​

Creating and Managing Collections

​

Overview

Collections are organized containers of information that AI agents can access to answer customer questions. They allow you to group related content together and make it easily accessible for intelligent responses.

​

What are Collections?

Collections serve as logical groupings of information within your knowledge base. They help organize content by topic, purpose, or type, making it easier for AI agents to find and use relevant information.

​

Collection Types

​

Regular Collections

Purpose: General knowledge and information storage Content Types:

• Company policies and procedures
• FAQ content
• Product documentation
• Support guides
• General business information

​

Product Collections

Purpose: E-commerce product information storage Content Types:

• Product catalogs
• Pricing information
• Product specifications
• Availability data
• Product-related support content

Note: Product collections are managed separately and used specifically for e-commerce integrations.

​

Creating New Collections

​

Navigation

Path: /playgrounds/collections → Click ”+ Collection” button → Fill out collection form

​

Step-by-Step Process

​

Step 1: Access Collections Management

1. Navigate to the collections list page
2. Click the ”+ Collection” button
3. Collection creation modal opens

​

Step 2: Collection Information

Required Fields:

• Collection Name: Unique identifier for the collection
• Collection Description: Detailed description for AI agents to understand the collection’s purpose

Collection Name Guidelines:

• Use English characters only
• Be descriptive and clear
• Avoid special characters
• Keep it concise but meaningful
• Examples: “Customer Support FAQ”, “Product Documentation”, “Company Policies”

Why English Only? Collection names are used to create AI functions that need to follow specific naming rules. English characters ensure compatibility with the AI system.

​

Step 3: Name Validation

Validation Process:

• System checks name uniqueness
• Validates naming conventions
• Provides suggestions if needed
• Real-time validation feedback

Validation Features:

• Automatic Checking: Real-time validation as you type
• Suggested Names: Alternative names if current one isn’t available
• Clear Feedback: Specific reasons if name is rejected

​

Step 4: Collection Creation

1. Review collection information
2. Click “OK” to create collection
3. System creates collection and redirects to content management
4. Collection appears in collections list

​

Collection Form Fields

​

Collection Name

Requirements:

• Must be unique within organization
• English characters only
• No special characters except spaces and hyphens
• Maximum length limitations apply

Best Practices:

• Use descriptive names that indicate content type
• Include context about the information contained
• Consider how AI agents will interpret the name
• Make it easy for team members to understand

​

Collection Description

Purpose: Helps AI agents understand when and how to use the collection Best Practices:

• Clearly explain what information is contained
• Describe when agents should use this collection
• Include any specific context or limitations
• Use clear, simple language

Examples:

• “This collection contains frequently asked questions about our products and services”
• “Company policies and procedures for employee reference”
• “Technical documentation for software products”

​

Managing Existing Collections

​

Collections List View

Available Information:

• Collection name and description
• Document count
• Collection type (regular vs. product)
• Creation date
• Last updated information

Actions Available:

• View collection details
• Edit collection information
• Add content to collections
• Delete collections (with restrictions)

​

Editing Collections

​

Accessing Edit Mode

1. Navigate to collections list
2. Click on collection name or edit button
3. Edit modal opens with current information

​

Editable Fields

• Collection Name: Can be updated if new name is valid
• Collection Description: Can be modified at any time

​

Edit Process

1. Modify desired fields
2. System validates changes
3. Click “OK” to save changes
4. Confirmation message appears

​

Collection Content Management

​

Navigation to Content

Path: Collections list → Click collection name → Content management interface

​

Content Management Interface

Features:

• Document list with search functionality
• Add content from multiple sources
• Edit existing content
• Delete outdated information
• Content organization tools

​

Adding Content to Collections

​

Data Source Options

​

1. URL Crawling

Purpose: Extract content from web pages Process:

1. Click “URL” data source
2. Enter list of URLs (one per line)
3. System crawls and extracts content
4. Content is processed and added to collection

URL Requirements:

• Must be publicly accessible
• Should contain relevant text content
• HTTPS preferred for security
• Avoid pages with heavy dynamic content

Best Practices:

• Use specific pages rather than entire websites
• Verify URLs are accessible before adding
• Include help pages, documentation, and FAQ pages
• Avoid duplicate content from multiple URLs

​

2. Document Upload

Purpose: Process and extract content from files Supported Formats: PDF, Microsoft Word (.doc, .docx), text files (.txt) Process:

1. Click “Docs” data source
2. Select files to upload
3. System extracts text content
4. Processed content added to collection

Upload Guidelines:

• Maximum file size limitations apply
• Ensure documents contain text (not just images)
• Use clear, well-formatted documents
• Remove sensitive information before upload

​

3. Notion Integration

Purpose: Import content from Notion workspaces Process:

1. Click “Notion” data source
2. Authenticate with Notion account
3. Select documents to import
4. System imports and processes content

Notion Requirements:

• Valid Notion account
• Proper permissions for documents
• Documents should be well-organized
• Consider privacy and access restrictions

​

4. Manual Entry

Purpose: Add custom content directly Process:

1. Click “Copy & Paste” data source
2. Enter content in text field
3. Provide title and context
4. Save content to collection

Manual Entry Best Practices:

• Provide clear, descriptive titles
• Include relevant context information
• Use proper formatting for readability
• Keep content focused and relevant

​

Content Processing

​

Automatic Processing

1. Text Extraction: System extracts text from various sources
2. Content Parsing: Breaks content into searchable chunks
3. Indexing: Creates searchable indexes
4. Semantic Analysis: Understands content meaning and context

​

Content Validation

• Quality Checks: Ensures content is readable and relevant
• Duplicate Detection: Identifies and handles duplicate content
• Format Validation: Ensures proper formatting for AI consumption
• Size Limitations: Enforces character and document limits

​

Collection Management Best Practices

​

Organization Strategy

​

Content Categorization

1. Topic-Based: Group by subject matter
2. Department-Based: Organize by business function
3. Customer-Facing: Separate internal vs. customer information
4. Frequency-Based: High-use vs. reference information

​

Naming Conventions

• Consistent Format: Use similar naming patterns
• Descriptive Names: Clear indication of content type
• Version Control: Include version numbers if needed
• Team Understanding: Names that team members can easily understand

​

Content Quality

​

Information Accuracy

• Regular Reviews: Periodic content audits
• Update Procedures: Clear processes for keeping content current
• Source Verification: Ensure information comes from reliable sources
• Version Control: Track changes and updates

​

Content Relevance

• Customer Focus: Prioritize information customers need
• Agent Capability: Ensure content matches agent abilities
• Business Alignment: Content supports business objectives
• Performance Monitoring: Track which content is most useful

​

Performance Optimization

​

Content Structure

• Clear Hierarchy: Logical organization of information
• Searchable Format: Content optimized for search and retrieval
• Contextual Information: Adequate context for AI understanding
• Concise Content: Focused, relevant information without unnecessary details

​

Regular Maintenance

• Content Audits: Regular review of all collection content
• Performance Analysis: Monitor which content performs best
• User Feedback: Collect feedback from agents and customers
• Continuous Improvement: Refine based on usage patterns

​

Common Issues and Troubleshooting

​

Collection Creation Issues

​

Name Validation Errors

Common Causes:

• Name already exists
• Invalid characters used
• Name too long or too short
• Non-English characters

Solutions:

1. Try alternative names
2. Use suggested names from system
3. Follow naming guidelines
4. Remove special characters

​

Content Upload Problems

Common Causes:

• File format not supported
• File size too large
• Network connectivity issues
• Content extraction failures

Solutions:

1. Convert files to supported formats
2. Compress large files or split content
3. Check internet connection
4. Retry upload after fixing issues

​

Content Management Issues

​

Poor Content Quality

Symptoms:

• AI agents provide irrelevant responses
• Information seems outdated or incorrect
• Agents can’t find relevant information

Solutions:

1. Review content accuracy and relevance
2. Improve content organization and structure
3. Add more context and detail to content
4. Remove outdated or incorrect information

​

Search and Retrieval Problems

Symptoms:

• Agents can’t find information that exists
• Irrelevant information returned for queries
• Inconsistent response quality

Solutions:

1. Improve content structure and formatting
2. Add relevant keywords and context
3. Review collection organization
4. Consider splitting large collections

​

Agent Integration Issues

​

Collections Not Available to Agents

Cause: Collection not properly connected to agents Solution:

1. Check agent configuration
2. Enable knowledge base capability
3. Connect specific collections to agents
4. Verify agent permissions

​

Poor Agent Responses

Cause: Content quality or organization issues Solution:

1. Review collection content quality
2. Improve content organization
3. Add more relevant information
4. Test agent responses and refine content