:

1. Create Test Questions: Develop questions that should be answerable from your knowledge sources
2. Test Knowledge Access: Verify the AI can find and use information from your sources
3. Check Response Quality: Review accuracy, relevance, and helpfulness of responses
4. Validate Content Coverage: Ensure the AI can handle questions across all knowledge areas

Sample Test Questions:

• "How do I reset my password?"
• "What are your business hours?"
• "How much does the premium plan cost?"
• "What's your return policy?"

Quality Assurance Checklist

Content Accuracy:

• All information is current and correct
• No outdated or conflicting information
• Technical details are accurate
• Contact information is up to date

Response Quality:

• Responses are factually correct
• Responses address the actual question
• Information is complete and helpful
• Responses are consistent for similar questions

Content Coverage:

• AI can handle common user questions
• All major topics are covered
• No significant knowledge gaps
• Content is appropriate for the target audience

Troubleshooting Common Issues

AI Can't Find Information:

• Check if the knowledge source was processed successfully
• Verify the source contains the expected content
• Ensure the source is included in the appropriate KSC
• Check source formatting and readability

Incorrect or Outdated Responses:

• Update the knowledge source with current information
• Remove or archive outdated content
• Verify source processing captured all content
• Check for conflicting information across sources

Poor Response Quality:

• Review and improve source content quality
• Ensure content is well-structured and clear
• Add more comprehensive information to sources
• Consider adding more specific sources for detailed topics

Step 5: Optimize and Maintain

Content Maintenance

Regular Updates:

• Schedule regular content reviews and updates
• Update information when products, policies, or procedures change
• Remove outdated content that could mislead users
• Add new content based on user questions and feedback

Quality Monitoring:

• Track which knowledge sources are most used
• Monitor response quality and user satisfaction
• Identify knowledge gaps based on user questions
• Gather feedback from users and support teams

Performance Optimization:

• Monitor KSC performance and response times
• Adjust source priorities based on usage patterns
• Optimize content structure for better AI understanding
• Consider splitting large sources into smaller, more focused ones

Advanced Configuration

Source Prioritization:

{
  "source_priority_config": {
    "high_priority": ["current-product-guide", "active-faqs"],
    "medium_priority": ["general-information", "contact-details"],
    "low_priority": ["archived-content", "legacy-documentation"]
  }
}

Content Filtering:

{
  "content_filtering": {
    "include_patterns": ["*.html", "*.pdf", "*.docx"],
    "exclude_patterns": ["*/draft/*", "*/archive/*"],
    "content_types": ["documentation", "faqs", "guides"],
    "date_range": {
      "start_date": "2023-01-01",
      "end_date": "current"
    }
  }
}

Next Steps

Once your knowledge sources are set up and tested:

1. Integrate the Chatbot - Add the chatbot to your applications
2. Configure the Design - Customize the visual appearance and user experience
3. Set up Continuous Improvement - Monitor and optimize performance

Related Documentation

• Knowledge Sources - Detailed knowledge source configuration
• Knowledge Source Configurations - Advanced KSC management
• Configuration Overview - All configuration options